fiscal federalism jonathan rodden stanford university august 8, 2011
TRANSCRIPT
Part 1:Broad Overview
• Intellectual history• From welfare economics and public choice to
political economy
• Stylized facts and trends • Partial decentralization• Incomplete contracts
• An example: The study of intergovernmental grants
Part 2: Macroeconomic management
• State/local budgets and the business cycle• Pro-cyclical fiscal flows and borrowing
• Fiscal discipline in multi-layered systems• The end of market discipline?
Welfare economics
• Coherent logic connecting Montesquieu, Rousseau, Tocqueville, Madison, Musgrave, Oates:• To achieve simultaneously the advantages of
large and small governmental units by solving the “assignment problem.”
• Oates: “The provision of public services should be located at the lowest level of government encompassing, in a spatial sense, the relevant benefits and costs.”
Welfare economics, cont.
• Assume that political leaders are benevolent despots who maximize the welfare of their constituents.
• Presumption in favor of decentralization because of:• stronger incentives• better information about preferences• above all, greater homogeneity of preferences
at lower levels of government
Competition and sorting
• Tiebout (1956): Key advantage of decentralization is the market analogy. • Citizen land-owners sort into communities that
offer desired levels of taxes and bundles of goods.
• Provides a powerful preference-revelation mechanism beyond voting and lobbying.
Competition as a restraint on government
• Leviathan (Hayek 1939, Brennan and Buchanan 1980)• Tax competition prevents revenue-hungry
politicians and bureaucrats from consuming too much.
• Persson and Tabellini (2000), Weingast (1995)• Decentralization with capital mobility allows
government to commit not to over-tax capital or over-regulate the economy.
Broad consensus circa 1990:
• Based on theory literature, virtual consensus about potential benefits of decentralization, especially in developing countries in late 1980s
What went wrong?
• The obvious things:• Corruption, clientelism, elite capture• Accountability problems• Challenges for safety nets and poverty
reduction• Macroeconomic management:
• Specifically, soft budget constraints and bailouts
What was the theory literature missing?
• Decentralization in practice rarely resembles the type of decentralization imagined in the theory literature.
• “Partial Decentralization” • Grants and shared taxes rather than
autonomous local taxation• Muddy division of authority• Politicized resource distribution
0
0.2
0.4
0.6
0.8
1
1965 1975 1985 1995
Percent of countries with elected subnational governments (43 countries)
local
regional
Correlates of expenditure decentralization
• Panizza 1999, Garrett and Rodden 2005:• Country size• GDP per capita• Democracy • Federal constitution• Ethno-linguistic heterogeneity?
01
23
Den
sity
0 .2 .4 .6 .8Share of total
State-local expenditure/total, GFSState-local own-source revenue/total, country sources
Kernel density of expenditure and tax decentralization in 40 countries, 1990s
01
02
03
0D
ensi
ty
0 .1 .2 .3State-local autonomous taxation/total revenue
Switzer-land USA Canada
Histogram, Subnational tax autonomy in OECD countries (full rate and base autonomy), 1990s
Source: OECD
Federalism vs. decentralization
• Federalism has roots in a bargain or contract• Coming together vs. holding together
• To be credible, the contract usually requires institutional protections:• Unit-based rather than population-based
representation• Supermajority requirements• Courts with judicial review• Explicit delimitation of powers & responsibilities,
residual powers
0
0.2
0.4
0.6
1965 1975 1985 1995
Percent of countries with decentralized primary education policy (43 countries)
Shared between center and subnational govts.
Shared between 2 or more subnational tiers
One subnational tier alone
Source: Henderson (2000)
Shared between center and subnational govts.
Shared between 2 or more subnational tiers
One subnational tier alone
Source: Henderson (2000)
0
0.2
0.4
0.6
1965 1975 1985 1995
Percent of countries with decentralized infastructure policy (43 countries)
Pathologies of partial fiscal decentralization
• Limited accountability• Local governments direct resources to clients
and blame higher-level governments for poor service provision.
• Offloading and unfunded mandates• Stringent conditions for grants• Incentives for local governments to hide
information and dissemble• Politicized transfers
Rethinking fiscal federalism in the last decade
• Motivations of politicians• Electoral and other political motivations
replace benevolent despots and Leviathans.
• Focus on institutions of representation• E.g., the nature of legislative bargaining:
Persson and Tabellini (1996), Inman and Rubinfeld (1997), Dixit and Londregan (1998), Besley and Coate (2003), Lockwood (2002).
Rethinking federalism, cont.
• Sharper focus on “fiscal interests”• Taxes and fees vs. grants• Types of grants, formulas
• Incomplete contracts• The ultimate locus of authority is often unclear
and contested.
• Principal-agent relationship• Focus on crafting better incentives for
subnational governments
Re-centralization?
• Central governments are seeking out new ways of structuring the principal-agent relationship• Replacing discretion with rules• Audits• Enhanced central monitoring and data
collection
• But challenges remain:• Example: difficulty of data collection in
decentralized environments
Motivation
• The trend toward fiscal decentralization is funded primarily by a combination of formulaic and discretionary transfers.
• Grants and fiscal flows shape incentives of regional governments and central legislators.
• By what logic are they distributed?
Studies of intergovernmental grants
• First generation: Welfare economics and fiscal flows
• Second generation: Political economy of fiscal flows• Partisan dictators• Legislative bargaining
• Fiscal flows and inter-regional redistribution: When and where are grants progressive?
• Representation and redistribution
Welfare economics
• Central government is a benevolent dictator
• Uses inter-regional fiscal flows to:• Capitalize on economies of scale in taxation
• Internalize externalities• Facilitate inter-governmental competition• Stabilize asymmetric shocks
Partisan dictators• Cox and McCubbins (1986):
• Core support• Key assumptions: Risk-averse incumbent
• Dixit and Londregan (1996):• “Swing voters”
• These theories are not necessarily about geography or districts, but the application is natural
• Partisan alignment
Empirical literature
• Scattered evidence for all these propositions• Usually an empirical focus on one relatively small,
discretionary part of the budget (e.g. environmental grants in Sweden, infrastructure grants in Spain).
• “Core vs. swing” debate unresolved: Basic story is that incumbents favor some combination of marginal and core districts, direct resources away from the opponent’s core support districts.
• Strong support for the partisan alignment hypothesis• Formulaic transfers are not immune
• Challenges:• Measuring ideological indifference• Endogeneity: Do fiscal flows actually buy votes?
Big questions left unanswered:
• What happens when we drop fixed effects and examine long-term cross-section variation?
• Are fiscal flows progressive?• When and where?• Implications for European idea of a “transfer union.”
Empirical analysis of fiscal flows
• MacDougall Report (1977)
• Renewed interest due to optimal currency area literature, e.g. Sala-i-Martín and Sachs (1992); Bayoumi and Masson (1995)
• Broadest comparative work builds on Bayoumi and Masson (1995): Espasa (2001); Barberán, Bosch, Castells, Espasa (2000); Bosch, Espasa, Sorribas (2002)
Income elasticity of fiscal flows
im
iii
m
i eY
Yba
B
B
ln10ln
income capitaper real Y
balance fiscal capitaper real B
regions all of average torefers
region torefers
m
i
-6-4
-20
2In
com
e e
last
icity
of g
ran
ts
ARG AUS BRA CAN DEU ESP EU IND USACountry
Income elasticity of grants in 9 federations, 1990-2005
To sum up:
• Considerable redistribution through intergovernmental grants in Canada, Spain, Germany, and Australia
• Very little redistribution in Argentina, Brazil, Mexico, Switzerland, the United States, and the EU.
• Why?
The representation of regions
• Some state receive far more representation per capita than others.
• There are good reasons to believe that over-represented states will do well in the game of legislative bargaining
Another possible explanation:
• Classic political economy theory about the income distribution:• Does the skew in the inter-regional income
distribution predict redistribution? • If the policy is set by the median state, we
should expect to see large redistribution when median state is poor relative to the mean.
0.5
11
.52
2.5
Den
sity
0 1 2 3 4Share of national average
GDP per capita Expenditures per capita
Brazil, 1990-2001
0.5
11
.52
2.5
Den
sity
0 1 2 3 4Share of national average
GDP per capita Expenditures per capita
Argentina, 1990-2001
0.5
11
.52
2.5
Den
sity
0 1 2 3 4Share of national average
GDP per capita Expenditures per capita
USA, 1990-1997
01
23
45
Den
sity
0 1 2 3 4Share of national average
GDP per capitaExpenditures per capita
Australia, 1990-2001
Note: NT dropped
01
23
45
Den
sity
0 1 2 3 4Share of national average
GDP per capitaExpenditures per capita
Canada, 1990-1997
01
23
45
Den
sity
0 1 2 3 4Share of national average
GDP per capitaExpenditures per capita
Germany, 1990
Note: city-states dropped
But what is the politically relevant income distribution?
• Perhaps in the parliamentary federations without much inter-provincial bargaining, the relevant distribution is the (highly skewed) inter-personal one, and high levels of inter-personal and inter-regional redistribution go hand in hand.
• But an interesting thing happens when the geography is divided into winner-take-all districts or states…
0.0
000
2.0
000
4.0
000
6D
ensi
ty
20000 40000 60000 80000Households: Median household income in 1999
U.S. House District MediansU.S. State Medians
Distribution of Median Income in U.S. Congressional Districts and U.S. States
Median/Mean Ratios:Individuals: .74Cong. Dist.: .95States: .98
A different perspective on unit-based vs. population-based representation
• Perhaps this helps explain why federations, and countries with single-member districts, demonstrate lower levels of redistribution
• The politically relevant median voter (the median income in the median state) is not very poor relative to the mean
A related observation:
• All of the redistributive federations are parliamentary systems with strong and disciplined political parties.
• The non-redistributive federations are presidential systems with weaker parties and region-based coalition building in the legislature, especially the upper chamber.
• A similar story emerges from Persson and Tabellini (1996), who show that inter-regional bargaining leads to lower levels of risk-sharing than majority rule
Summing up:
• The “first generation” literature taught lessons about the optimal distribution of authority that are still relevant
• But it ignored questions related to institutional design and political economy
• After addressing these questions, we now know more about the incentives generated by fiscal and political institutions for voters, creditors, elected officials, and bureaucrats.
• This helps provide a clearer sense of the conditions under which decentralization might facilitate or undermine service delivery and macroeconomic management.
Summing up (cont.):
• Much literature now focuses on strategies to minimize the “dangers” of decentralization
• Not much focus on the impact of decentralization per se
• Instead, focus on incentives created by the intergovernmental framework
• Transition from observational to experimental empirical research
Looking ahead
• Macroeconomic management in a world of:• Partial decentralization• Incomplete contracts• Politicized transfers
Overview
• Fiscal federalism and the business cycle
• Fiscal adjustment in a multi-layered system: the moral hazard problem
• Paths to fiscal discipline• Hierarchy• Markets
• Can market discipline survive?
Some important questions:
• How do local budgets respond to the business cycle?
• With what implications for macroeconomic management?
• If central government attempts to generate fiscal stimulus during recession, to what extent do credit-constrained subnational governments undermine this?
Provincial-level time series data (real local currency units per capita):
• Variables:• Total revenue
• Grants (discretionary and formulaic)• Total taxes and fees
• Total expenditures• Deficit • Provincial GDP
• Federations: • Canada, USA, Germany, Australia, Argentina,
Brazil, India
What should we expect?
• Revenues:• Highly pro-cyclical taxes• Grants?
• First generation fiscal federalism literature seems to imply counter-cyclical grants
• Literature on optimal currency unions• But second generation political economy
perspective leads to skepticism
What should we expect?
• Expenditures and borrowing• Barriers to borrowing (and saving):
• USA: Balanced budget rules and revenue restrictions
• Canada is at the opposite extreme: No centrally- or self-imposed restrictions
• Centrally-imposed and cooperative restrictions in each of the other federations• But many loopholes (e.g. German “golden rules,”
Brazilian Senate oversight)
• Voracity effect, credit crunch problem
Summing up:
• Provincial fiscal behavior is highly pro-cyclical everywhere
• Grants do not smooth symmetric shocks in federations (except perhaps Australia)
• Some modest smoothing through saving and borrowing in OECD federations
• But ultimately, expenditures are generally pro-cyclical, which complicates efforts at stimulus. • See Aizenman & Pasricha (2011).
Dynamic Bailout Game
C
hanc
e
p
1-p
Adjust
Unsustainable Borrowing
Adjust
Unsustainable Borrowing
Bailout
No Bailout
Bailout
No Bailout
No Bailout
No Bailout
Bailout
Bailout
EA
EA
EB
EB
LB
D
LB
D
SNG
SNG
CG
CG
C
G
SNG
CG
CG
LA
LA
Adjust
Debt Crisis
Adjust
Debt Crisis
SNG
Usng(EB) = 1 >Usng(LB)> Usng(EA)> Usng (LA)>Usng(D) = 0.
Ucgr(EA) = 1 > Ucgr(LA)> Ucgr(D)> Ucgr(EB)> Ucgr(LB) = 0. Ucgi(EA) = 1 > Ucgi(LA)> Ucgi(EB)> Ucgi(LB)> Ucgi(D) = 0.
Dynamic bailout game
• First, consider equilibria under perfect information• If p=1 (SNG believes with certainty that center is
resolute), adjust immediately. SNG is a sovereign. Market discipline.
• If p=0, SNG avoids adjustment and immediate bailout ensues.
• Under incomplete information, SNGs are semi-sovereigns• No separating equilibrium in pure strategies. SNG
cannot ascertain center’s type after first round.• This can lead to “resolve-testing” equilibria.
What drives perceptions of the center’s resolve?
• Basic powers and obligations of the center
• Externalities and the structure of jurisdictions
• Identity of debt holders
• Legislative representation of solvent and insolvent states
• Court decisions
• Revenue sources and autonomy
When the center cannot commit:
• Fully credible commitment is rare
• Unitary systems: Lack of commitment is common knowledge, and center confronts moral hazard problem through hierarchy:• Strict debt limits, administrative controls• Centralized credit allocation
• Empirical regularity: transfer-dependence is associated with borrowing restrictions (e.g. Von Hagen and Eichengreen 1995)
Commitment problems in federations
• Recall that federations emerged from historical bargains with institutional legacies that make hierarchical control difficult:• Brazilian Senate• EU Council of Ministers
• Dysfunctional federalism: Center can neither commit nor regulate• Argentina and Brazil early 1990s• European Union today
What went wrong in Europe?
• Half-hearted attempts at markets (no-bailout clause) and hierarchy (excessive deficit procedure).• The latter undermined the former
• The identity of debt holders, externalities
• Banking sector
• Uncertainty• Both about bailouts and defaults
51
01
51
0-y
ear
bo
nd
yie
ld
February 7 1992 January 1 1999
Belgium GermanyItaly France
Netherlands PortugalSpain SwedenDenmark UK
European Monetary Union and the Convergence of bond yields
Debt crisis and divergence in 2010
24
68
10
12
10-y
ear
bo
nd
yie
ld
April 2009 May 10 2010 Nov 12 2010
Belgium GermanyItaly FranceGreece Netherlands
Portugal SpainSweden UKIreland
The way forward in Europe
• A moment for centralization?
• Can market discipline function again?• Toward an orderly default procedure• European bankruptcy?
What about the United States?
• On one hand, some market analysts believe default is imminent.
• On the other hand, Roubini and Buffet tell us that in the wake of Bear Stearns and GM, federal government already provides an implicit guarantee.
Debt/GDP ratios for European countries and U.S. states
AUT
BEL
BUL
CYP
CZR
DEN
DEU
ESP
EST
FIN
FRA
GRE
HUN
IRE
ITA
LTA
LUX
LVA
MAL
NET
POL
POR
ROM
SLJSVK
SWE
UK
AKAL
AR
AZCACO
CT
DCDE
FL
GA
HI
ID
IL IN
IO
KS
KY
LA
MA
MDME
MIMNMO
MS
MT
NCND
NENHNJNMNV
NY
OHOK
OR
PARI
SC
SDTN
TXUTVA
VTWA
WIWV
WY
05
01
00
150
Deb
t/G
DP
200
9
Europe USA
But this is deceptive
• Unfunded pension liabilities
• Implicit responsibility for municipal debt
• Insolvency is probably not imminent, but if conditions deteriorate, will states be allowed to default? Will the federal government provide a bailout?
Market discipline in U.S. federalism
• Begins with aftermath of 1840s debt crisis
• Throughout 20th century, rapid response to negative revenue shocks, especially in states with most stringent balanced budget requirements (e.g. Poterba 1995).
• Bond yields and ratings quickly very responsive to changes in debt/GSP ratio and other indicators
• Default extremely rare
• No federal debt assumptions
But much has changed
• Beginning with the New Deal, creeping centralization.
• States are increasingly used as agents of the federal government
0.0
5.1
.15
.2.2
5F
ed
era
l gra
nts
as
sh
are
of cu
rre
nt e
xpe
nd
iture
s
1920 1940 1960 1980 2000 2020year
Source: BEA
Federal grants as share of state-local current expenditures
Response to recent recessions:
• Inefficient and painful expenditure cuts
• Requests for implicit bailouts:• Medicaid assistance• Infrastructure stimulus• Build America bonds
• These are delayed, ad hoc, and politicized.
• They send the wrong signals to market actors.
• But do they spell the end of market discipline?
Can market discipline survive?
• There are good reasons for optimism. • Identity of bond-holders• Representation of (potentially) insolvent states
• The most important question is not whether bailouts are possible, but whether states and creditors are sufficiently uncertain.
• States and their creditors are not behaving as if bailouts are imminent.
• States are making better progress than the federal government
Bolstering market discipline
• Reduce unfunded mandates
• Avoid policies that make states responsible for municipalities
• Embed automatic stabilizers into transfer system
• Keep state and federal obligations as separate as possible
• Orderly default procedure
Summary
• Fiscal federalism creates serious challenges for macroeconomic management, especially in the wake of fiscal crisis
• Subnational governments are font-line service providers, often responsible for providing unemployment insurance, health services, safety net, not to mention education, police, fire protection, infrastructure.
• Own-source revenues are extremely pro-cyclical, and grants are not much better. Most subnational governments are highly credit-constrained, and cannot easily borrow to smooth shocks.
Summary (cont.)
• Subnational governments are thus largely reliant on the central government for stabilization (as prescribed in the first-generation normative literature).
• But central governments, and the intergovernmental fiscal framework, are often not up to the task.
• This is an important area for reform
• Even in unitary systems, there is an ongoing struggle to improve incentives associated with partial decentralization
• But the largest challenges appear to be in federations and quasi-federations, where institutions, along with ethnic and regional tensions, undermine both hierarchical and market-based forms of fiscal discipline.
Implications for the IMF
• Conditions, targets, monitoring must be sensitive to activities and obligations of subnational governments
• It is important to assess the basic incentives created by the intergovernmental framework. Things are often not as they appear on paper, and it is crucial to understand the political incentive structure.
• Need for further collaborative research