(2011), partisan alignment, clientelism and intergovernmental transfers - evidence from spanish...

30
 1 Partisan alignment, clientelism and intergovernmental transfers: evidence from Spanish municipalities  Marta Curto-Grau, Albert Solé-Ollé and Pilar Sorribas-Nava rro Universitat de Barcelona & Institut d’Economia de Barcelona This draft: May 2011 Abstract In this study we provide additional and robust evidence of the impact of partisan alignment on intergovernmental transfers to municipalities. In addition, we study several mechanisms through which this impact could be affected by clientelism. The empirical strategy consists on applying a regression discontinuity design on capital grants to Spanish municipalities over the period 2002- 2007. The results show that municipalities politically aligned with upper tiers do benefit from larger transfers, especially in the case o f regional grants. When clientelist tactics are introduced in our framework, the allocation of grants is no longer based only on partisan politics but also on two other factors: the political credit that “political machines” can claim and their ability to monitor who their constituents vote for. In this sense, we find tha t in smaller municipalities – where monitoring costs are lower- the discontinuity o n transfers due to partisan ties is larger. Keywords: intergovernmental transfers, partisan alignment, clientelism, regression discontinuity JEL classification: H77, D72  This paper has benefited from the financial support of ECO2009-12680/ECON (Spanish Ministry of Education and Science) and project 2009 SGR 102 (Generalitat de Catalunya). We are also grateful to Jonathan Rodden for his helpful co mments. E-mail addresses: [email protected] (M. Curto-Grau), [email protected] (A. Solé-Ollé), [email protected] (P. Sorribas-Navarro).

Upload: allanshariff

Post on 03-Jun-2018

222 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 130

1

Partisan alignment clientelism and intergovernmental transfers evidence from Spanish municipalities

lowastlowastlowastlowast

Marta Curto-Grau Albert Soleacute-Olleacute and Pilar Sorribas-Navarro

Universitat de Barcelona amp Institut drsquoEconomia de Barcelona

This draft May 2011

Abstract

In this study we provide additional and robust evidence of the impact of partisan alignment on

intergovernmental transfers to municipalities In addition we study several mechanisms throughwhich this impact could be affected by clientelism The empirical strategy consists on applying aregression discontinuity design on capital grants to Spanish municipalities over the period 2002-2007 The results show that municipalities politically aligned with upper tiers do benefit fromlarger transfers especially in the case of regional grants When clientelist tactics are introduced inour framework the allocation of grants is no longer based only on partisan politics but also ontwo other factors the political credit that ldquopolitical machinesrdquo can claim and their ability tomonitor who their constituents vote for In this sense we find that in smaller municipalities ndashwhere monitoring costs are lower- the discontinuity on transfers due to partisan ties is larger

Keywords intergovernmental transfers partisan alignment clientelism regression discontinuity

JEL classification H77 D72

lowast This paper has benefited from the financial support of ECO2009-12680ECON (Spanish Ministryof Education and Science) and project 2009 SGR 102 (Generalitat de Catalunya) We are alsograteful to Jonathan Rodden for his helpful commentsE-mail addresses mcurtoubedu (M Curto-Grau) asoleubedu (A Soleacute-Olleacute)

psorribasubedu (P Sorribas-Navarro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 230

2

1 Introduction

A good process of fiscal decentralization has to ensure that constituent units within a

federal system are provided with enough resources to meet their expenditure responsibilities

In most federations local governments have a weak tax base so in order to provide publicgoods and services to their citizens they rely significantly on transfers from upper-level

governments According to the normative criterion of fiscal federalism the allocation of

grants has to guarantee economic efficiency and equity amongst the members of the

federation (Musgrave 1959 1983 Oates 1972) Nonetheless the positive approach claims

that transfers are distributed by non-benevolent agents who are not driven by economic but

political interests (Grossman 1994) In line with this last view our paper evaluates to which

extent the distribution of grants is largely conditional on the partisan alignment betweendonor and recipient

The political economy literature has identified four main political determinants of the

regional allocation of national resources The first one is legislative representation

(Ansolabehere Gerber and Snyder 2003 Porto and Sanguinetti 2003) A second factor is

the share of ldquoswing votersrdquo versus ldquocore votersrdquo in constituencies This idea has been

analyzed in different ways through theoretical papers based on electoral competition

(Lindbeck and Weibull 1987 Dixit and Londregan 1998 Cox and McCubbis 1986)

through theoretical papers based on legislative bargaining (Weingast Shepsle and Johnsen

1981) through empirical studies (Schady 2000 Case 2001 Stroumlmberg 2001 Johansson

2003) which have provided mixed and inconclusive evidence until now Another determinant

is the citizensrsquo level of information and participation in the electoral system (Besley and

Burgess 2002 Stroumlmberg 2004) And last but not least it is also relevant the party affiliation

of the different government tiers (Grossman 1994 Levitt and Snyder 1995 Worthington and

Dollery 1998 Khemani 2003 Arulampalam et al 2009) Two main problems arise fromthis literature The first one is that measuring the concept of core and swing voters has proven

to be a difficult task The second issue is that whenever election results are used as variables

to explain the regional distribution of transfers one should correct for endogeneity problems

Our study presents an empirical analysis that aims at overcoming the two

aforementioned problems Firstly instead of taking a ldquoswing vs core voterrdquo approach the

explanatory variable for intergovernmental transfers is the political alignment between

municipalities and their upper-tiers There are two main reasons to believe that the

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 330

3

distribution of transfers may be biased towards co-partisans On the one hand following the

intuition beneath Diaz-Cayeros Magaloni and Weingast (2006) in certain circumstances the

incumbent party may have strong incentives to use a mechanism of punishments and rewards

to maintain electoral support Hence municipalities who defect to the opposition (ie

unaligned municipalities) should receive lower transfers as means of punishment This idea is

at odds with the swing voter models of Dixit and Londregan and Lindbeck and Weibull On

the other hand the distribution of grants on partisan grounds is also influenced by the fact that

voters have incomplete information about which tier of government is the source of the grant

In this situation the grantor may not be able to claim much (or even none) political credit for

the grant If there is no credit leakage (ie the grantor cannot claim his credit) he finds it more

profitable to distribute grants to aligned municipalities However if the leakage is large

enough the incumbent grantor may consider sending funds to unaligned municipalities as he

could still get some electoral reward from that The empirical evidence extracted from some

of the above mentioned studies concludes that the allocation of intergovernmental grants is

indeed skewed in favor of aligned municipalities For instance Khemani (2003) finds that

transfers subject to less stringent rules (more discretionary) are positively affected by

alignment while Arulampalam et al (2009) observe that being an aligned and swing state

translates into 16 higher center-state transfers (compared to being unaligned and non-

swing)

The second aforesaid concern that we want to solve is to isolate the effect that

alignment per se has on transfers and to ensure that causality runs in the desired direction To

do that we apply a regression discontinuity design (RDD)1 which allows us to benefit from

the relatively milder assumptions that it requires

The empirical strategy of our paper relies on data on capital transfers received by

Spanish municipalities during two legislatures (2000-2003 and 2004-2007) from threedifferent upper-level administrations (central regional and upper-local) Although there are

two recent contributions studying the effect of partisan alignment on intergovernmental

transfers in Spain Soleacute-Olleacute and Sorribas Navarro (2008a 2008b)2 their main limitation is

the inability to disentangle if larger transfers are explained by higher political support (larger

1 This strategy is described in detail in Lee and Lemieux (2010)2 Both studies are based on a model of political competition where grants to aligned municipalities are assumedto be capable to extract more votes than grants to municipalities represented by opposition parties As proven in

Soleacute-Olleacute and Sorribas-Navarro (2008a) the relatively higher electoral reward of politically alignedmunicipalities leads upper layer governments to transfer larger grants to them

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 430

4

winning margins) or merely by partisan alignment Furthermore the RD estimates that we

obtain are more credible than the ones from ldquonatural experimentsrdquo 3 like the differences-in-

differences used in Soleacute-Olleacute and Sorribas-Navarro (2008a) because the RDD can be

considered as a local randomized experiment in which the (randomized) variation observed in

the political alignment of municipalities is imprecisely controlled by individuals Therefore

the contribution of our paper is not merely to present additional empirical evidence on the

Spanish case but more importantly to provide more robust results through a neat explanatory

variable (partisan alignment)

The clear advantages of the RDD that we have exposed have made it a popular

econometric tool for researchers In the political economy field it has not been widely used

but there are a few studies that apply this strategy to examine the effects of a number ofvariables on government expenditures Albouy (2009) for instance analyses the effect that

the party-identity of the legislator has on the allocation of federal expenditures Ferreira and

Gyourko (2009) test for US municipalities if being the mayor a Democrat or a Republican

affects the allocation of local public spending For Sweden Pettersson-Lidbom (2008)

presents evidence that party control has a causal effect on economic outcomes more

precisely on the level of public expenditures and tax revenues Additionally there are two

applied studies that we consider to be close to ours as they both apply a RDD to examine theeffects of partisan alignment on intergovernmental transfers Migueis (2010) for Portugal

and Brollo and Nannicini (2010) for Brazil We consider we improve upon existing literature

for several reasons First of all while Migueis (2010) considers the causal effect of alignment

on grants to be homogeneous we take a more complex approach by testing the possibility that

this effect is heterogeneous This strategy is also found in the paper by Brollo and Nannicini

where they focus on two dimensions which could affect the impact of alignment on transfers

the value of political capital for the central government and the impact of transfers on

electoral outcomes However our study differs from the latter because we center in a very

specific and different source of heterogeneity which may be more relevant for the Spanish

case clientelism Clientelism is a salient issue in the Spanish socio-economic context and we

think it may be strategically used jointly with partisan tactics to influence electoral results

When clientelistic strategies are accounted for in our benchmark ldquopolitical machinesrdquo not

only allocate transfers based on partisan ties but also on the political credit they can claim and

their ability to monitor their constituentsrsquo votes To undertake this part of our study we draw

3 The formal proof of the lsquosuperiorityrsquo of RDD over lsquonatural experimentsrsquo is in Lee (2008)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 530

5

on ideas from Diaz-Cayeros Magaloni and Weingast (2006) Stokes (2005 2007) Brusco

Nazareno and Stokes (2004) and Soleacute-Olleacute and Sorribas-Navarro (2008a) Based on these

studies we contrast four hypotheses We examine the possibility that the treatment effect of

political alignment on transfers may be larger in poorer municipalities smaller

municipalities municipalities aligned at all levels of government and municipalities with

lower reliance on intergovernmental transfers (eg those with a lower debt burden)

Moreover traditionally Spanish left-wing parties have been accused more often of using

clientelist strategies than right-wing parties so we also test if the political color of the grantor

has an impact on the causal effect of alignment on grants Finally the last hypothesis we

contrast is whether a majority government is more able to allocate transfers on partisan

grounds than a coalition government

To summarize the findings of our paper our first set of results show through RD

estimates that being an aligned municipality has a positive effect on the amount of transfers

received by lower-level governments This impact is the largest in the case of transfers

allocated by the regional governments while the lowest impact is found for central

government transfers Thanks to an increase on the sample size and to the use of the RDD we

consider these results to be more robust than those in previous studies Concerning the impact

of clientelism on the causal impact of alignment we do not find any evidence of it when welook at central transfers However for regional and upper-level transfers we do find evidence

that the sharp increase on transfers observed in aligned municipalities is greater for smaller

municipalities This effect is also larger for left-wing grantors (compared to right-wing ones)

at the regional level

The remaining paper is structured as follows In section 2 we establish the relevance

of clientelism for our analysis Section 31 contains some brief background information on

Spanish institutions (electoral processes) and intergovernmental grants In section 32 wediscuss the empirical strategy Section 4 presents our main findings as well as the tests that

validate our analysis We conclude with section 5

2 The role of clientelism in Spain heterogeneous effects of partisan alignment

A large part of the Spanish public opinion presumes that the allocation of public

spending in the country is largely influenced by clientelistic ties between voters and

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 630

6

politicians Political clientelism is often cited by Spanish newspapers and its persistence in

this country has its strong roots in the well-known caciquismo of Restoration times -during

the late 19th and early 20th century4 Due to socio-economic changes nowadays traditional

clientelism (based on a peasant clientele system) has been replaced by a modern form of

territorial clientelism also known as ldquobroker clientelismrdquo In her study of Spanish clientelism

Blakeley (2001) explains how the Partido Popular (peoplersquos party) has strategically targeted

Galicia for decades while the Partido Socialista Obrero Espantildeol (socialist party) has

traditionally done so in Andalusia The Andalusian case is frequently illustrated through the

agricultural unemployment payment scheme (PER5) while Galician clientelism is especially

in the form of employment to private business closely linked with the Partido Popular6 In her

text Blakeley also notices the wide-spread accusations of clientelism posed to a long-lasting

ruling party in Catalonia Convergegravencia i Unioacute

There are two dimensions that one should take into account when distinguishing

between clientelism and other related concepts such as pork-barreling and rent-seeking the

scope (how narrow the group of beneficiaries is) and the directedness of the exchange 7

Hence clientelism is based on a quid pro quo arrangement between the patron (politician) and

the client (the voter) where they exchange goods for votes this is made ldquonot simply to seek

but rather to directly reward [electoral] supportrdquo All in all ldquoeven a very lsquobroadrsquo policy maybe lsquoclientelisticrsquo while even very narrow targeting may notrdquo So even broad project grants

such as the ones financed through intergovernmental transfers can be used by governments to

tie the hands of their electorate8

4 For a broader description of the transition from traditional clientelism to partisan clientelism in Spain seeCazorla (1994)5

This agrarian subsidy was implemented by the Andalusian regional government ( Junta de Andaluciacutea) In 1996it was renamed to AEPSA and it extended its scope to other autonomous communities although Andalusia stillreceives a large share of it6 Hopkin (2001) and Corzo (2002) also cite the Galician and Andalusian cases as examples of ldquomodernrdquoclientelism in Spain Robles Egea(2003) focuses on political clientelism in Andalusia7 See Carroll and Lyne (2006) for further discussion on this issue8 We quote two examples extracted from Spanish newspapers to illustrate how transfers can be used withclientelistic purposes both refer to clientelism at the upper-local level The first one is a statement made by EnricMorera a deputy in Les Corts Valencianes (the Valencian regional government)

ldquo(hellip) we have three diputaciones with no competencies who only serve to provide jobs for advisors andtrustworthy people who mandate and in addition the public funds they have at their disposal are used as a toolto preform political clientelism and propaganda of those who are in powerrdquo El Periogravedic 17012011

(httpwwwelperiodiccomnoticias102792_morera-psoe-hablan-acabar-autonomias-pero-diputaciones-nidos-deficit-clientelismo-politicohtml)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 730

7

To present the above ideas in a formal way we present a model borrowing ideas from

Stokes (2005 2007) and Diaz-Cayeros Magaloni and Weingast (2006) We start with a one-

shot game with two actors voters and parties Voters have ideology xi there are two parties 1

(incumbent) or 2 (opposition) who compete to head an upper-tier government (in our study

central regional or upper-local levels) Votersrsquo preferences are as follow

ui= αi times (-12(vi-xi)2) + (1- αi ) times gi(θ micro)

where vi=x1x2 is the citizenrsquos decision to vote for the clientelist party or the

opposition xi represents the voterrsquos ideology whose weight is α ϵ[01] Voters also get utility

from a potential reward gi=0 g(θ micro) depending on whether the clientelist party wins or

not In our study the reward corresponds to intergovernmental transfers As in Diaz-Cayeros

Magaloni and Weingast (2006) failure to support the clientelist party translates into lack of

reward ie punishment through lower funds Furthermore the reward may in turn be

influenced by two factors the political credit (θ) that the grantor can claim when giving the

reward and the partyrsquos ability (micro) to observe the citizenrsquos vote The reward function is

increasing in both factors Stokesrsquo model presumes that ldquoa personrsquos vote is (hellip) perfectly

observable by political partiesrdquo but in a democracy with secret ballot elections like Spain

this is very unlikely to occur However as pointed by Brusco Nazareno and Stokes (2004)

clientelistic strategies can still work under those circumstances This is possible because

ldquoclientelist parties compensate for the inability to observe the vote directly by observing a

range of other actions and behaviors (hellip) that allow party operatives to make good guessesrdquo

In multi-government systems the lower level representatives (the mayors) may be seen as

intermediaries of the upper-level governments and their proximity to citizens makes them

more capable to perceive their constituentsrsquo actions

All in all the timing of the game can be easily summarized in two stages At the

beginning of the first stage t=1 electoral campaigns take place and politicians signal their

The second example refers to a complaint raised by a representative in the diputacioacuten de Castilla-la-ManchaMario Gonzaacutelez Somoano (peoplesrsquo party) to the representative of public works of the diputacioacuten AacutengelMoreno(socialist party) Gonzaacutelez Somoano complains because in his village Gascuentildea there is an emergencysituation as one of the streets is in an extremely bad condition and it needs to be asphalted

ldquoGonzaacutelez Somoano replied that (hellip) they had undertaken asphalt works in neighboring villages of Gascuentildeabut governed by socialist mayors accusing of acting more in lsquotheir villagesrsquo than in others and reminding that allthe roads from Cantalojas (municipality whose mayor is Moreno) had been asphalted the previous yearrdquoGuadaqueacute17072009(httpwwwguadaquecomindexphpoption=com_contentampview=articleampcatid=33Aprovinciaampid=36823Ael-pp-acusa-al-psoe-de-clientelismo-y-rafael-esteban-llama-al-portavoz-

payasoampItemid=54)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 830

8

willingness to provide rewards (g) During these campaigns candidates express promises

which often have implicit threats indicating that the only way to ensure they get the public

goodsservices desired is to vote for them and not for the opposition party In order to target

their voters more accurately candidates often promise specific goods such as infrastructures 9

In this same stage voters express to candidates their willingness to vote for them in exchange

for goods and at the end of the period they cast their vote they have two possibilities either

they renege on their promise or they do not At a second stage t=2 parties take action and

implement public policies Similar to Stokesrsquo analyses there are four possible situations with

different payoffs ndashsummarized in the table below- If the voter complies and votes the

clientelist party he wins and gets the votes of its constituents (vgt0) but if the voter reneges

this party does not get the vote (v=0) and the party loses the elections In this case the party

can still decide whether to pay a reward to the voter or not On the contrary if the party wins

it has two possible options either he complies and rewards its voters (ggt0) or it reneges

(g=0)

PARTY

Reward No Reward

V O T E R

Comply αtimes(-12(xi-x1)2) +(1 α)times g(θ micro) v- g(θ micro) αtimes(-12(xi-x1)

2) v

Defect αtimes(-12(xi-x2)2) +(1 α)times g(θ micro) - g(θ micro) αtimes(-12(xi-x2)

2) 0

As explained the amount of the reward (g) is conditional on both θ and micro So we

would be in the situation ldquoComply-No Reward rdquo whenever the system is totally opaque and the

winning party can hardly be monitored (ie micro rarr0) andor the grantor cannot claim any credit

from the reward (θ=0) Moreover note that to overcome commitment problems the above

model needs to be a repeated game in which micro ne0

From the formal description we can derive four hypotheses on how clientelism could

smooth or exacerbate the impact of political alignment on transfers There may be a chance

that the treatment effect of political alignment on transfers is larger in

H1 smaller municipalities In this case monitoring the voters may be easier (we

expect micro to be higher)

9 In Spain local infrastructures (like sports centers swimming pools roads hospitals etc) are largely financedthrough capital transfers from upper-level governments

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 2: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 230

2

1 Introduction

A good process of fiscal decentralization has to ensure that constituent units within a

federal system are provided with enough resources to meet their expenditure responsibilities

In most federations local governments have a weak tax base so in order to provide publicgoods and services to their citizens they rely significantly on transfers from upper-level

governments According to the normative criterion of fiscal federalism the allocation of

grants has to guarantee economic efficiency and equity amongst the members of the

federation (Musgrave 1959 1983 Oates 1972) Nonetheless the positive approach claims

that transfers are distributed by non-benevolent agents who are not driven by economic but

political interests (Grossman 1994) In line with this last view our paper evaluates to which

extent the distribution of grants is largely conditional on the partisan alignment betweendonor and recipient

The political economy literature has identified four main political determinants of the

regional allocation of national resources The first one is legislative representation

(Ansolabehere Gerber and Snyder 2003 Porto and Sanguinetti 2003) A second factor is

the share of ldquoswing votersrdquo versus ldquocore votersrdquo in constituencies This idea has been

analyzed in different ways through theoretical papers based on electoral competition

(Lindbeck and Weibull 1987 Dixit and Londregan 1998 Cox and McCubbis 1986)

through theoretical papers based on legislative bargaining (Weingast Shepsle and Johnsen

1981) through empirical studies (Schady 2000 Case 2001 Stroumlmberg 2001 Johansson

2003) which have provided mixed and inconclusive evidence until now Another determinant

is the citizensrsquo level of information and participation in the electoral system (Besley and

Burgess 2002 Stroumlmberg 2004) And last but not least it is also relevant the party affiliation

of the different government tiers (Grossman 1994 Levitt and Snyder 1995 Worthington and

Dollery 1998 Khemani 2003 Arulampalam et al 2009) Two main problems arise fromthis literature The first one is that measuring the concept of core and swing voters has proven

to be a difficult task The second issue is that whenever election results are used as variables

to explain the regional distribution of transfers one should correct for endogeneity problems

Our study presents an empirical analysis that aims at overcoming the two

aforementioned problems Firstly instead of taking a ldquoswing vs core voterrdquo approach the

explanatory variable for intergovernmental transfers is the political alignment between

municipalities and their upper-tiers There are two main reasons to believe that the

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 330

3

distribution of transfers may be biased towards co-partisans On the one hand following the

intuition beneath Diaz-Cayeros Magaloni and Weingast (2006) in certain circumstances the

incumbent party may have strong incentives to use a mechanism of punishments and rewards

to maintain electoral support Hence municipalities who defect to the opposition (ie

unaligned municipalities) should receive lower transfers as means of punishment This idea is

at odds with the swing voter models of Dixit and Londregan and Lindbeck and Weibull On

the other hand the distribution of grants on partisan grounds is also influenced by the fact that

voters have incomplete information about which tier of government is the source of the grant

In this situation the grantor may not be able to claim much (or even none) political credit for

the grant If there is no credit leakage (ie the grantor cannot claim his credit) he finds it more

profitable to distribute grants to aligned municipalities However if the leakage is large

enough the incumbent grantor may consider sending funds to unaligned municipalities as he

could still get some electoral reward from that The empirical evidence extracted from some

of the above mentioned studies concludes that the allocation of intergovernmental grants is

indeed skewed in favor of aligned municipalities For instance Khemani (2003) finds that

transfers subject to less stringent rules (more discretionary) are positively affected by

alignment while Arulampalam et al (2009) observe that being an aligned and swing state

translates into 16 higher center-state transfers (compared to being unaligned and non-

swing)

The second aforesaid concern that we want to solve is to isolate the effect that

alignment per se has on transfers and to ensure that causality runs in the desired direction To

do that we apply a regression discontinuity design (RDD)1 which allows us to benefit from

the relatively milder assumptions that it requires

The empirical strategy of our paper relies on data on capital transfers received by

Spanish municipalities during two legislatures (2000-2003 and 2004-2007) from threedifferent upper-level administrations (central regional and upper-local) Although there are

two recent contributions studying the effect of partisan alignment on intergovernmental

transfers in Spain Soleacute-Olleacute and Sorribas Navarro (2008a 2008b)2 their main limitation is

the inability to disentangle if larger transfers are explained by higher political support (larger

1 This strategy is described in detail in Lee and Lemieux (2010)2 Both studies are based on a model of political competition where grants to aligned municipalities are assumedto be capable to extract more votes than grants to municipalities represented by opposition parties As proven in

Soleacute-Olleacute and Sorribas-Navarro (2008a) the relatively higher electoral reward of politically alignedmunicipalities leads upper layer governments to transfer larger grants to them

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 430

4

winning margins) or merely by partisan alignment Furthermore the RD estimates that we

obtain are more credible than the ones from ldquonatural experimentsrdquo 3 like the differences-in-

differences used in Soleacute-Olleacute and Sorribas-Navarro (2008a) because the RDD can be

considered as a local randomized experiment in which the (randomized) variation observed in

the political alignment of municipalities is imprecisely controlled by individuals Therefore

the contribution of our paper is not merely to present additional empirical evidence on the

Spanish case but more importantly to provide more robust results through a neat explanatory

variable (partisan alignment)

The clear advantages of the RDD that we have exposed have made it a popular

econometric tool for researchers In the political economy field it has not been widely used

but there are a few studies that apply this strategy to examine the effects of a number ofvariables on government expenditures Albouy (2009) for instance analyses the effect that

the party-identity of the legislator has on the allocation of federal expenditures Ferreira and

Gyourko (2009) test for US municipalities if being the mayor a Democrat or a Republican

affects the allocation of local public spending For Sweden Pettersson-Lidbom (2008)

presents evidence that party control has a causal effect on economic outcomes more

precisely on the level of public expenditures and tax revenues Additionally there are two

applied studies that we consider to be close to ours as they both apply a RDD to examine theeffects of partisan alignment on intergovernmental transfers Migueis (2010) for Portugal

and Brollo and Nannicini (2010) for Brazil We consider we improve upon existing literature

for several reasons First of all while Migueis (2010) considers the causal effect of alignment

on grants to be homogeneous we take a more complex approach by testing the possibility that

this effect is heterogeneous This strategy is also found in the paper by Brollo and Nannicini

where they focus on two dimensions which could affect the impact of alignment on transfers

the value of political capital for the central government and the impact of transfers on

electoral outcomes However our study differs from the latter because we center in a very

specific and different source of heterogeneity which may be more relevant for the Spanish

case clientelism Clientelism is a salient issue in the Spanish socio-economic context and we

think it may be strategically used jointly with partisan tactics to influence electoral results

When clientelistic strategies are accounted for in our benchmark ldquopolitical machinesrdquo not

only allocate transfers based on partisan ties but also on the political credit they can claim and

their ability to monitor their constituentsrsquo votes To undertake this part of our study we draw

3 The formal proof of the lsquosuperiorityrsquo of RDD over lsquonatural experimentsrsquo is in Lee (2008)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 530

5

on ideas from Diaz-Cayeros Magaloni and Weingast (2006) Stokes (2005 2007) Brusco

Nazareno and Stokes (2004) and Soleacute-Olleacute and Sorribas-Navarro (2008a) Based on these

studies we contrast four hypotheses We examine the possibility that the treatment effect of

political alignment on transfers may be larger in poorer municipalities smaller

municipalities municipalities aligned at all levels of government and municipalities with

lower reliance on intergovernmental transfers (eg those with a lower debt burden)

Moreover traditionally Spanish left-wing parties have been accused more often of using

clientelist strategies than right-wing parties so we also test if the political color of the grantor

has an impact on the causal effect of alignment on grants Finally the last hypothesis we

contrast is whether a majority government is more able to allocate transfers on partisan

grounds than a coalition government

To summarize the findings of our paper our first set of results show through RD

estimates that being an aligned municipality has a positive effect on the amount of transfers

received by lower-level governments This impact is the largest in the case of transfers

allocated by the regional governments while the lowest impact is found for central

government transfers Thanks to an increase on the sample size and to the use of the RDD we

consider these results to be more robust than those in previous studies Concerning the impact

of clientelism on the causal impact of alignment we do not find any evidence of it when welook at central transfers However for regional and upper-level transfers we do find evidence

that the sharp increase on transfers observed in aligned municipalities is greater for smaller

municipalities This effect is also larger for left-wing grantors (compared to right-wing ones)

at the regional level

The remaining paper is structured as follows In section 2 we establish the relevance

of clientelism for our analysis Section 31 contains some brief background information on

Spanish institutions (electoral processes) and intergovernmental grants In section 32 wediscuss the empirical strategy Section 4 presents our main findings as well as the tests that

validate our analysis We conclude with section 5

2 The role of clientelism in Spain heterogeneous effects of partisan alignment

A large part of the Spanish public opinion presumes that the allocation of public

spending in the country is largely influenced by clientelistic ties between voters and

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 630

6

politicians Political clientelism is often cited by Spanish newspapers and its persistence in

this country has its strong roots in the well-known caciquismo of Restoration times -during

the late 19th and early 20th century4 Due to socio-economic changes nowadays traditional

clientelism (based on a peasant clientele system) has been replaced by a modern form of

territorial clientelism also known as ldquobroker clientelismrdquo In her study of Spanish clientelism

Blakeley (2001) explains how the Partido Popular (peoplersquos party) has strategically targeted

Galicia for decades while the Partido Socialista Obrero Espantildeol (socialist party) has

traditionally done so in Andalusia The Andalusian case is frequently illustrated through the

agricultural unemployment payment scheme (PER5) while Galician clientelism is especially

in the form of employment to private business closely linked with the Partido Popular6 In her

text Blakeley also notices the wide-spread accusations of clientelism posed to a long-lasting

ruling party in Catalonia Convergegravencia i Unioacute

There are two dimensions that one should take into account when distinguishing

between clientelism and other related concepts such as pork-barreling and rent-seeking the

scope (how narrow the group of beneficiaries is) and the directedness of the exchange 7

Hence clientelism is based on a quid pro quo arrangement between the patron (politician) and

the client (the voter) where they exchange goods for votes this is made ldquonot simply to seek

but rather to directly reward [electoral] supportrdquo All in all ldquoeven a very lsquobroadrsquo policy maybe lsquoclientelisticrsquo while even very narrow targeting may notrdquo So even broad project grants

such as the ones financed through intergovernmental transfers can be used by governments to

tie the hands of their electorate8

4 For a broader description of the transition from traditional clientelism to partisan clientelism in Spain seeCazorla (1994)5

This agrarian subsidy was implemented by the Andalusian regional government ( Junta de Andaluciacutea) In 1996it was renamed to AEPSA and it extended its scope to other autonomous communities although Andalusia stillreceives a large share of it6 Hopkin (2001) and Corzo (2002) also cite the Galician and Andalusian cases as examples of ldquomodernrdquoclientelism in Spain Robles Egea(2003) focuses on political clientelism in Andalusia7 See Carroll and Lyne (2006) for further discussion on this issue8 We quote two examples extracted from Spanish newspapers to illustrate how transfers can be used withclientelistic purposes both refer to clientelism at the upper-local level The first one is a statement made by EnricMorera a deputy in Les Corts Valencianes (the Valencian regional government)

ldquo(hellip) we have three diputaciones with no competencies who only serve to provide jobs for advisors andtrustworthy people who mandate and in addition the public funds they have at their disposal are used as a toolto preform political clientelism and propaganda of those who are in powerrdquo El Periogravedic 17012011

(httpwwwelperiodiccomnoticias102792_morera-psoe-hablan-acabar-autonomias-pero-diputaciones-nidos-deficit-clientelismo-politicohtml)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 730

7

To present the above ideas in a formal way we present a model borrowing ideas from

Stokes (2005 2007) and Diaz-Cayeros Magaloni and Weingast (2006) We start with a one-

shot game with two actors voters and parties Voters have ideology xi there are two parties 1

(incumbent) or 2 (opposition) who compete to head an upper-tier government (in our study

central regional or upper-local levels) Votersrsquo preferences are as follow

ui= αi times (-12(vi-xi)2) + (1- αi ) times gi(θ micro)

where vi=x1x2 is the citizenrsquos decision to vote for the clientelist party or the

opposition xi represents the voterrsquos ideology whose weight is α ϵ[01] Voters also get utility

from a potential reward gi=0 g(θ micro) depending on whether the clientelist party wins or

not In our study the reward corresponds to intergovernmental transfers As in Diaz-Cayeros

Magaloni and Weingast (2006) failure to support the clientelist party translates into lack of

reward ie punishment through lower funds Furthermore the reward may in turn be

influenced by two factors the political credit (θ) that the grantor can claim when giving the

reward and the partyrsquos ability (micro) to observe the citizenrsquos vote The reward function is

increasing in both factors Stokesrsquo model presumes that ldquoa personrsquos vote is (hellip) perfectly

observable by political partiesrdquo but in a democracy with secret ballot elections like Spain

this is very unlikely to occur However as pointed by Brusco Nazareno and Stokes (2004)

clientelistic strategies can still work under those circumstances This is possible because

ldquoclientelist parties compensate for the inability to observe the vote directly by observing a

range of other actions and behaviors (hellip) that allow party operatives to make good guessesrdquo

In multi-government systems the lower level representatives (the mayors) may be seen as

intermediaries of the upper-level governments and their proximity to citizens makes them

more capable to perceive their constituentsrsquo actions

All in all the timing of the game can be easily summarized in two stages At the

beginning of the first stage t=1 electoral campaigns take place and politicians signal their

The second example refers to a complaint raised by a representative in the diputacioacuten de Castilla-la-ManchaMario Gonzaacutelez Somoano (peoplesrsquo party) to the representative of public works of the diputacioacuten AacutengelMoreno(socialist party) Gonzaacutelez Somoano complains because in his village Gascuentildea there is an emergencysituation as one of the streets is in an extremely bad condition and it needs to be asphalted

ldquoGonzaacutelez Somoano replied that (hellip) they had undertaken asphalt works in neighboring villages of Gascuentildeabut governed by socialist mayors accusing of acting more in lsquotheir villagesrsquo than in others and reminding that allthe roads from Cantalojas (municipality whose mayor is Moreno) had been asphalted the previous yearrdquoGuadaqueacute17072009(httpwwwguadaquecomindexphpoption=com_contentampview=articleampcatid=33Aprovinciaampid=36823Ael-pp-acusa-al-psoe-de-clientelismo-y-rafael-esteban-llama-al-portavoz-

payasoampItemid=54)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 830

8

willingness to provide rewards (g) During these campaigns candidates express promises

which often have implicit threats indicating that the only way to ensure they get the public

goodsservices desired is to vote for them and not for the opposition party In order to target

their voters more accurately candidates often promise specific goods such as infrastructures 9

In this same stage voters express to candidates their willingness to vote for them in exchange

for goods and at the end of the period they cast their vote they have two possibilities either

they renege on their promise or they do not At a second stage t=2 parties take action and

implement public policies Similar to Stokesrsquo analyses there are four possible situations with

different payoffs ndashsummarized in the table below- If the voter complies and votes the

clientelist party he wins and gets the votes of its constituents (vgt0) but if the voter reneges

this party does not get the vote (v=0) and the party loses the elections In this case the party

can still decide whether to pay a reward to the voter or not On the contrary if the party wins

it has two possible options either he complies and rewards its voters (ggt0) or it reneges

(g=0)

PARTY

Reward No Reward

V O T E R

Comply αtimes(-12(xi-x1)2) +(1 α)times g(θ micro) v- g(θ micro) αtimes(-12(xi-x1)

2) v

Defect αtimes(-12(xi-x2)2) +(1 α)times g(θ micro) - g(θ micro) αtimes(-12(xi-x2)

2) 0

As explained the amount of the reward (g) is conditional on both θ and micro So we

would be in the situation ldquoComply-No Reward rdquo whenever the system is totally opaque and the

winning party can hardly be monitored (ie micro rarr0) andor the grantor cannot claim any credit

from the reward (θ=0) Moreover note that to overcome commitment problems the above

model needs to be a repeated game in which micro ne0

From the formal description we can derive four hypotheses on how clientelism could

smooth or exacerbate the impact of political alignment on transfers There may be a chance

that the treatment effect of political alignment on transfers is larger in

H1 smaller municipalities In this case monitoring the voters may be easier (we

expect micro to be higher)

9 In Spain local infrastructures (like sports centers swimming pools roads hospitals etc) are largely financedthrough capital transfers from upper-level governments

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 3: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 330

3

distribution of transfers may be biased towards co-partisans On the one hand following the

intuition beneath Diaz-Cayeros Magaloni and Weingast (2006) in certain circumstances the

incumbent party may have strong incentives to use a mechanism of punishments and rewards

to maintain electoral support Hence municipalities who defect to the opposition (ie

unaligned municipalities) should receive lower transfers as means of punishment This idea is

at odds with the swing voter models of Dixit and Londregan and Lindbeck and Weibull On

the other hand the distribution of grants on partisan grounds is also influenced by the fact that

voters have incomplete information about which tier of government is the source of the grant

In this situation the grantor may not be able to claim much (or even none) political credit for

the grant If there is no credit leakage (ie the grantor cannot claim his credit) he finds it more

profitable to distribute grants to aligned municipalities However if the leakage is large

enough the incumbent grantor may consider sending funds to unaligned municipalities as he

could still get some electoral reward from that The empirical evidence extracted from some

of the above mentioned studies concludes that the allocation of intergovernmental grants is

indeed skewed in favor of aligned municipalities For instance Khemani (2003) finds that

transfers subject to less stringent rules (more discretionary) are positively affected by

alignment while Arulampalam et al (2009) observe that being an aligned and swing state

translates into 16 higher center-state transfers (compared to being unaligned and non-

swing)

The second aforesaid concern that we want to solve is to isolate the effect that

alignment per se has on transfers and to ensure that causality runs in the desired direction To

do that we apply a regression discontinuity design (RDD)1 which allows us to benefit from

the relatively milder assumptions that it requires

The empirical strategy of our paper relies on data on capital transfers received by

Spanish municipalities during two legislatures (2000-2003 and 2004-2007) from threedifferent upper-level administrations (central regional and upper-local) Although there are

two recent contributions studying the effect of partisan alignment on intergovernmental

transfers in Spain Soleacute-Olleacute and Sorribas Navarro (2008a 2008b)2 their main limitation is

the inability to disentangle if larger transfers are explained by higher political support (larger

1 This strategy is described in detail in Lee and Lemieux (2010)2 Both studies are based on a model of political competition where grants to aligned municipalities are assumedto be capable to extract more votes than grants to municipalities represented by opposition parties As proven in

Soleacute-Olleacute and Sorribas-Navarro (2008a) the relatively higher electoral reward of politically alignedmunicipalities leads upper layer governments to transfer larger grants to them

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 430

4

winning margins) or merely by partisan alignment Furthermore the RD estimates that we

obtain are more credible than the ones from ldquonatural experimentsrdquo 3 like the differences-in-

differences used in Soleacute-Olleacute and Sorribas-Navarro (2008a) because the RDD can be

considered as a local randomized experiment in which the (randomized) variation observed in

the political alignment of municipalities is imprecisely controlled by individuals Therefore

the contribution of our paper is not merely to present additional empirical evidence on the

Spanish case but more importantly to provide more robust results through a neat explanatory

variable (partisan alignment)

The clear advantages of the RDD that we have exposed have made it a popular

econometric tool for researchers In the political economy field it has not been widely used

but there are a few studies that apply this strategy to examine the effects of a number ofvariables on government expenditures Albouy (2009) for instance analyses the effect that

the party-identity of the legislator has on the allocation of federal expenditures Ferreira and

Gyourko (2009) test for US municipalities if being the mayor a Democrat or a Republican

affects the allocation of local public spending For Sweden Pettersson-Lidbom (2008)

presents evidence that party control has a causal effect on economic outcomes more

precisely on the level of public expenditures and tax revenues Additionally there are two

applied studies that we consider to be close to ours as they both apply a RDD to examine theeffects of partisan alignment on intergovernmental transfers Migueis (2010) for Portugal

and Brollo and Nannicini (2010) for Brazil We consider we improve upon existing literature

for several reasons First of all while Migueis (2010) considers the causal effect of alignment

on grants to be homogeneous we take a more complex approach by testing the possibility that

this effect is heterogeneous This strategy is also found in the paper by Brollo and Nannicini

where they focus on two dimensions which could affect the impact of alignment on transfers

the value of political capital for the central government and the impact of transfers on

electoral outcomes However our study differs from the latter because we center in a very

specific and different source of heterogeneity which may be more relevant for the Spanish

case clientelism Clientelism is a salient issue in the Spanish socio-economic context and we

think it may be strategically used jointly with partisan tactics to influence electoral results

When clientelistic strategies are accounted for in our benchmark ldquopolitical machinesrdquo not

only allocate transfers based on partisan ties but also on the political credit they can claim and

their ability to monitor their constituentsrsquo votes To undertake this part of our study we draw

3 The formal proof of the lsquosuperiorityrsquo of RDD over lsquonatural experimentsrsquo is in Lee (2008)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 530

5

on ideas from Diaz-Cayeros Magaloni and Weingast (2006) Stokes (2005 2007) Brusco

Nazareno and Stokes (2004) and Soleacute-Olleacute and Sorribas-Navarro (2008a) Based on these

studies we contrast four hypotheses We examine the possibility that the treatment effect of

political alignment on transfers may be larger in poorer municipalities smaller

municipalities municipalities aligned at all levels of government and municipalities with

lower reliance on intergovernmental transfers (eg those with a lower debt burden)

Moreover traditionally Spanish left-wing parties have been accused more often of using

clientelist strategies than right-wing parties so we also test if the political color of the grantor

has an impact on the causal effect of alignment on grants Finally the last hypothesis we

contrast is whether a majority government is more able to allocate transfers on partisan

grounds than a coalition government

To summarize the findings of our paper our first set of results show through RD

estimates that being an aligned municipality has a positive effect on the amount of transfers

received by lower-level governments This impact is the largest in the case of transfers

allocated by the regional governments while the lowest impact is found for central

government transfers Thanks to an increase on the sample size and to the use of the RDD we

consider these results to be more robust than those in previous studies Concerning the impact

of clientelism on the causal impact of alignment we do not find any evidence of it when welook at central transfers However for regional and upper-level transfers we do find evidence

that the sharp increase on transfers observed in aligned municipalities is greater for smaller

municipalities This effect is also larger for left-wing grantors (compared to right-wing ones)

at the regional level

The remaining paper is structured as follows In section 2 we establish the relevance

of clientelism for our analysis Section 31 contains some brief background information on

Spanish institutions (electoral processes) and intergovernmental grants In section 32 wediscuss the empirical strategy Section 4 presents our main findings as well as the tests that

validate our analysis We conclude with section 5

2 The role of clientelism in Spain heterogeneous effects of partisan alignment

A large part of the Spanish public opinion presumes that the allocation of public

spending in the country is largely influenced by clientelistic ties between voters and

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 630

6

politicians Political clientelism is often cited by Spanish newspapers and its persistence in

this country has its strong roots in the well-known caciquismo of Restoration times -during

the late 19th and early 20th century4 Due to socio-economic changes nowadays traditional

clientelism (based on a peasant clientele system) has been replaced by a modern form of

territorial clientelism also known as ldquobroker clientelismrdquo In her study of Spanish clientelism

Blakeley (2001) explains how the Partido Popular (peoplersquos party) has strategically targeted

Galicia for decades while the Partido Socialista Obrero Espantildeol (socialist party) has

traditionally done so in Andalusia The Andalusian case is frequently illustrated through the

agricultural unemployment payment scheme (PER5) while Galician clientelism is especially

in the form of employment to private business closely linked with the Partido Popular6 In her

text Blakeley also notices the wide-spread accusations of clientelism posed to a long-lasting

ruling party in Catalonia Convergegravencia i Unioacute

There are two dimensions that one should take into account when distinguishing

between clientelism and other related concepts such as pork-barreling and rent-seeking the

scope (how narrow the group of beneficiaries is) and the directedness of the exchange 7

Hence clientelism is based on a quid pro quo arrangement between the patron (politician) and

the client (the voter) where they exchange goods for votes this is made ldquonot simply to seek

but rather to directly reward [electoral] supportrdquo All in all ldquoeven a very lsquobroadrsquo policy maybe lsquoclientelisticrsquo while even very narrow targeting may notrdquo So even broad project grants

such as the ones financed through intergovernmental transfers can be used by governments to

tie the hands of their electorate8

4 For a broader description of the transition from traditional clientelism to partisan clientelism in Spain seeCazorla (1994)5

This agrarian subsidy was implemented by the Andalusian regional government ( Junta de Andaluciacutea) In 1996it was renamed to AEPSA and it extended its scope to other autonomous communities although Andalusia stillreceives a large share of it6 Hopkin (2001) and Corzo (2002) also cite the Galician and Andalusian cases as examples of ldquomodernrdquoclientelism in Spain Robles Egea(2003) focuses on political clientelism in Andalusia7 See Carroll and Lyne (2006) for further discussion on this issue8 We quote two examples extracted from Spanish newspapers to illustrate how transfers can be used withclientelistic purposes both refer to clientelism at the upper-local level The first one is a statement made by EnricMorera a deputy in Les Corts Valencianes (the Valencian regional government)

ldquo(hellip) we have three diputaciones with no competencies who only serve to provide jobs for advisors andtrustworthy people who mandate and in addition the public funds they have at their disposal are used as a toolto preform political clientelism and propaganda of those who are in powerrdquo El Periogravedic 17012011

(httpwwwelperiodiccomnoticias102792_morera-psoe-hablan-acabar-autonomias-pero-diputaciones-nidos-deficit-clientelismo-politicohtml)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 730

7

To present the above ideas in a formal way we present a model borrowing ideas from

Stokes (2005 2007) and Diaz-Cayeros Magaloni and Weingast (2006) We start with a one-

shot game with two actors voters and parties Voters have ideology xi there are two parties 1

(incumbent) or 2 (opposition) who compete to head an upper-tier government (in our study

central regional or upper-local levels) Votersrsquo preferences are as follow

ui= αi times (-12(vi-xi)2) + (1- αi ) times gi(θ micro)

where vi=x1x2 is the citizenrsquos decision to vote for the clientelist party or the

opposition xi represents the voterrsquos ideology whose weight is α ϵ[01] Voters also get utility

from a potential reward gi=0 g(θ micro) depending on whether the clientelist party wins or

not In our study the reward corresponds to intergovernmental transfers As in Diaz-Cayeros

Magaloni and Weingast (2006) failure to support the clientelist party translates into lack of

reward ie punishment through lower funds Furthermore the reward may in turn be

influenced by two factors the political credit (θ) that the grantor can claim when giving the

reward and the partyrsquos ability (micro) to observe the citizenrsquos vote The reward function is

increasing in both factors Stokesrsquo model presumes that ldquoa personrsquos vote is (hellip) perfectly

observable by political partiesrdquo but in a democracy with secret ballot elections like Spain

this is very unlikely to occur However as pointed by Brusco Nazareno and Stokes (2004)

clientelistic strategies can still work under those circumstances This is possible because

ldquoclientelist parties compensate for the inability to observe the vote directly by observing a

range of other actions and behaviors (hellip) that allow party operatives to make good guessesrdquo

In multi-government systems the lower level representatives (the mayors) may be seen as

intermediaries of the upper-level governments and their proximity to citizens makes them

more capable to perceive their constituentsrsquo actions

All in all the timing of the game can be easily summarized in two stages At the

beginning of the first stage t=1 electoral campaigns take place and politicians signal their

The second example refers to a complaint raised by a representative in the diputacioacuten de Castilla-la-ManchaMario Gonzaacutelez Somoano (peoplesrsquo party) to the representative of public works of the diputacioacuten AacutengelMoreno(socialist party) Gonzaacutelez Somoano complains because in his village Gascuentildea there is an emergencysituation as one of the streets is in an extremely bad condition and it needs to be asphalted

ldquoGonzaacutelez Somoano replied that (hellip) they had undertaken asphalt works in neighboring villages of Gascuentildeabut governed by socialist mayors accusing of acting more in lsquotheir villagesrsquo than in others and reminding that allthe roads from Cantalojas (municipality whose mayor is Moreno) had been asphalted the previous yearrdquoGuadaqueacute17072009(httpwwwguadaquecomindexphpoption=com_contentampview=articleampcatid=33Aprovinciaampid=36823Ael-pp-acusa-al-psoe-de-clientelismo-y-rafael-esteban-llama-al-portavoz-

payasoampItemid=54)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 830

8

willingness to provide rewards (g) During these campaigns candidates express promises

which often have implicit threats indicating that the only way to ensure they get the public

goodsservices desired is to vote for them and not for the opposition party In order to target

their voters more accurately candidates often promise specific goods such as infrastructures 9

In this same stage voters express to candidates their willingness to vote for them in exchange

for goods and at the end of the period they cast their vote they have two possibilities either

they renege on their promise or they do not At a second stage t=2 parties take action and

implement public policies Similar to Stokesrsquo analyses there are four possible situations with

different payoffs ndashsummarized in the table below- If the voter complies and votes the

clientelist party he wins and gets the votes of its constituents (vgt0) but if the voter reneges

this party does not get the vote (v=0) and the party loses the elections In this case the party

can still decide whether to pay a reward to the voter or not On the contrary if the party wins

it has two possible options either he complies and rewards its voters (ggt0) or it reneges

(g=0)

PARTY

Reward No Reward

V O T E R

Comply αtimes(-12(xi-x1)2) +(1 α)times g(θ micro) v- g(θ micro) αtimes(-12(xi-x1)

2) v

Defect αtimes(-12(xi-x2)2) +(1 α)times g(θ micro) - g(θ micro) αtimes(-12(xi-x2)

2) 0

As explained the amount of the reward (g) is conditional on both θ and micro So we

would be in the situation ldquoComply-No Reward rdquo whenever the system is totally opaque and the

winning party can hardly be monitored (ie micro rarr0) andor the grantor cannot claim any credit

from the reward (θ=0) Moreover note that to overcome commitment problems the above

model needs to be a repeated game in which micro ne0

From the formal description we can derive four hypotheses on how clientelism could

smooth or exacerbate the impact of political alignment on transfers There may be a chance

that the treatment effect of political alignment on transfers is larger in

H1 smaller municipalities In this case monitoring the voters may be easier (we

expect micro to be higher)

9 In Spain local infrastructures (like sports centers swimming pools roads hospitals etc) are largely financedthrough capital transfers from upper-level governments

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 4: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 430

4

winning margins) or merely by partisan alignment Furthermore the RD estimates that we

obtain are more credible than the ones from ldquonatural experimentsrdquo 3 like the differences-in-

differences used in Soleacute-Olleacute and Sorribas-Navarro (2008a) because the RDD can be

considered as a local randomized experiment in which the (randomized) variation observed in

the political alignment of municipalities is imprecisely controlled by individuals Therefore

the contribution of our paper is not merely to present additional empirical evidence on the

Spanish case but more importantly to provide more robust results through a neat explanatory

variable (partisan alignment)

The clear advantages of the RDD that we have exposed have made it a popular

econometric tool for researchers In the political economy field it has not been widely used

but there are a few studies that apply this strategy to examine the effects of a number ofvariables on government expenditures Albouy (2009) for instance analyses the effect that

the party-identity of the legislator has on the allocation of federal expenditures Ferreira and

Gyourko (2009) test for US municipalities if being the mayor a Democrat or a Republican

affects the allocation of local public spending For Sweden Pettersson-Lidbom (2008)

presents evidence that party control has a causal effect on economic outcomes more

precisely on the level of public expenditures and tax revenues Additionally there are two

applied studies that we consider to be close to ours as they both apply a RDD to examine theeffects of partisan alignment on intergovernmental transfers Migueis (2010) for Portugal

and Brollo and Nannicini (2010) for Brazil We consider we improve upon existing literature

for several reasons First of all while Migueis (2010) considers the causal effect of alignment

on grants to be homogeneous we take a more complex approach by testing the possibility that

this effect is heterogeneous This strategy is also found in the paper by Brollo and Nannicini

where they focus on two dimensions which could affect the impact of alignment on transfers

the value of political capital for the central government and the impact of transfers on

electoral outcomes However our study differs from the latter because we center in a very

specific and different source of heterogeneity which may be more relevant for the Spanish

case clientelism Clientelism is a salient issue in the Spanish socio-economic context and we

think it may be strategically used jointly with partisan tactics to influence electoral results

When clientelistic strategies are accounted for in our benchmark ldquopolitical machinesrdquo not

only allocate transfers based on partisan ties but also on the political credit they can claim and

their ability to monitor their constituentsrsquo votes To undertake this part of our study we draw

3 The formal proof of the lsquosuperiorityrsquo of RDD over lsquonatural experimentsrsquo is in Lee (2008)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 530

5

on ideas from Diaz-Cayeros Magaloni and Weingast (2006) Stokes (2005 2007) Brusco

Nazareno and Stokes (2004) and Soleacute-Olleacute and Sorribas-Navarro (2008a) Based on these

studies we contrast four hypotheses We examine the possibility that the treatment effect of

political alignment on transfers may be larger in poorer municipalities smaller

municipalities municipalities aligned at all levels of government and municipalities with

lower reliance on intergovernmental transfers (eg those with a lower debt burden)

Moreover traditionally Spanish left-wing parties have been accused more often of using

clientelist strategies than right-wing parties so we also test if the political color of the grantor

has an impact on the causal effect of alignment on grants Finally the last hypothesis we

contrast is whether a majority government is more able to allocate transfers on partisan

grounds than a coalition government

To summarize the findings of our paper our first set of results show through RD

estimates that being an aligned municipality has a positive effect on the amount of transfers

received by lower-level governments This impact is the largest in the case of transfers

allocated by the regional governments while the lowest impact is found for central

government transfers Thanks to an increase on the sample size and to the use of the RDD we

consider these results to be more robust than those in previous studies Concerning the impact

of clientelism on the causal impact of alignment we do not find any evidence of it when welook at central transfers However for regional and upper-level transfers we do find evidence

that the sharp increase on transfers observed in aligned municipalities is greater for smaller

municipalities This effect is also larger for left-wing grantors (compared to right-wing ones)

at the regional level

The remaining paper is structured as follows In section 2 we establish the relevance

of clientelism for our analysis Section 31 contains some brief background information on

Spanish institutions (electoral processes) and intergovernmental grants In section 32 wediscuss the empirical strategy Section 4 presents our main findings as well as the tests that

validate our analysis We conclude with section 5

2 The role of clientelism in Spain heterogeneous effects of partisan alignment

A large part of the Spanish public opinion presumes that the allocation of public

spending in the country is largely influenced by clientelistic ties between voters and

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 630

6

politicians Political clientelism is often cited by Spanish newspapers and its persistence in

this country has its strong roots in the well-known caciquismo of Restoration times -during

the late 19th and early 20th century4 Due to socio-economic changes nowadays traditional

clientelism (based on a peasant clientele system) has been replaced by a modern form of

territorial clientelism also known as ldquobroker clientelismrdquo In her study of Spanish clientelism

Blakeley (2001) explains how the Partido Popular (peoplersquos party) has strategically targeted

Galicia for decades while the Partido Socialista Obrero Espantildeol (socialist party) has

traditionally done so in Andalusia The Andalusian case is frequently illustrated through the

agricultural unemployment payment scheme (PER5) while Galician clientelism is especially

in the form of employment to private business closely linked with the Partido Popular6 In her

text Blakeley also notices the wide-spread accusations of clientelism posed to a long-lasting

ruling party in Catalonia Convergegravencia i Unioacute

There are two dimensions that one should take into account when distinguishing

between clientelism and other related concepts such as pork-barreling and rent-seeking the

scope (how narrow the group of beneficiaries is) and the directedness of the exchange 7

Hence clientelism is based on a quid pro quo arrangement between the patron (politician) and

the client (the voter) where they exchange goods for votes this is made ldquonot simply to seek

but rather to directly reward [electoral] supportrdquo All in all ldquoeven a very lsquobroadrsquo policy maybe lsquoclientelisticrsquo while even very narrow targeting may notrdquo So even broad project grants

such as the ones financed through intergovernmental transfers can be used by governments to

tie the hands of their electorate8

4 For a broader description of the transition from traditional clientelism to partisan clientelism in Spain seeCazorla (1994)5

This agrarian subsidy was implemented by the Andalusian regional government ( Junta de Andaluciacutea) In 1996it was renamed to AEPSA and it extended its scope to other autonomous communities although Andalusia stillreceives a large share of it6 Hopkin (2001) and Corzo (2002) also cite the Galician and Andalusian cases as examples of ldquomodernrdquoclientelism in Spain Robles Egea(2003) focuses on political clientelism in Andalusia7 See Carroll and Lyne (2006) for further discussion on this issue8 We quote two examples extracted from Spanish newspapers to illustrate how transfers can be used withclientelistic purposes both refer to clientelism at the upper-local level The first one is a statement made by EnricMorera a deputy in Les Corts Valencianes (the Valencian regional government)

ldquo(hellip) we have three diputaciones with no competencies who only serve to provide jobs for advisors andtrustworthy people who mandate and in addition the public funds they have at their disposal are used as a toolto preform political clientelism and propaganda of those who are in powerrdquo El Periogravedic 17012011

(httpwwwelperiodiccomnoticias102792_morera-psoe-hablan-acabar-autonomias-pero-diputaciones-nidos-deficit-clientelismo-politicohtml)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 730

7

To present the above ideas in a formal way we present a model borrowing ideas from

Stokes (2005 2007) and Diaz-Cayeros Magaloni and Weingast (2006) We start with a one-

shot game with two actors voters and parties Voters have ideology xi there are two parties 1

(incumbent) or 2 (opposition) who compete to head an upper-tier government (in our study

central regional or upper-local levels) Votersrsquo preferences are as follow

ui= αi times (-12(vi-xi)2) + (1- αi ) times gi(θ micro)

where vi=x1x2 is the citizenrsquos decision to vote for the clientelist party or the

opposition xi represents the voterrsquos ideology whose weight is α ϵ[01] Voters also get utility

from a potential reward gi=0 g(θ micro) depending on whether the clientelist party wins or

not In our study the reward corresponds to intergovernmental transfers As in Diaz-Cayeros

Magaloni and Weingast (2006) failure to support the clientelist party translates into lack of

reward ie punishment through lower funds Furthermore the reward may in turn be

influenced by two factors the political credit (θ) that the grantor can claim when giving the

reward and the partyrsquos ability (micro) to observe the citizenrsquos vote The reward function is

increasing in both factors Stokesrsquo model presumes that ldquoa personrsquos vote is (hellip) perfectly

observable by political partiesrdquo but in a democracy with secret ballot elections like Spain

this is very unlikely to occur However as pointed by Brusco Nazareno and Stokes (2004)

clientelistic strategies can still work under those circumstances This is possible because

ldquoclientelist parties compensate for the inability to observe the vote directly by observing a

range of other actions and behaviors (hellip) that allow party operatives to make good guessesrdquo

In multi-government systems the lower level representatives (the mayors) may be seen as

intermediaries of the upper-level governments and their proximity to citizens makes them

more capable to perceive their constituentsrsquo actions

All in all the timing of the game can be easily summarized in two stages At the

beginning of the first stage t=1 electoral campaigns take place and politicians signal their

The second example refers to a complaint raised by a representative in the diputacioacuten de Castilla-la-ManchaMario Gonzaacutelez Somoano (peoplesrsquo party) to the representative of public works of the diputacioacuten AacutengelMoreno(socialist party) Gonzaacutelez Somoano complains because in his village Gascuentildea there is an emergencysituation as one of the streets is in an extremely bad condition and it needs to be asphalted

ldquoGonzaacutelez Somoano replied that (hellip) they had undertaken asphalt works in neighboring villages of Gascuentildeabut governed by socialist mayors accusing of acting more in lsquotheir villagesrsquo than in others and reminding that allthe roads from Cantalojas (municipality whose mayor is Moreno) had been asphalted the previous yearrdquoGuadaqueacute17072009(httpwwwguadaquecomindexphpoption=com_contentampview=articleampcatid=33Aprovinciaampid=36823Ael-pp-acusa-al-psoe-de-clientelismo-y-rafael-esteban-llama-al-portavoz-

payasoampItemid=54)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 830

8

willingness to provide rewards (g) During these campaigns candidates express promises

which often have implicit threats indicating that the only way to ensure they get the public

goodsservices desired is to vote for them and not for the opposition party In order to target

their voters more accurately candidates often promise specific goods such as infrastructures 9

In this same stage voters express to candidates their willingness to vote for them in exchange

for goods and at the end of the period they cast their vote they have two possibilities either

they renege on their promise or they do not At a second stage t=2 parties take action and

implement public policies Similar to Stokesrsquo analyses there are four possible situations with

different payoffs ndashsummarized in the table below- If the voter complies and votes the

clientelist party he wins and gets the votes of its constituents (vgt0) but if the voter reneges

this party does not get the vote (v=0) and the party loses the elections In this case the party

can still decide whether to pay a reward to the voter or not On the contrary if the party wins

it has two possible options either he complies and rewards its voters (ggt0) or it reneges

(g=0)

PARTY

Reward No Reward

V O T E R

Comply αtimes(-12(xi-x1)2) +(1 α)times g(θ micro) v- g(θ micro) αtimes(-12(xi-x1)

2) v

Defect αtimes(-12(xi-x2)2) +(1 α)times g(θ micro) - g(θ micro) αtimes(-12(xi-x2)

2) 0

As explained the amount of the reward (g) is conditional on both θ and micro So we

would be in the situation ldquoComply-No Reward rdquo whenever the system is totally opaque and the

winning party can hardly be monitored (ie micro rarr0) andor the grantor cannot claim any credit

from the reward (θ=0) Moreover note that to overcome commitment problems the above

model needs to be a repeated game in which micro ne0

From the formal description we can derive four hypotheses on how clientelism could

smooth or exacerbate the impact of political alignment on transfers There may be a chance

that the treatment effect of political alignment on transfers is larger in

H1 smaller municipalities In this case monitoring the voters may be easier (we

expect micro to be higher)

9 In Spain local infrastructures (like sports centers swimming pools roads hospitals etc) are largely financedthrough capital transfers from upper-level governments

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 5: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 530

5

on ideas from Diaz-Cayeros Magaloni and Weingast (2006) Stokes (2005 2007) Brusco

Nazareno and Stokes (2004) and Soleacute-Olleacute and Sorribas-Navarro (2008a) Based on these

studies we contrast four hypotheses We examine the possibility that the treatment effect of

political alignment on transfers may be larger in poorer municipalities smaller

municipalities municipalities aligned at all levels of government and municipalities with

lower reliance on intergovernmental transfers (eg those with a lower debt burden)

Moreover traditionally Spanish left-wing parties have been accused more often of using

clientelist strategies than right-wing parties so we also test if the political color of the grantor

has an impact on the causal effect of alignment on grants Finally the last hypothesis we

contrast is whether a majority government is more able to allocate transfers on partisan

grounds than a coalition government

To summarize the findings of our paper our first set of results show through RD

estimates that being an aligned municipality has a positive effect on the amount of transfers

received by lower-level governments This impact is the largest in the case of transfers

allocated by the regional governments while the lowest impact is found for central

government transfers Thanks to an increase on the sample size and to the use of the RDD we

consider these results to be more robust than those in previous studies Concerning the impact

of clientelism on the causal impact of alignment we do not find any evidence of it when welook at central transfers However for regional and upper-level transfers we do find evidence

that the sharp increase on transfers observed in aligned municipalities is greater for smaller

municipalities This effect is also larger for left-wing grantors (compared to right-wing ones)

at the regional level

The remaining paper is structured as follows In section 2 we establish the relevance

of clientelism for our analysis Section 31 contains some brief background information on

Spanish institutions (electoral processes) and intergovernmental grants In section 32 wediscuss the empirical strategy Section 4 presents our main findings as well as the tests that

validate our analysis We conclude with section 5

2 The role of clientelism in Spain heterogeneous effects of partisan alignment

A large part of the Spanish public opinion presumes that the allocation of public

spending in the country is largely influenced by clientelistic ties between voters and

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 630

6

politicians Political clientelism is often cited by Spanish newspapers and its persistence in

this country has its strong roots in the well-known caciquismo of Restoration times -during

the late 19th and early 20th century4 Due to socio-economic changes nowadays traditional

clientelism (based on a peasant clientele system) has been replaced by a modern form of

territorial clientelism also known as ldquobroker clientelismrdquo In her study of Spanish clientelism

Blakeley (2001) explains how the Partido Popular (peoplersquos party) has strategically targeted

Galicia for decades while the Partido Socialista Obrero Espantildeol (socialist party) has

traditionally done so in Andalusia The Andalusian case is frequently illustrated through the

agricultural unemployment payment scheme (PER5) while Galician clientelism is especially

in the form of employment to private business closely linked with the Partido Popular6 In her

text Blakeley also notices the wide-spread accusations of clientelism posed to a long-lasting

ruling party in Catalonia Convergegravencia i Unioacute

There are two dimensions that one should take into account when distinguishing

between clientelism and other related concepts such as pork-barreling and rent-seeking the

scope (how narrow the group of beneficiaries is) and the directedness of the exchange 7

Hence clientelism is based on a quid pro quo arrangement between the patron (politician) and

the client (the voter) where they exchange goods for votes this is made ldquonot simply to seek

but rather to directly reward [electoral] supportrdquo All in all ldquoeven a very lsquobroadrsquo policy maybe lsquoclientelisticrsquo while even very narrow targeting may notrdquo So even broad project grants

such as the ones financed through intergovernmental transfers can be used by governments to

tie the hands of their electorate8

4 For a broader description of the transition from traditional clientelism to partisan clientelism in Spain seeCazorla (1994)5

This agrarian subsidy was implemented by the Andalusian regional government ( Junta de Andaluciacutea) In 1996it was renamed to AEPSA and it extended its scope to other autonomous communities although Andalusia stillreceives a large share of it6 Hopkin (2001) and Corzo (2002) also cite the Galician and Andalusian cases as examples of ldquomodernrdquoclientelism in Spain Robles Egea(2003) focuses on political clientelism in Andalusia7 See Carroll and Lyne (2006) for further discussion on this issue8 We quote two examples extracted from Spanish newspapers to illustrate how transfers can be used withclientelistic purposes both refer to clientelism at the upper-local level The first one is a statement made by EnricMorera a deputy in Les Corts Valencianes (the Valencian regional government)

ldquo(hellip) we have three diputaciones with no competencies who only serve to provide jobs for advisors andtrustworthy people who mandate and in addition the public funds they have at their disposal are used as a toolto preform political clientelism and propaganda of those who are in powerrdquo El Periogravedic 17012011

(httpwwwelperiodiccomnoticias102792_morera-psoe-hablan-acabar-autonomias-pero-diputaciones-nidos-deficit-clientelismo-politicohtml)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 730

7

To present the above ideas in a formal way we present a model borrowing ideas from

Stokes (2005 2007) and Diaz-Cayeros Magaloni and Weingast (2006) We start with a one-

shot game with two actors voters and parties Voters have ideology xi there are two parties 1

(incumbent) or 2 (opposition) who compete to head an upper-tier government (in our study

central regional or upper-local levels) Votersrsquo preferences are as follow

ui= αi times (-12(vi-xi)2) + (1- αi ) times gi(θ micro)

where vi=x1x2 is the citizenrsquos decision to vote for the clientelist party or the

opposition xi represents the voterrsquos ideology whose weight is α ϵ[01] Voters also get utility

from a potential reward gi=0 g(θ micro) depending on whether the clientelist party wins or

not In our study the reward corresponds to intergovernmental transfers As in Diaz-Cayeros

Magaloni and Weingast (2006) failure to support the clientelist party translates into lack of

reward ie punishment through lower funds Furthermore the reward may in turn be

influenced by two factors the political credit (θ) that the grantor can claim when giving the

reward and the partyrsquos ability (micro) to observe the citizenrsquos vote The reward function is

increasing in both factors Stokesrsquo model presumes that ldquoa personrsquos vote is (hellip) perfectly

observable by political partiesrdquo but in a democracy with secret ballot elections like Spain

this is very unlikely to occur However as pointed by Brusco Nazareno and Stokes (2004)

clientelistic strategies can still work under those circumstances This is possible because

ldquoclientelist parties compensate for the inability to observe the vote directly by observing a

range of other actions and behaviors (hellip) that allow party operatives to make good guessesrdquo

In multi-government systems the lower level representatives (the mayors) may be seen as

intermediaries of the upper-level governments and their proximity to citizens makes them

more capable to perceive their constituentsrsquo actions

All in all the timing of the game can be easily summarized in two stages At the

beginning of the first stage t=1 electoral campaigns take place and politicians signal their

The second example refers to a complaint raised by a representative in the diputacioacuten de Castilla-la-ManchaMario Gonzaacutelez Somoano (peoplesrsquo party) to the representative of public works of the diputacioacuten AacutengelMoreno(socialist party) Gonzaacutelez Somoano complains because in his village Gascuentildea there is an emergencysituation as one of the streets is in an extremely bad condition and it needs to be asphalted

ldquoGonzaacutelez Somoano replied that (hellip) they had undertaken asphalt works in neighboring villages of Gascuentildeabut governed by socialist mayors accusing of acting more in lsquotheir villagesrsquo than in others and reminding that allthe roads from Cantalojas (municipality whose mayor is Moreno) had been asphalted the previous yearrdquoGuadaqueacute17072009(httpwwwguadaquecomindexphpoption=com_contentampview=articleampcatid=33Aprovinciaampid=36823Ael-pp-acusa-al-psoe-de-clientelismo-y-rafael-esteban-llama-al-portavoz-

payasoampItemid=54)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 830

8

willingness to provide rewards (g) During these campaigns candidates express promises

which often have implicit threats indicating that the only way to ensure they get the public

goodsservices desired is to vote for them and not for the opposition party In order to target

their voters more accurately candidates often promise specific goods such as infrastructures 9

In this same stage voters express to candidates their willingness to vote for them in exchange

for goods and at the end of the period they cast their vote they have two possibilities either

they renege on their promise or they do not At a second stage t=2 parties take action and

implement public policies Similar to Stokesrsquo analyses there are four possible situations with

different payoffs ndashsummarized in the table below- If the voter complies and votes the

clientelist party he wins and gets the votes of its constituents (vgt0) but if the voter reneges

this party does not get the vote (v=0) and the party loses the elections In this case the party

can still decide whether to pay a reward to the voter or not On the contrary if the party wins

it has two possible options either he complies and rewards its voters (ggt0) or it reneges

(g=0)

PARTY

Reward No Reward

V O T E R

Comply αtimes(-12(xi-x1)2) +(1 α)times g(θ micro) v- g(θ micro) αtimes(-12(xi-x1)

2) v

Defect αtimes(-12(xi-x2)2) +(1 α)times g(θ micro) - g(θ micro) αtimes(-12(xi-x2)

2) 0

As explained the amount of the reward (g) is conditional on both θ and micro So we

would be in the situation ldquoComply-No Reward rdquo whenever the system is totally opaque and the

winning party can hardly be monitored (ie micro rarr0) andor the grantor cannot claim any credit

from the reward (θ=0) Moreover note that to overcome commitment problems the above

model needs to be a repeated game in which micro ne0

From the formal description we can derive four hypotheses on how clientelism could

smooth or exacerbate the impact of political alignment on transfers There may be a chance

that the treatment effect of political alignment on transfers is larger in

H1 smaller municipalities In this case monitoring the voters may be easier (we

expect micro to be higher)

9 In Spain local infrastructures (like sports centers swimming pools roads hospitals etc) are largely financedthrough capital transfers from upper-level governments

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 6: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 630

6

politicians Political clientelism is often cited by Spanish newspapers and its persistence in

this country has its strong roots in the well-known caciquismo of Restoration times -during

the late 19th and early 20th century4 Due to socio-economic changes nowadays traditional

clientelism (based on a peasant clientele system) has been replaced by a modern form of

territorial clientelism also known as ldquobroker clientelismrdquo In her study of Spanish clientelism

Blakeley (2001) explains how the Partido Popular (peoplersquos party) has strategically targeted

Galicia for decades while the Partido Socialista Obrero Espantildeol (socialist party) has

traditionally done so in Andalusia The Andalusian case is frequently illustrated through the

agricultural unemployment payment scheme (PER5) while Galician clientelism is especially

in the form of employment to private business closely linked with the Partido Popular6 In her

text Blakeley also notices the wide-spread accusations of clientelism posed to a long-lasting

ruling party in Catalonia Convergegravencia i Unioacute

There are two dimensions that one should take into account when distinguishing

between clientelism and other related concepts such as pork-barreling and rent-seeking the

scope (how narrow the group of beneficiaries is) and the directedness of the exchange 7

Hence clientelism is based on a quid pro quo arrangement between the patron (politician) and

the client (the voter) where they exchange goods for votes this is made ldquonot simply to seek

but rather to directly reward [electoral] supportrdquo All in all ldquoeven a very lsquobroadrsquo policy maybe lsquoclientelisticrsquo while even very narrow targeting may notrdquo So even broad project grants

such as the ones financed through intergovernmental transfers can be used by governments to

tie the hands of their electorate8

4 For a broader description of the transition from traditional clientelism to partisan clientelism in Spain seeCazorla (1994)5

This agrarian subsidy was implemented by the Andalusian regional government ( Junta de Andaluciacutea) In 1996it was renamed to AEPSA and it extended its scope to other autonomous communities although Andalusia stillreceives a large share of it6 Hopkin (2001) and Corzo (2002) also cite the Galician and Andalusian cases as examples of ldquomodernrdquoclientelism in Spain Robles Egea(2003) focuses on political clientelism in Andalusia7 See Carroll and Lyne (2006) for further discussion on this issue8 We quote two examples extracted from Spanish newspapers to illustrate how transfers can be used withclientelistic purposes both refer to clientelism at the upper-local level The first one is a statement made by EnricMorera a deputy in Les Corts Valencianes (the Valencian regional government)

ldquo(hellip) we have three diputaciones with no competencies who only serve to provide jobs for advisors andtrustworthy people who mandate and in addition the public funds they have at their disposal are used as a toolto preform political clientelism and propaganda of those who are in powerrdquo El Periogravedic 17012011

(httpwwwelperiodiccomnoticias102792_morera-psoe-hablan-acabar-autonomias-pero-diputaciones-nidos-deficit-clientelismo-politicohtml)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 730

7

To present the above ideas in a formal way we present a model borrowing ideas from

Stokes (2005 2007) and Diaz-Cayeros Magaloni and Weingast (2006) We start with a one-

shot game with two actors voters and parties Voters have ideology xi there are two parties 1

(incumbent) or 2 (opposition) who compete to head an upper-tier government (in our study

central regional or upper-local levels) Votersrsquo preferences are as follow

ui= αi times (-12(vi-xi)2) + (1- αi ) times gi(θ micro)

where vi=x1x2 is the citizenrsquos decision to vote for the clientelist party or the

opposition xi represents the voterrsquos ideology whose weight is α ϵ[01] Voters also get utility

from a potential reward gi=0 g(θ micro) depending on whether the clientelist party wins or

not In our study the reward corresponds to intergovernmental transfers As in Diaz-Cayeros

Magaloni and Weingast (2006) failure to support the clientelist party translates into lack of

reward ie punishment through lower funds Furthermore the reward may in turn be

influenced by two factors the political credit (θ) that the grantor can claim when giving the

reward and the partyrsquos ability (micro) to observe the citizenrsquos vote The reward function is

increasing in both factors Stokesrsquo model presumes that ldquoa personrsquos vote is (hellip) perfectly

observable by political partiesrdquo but in a democracy with secret ballot elections like Spain

this is very unlikely to occur However as pointed by Brusco Nazareno and Stokes (2004)

clientelistic strategies can still work under those circumstances This is possible because

ldquoclientelist parties compensate for the inability to observe the vote directly by observing a

range of other actions and behaviors (hellip) that allow party operatives to make good guessesrdquo

In multi-government systems the lower level representatives (the mayors) may be seen as

intermediaries of the upper-level governments and their proximity to citizens makes them

more capable to perceive their constituentsrsquo actions

All in all the timing of the game can be easily summarized in two stages At the

beginning of the first stage t=1 electoral campaigns take place and politicians signal their

The second example refers to a complaint raised by a representative in the diputacioacuten de Castilla-la-ManchaMario Gonzaacutelez Somoano (peoplesrsquo party) to the representative of public works of the diputacioacuten AacutengelMoreno(socialist party) Gonzaacutelez Somoano complains because in his village Gascuentildea there is an emergencysituation as one of the streets is in an extremely bad condition and it needs to be asphalted

ldquoGonzaacutelez Somoano replied that (hellip) they had undertaken asphalt works in neighboring villages of Gascuentildeabut governed by socialist mayors accusing of acting more in lsquotheir villagesrsquo than in others and reminding that allthe roads from Cantalojas (municipality whose mayor is Moreno) had been asphalted the previous yearrdquoGuadaqueacute17072009(httpwwwguadaquecomindexphpoption=com_contentampview=articleampcatid=33Aprovinciaampid=36823Ael-pp-acusa-al-psoe-de-clientelismo-y-rafael-esteban-llama-al-portavoz-

payasoampItemid=54)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 830

8

willingness to provide rewards (g) During these campaigns candidates express promises

which often have implicit threats indicating that the only way to ensure they get the public

goodsservices desired is to vote for them and not for the opposition party In order to target

their voters more accurately candidates often promise specific goods such as infrastructures 9

In this same stage voters express to candidates their willingness to vote for them in exchange

for goods and at the end of the period they cast their vote they have two possibilities either

they renege on their promise or they do not At a second stage t=2 parties take action and

implement public policies Similar to Stokesrsquo analyses there are four possible situations with

different payoffs ndashsummarized in the table below- If the voter complies and votes the

clientelist party he wins and gets the votes of its constituents (vgt0) but if the voter reneges

this party does not get the vote (v=0) and the party loses the elections In this case the party

can still decide whether to pay a reward to the voter or not On the contrary if the party wins

it has two possible options either he complies and rewards its voters (ggt0) or it reneges

(g=0)

PARTY

Reward No Reward

V O T E R

Comply αtimes(-12(xi-x1)2) +(1 α)times g(θ micro) v- g(θ micro) αtimes(-12(xi-x1)

2) v

Defect αtimes(-12(xi-x2)2) +(1 α)times g(θ micro) - g(θ micro) αtimes(-12(xi-x2)

2) 0

As explained the amount of the reward (g) is conditional on both θ and micro So we

would be in the situation ldquoComply-No Reward rdquo whenever the system is totally opaque and the

winning party can hardly be monitored (ie micro rarr0) andor the grantor cannot claim any credit

from the reward (θ=0) Moreover note that to overcome commitment problems the above

model needs to be a repeated game in which micro ne0

From the formal description we can derive four hypotheses on how clientelism could

smooth or exacerbate the impact of political alignment on transfers There may be a chance

that the treatment effect of political alignment on transfers is larger in

H1 smaller municipalities In this case monitoring the voters may be easier (we

expect micro to be higher)

9 In Spain local infrastructures (like sports centers swimming pools roads hospitals etc) are largely financedthrough capital transfers from upper-level governments

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 7: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 730

7

To present the above ideas in a formal way we present a model borrowing ideas from

Stokes (2005 2007) and Diaz-Cayeros Magaloni and Weingast (2006) We start with a one-

shot game with two actors voters and parties Voters have ideology xi there are two parties 1

(incumbent) or 2 (opposition) who compete to head an upper-tier government (in our study

central regional or upper-local levels) Votersrsquo preferences are as follow

ui= αi times (-12(vi-xi)2) + (1- αi ) times gi(θ micro)

where vi=x1x2 is the citizenrsquos decision to vote for the clientelist party or the

opposition xi represents the voterrsquos ideology whose weight is α ϵ[01] Voters also get utility

from a potential reward gi=0 g(θ micro) depending on whether the clientelist party wins or

not In our study the reward corresponds to intergovernmental transfers As in Diaz-Cayeros

Magaloni and Weingast (2006) failure to support the clientelist party translates into lack of

reward ie punishment through lower funds Furthermore the reward may in turn be

influenced by two factors the political credit (θ) that the grantor can claim when giving the

reward and the partyrsquos ability (micro) to observe the citizenrsquos vote The reward function is

increasing in both factors Stokesrsquo model presumes that ldquoa personrsquos vote is (hellip) perfectly

observable by political partiesrdquo but in a democracy with secret ballot elections like Spain

this is very unlikely to occur However as pointed by Brusco Nazareno and Stokes (2004)

clientelistic strategies can still work under those circumstances This is possible because

ldquoclientelist parties compensate for the inability to observe the vote directly by observing a

range of other actions and behaviors (hellip) that allow party operatives to make good guessesrdquo

In multi-government systems the lower level representatives (the mayors) may be seen as

intermediaries of the upper-level governments and their proximity to citizens makes them

more capable to perceive their constituentsrsquo actions

All in all the timing of the game can be easily summarized in two stages At the

beginning of the first stage t=1 electoral campaigns take place and politicians signal their

The second example refers to a complaint raised by a representative in the diputacioacuten de Castilla-la-ManchaMario Gonzaacutelez Somoano (peoplesrsquo party) to the representative of public works of the diputacioacuten AacutengelMoreno(socialist party) Gonzaacutelez Somoano complains because in his village Gascuentildea there is an emergencysituation as one of the streets is in an extremely bad condition and it needs to be asphalted

ldquoGonzaacutelez Somoano replied that (hellip) they had undertaken asphalt works in neighboring villages of Gascuentildeabut governed by socialist mayors accusing of acting more in lsquotheir villagesrsquo than in others and reminding that allthe roads from Cantalojas (municipality whose mayor is Moreno) had been asphalted the previous yearrdquoGuadaqueacute17072009(httpwwwguadaquecomindexphpoption=com_contentampview=articleampcatid=33Aprovinciaampid=36823Ael-pp-acusa-al-psoe-de-clientelismo-y-rafael-esteban-llama-al-portavoz-

payasoampItemid=54)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 830

8

willingness to provide rewards (g) During these campaigns candidates express promises

which often have implicit threats indicating that the only way to ensure they get the public

goodsservices desired is to vote for them and not for the opposition party In order to target

their voters more accurately candidates often promise specific goods such as infrastructures 9

In this same stage voters express to candidates their willingness to vote for them in exchange

for goods and at the end of the period they cast their vote they have two possibilities either

they renege on their promise or they do not At a second stage t=2 parties take action and

implement public policies Similar to Stokesrsquo analyses there are four possible situations with

different payoffs ndashsummarized in the table below- If the voter complies and votes the

clientelist party he wins and gets the votes of its constituents (vgt0) but if the voter reneges

this party does not get the vote (v=0) and the party loses the elections In this case the party

can still decide whether to pay a reward to the voter or not On the contrary if the party wins

it has two possible options either he complies and rewards its voters (ggt0) or it reneges

(g=0)

PARTY

Reward No Reward

V O T E R

Comply αtimes(-12(xi-x1)2) +(1 α)times g(θ micro) v- g(θ micro) αtimes(-12(xi-x1)

2) v

Defect αtimes(-12(xi-x2)2) +(1 α)times g(θ micro) - g(θ micro) αtimes(-12(xi-x2)

2) 0

As explained the amount of the reward (g) is conditional on both θ and micro So we

would be in the situation ldquoComply-No Reward rdquo whenever the system is totally opaque and the

winning party can hardly be monitored (ie micro rarr0) andor the grantor cannot claim any credit

from the reward (θ=0) Moreover note that to overcome commitment problems the above

model needs to be a repeated game in which micro ne0

From the formal description we can derive four hypotheses on how clientelism could

smooth or exacerbate the impact of political alignment on transfers There may be a chance

that the treatment effect of political alignment on transfers is larger in

H1 smaller municipalities In this case monitoring the voters may be easier (we

expect micro to be higher)

9 In Spain local infrastructures (like sports centers swimming pools roads hospitals etc) are largely financedthrough capital transfers from upper-level governments

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 8: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 830

8

willingness to provide rewards (g) During these campaigns candidates express promises

which often have implicit threats indicating that the only way to ensure they get the public

goodsservices desired is to vote for them and not for the opposition party In order to target

their voters more accurately candidates often promise specific goods such as infrastructures 9

In this same stage voters express to candidates their willingness to vote for them in exchange

for goods and at the end of the period they cast their vote they have two possibilities either

they renege on their promise or they do not At a second stage t=2 parties take action and

implement public policies Similar to Stokesrsquo analyses there are four possible situations with

different payoffs ndashsummarized in the table below- If the voter complies and votes the

clientelist party he wins and gets the votes of its constituents (vgt0) but if the voter reneges

this party does not get the vote (v=0) and the party loses the elections In this case the party

can still decide whether to pay a reward to the voter or not On the contrary if the party wins

it has two possible options either he complies and rewards its voters (ggt0) or it reneges

(g=0)

PARTY

Reward No Reward

V O T E R

Comply αtimes(-12(xi-x1)2) +(1 α)times g(θ micro) v- g(θ micro) αtimes(-12(xi-x1)

2) v

Defect αtimes(-12(xi-x2)2) +(1 α)times g(θ micro) - g(θ micro) αtimes(-12(xi-x2)

2) 0

As explained the amount of the reward (g) is conditional on both θ and micro So we

would be in the situation ldquoComply-No Reward rdquo whenever the system is totally opaque and the

winning party can hardly be monitored (ie micro rarr0) andor the grantor cannot claim any credit

from the reward (θ=0) Moreover note that to overcome commitment problems the above

model needs to be a repeated game in which micro ne0

From the formal description we can derive four hypotheses on how clientelism could

smooth or exacerbate the impact of political alignment on transfers There may be a chance

that the treatment effect of political alignment on transfers is larger in

H1 smaller municipalities In this case monitoring the voters may be easier (we

expect micro to be higher)

9 In Spain local infrastructures (like sports centers swimming pools roads hospitals etc) are largely financedthrough capital transfers from upper-level governments

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 9: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 930

9

H2 municipalities aligned at all levels The credit loss is also lower in this situation (θ

is higher)

H3 in poorer municipalities This is due to the fact that for poor people the benefits

extracted from the reward outweight the disutility from voting a party whose ideologyis far from ours (αpoor 983100 αrich) Thus vote buying is easier in this case

H4 municipalities that are more reliant on intergovernmental transfers This

assumption is closely linked with the previous hypothesis In municipalities with a

large debt burden their ldquopatron-dependencyrdquo makes voters value more the reward

they would get from voting the clientelistic party than their own ideology

Besides the above-mentioned hypotheses we are also interested in analyzing two otherfactors which may affect the use of clientelism the political color of the grantor and its

relative ldquofreedomrdquo for distributing public resources

H5 in Spain left-wing grantors are thought to make more use of clientelist tactics

than right-wing parties so they may allocate larger transfers to municipalities

politically aligned with them

H6 compared to a coalition government a majority government may be more able to

divert resources to their aligned municipalities because it enjoys more freedom of

action

All these hypotheses are contrasted in section 42

3 Empirical Analysis

31 Intergovernmental transfers and elections in Spain

Intergovernmental transfers

During the last 25 years Spain has been involved in an increasing process of fiscal

and political decentralization despite being constitutionally defined as a unitary country10 The

10 The Spanish Constitution contains several provisions that promote federalism For instance according toArticle 137 ldquo[t]he State is organized territorially into municipalities provinces and any AutonomousCommunities that may be constituted All these bodies shall enjoy self-government for the management of theirrespective interestsrdquo

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 10: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1030

10

central government (the Cortes Generales) consists of the Congress of Deputies (Congreso de

los Diputados) and the Senate (Senado) At the regional level Spain is composed of

seventeen autonomous communities (AC) and two autonomous cities At the local level 85

(out of over 8000) of municipalities do not have more than 5000 inhabitants which makes it

often convenient to have upper-local administrations in order to take advantage of economies

of scale or to use inter-municipal cooperation as a tool for a better governance This upper-

local level is known as proviacutencia (province) and its government Diputacioacuten There is a total

of fifty provinces spread amongst the ACs but seven of these regions consist only of one

province

Municipalities are the administration which is closer to citizens and in Spain to

provide them with public goods and services their financing comes mainly from own sourcerevenues and specially from intergovernmental transfers11 The formulas used in the

allocation of current revenues reduce the chances of using such funds in a discretionary way

However the distribution of resources for capital spending is subject to less stringent rules

which may give rise to different sorts of tactical behavior on the governmentrsquos side This is

one of the main reasons for focusing our analysis on capital transfers This type of transfers is

meant to finance a large part of earmarked project suggested by municipalities to their upper

tiers depending on their infrastructure needs A large share of the capital grants received bylocal governments come from the regional and upper-local tiers as shown in table 1

Table 1 here

Electoral processes in Spain

In Spain general regional and municipal elections take place as a rule every four

years On the years analyzed local elections were held in 1999 and 2003 regional elections in

most of ACs were as well in 1999 and 2003 while general elections were in 2000 and 2004Electoral districts are equivalent to provinces and representatives in each province are elected

from a closed list presented by each party Although the political scene is highly polarized (by

the two main parties PSOE and PP) in general elections in parliaments we can also find the

presence of several regional parties as well as a limited number of minority parties hence

ensuring plurality Representatives of upper-local governments ( Diputaciones) are elected

11 In 2006 intergovernmental transfers accounted for 13 of current revenues In turn 23 of these transfers werecurrent grants while the remaining were capital grants

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 11: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1130

11

indirectly as an outcome of local elections This reduces accountability and may generate

undesired incentives

32 Empirical strategy

Econometric specification

To obtain estimates of the impact of partisan alignment on intergovernmental transfers

we use the regression discontinuity technique reviewed extensively in Lee and Lemieux

(2010) The intuition beneath this strategy is that in the neighborhood of a cutoff point of an

assignment variable there exists a discontinuity (a sharp change) both in the assignment

variable and in average outcomes The lsquoRDrsquo estimates are seen as a LATE (local average

treatment effect) derived from the comparison of very similar individuals whose maindifference is to be in a different side of the cutoff (ie being eligible or non-eligible for

treatment) In terms of internal validity this econometric technique is considered to be the

best of all quasi-experimental methods Another advantage of the RDD is that it requires less-

stringent identification assumptions The main disadvantage of using this technique is the loss

of external validity

In our particular study capital transfers are conditional on treatment status where

lsquotreatmentrsquo implies that the local government is aligned with the upper level (central regional

or upper-local) As shown in equation (1) below there are two potential outcomes for each

municipality depending on its treatment status

where gi represents per capita grants to local governments and Alignment i is a binary variable

to indicate treatment status

The assignment variable that determines the treatment and control groups is the

electoral margin This variable Margini is computed as follows

if the party in government at the upper level is a right-wing party the variable

margin is the difference between the share of votes obtained by right-wing minus left-

wing parties in local elections in municipality i

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 12: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1230

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 13: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1330

13

elections in 1999 affects the aggregate transfers in 2002 and 2003 while alignment from

local elections in 2003 does so to the aggregate transfers of 2006 and 2007 Information on

intergovernmental grants has been obtained from budgetary data from the Ministry of

Economics and Finance and transfers are used in per capita terms computed with population

data from the National Institute of Statistics (INE)

Alignment assignment variable and controls

To compute alignment between local governments and each upper-tier (central

regional and upper-local) we use electoral data from the Ministry of Territorial Policy and

Public Administration and the Ministry of Interior It is easy to define alignment when there

is only a single-government controlling the upper and lower-tiers In such situation a

government is politically aligned when the parties at the two levels coincide However when

governments are formed by coalitions the definition of alignment has to be broadened We

use the classification of alignment developed in Soleacute-Olleacute and Sorribas-Navarro (2008a) in

which four different types of alignment are distinguished depending on whether the

alignment is between leaders (be it single party leaders or coalition leaders) or between

coalition partners We focus our attention in two out of their four types of alignment type-a

and type-b12 As we limit to these two types we consider a municipality to be aligned if it

meets one of these conditions (a) the same party controls a single-party government at both

layers (b) a party controls a single-party government at one layer and leads a coalition at the

other layer or a party leads a coalition at both layers We have restricted the definition of

alignment to these types for two reasons On the one hand coalition leaders are expected to

attract more funds than coalition partners On the other hand this narrower definition

represents better the idea of partisan alignment as it only takes into account the party in

control of the government In the situation in which municipalities are type-c and type-d

aligned we have two options either we drop these observations (they account for about frac14 ofthe sample) or we consider them to be unaligned Considering them simply as aligned

municipalities would introduce an important error as equation (2) would no longer be valid

The results presented in section 4 are based on a sample where municipalities with type-c and

type-d alignment are disregarded but we also comment the validity of the results when they

are considered to be unaligned

12 Type-c and type-d alignment are the ones that take into account coalition partners when computing alignment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 14: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1430

14

In the first mayoral term around 41 of municipalities were aligned with the central

level 58 with the regional level and 60 with the upper-local level while in the second

term the shares were 45 58 and 63 respectively

In section 31 we argued that the assignment variable of this model which indicateswhether a municipality is aligned or unaligned is the electoral margin that upper-tier

governments obtain in municipal elections The validity of this variable is discussed in

further detail in section 5 To construct the Margin variable we use the municipal electoral

outcomes of 1999 and 2003 to compute the difference in vote shares between right and left

wing parties

In order to provide more robust estimates we include three additional control

variables population density debt burden over current revenues and property tax rates The

effect of population density on capital grants may be negative or positive because this type of

grants tend to favor small municipalities (Soleacute-Olleacute and Bosch 2005) but at the same time

they are biased towards more extensive municipalities were urban sprawl generates higher

expenditure needs The property tax rate can also have a positive or negative effect as

explained in Soleacute-Olleacute and Sorribas-Navarro (2008a) Lastly the ratio of debt burden over

current revenues is included as a proxy for fiscal effort

Sample

Out of the approximately 8000 Spanish municipalities in order to have information

on all the above variables for the two terms analyzed we limit our sample to 1828

municipalities for the analysis of regional and central transfers In the case of upper-local

transfers the sample reduces to 1607 municipalities given that we omit the autonomous

regions with only one province All the municipalities of our study have more than 1000

inhabitants due to restrictions on socio-economic data We also had to restrict our sample tothose municipalities for which we had disaggregated data on transfers We end up with a

representative sample of around 70 of Spanish municipalities of over 1000 inhabitants

Table 2 contains descriptive statistics of all the variables

Table 2 here

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 15: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1530

15

4 Results

41 Discontinuity on transfers homogeneous treatment effects

One of the main advantages of the RD technique is that it is very intuitive since it can

easily be conveyed by a picture The first graphics to present are the ones showing the

ldquojumprdquo on the assignment variable To graph a plot of the assignment variable ( Margin)

against the forcing variable ( Alignment ) the sample is divided into a number of bins of equal

size Then we plot the average value of the assignment and forcing variable in each bin and

we fit a polynomial model for a better visual inspection

If the bins are too narrow the estimates are quite imprecise and if they are too wide

estimates may be biased So to choose the right width of the bin we conduct an F-test fornested models comparing the fit of a regression with Krsquo bins with the fit of a regression with

2Krsquo bins The choice of the optimal polynomial is made upon evaluation of the Akaikersquos

information criterion Figures 1 2 and 3 show these plots for regional and upper-local

transfers and the bin width used is the optimal one indicated from the F-test

As explained in section 31 a relatively simple way to estimate equation (4) is to run a

TSLS instrumenting the forcing variable ( Alignment ) with a dummy (d ) of the assignment

variable ( Margin) We present the first stage estimates in table 3 The results (jointly with a

test for weak instruments) prove that a dummy equal to one when Margin is positive is a

strong instrument of the Alignment variable Panel A and C do not contain neither territorial

nor time effects while Panel B and D do so Panel C and D contain control variables (debt

burden property tax rate and population density) Figures in bold correspond to the estimates

of the optimal polynomial

Table 3 here

Both the graphs and the figures on table 3 confirm the existence of a sharp change on

the assignment variable at the threshold To demonstrate whether the same occurs with the

outcome variable (per capita grants) we plot the assignment variable against the outcome

variable following the same procedure than before Not observing a graphical ldquojumprdquo on

transfers rules out any possibility of having a significant treatment effect of alignment

Figures 4 5 and 6 show the causal relationship between alignment and transfers at central

regional and upper-local levels From these three pictures we would expect the treatment

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 16: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1630

16

effect of alignment on transfers to be the highest in the case of regional transfers while the

impact is clearly reduced for central and upper-local transfers

RD estimates are shown in tables 4 and 513 Table 4 presents the reduced form (ie

intention to treat) estimates while table 5 shows IV estimates The table is structured in fourpanels (A B C and D) which have the same structure than the ones in table 3 The choice of

the optimal polynomial is based on the AIC values in the case of the reduced form and on

an F-test for joint significance of the polynomial in the case of the IV regression

Table 4 and 5 here

The results on table 4 and 5 confirm the ideas extracted from the previous

graphs the amount of per capita grants received by municipalities is statistically larger foraligned municipalities (relative to unaligned ones) In turn the estimates show that the

impact of political alignment is greater in the case of regional transfers This is consistent

with the findings of Soleacute-Olleacute and Sorribas-Navarro (2008a) who show that ldquoa municipality

aligned with Central Regional and Upper-local grantors will receive additional grants of 51

12 and 57 euros per capita respectivelyrdquo

As previously noted the sample from which we extract the results in tables 4 and 5

disregards those municipalities with alignment type-c and type-d In addition we have

obtained the RD estimates from a sample where these municipalities are considered

unaligned and we find that the significance and sign of the coefficients remain unaltered 14

42 Discontinuity on transfers heterogeneous treatment effects

In section 2 we have argued why we believe clientelism could be an important source

of heterogeneity for the treatment effect of partisan alignment on transfers To test this idea

we focus on six different dimensions the size of the municipality the alignment at all levelsof government the poverty level the reliance on intergovernmental transfers the political

color of the grantor and the ldquomajorityrdquo status of the grantor government In the first case we

use population data to split the observations into two subsamples municipalities below and

13 We have performed the same analysis using as a dependent variable the logarithm of capital transfers percapita Although we do not show the results for the sake of brevity we can confirm the overall result holdpartisan alignment has a positive and significant impact on the distribution of central regional and upper-local

transfers14 Results are available upon request

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 17: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1730

17

over 5000 inhabitants15 and then we use a Wald test to establish whether the differences are

statistically significant In the second case we follow a similar procedure but we separate the

sample into municipalities that are aligned at all levels of government and municipalities who

are not That is the control group in both subsamples is the same (unaligned municipalities)

but the treatment group in the first subsample contains only municipalities aligned with all the

upper tiers (central regional and upper-local)

To study the third dimension of interest we use unemployment rates as a proxy for

poverty levels we separate the sample into municipalities with unemployment rates over and

below the average at time t of the region (province or AC) they belong to To test the fourth

hypothesis we use the average debt burden of the municipalities of our study to divide the

sample into those who are indebted over or below the average Finally to study the impact ofthe political color of the grantor the sample is split into municipalities with a left-wing grantor

and municipalities with a right-wing grantor while to study the hypothesis of the ldquomajority

statusrdquo effect we divide the sample into majority and coalition grantors

The results of our empirical analysis are in table 6 where we present the reduced form

estimates of each subsample taken into account the six dimensions we have discussed The

regressions in Panel 1 -where the dependent variable is central transfers per capita- show that

the casual effect of alignment on transfers is homogeneous amongst municipalities and that

clientelism does not have a significant impact In the case of regional transfers (Panel 2 in

table 6) we do find empirical evidence that for smaller municipalities the impact of partisan

alignment on grants is larger than for bigger municipalities and this impact is also larger when

the grantor is left-wing (compared to a right-wing grantor) Finally Panel 3 presents the

results for upper-local transfers where we also observe that being a small municipality

compared to being a big one translates into a larger treatment effect of alignment on transfers

43 Validity tests

There are two main concerns that may question the validity of our results The first

one is the possibility of manipulation of the forcing variable ( Margin) If this was the case it

would mean that political parties can influence in some way the electoral results in close

15 We could split the sample into municipalities with population above or below the mean but for the Spanishcase this separation is less relevant given that a large share of municipalities are small municipalities and wewould end up with two subsamples where one of them is not representative We have chosen the 5000inhabitants threshold because this figure is also used to allocate competencies to municipalities so city councilsabove or below this threshold may have similar competencies to fulfill

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 18: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1830

18

races In this situation we would observe an abnormal concentration of observations near the

threshold In Figure 7 we present the histogram of the assignment variable to show that this is

not the case and that the results of elections in close races in Spain cannot be precisely

controlled by agents that is they are ldquoas good as randomizedrdquo

The second concern that may affect the validity of the results is the possibility that

other covariates experience a sharp change (a discontinuity) at the cutoff To rule out this

option we estimate the 2SLS coefficients of equation (4) but we switch the dependent variable

to different covariates population density (columns 1 and 4 in table 6) unemployment rates

(columns 2 and 5 in table 6) and debt burden (columns 3 and 6 in table 6) The figures in table

6 show that in any case there exists a discontinuity for these covariates

5 Conclusions

In a system of multi-level governments the normative rationale beneath the

distribution of intergovernmental transfers may be offset by the effects of partisan ties

between grantor and recipient This has been claimed in several analyses in the political

economy field and in this paper we have presented empirical evidence of the existence of this

phenomenon in Spain Our results are obtained from applying a regression discontinuity

design in a database of around 2000 Spanish municipalities over two mayoral terms 2000-

2003 and 2004-2007 The use of this econometric technique allows us to obtain more robust

estimates than those in previous studies Furthermore we believe we improve upon existing

literature by linking partisan ties with clientelism a salient issue in Spain Adding this factor

in our analysis makes it possible to study a possible heterogeneous effect of partisan

alignment on grants

The findings presented in this paper have shown that aligned municipalities which had

close electoral races do indeed receive larger transfers (in the last two years of the electoral

term) than unaligned municipalities This discontinuity in the amount of transfers is

especially sharp in the case of regional transfers compared to central and upper-local grants

This confirms the findings of previous studies but in addition we believe we are able to

isolate better the effect of partisan ties per ser Lastly we also find evidence that the

discontinuity in grants is larger for smaller municipalities (in the case of regional and upper-

local grants) and for municipalities with a left-wing grantor at the regional level

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 19: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 1930

19

References

Albouy David 2009 ldquoPartisan Representation in Congress and the Geographic

Distribution of Federal Fundsrdquo National Bureau of Economic Research Working Paper

15224

Ansolabehere S A Gerber and J Snyder 2002 ldquoEqual votes equal money court-

ordered redistricting and public expenditures in the American statesrdquo American Political

Science Review 96 767-777

Arulampalam W Dasgupta S Dhillon A and Dutta B 2009 ldquoElectoral Goals and

Center-state Transfers A Theoretical Model and Empirical Evidence from Indiardquo Journal of

Development Economics 88 pp 103ndash119

Blakeley G (2001) Clientelism in the building of state and civil society in Spain In

Piattoni Simona ed Clientelism interests and democratic representation the European

experience in historical and comparative perspective Cambridge Studies in Comparative

Politics Cambridge UK Cambridge University Press

Brollo F and Nannicini T 2010 ldquoTying Your Enemyrsquos Hands in Close Races The

Politics of Federal Transfers in Brazilrdquo IGIER Working Paper

Brusco V Nazareno M and Stokes S 2004 Selective Incentives and Electoral

Mobilization Evidence from Argentina Chicago Center on Democracy Working Paper 26

Besley T and Burgess R 2002 ldquoThe Political Economy of Government

Responsiveness Theory and Evidence from Indiardquo Quarterly Journal of Economics 117(4)

1415-1452

Case A 2001 ldquoElection goals and income redistribution recent evidence fromAlbaniardquo European Economic Review 45 405ndash423

Cazorla J 1994 ldquoDel clientelismo tradicional al clientelismo de partido evolucioacuten y

caracteriacutesticasrdquo Barcelona ICPS Working paper 86

Carroll R A and Lyne M M 2006 Rent-seeking Pork-barreling and Clientelism

Integrating the Study of Political Market Failure Paper presented at the annual meeting of

the American Political Science Association

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 20: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2030

20

Corzo Fernaacutendez S (2002) ldquoEl clientelismo poliacutetico como intercambiordquo WP 206

Institut de Ciegravencies Poliacutetiques i Socials

Cox G McCubbins M 1986 ldquoElectoral politics as a redistributive gamerdquo The

Journal of Politics 48 370ndash389

Diaz-Cayeros A Magaloni B and Weingast B 2006 ldquoTragic Brilliance

Equilibrium Party Hegemony in MexicordquoWorking Paper Hoover Institution

Robles Egea A 2003 ldquoEl clientelismo poliacutetico y la democracia en Andaluciacuteardquo in

Anuario de derecho parlamentario

Ferreira F and Gyourko J 2009 ldquoDo Political Parties Matter Evidence from US

Citiesrdquo Quarterly Journal of Economics 124(1) 399ndash422

Grossman P 1994 ldquoA political theory of intergovernmental grantsrdquo Public Choice

78 295ndash303

Hopkin J (2001) ldquoA lsquoSouthern model of electoral mobilisation Clientelism and

electoral politics in post-Franco Spainrdquo [online] London LSE Research Online

Johansson E 2003 ldquoIntergovernmental grants as a tactical instrument empirical

evidence from Swedish municipalitiesrdquo Journal of Public Economics 87 883-915

Khemani S 2003 ldquoPartisan Politics and Intergovernmental Transfers in India

Working Paper vol 3016 Development Research Group The World Bank

Lee David S 2008 ldquoRandomized Experiments from Non-random Selection in US

House Electionsrdquo Journal of Econometrics 142(2) 675ndash97

Levitt S Snyder J 1995 ldquoPolitical parties and the distribution of federal outlaysrdquoAmerican Journal of Political Science 39 958ndash980

Lindbeck A Weibull J 1987 Balanced budget redistribution as the outcome of

political competition Public Choice 52 273ndash297

Migueis M 2009 ldquoThe Effect of Political Alignment on Transfers to Portuguese

Municipalitiesrdquo Working Paper

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 21: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2130

21

Musgrave R 1959 The Theory of Public Finance A Study in Public Economics New

York McGraw Hill

Musgrave R 1983 ldquoWho should tax where and whatrdquo In C McLure Jr ed Tax

Assignment in Federal Countries Centre for Research on Federal Financial RelationsCanberra Australian National University

Oates W 1972 Fiscal Federalism New York Harcourt-Brace-Jovanovich

Porto A and P Sanguinetti 2003 ldquoPolitical determinants of intergovernmental

grants evidence from Argentinardquo Economics and Politics 13 237-56

Schady N 2000 ldquoThe political economy of expenditures by the Peruvian Social Fund

(FONCODES) 1991-1995rdquo American Political Science Review 94 289-304

Stokes S 2005 ldquoPerverse Accountability A Formal Model of Machine Politics with

evidence from Argentinardquo American Political Science Review 99(3)315-325

Soleacute-Olleacute A Bosch N 2005 ldquoOn the relationship between authority size and the

cost of providing local services lessons for the design of intergovernmental transfers in

Spainrdquo Public Finance Review 33 343ndash384

Soleacute-Olleacute A and Sorribas-Navarro P 2008a ldquoThe effects of partisan alignment on

theallocation of intergovernmental transfers Differences-in-differences estimates for

Spainrdquo Journal of Public Economics 92 (12) 2302-2319

Soleacute-Olleacute A and Sorribas-Navarro P 2008b ldquoDoes partisan alignment affect the

electoral reward of intergovernmental transfersrdquo CESIFO Working Paper no2335

Stroumlmberg D 2001 ldquoRadios impact on public spendingrdquo Institute for International

Economic Studies Stockholm University mimeo wwwiiessuse~stromberRadiopdf

Pettersson-Lidbom Per 2008 ldquoDo Parties Matter for Economic Outcomes A

Regression-Discontinuity Approachrdquo Journal of the European Economic Association 6(5)

1037ndash56

Weingast B K Shepsle and C Johnsen 1981 ldquoThe political economy of benefits

and costs a neoclassical approach to distributive politicsrdquo Journal of Political Economy 89

642-64

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 22: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2230

22

Worthington A Dollery B 1998 ldquoThe political determination of intergovernmental

grants in Australiardquo Public Choice 94 299ndash315

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 23: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2330

23

Table 1 Capital transfers to Spanish municipalities (2002-2007)

Transfers to local governments 2002 2003 2004 2005 2006 2007

From State 13 14 11 10 10 11From Autonomous Communities (AC) 46 47 48 49 52 55From Diputaciones (Upper-local) 20 20 19 19 18 18

Source Ministry of Economics and Finance

Table 2 Descriptive statistics and data sources

Variable Description Mean

(SD) Source

Central grants

Per capita capital grants assigned fromcentral government (item 72 of the revenuebudget)

3305(7826)

Ministry of Economics andFinance

Regional grants

Per capita capital grants assigned fromregional government (item 75 of therevenue budget)

15994(19654)

Upper-local grants

Per capita capital grants assigned fromupper- local government (item 761 of therevenue budget)

7607(9482)

Debt burden

Debt service (capital item 9 of the spendingbudget +interest item 3) as a share ofcurrent revenues

0062(008)

Margin

Vote share of right-wing parties if upper-tieris right-wing vote share of left-wing parties

if upper-tier is left-wing 0 otherwise

009(035)

Ministry of TerritorialPolicy and Public

Administration andMinistry of Interior

Population density Population Km2 41476(138892)

National Institute ofStatistics (INE)

Property tax rate Nominal property tax rate (IBI) 059(016)

Cadastre (Catastro)

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 24: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2430

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 25: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2530

25

Table 4 Electoral margin and intergovernmental transfers (reduced form)

(1) (2) (3) (4) (5) (8) (9) (10) (11) (12)Polynomiald

0 1 2 3 4 0 1 2 3 4

Panel A Panel B

Central

d(C) 6630 3882 4683 1284 1283 3974 7030 7427 1140 1145

(2125) (3591) (3585) (4392) (4402) (2070) (3379) (3371) (4108) (4129) Regional

d(R) 6034 4755 6160 5112 4968 5780 4876 5717 4845 4799(6318) (1067) (1107) (1311) (1335) (6234) (9986) (1028) (1251) (1270)

Upper-local

d(UL) 1280 6434 1721 1565 1923 2051 1139 1923 1779 1795(3561) (5584) (5793) (6635) (6847) (3744) (5189) (5305) (6172) (6341)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls No No No No No No No No No NoPanel C Panel D

Central

d(C) 6665 4369 8510 1084 1067 4653 6964 8679 1046 1014(2177) (3748) (3666) (4581) (4585) (2159) (3494) (3473) (4263) (4257)

Regional

d(R) 5831 4433 5724 5220 5195 5441 4745 5591 5273 5302(6018) (1025) (1040) (1275) (1284) (5896) (9826) (9965) (1233) (1240)

Upper-local

d(UL) 1120 4322 1472 1299 1661 2026 1164 1818 1664 1721(3546) (5540) (5709) (6605) (6787) (3656) (5080) (5211) (6076) (6251)

Territoriald

No No No No No Yes Yes Yes Yes Yes

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports reduced form estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) D() is an indicator for electoral margin below the eligibility threshold (5) no interactionsof the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with no controls) and3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (with controlvariables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

Table 5 Intergovernmental transfers and partisan alignment (2SLS estimates)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)Polynomial d 0 1 2 3 4 0 1 2 3 4

Panel A Panel BCentral

Alignment(C) 8831 5929 7157 2334 2336 5383 1081 1142 2111 2122(2829) (5480) (5475) (8028) (8069) (2795) (5188) (5177) (7644) (7697)

Regional

Alignment (R) 7843 6996 9687 9329 9216 7472 7440 9102 9137 9106(8212) (1561) (1727) (2376) (2462) (7990) (1504) (1610) (2324) (2376)

Upper-local

Alignment

(UL)

1671 9933 2825 2926 3663 2627 1838 3206 3452 3519

(4650) (8611) (9554) (1241) (1308) (4794) (8352) (8865) (1200) (1248)Territorial d No No No No No Yes Yes Yes Yes YesControls No No No No No No No No No No

Panel C Panel D

Central

Alignment (C) 8705 6424 1257 1915 1877 6119 1044 1304 1888 1822(2840) (5506) (5416) (8112) (8091) (2829) (5229) (5214) (7718) (7667)

Regional

Alignment (R) 7590 6533 9043 9626 9738 7011 7317 8986 1006 1018(7799) (1495) (1622) (2322) (2376) (7539) (1489) (1574) (2311) (2340)

Upper-local

Alignment

(UL)

1461 6660 2411 2420 3374 2597 1880 3026 3223 3371

(4619) (8520) (9359) (1227) (1600) (4677) (8169) (8675) (1175) (1226)Territorial d No No No No No Yes Yes Yes Yes YesControls Yes Yes Yes Yes Yes Yes Yes Yes Yes YesNote (1)The table reports 2SLS estimates of the effect of partisan alignment on intergovernmental transfers) (2) robust standard errors in parentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 if the municipality is aligned with the upper tier (5)no interactions of the polynomials with the eligibility indicator (6) regressions for central and regional transfers are on a sample of 3657 observations (with nocontrols) and 3477 (with control variables) while regressions for upper-local transfers are on a sample of 3215 observations (with no controls) and 3208 (withcontrol variables) (7) panel A and C do not contain neither time nor territorial effects while panel B and D contain both

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 26: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2630

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 27: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2730

27

Table 7 Discontinuity in other covariates (2SLS estimates)

(1) (2) (3) (5) (6) (7)

Panel A Panel BCentral

Alignment(C) -2312 -935e-05 000101 -1636 -0000987 -000135(1589) (000187) (000975) (1622) (000172) (000976)

Regional

Alignment (R) -1876 -000391 -000395 -1873 -000257 0000523(1746) (000084) (000900) (1745) (000164) (000883)

Upper-local

Alignment (UL) -4324 -000569 -00120 -2298 -000204 -000907(1162) (000017) (00104) (2019) (000170) (00104)

Territorial d No No No Yes Yes YesControls No No No No No NoNote (1)The table reports 2SLS estimates of the effect of partisan alignment on several covariates (2) robust standard errors inparentheses plt001 plt005 plt01 (3) SE clustered by municipality (4) Alignment() is a binary variable equal to 1 ifthe municipality is aligned with the upper tier (5) no interactions of the polynomials with the eligibility indicator (6) dependentvariable for columns 1 and 5 population density for columns 2 and 6 unemployment level for columns 3 and 7 debt burden forcolumns 4 and 8 level of studies

Figure 1

0

2

4

6

8

1

A l i g n m e n

t C e n

t r a l - L o

c a

l

-5 0 5

Electoral marginNote bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 28: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2830

28

Figure 2

Figure 3

0

2

4

6

8

1

A l i g n m e n

t R e g

i o n a

l - L o c a

l

-5 0 5Electoral margin

Note bin window=5 percent

0

2

4

6

8

1

A l i g n

m e n t U p p e r L o c a l - L o c a l

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 29: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 2930

29

Figure 4

Figure 5

0

2 0

4 0

6 0

8 0

C e n t r a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

5 0

1

0 0

1 5 0

2 0 0

R e g

i o n a

l t r a n s

f e r s

-5 0 5Electoral margin

Note bin window=5 percent

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin

Page 30: (2011), Partisan alignment, clientelism and intergovernmental transfers - evidence from Spanish municipalities.pdf

8122019 (2011) Partisan alignment clientelism and intergovernmental transfers - evidence from Spanish municipalitiespdf

httpslidepdfcomreaderfull2011-partisan-alignment-clientelism-and-intergovernmental-transfers-evidence 3030

30

Figure 6

Figure 7

4 0

6 0

8 0

1 0 0

1 2 0

1 4 0

U p p e r - l o c a l t r a n s f e r s

-5 0 5Electoral margin

Note bin window=5 percent

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Central govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin State govt party

0

5

1

1 5

F r e q u e n c y

-1 -5 0 5 1Margin Upper-local govt party

Note bin=5 percent

Figure Histograms of the margin