feature interaction: a critical review and considered forecast

21
Alternative a ordability measures for evaluating public utility markets reforms Raaele Miniaci Carlo Scarpa Paola Valbonesi § Preliminary Version June 25, 2007 Abstract In this paper we rstly present a survey of the dierent standard mea- sures of aordability proposed in the literature to study the eects of price reforms in the public utility markets on household budgets. We then sug- gest a new measure that overcomes the main theoretical and practical diculties faced by the standard practice. Finally, we provide the ap- plication of this new measure to evaluate the distributional eects from liberalization reform in three Italian public utility markets (electricity, gas and water sectors) and to compare the results obtained with those from standard aordability measures. Keywords: Aordability, Public Utilities, Regulation, Gas, Water, Electricity JEL: D12, D63, I32 1 We appreciate comments received at the University of Milano and Brescia. We gratefully acknowledge nancial support from the MIUR (Con 2004 - No. 2004134814_005 and Con 2006 - No. 2006130472_003 grants). Responsibility for errors is our own. Dipartimento di Scienze Economiche, Università di Brescia - [email protected] Dipartimento di Scienze Economiche, Università di Brescia - [email protected] § Dipartimento di Scienze Economiche, Università di Padova - [email protected] 1

Upload: leicester

Post on 21-Nov-2023

0 views

Category:

Documents


0 download

TRANSCRIPT

Alternative affordability measuresfor evaluating public utility markets reforms∗

Raffaele Miniaci† Carlo Scarpa‡ Paola Valbonesi§

Preliminary Version

June 25, 2007

Abstract

In this paper we firstly present a survey of the different standard mea-sures of affordability proposed in the literature to study the effects of pricereforms in the public utility markets on household budgets. We then sug-gest a new measure that overcomes the main theoretical and practicaldifficulties faced by the standard practice. Finally, we provide the ap-plication of this new measure to evaluate the distributional effects fromliberalization reform in three Italian public utility markets (electricity, gasand water sectors) and to compare the results obtained with those fromstandard affordability measures.

Keywords: Affordability, Public Utilities, Regulation, Gas,Water, Electricity

JEL: D12, D63, I32

1

∗We appreciate comments received at the University of Milano and Brescia. We gratefullyacknowledge financial support from the MIUR (Cofin 2004 - No. 2004134814_005 and Cofin2006 - No. 2006130472_003 grants). Responsibility for errors is our own.

†Dipartimento di Scienze Economiche, Università di Brescia - [email protected]‡Dipartimento di Scienze Economiche, Università di Brescia - [email protected]§Dipartimento di Scienze Economiche, Università di Padova - [email protected]

1

IntroductionDifferent waves of liberalization reforms in public utility markets (i.e. gas, elec-tricity and water services) have occurred since the beginning of the last decadein many European countries, extending competition to previous monopolisticsegments to promote efficiency. Competition and better regulation - changingthe level of prices - affect different consumers differently as demand for publicutilities increases with income, but at a decreasing rate. International agencies,national governments and consumers unions are very interested in using simpleindexes, able to give a proper picture of how these reforms affect the welfare ofthe consumers, and particularly the most vulnerable part of the population.This concern specifically arises from two sources. The first one is the ob-

servation that the short run demand function for these essential goods is char-acterized both by low price and income elasticities, as that there are no closesubstitutes for them. Thus, any reform that alters the utility’s tariff schemeaffects the affordability of the service itself, and its effect can be particularlyrelevant for low-income consumers1 and vulnerable households.2

The second source of concern comes from the observation that the consump-tion of these utilities is necessary for individual well-being, as well as it producespositive externalities on the whole society (Barr, 1992). It is beyond questionthat continuous and adequate supply of water, sewerage and electricity to everyhousehold are essential for healthy living in our society and - in the meantime -affect public health and human productivity. Where gas is used for heating orcooking, it acquires a similar status.3 In this perspective, some contributionsto public economic literature include these utilities in the class of merit goods(Sandmo, 1983; Besley, 1988). It is well known that behind the definition ofmerit goods lies a social planner with a paternalistic approach, that is, one whobelieves that a consumer’s preferences for the utilities considered are a “faulty”representation of his well-being. Thus, the possibility that the consumer choosesa wrong bundle entitles the social planner to affect his choice, i.e. monitoringthe affordability in public utility and, consequently, targeting policies aimed atreducing it.The effects of liberalization reforms are routinely verified through affordabil-

ity measures which usually convey the information on the consumer’s resourcesthat are necessary to buy the utility service before and after the reform. Specifi-cally, the currently adopted affordability measures approach one of the followingtwo facets: a) how many households are able to afford a minimum quantity ofthe utility considered; b) how many households spend more than a “reasonable

1The distributional impact of reforms is assessed by Wolak (1996) in US telecomunication,by Waddams Price and Hancock (1998) and Waddams Price (2005) in UK energy markets;by Gomez-Lobo (1996) in the UK gas market; by Miniaci, Scarpa and Valbonesi (2006) inItalian energy and water markets.

2 In fuel poverty literature, a household is defined vulnerable as it contains children, elderlypeople, disabled or long-term sick.

3This observation leaves room to the debate on the desiderability of in-kind transfers forthe consumption of such goods (see for a survey: Balestrino, 1999) and to the literature onpublic provision of private goods (Bergstorm, Blume and Varian, 1986).

2

amount of money” for the selected utility. No single approach tells the wholestory and both the above are characterized by pros and cons.The former notion - the more widely used in the empirical analyses on af-

fordability of public utility - often takes the level of household consumptionbefore the liberalization reform as the one that consumers should be able toafford even after the reform.4 In many cases, this choice is far from sensible:consider for instance Eastern European countries where access to public utilitieswas almost free of charge, and therefore an inefficient level of consumption wassubsidized. Following this approach, some households classified as “with afford-ability problems” might actually love over-heating their homes, re-fill every daytheir tubs or refrigerating “too much” their houses. On the opposite side, someof the households classified as “not suffering any affordability problem” may infact be under-heating their homes due to their binding budget constraints.In some (few) cases, the quantity with respect to which affordability is de-

fined, is identified in a more normative and compelling way. For instance, theBritish government definition of heating affordability is based on the amountof fuel necessary to reach a fixed standard of heating suggested by the WorldHealth Organization. The standard depends on the family type (in particular,on the presence of pensioners and children) and the heating efficiency of thehouse is taken as given in the short run. By doing so, it is recognized thatneeds vary across population, that the utility services are inputs in a home pro-duction function with given technology - and economies of scale - and that theaffordability issue can be solved (in the medium term) also by improving on thetechnological efficiency of the households.5

The latter approach in affordability measurement - focussing on expenditurerather than on quantities - defines a household as having an affordability prob-lem if the ratio between its actual expenditure for utilities and its income (ortotal expenditure) is above a given threshold6. This threshold is identified withcriteria of "reasonability", and it is often set as the average (or median) budgetshare devoted to the utilities by the poor households. But this criterion doesnot help to tell the full story and to avoid the mis-classification issue discussedabove. For instance, high and medium-level income consumers with high con-sumption of utilities (i.e. a large house to warm up or the swimming pool tofill up) could easily be identified as having affordability problems. Thus, thisapproach always raises the question on what expenditure ratio is adequate andacceptable, that is, on how the critical threshold should be fixed.This calls for attention on the desirable properties of an affordability index,

to be adopted in evaluating the effects of reforms on the consumption of theseessential services. Even if the debate on the properties of different welfareand poverty measures appears well developed in the poverty literature and -specifically - in housing economics, this is not the case for affordability measuresadopted for public utility sectors. The present paper attempts to fill this gap.In what follows, we first discuss different affordability indexes usually adopted4See Sefton (2002, 2005).5 See about: DEFRA (2001) and (2004).6Frankahauser and Tepic (2006, forthcoming).

3

in the applied literature to study how they capture the effects of price reforms inpublic utility markets. We then refer to the ongoing debate in the housing eco-nomics literature to suggest an affordability measure for utility sectors, whichovercomes the main theoretical and practical difficulties faced by the standardpractice. In particular, we propose to consider the affordability problem lookingat the residual income, namely the income left to the household once it has pur-chased a standard quantity of utility services. If this residual income falls abovethe absolute poverty line (net of public utilities), then we claim the householdhas no affordability problems, no matter what its actual utility budget share is.The residual income approach to the affordability of public utilities has dif-

ferent and relevant advantages. First of all, it explicitly requires to define aconsistent standard level of consumption - whose identification might be a re-sult of an (informed) interaction among different agents participating to the re-form of the utility markets (i.e. providers, government, consumers unions, ...).Moreover, it can rely on the results obtained by the well established literatureon poverty indexes. Finally, it does not require the estimation of the demandfunctions for utility services, and it better addresses the source of household’saffordability.We finally provide the application of this new measure to evaluate the dis-

tributional effects from liberalization reforms in Italian public utility electricity,gas and water markets an to compare the results obtained with this index, tothose which would emerge from standard affordability measures. We find thatin Italy affordability problems do not seems to have been made more seriousby the recent reforms, but, depending on the measure adopted, very differentaffordability issues are highlighted. Specifically, the residual income approach toaffordability better helps to identify those households, which are more likely toface affordability problems due to inefficiency of their technological endowment(type of house, electrical appliances, etc.).The paper is organized as follows. In Section 2, we review the standard

affordability measures adopted in applied literature; in Section 3, we discusshow the residual income can be used to describe the affordability issue in thepublic utility sectors. In Section 4, we provide a preliminary application of thisapproach to the case of Italy; finally, in Section 5, we discuss the results fromthe case study along with the comparison of the residual income index for withother affordability measures.

2 Standard affordability measures for public util-ity markets

In this Section we briefly present the affordability measures commonly adoptedto evaluate the effects of liberalization reforms in public utility markets: we startwith the basic headcount index, and we then move to illustrate the poverty gapindex, defined both in monetary terms and in terms of budget shares.Define xuh as the actual observed expenditure for utilities (u) of household

4

h, xch its expenditure for the remaining other goods, and xh = xuh+xch the totalexpenditure. The standard practice is to define an household h to suffer anaffordability problem if its expenditure share for utilities - that is, rh = xuh/xh- is greater than a given critical threshold ru. A basic headcount index ofaffordability is therefore simply given by the fraction of households with rh > ru

HI =1

N

Xh

1 (rh > ru) (1)

where N is the total number of households, 1 (·) equals one if its argument istrue, zero otherwise. To illustrate why this index can hardly be considered asensible measure of affordability we can refer to Figure 1. Consider the absolutepoverty line xp = puq

up + pcqcp where qup and qcp are the minimum quantities

respectively of utilities and of other goods necessary to reach a decent standardof living; pu and pc are their corresponding market prices. Notice that qup andqcp may depend on demographic characteristic of the households, but they areset in absolute value, irrespective of the households’ income. All householdswhose total expenditure xh = puq

uh + pcq

ch falls below the poverty line are poor

households (area A+B), but among them, only those whose ratio rh = xuh/xhfalls above the ray ru are classified as facing affordability problems (area A). Theremaining poor households (B) are considered not having affordability problems,even if they cannot afford to heat their houses (xh < puq

up). According to thisdefinition, households without access to the public utility markets cannot faceany affordability problem7 . Moreover, a not negligible fraction of non-poorhouseholds may be classified as facing affordability problems (C), even thoughtheir total expenditure is well above the poverty line xp.

As Figure 1 above shows, such a measure of affordability lacks specific con-sideration of the households’ ability to pay. The (little) attention paid to thechoice of the critical threshold ru neglects that ru implicitly identifies the quan-tity of public utilities8

qu (xh; ru, pu) = ru (xh/pu)

below which the expenditure for public utility is affordable, with qu (xh; ru, pu)

increasing with xh and decreasing with pu.9 That is, if the policy maker eval-uates the effects of a tariffs change by simply looking at variations of HI,

7Notice that this is a particularly relevant problems in countries where utility services arenot provided in the 100% of the territory (limited coverage). However, in what follows, we donot apecifically address to the issue of affordability of access in public utility, a relevant issuefor developing countries. For a survey on access to utilities by the poor in Latin America, seeUgaz and Waddams, 2003.

8Frankhauser and Tepic (2006, forthcoming) collected a list of ru adopted by differentgovernments and international institutions for measuring affordability in electricity, heatingand water: although no universal benchmark arises, their table highlights that a threshold ofabout 25% of household expenditure in the three mentioned utilities could considered as thewidely adopted.

9 See affordability condition A2 in Thalmann (2003, p.295), which defines standars housingexpenditure, irrespective of household’s budget.

5

Figure 1:

then she worries about the “affordability” of large quantities for the rich andsmall quantities for the poor. Notice that for the poor it might well be thatqu (xh; r

u, pu) < qup, i.e. the quantity implicitly defined by the choice of ru maybe insufficient for a decent standard of living. Furthermore, in many empiricalapplications of this kind of affordability measure, heterogeneity in households’composition and income, disparities in regional infrastructures and differencesin quality of the items consumed appear to be neglected10.Moreover, the HI index does not capture the depth of households problems.

A more sophisticated index resembles the poverty gap index and it takes intoconsideration the distance between the actual share rh and the critical thresholdru:

PGI =1

N

Xh

1 (rh > ru)× (rh − ru)α

α ≥ 0 (2)

where α represents the policy-maker’s concern for the depth of affordabilityproblems (with α = 0, PGI = HI). With α > 2, the PGI satisfies Sen’smonotonicity and transfer axioms and Kakwani transfer sensitivity axiom.11

The index (2) is also “additively separable”, meaning that it can be calculatedfor different household types and aggregated using a weighted sum, where theweights are given by the fraction of households falling in each type relative tothe total number of households.Similar indexes can be defined considering the budget share which would be

devoted to public utilities, that is, a standard quantity considered necessary for

10 In particular, Chaplin and Freeman (1999) have shown that the British National HousingFederation description of affordability fails in measuring accurately the affordability situationof individual households within groups of households.11For the analysis on the Sen approach to poverty measurement and the studies related, see

Foster et al. (1984); Chaplin and Freeman (1999).

6

a decent standard of living is exogenously defined:

r∗h =puq

up

xh< ru ⇔ puq

up < ruxh (3)

We therefore have

HI∗ =1

N

Xh

1 (r∗h > ru) (4)

PGI∗ =1

N

Xh

1 (r∗h > ru)× (r∗h − ru)α

α ≥ 0 (5)

The indexes (4) and (5) stem from a potential affordability approach,12 an ap-proach which has relevant advantages. First of all, it refers to a basket of publicutilities which explicitly identifies the minimum standard of living the policymaker is targeting.13 Moreover, this index does not classify households withoverabundant level of consumption and high income as families with affordabil-ity problems, while it can correctly recognize an affordability problem for thosehouseholds whose actual consumption is lower than the minimum quantity (i.e.those with quh < qup). Finally, it does not require one to forecast households’reaction to a price change, as qup is absolutely defined.This approach has been adopted by the British government to assess fuel

poverty, with the standard quantity qup defined in absolute terms, and functionof the demographic characteristics of the households and their housing condi-tions (see Sefton and Chesshire, 2005). Notice that if qup differs from the oneused to define the standard absolute poverty line, and/or ru 6= puq

up/xp, thenthe potential affordability condition above implicitly defines a new poverty line(say xp0 = puq

up/ru). In other words, when the policy maker uses the potentialaffordability definition to identify the subgroups of households in need, she issimply searching for a (household specific) income threshold (a multiple of thestandard poverty line) below which households deserve some sort of benefit.However, if not combined with other criteria, this approach would tend to

implement “general income support” policies. In other terms, given the standardbasket of utility services and its cost, the simplest way to overcome affordabilityproblems would be to increase the income of the households below this threshold.Sefton (2002) refers to a variation of the PGI∗ index, where the depth is

considered not in terms of budget shares, but in monetary terms:

PGI∗0 =1

N

Xh

1 (r∗h > ru)× (puqup − ruxh) (6)

This variant goes toward considering the ability to pay for the minimum stan-dard consumption as crucial for the definition of affordability. However, it does

12For a discussion, see the condition of potential affordability PA2 in Thalmann (2003, p.296).13 In other words, this index considers the consumption of a minimum level of public utility

as the consumption of a merit good (see about Hancock, 1993).

7

not take into account that public utilities, although essential, are not the onlygoods households need in order to live decently, and that families should alsoafford a standard basket of other goods defined in absolute terms xcp = pcq

cp.This brings us to consider the “residual income” as crucial to measure afford-ability in public utility markets.

3 The “residual” income as a measure of afford-ability for public utility services

The approach of the “residual” income to measure affordability has been firstlyapplied in housing economics. Quoting Thalmann (2003, p.294) “(h)ousing isnot affordable for a household if it excessively crowds out other expenditure.Obviously housing expenditure always crowds out non/housing expenditure. Theterm ‘excessively’ is key”. Considering the affordability issue with respect topublic utility expenditure, we can say that there is an affordability problem if- after having paid the public utility bills - the households do not have enoughmoney to finance a minimum level of the other goods’ consumption. But whatlevel of consumption in public utilities and in other goods do we have to referto? If we consider the actual level of consumption quh , household h faces anaffordability problem if its residual income xh−puquh is lower than the monetaryvalue of minimum level of consumption of the other goods:14

xh − puquh < pcq

pc (7)

This mimics the shelter poverty definition of Stone (1993), and - rewordingKutty (2005) - let’s call it “public utility induced poverty”. Following thisdefinition, we can observe in Figure 2 that poor households falling in E and Fareas, as well as non-poor households belonging to the A area, face what wecalled public utility induced poverty.Equation (7) can fruitfully be used to analyze the short term effects of price

reforms in public utility, as in those markets it is unlikely to expect a promptadjustment of quh in response to a variation of the price pu. But there aretwo important caveats to comment here. As in the case of the affordabilitymeasures based on the bills-to-income-ratios rh, if we do not introduce an upperlimit to quh , definition (7) suffer of the so called “perverse” preferences problem(Hancock, 1993): some households may have preferences such that - even thoughthey can afford a higher quantity of other goods - they prefer a consumptionlevel lower than the minimum standard. Again, without a lower limit, somehouseholds could “solve” their public utility induced poverty by reducing to zeropublic utilities’ consumption. As can be observed in Figure 2, poor householdsrepresented in D area, cannot afford the minimum standard qup but are not inpublic utility induced poverty.

14This condition corresponds to Definition 2 in Hancock (1993, p.135), or the affordabilitycondition A1 in Thalmann (2003, p.294).

8

Figure 2:

We can refine the previous (7) by introducing a requirement of a minimumstandard of public utilities’ consumption (see definition 3 in Hancock, 1993):public utilities are unaffordable if⎧⎨⎩ xh − puq

uh < pcq

cp

orxh − pcq

ch < puq

up, qch > qcp(8)

With this refinement, now households falling in C and D areas in Figure 2 jointhe subset of the households in public utility induced poverty.In order to precisely evaluate the effects of price reforms in public utility

markets and to correctly address any "mitigating" policy, it is important torecognize that an affordability problem can stem from different reasons. Con-sidering household groups as they results from Figure 2, we can observe thatfamilies in A area are not poor, but they are over-consuming public utilities:for them it is relevant to recognize whether this is due to their preferences orit is because they have some binding non-monetary constraint (for instance,they are locked in with the heating technology). In this latter hypothesis, anaddressed programme would specifically offer incentives to increase energy ef-ficiency through the adoption of new technology. Moreover, Households in Care not poor, but they are under-consuming public utilities: again, this canbe a matter of tastes, or because they do not have access to the desired (andmonetary affordable) quantity of public utilities.In other words, when the policy maker refers to condition (8) it identifies

three types of consumers with affordability problems: (i) the (absolutely) poor,they cannot afford the minimum quantities (qcp, qup) and general income sup-port to them may be advisable; (ii) those “over-consuming” and (iii) those“under-consuming” public utilities because of non-monetary constraints, to-

9

wards which policies aimed at relaxing these constraints may be effective. Forinstance, if the housing conditions of household h are such that the consumptionof the quantity quh > qup is necessary to reach a comfortable temperature forits home and its monetary income mh is so low that mh − puq

uh < pcq

cp, thehousehold cannot afford the minimum basket of other goods once it has paid itsheating bill. This is a case in which the poor housing conditions make the house-hold fall into public induced utility poverty, even though it is not necessarilyabsolutely poor. The policy maker can choose between two different strategiesto relax the constraint faced by the household and to make the couple (qcp, quh)affordable: i) it can pay a cash benefit of s = pcq

cp− (mh − puquh), or ii) it can

finance a price reduction to pu ≤ (mh − pcqcp) /quh. Notice that both strategies

are approximately implemented in practice by the policy makers by designingfuel stamps program, and/or by financing lower tariffs for households satisfyingspecific entitlement requirements.Once a definition of poverty has been chosen, either a headcount index

and/or a poverty gap index can be estimated. If the definition (8) is adoptedwe can have

Iuh = 1 (xh − puquh < pcq

cp)

Ich = 1 (xh − pcqch < puq

up)

HI∗∗ =1

N

Xh

Iuh + Ich − IchIuh

PGI∗∗ =1

N

Xh

Iuh ×µ1− pcq

ch

pcqcp

¶α+ Ich ×

µ1− puq

uh

puqup

¶αα ≥ 0

The data requirements for estimating the incidence of the public utility in-duced poverty depend upon the poverty concept we refer to. If the definition(7) is adopted, the information set should include household income/total con-sumption expenditure (xh), public utilities expenditure (puquh) and an estimateof the value of the non-public utility component of the official absolute povertyline (pcqcp). When we turn to the definition (8), data on the expenditure fornon-public utility goods at the household level (pcqch) and the value of the publicutility component of the official absolute poverty line (puqup) are also required.Notice that no information on quantities or prices are required in practice. Thisis relevant for practical purposes, as most of the surveys on family expendituredo not record quantities and/or prices. Furthermore, public utilities are oftencharacterized by non linear tariff schemes, that could make the price of qup dif-ferent from the price of quh: this future is irrelevant as far as the scope of theanalysis is to provide a measure of the affordability at given point in time orover time. Measurement issues may rise when decomposition of the affordabilityindexes are considered by area or household types. In these cases some notes ofcaution should be taken into consideration. For practical purposes computationof affordability indexes can be obtain to a detail dictated by the availability of areasonable measure of the reference expenditure (pcqcp, puqup) for area and/orhousehold type. Taking into account household size is usually not a practical

10

problem, while controlling for area it may be if the absolute poverty line is notregional specific.

4 A possible application: Italy 2002-2004The privatization and liberalization of utility sectors have started in Italy withsome caution, given that Italy has been traditionally attached to public owner-ship of public utilities and direct control of services. In all the three utilities hereinvestigated new sectoral laws have been introduced to give companies betterincentives for efficient production and pricing; but their implementation have -in all the three utilities - very slow start given specific inertia from each sector’spast organization. In what follows, we briefly focus on price reforms in theItalian water, electricity and natural gas sectors15 .In the water sector, the new tariff is adjusted by the local regulatory author-

ity following a price-cap mechanism which incorporates operating costs (net ofa 0.5-2% annual efficiency gain), depreciation of assets and investments at themaximum rates by law, and 7% return on investments. This two part tariff ap-plies the “full cost recovery principle”, which - along with the high investmentplanned for the whole integrated water service - determine increases in tariffs.Given the greater market orientation of the sector, cross subsidies - which werewidespread in past management by municipalities - are no longer allowed: this,in turn, may partially compensate consumers by saving on of municipal costs(i.e.: less local taxes or greater supply of other public utility services), but callsfor investigation on distributive effects.In the Italian natural gas sector, the presence of a dominant and vertically

integrated firm (Eni) with long term contracts still affect the transition towarda competitive market. Legal unbundling and mild antitrust ceilings have beenadopted to limit the incumbent market power. Moreover, since 2001, tariffs areregulated with a price-cap (RPI-x) by the national regulatory Authority (Aeeg)according to a non discriminatory and cost reflective standard. The price-capformula contains elements of price increase which aim to compensate firms for“unpredictable” events, to reward them for their activities of demand controland for quality improvements.The Italian electricity sector - previously characterized by the presence of a

vertically integrated dominant firm (Enel) - has the transmission network withopen access to third parties on the basis of conditions set by the regulatoryAuthority. Since 2004 the wholesale market has been organized as a competitivePool market, and in the upstream market no firm is allowed to own more than50% of total installed power or to sell more than 50% of total energy, includingimports.Prices are free in the wholesale segment and in the sale to “eligible” cus-

tomers. Since January 2000, other prices are regulated according to a RPI-xsystem with an automatic pass-through of wholesale price increases to final

15For a survey on the fundamental features of the reforms specifically belonging to theItalian water, electricity and gas sectors, see Miniaci et al. (2005).

11

prices to non-eligible customers. Distributors selling energy to franchise (noneligible) customers must buy the energy for these customers through a SingleBuyer, which is also part of the State owned Grtn group. The thresholds foreligibility were established to accelerate the process of market opening relativeto the dates set in the Directive. At present all non domestic customers areeligible, while domestic ones still face a uniform national tariff, updated on thebasis of a RPI-x scheme. Until 1999, the previous price system has been simply“cleaned”, eliminating some subsidies and clarifying the complex structure ofcharges and surcharges.Our empirical analysis on affordability in the above Italian utilities works

up from the data sources of ISTAT Surveys on Family Budgets (SFB) from2002 to 2004. These surveys (which correspond to the British FES and theCEX in the US, with independent samples of about 20,000 households per year,representative of the Italian population) provide detailed information on ex-penditure and demographics, and some information on stock of durables andhousing conditions. The data are collected through a face to face interview(plus a weekly diary) during which the households are asked if the house theylive in has potable water, electricity and heating (if yes, the technology adoptedand fuel used). Households should provide information about the amount ofthe latest bill for electricity and natural gas, and on the expenditure during thethree months before the interview for water, other fuels (LPG, kerosene, dieseloil, coal and wood) and centralized heating. Data on ordinary and extraordi-nary maintenance works are collected for the three months prior the interview.Information on main and secondary home of residence are clearly separable.Given that heating fuels consumption is strongly correlated with climate

conditions, we present many of the statistics conditional on climatic regions.We group the 20 administrative regions in four classes:1. warm regions, with average degrees-days index not greater than 1300

(Campania, Sicily and Sardinia, 19.2% of Italian households)2. tepid regions, with average degrees-days index between 1300 and 1800

(Liguria, Lazio, Puglia, Calabria, 21.6% of Italian households)3. cool regions, with average degrees-days index between 1800 and 2300

(Tuscany, Umbria, Marche, Abruzzi, Molise and Basilicata, 12.9% of Italianhouseholds)4. cold regions, with average degrees-days index above 2300 (Piedmont,

Valle d’Aosta, Lombardy, Trentino Alto Adige, Veneto, Friuli Venezia Giuliaand Emilia Romagna, 46.4% of Italian households)Table 1 highlights that the living standards of the areas are strongly differ-

entiated: in 2002, conditional on family size, the average family expenditure inthe northern cold regions is 35% to 55% higher than in the (southern) warmpart of Italy. These differences are not necessarily reflected in the expenditureon utility services: in 2002, families in warm regions spent almost the sameamount for water as households in the cold area, and about 25% more in elec-tricity. When we consider heating fuels expenditure, an opposite effect emerges:the average per family expenditure for natural gas and other fuels in the coldarea was about 3 times that in the warm area. Conditioning on family size

12

is crucial to understand to what extent households can exploit “economies ofscale”. This is particularly true for the utilities, as they are mainly devoted tothe production of (intra-household) public goods, such as lighting and heating.Considering Italy as a whole, the 2002 per capita expenditure for water in a 4member family was only 70% of the corresponding per capita expenditure of acouple. The same ratio was 63% for electricity, 56% for natural gas and otherfuels. But the total expenditure of larger households is higher, and all utilitieshave increasing block tariffs, so that larger households may face higher marginalprices. This raises a fairness (and an efficiency) issue for the block tariffs whichdo not take into account family size.Table 1 clearly indicates that an approach to the affordability problem based

on the adoption of an unified national criterion for all climatic regions runs therisk of introducing large biases. This would happen, for instance, if one wereto use the criterion of the Italian Poverty Commission in defining the absolutepoverty line, which includes - among other things - the minimum expenditurefor heating and electricity. Such Italian Commission fixed this minimum expen-diture equal to the first quartile of the national distribution of the expenditurefor natural gas and electricity, conditional on household size. However, what isacceptable in a warm region may not be acceptable in a cold area: householdsin the cold area spend much more for heating than families in the southernregions. We therefore suggest regional specific reference expenditure in order todefine more plausible minimum thresholds.Table 2 compares the national official reference expenditure for heating, elec-

tricity and other goods (estimated in 1997, their monetary values are updatedwith the relative consumer price indexes), with our estimates of the referenceexpenditure, computed as the first quartile of the regional 2002 distribution ofthe expenditure for the utility, conditional on household size. With our defi-nition, notice that, for instance, the amount considered socially acceptable forheating in the cool area is more than twice the corresponding amount for thewarm area.In order to compute the affordability indexes HI and HI∗ based on the bill-

to-income ratios, we need to define the critical threshold (ru). As the two indexesrefer to different quantities of consumption, we define two different thresholdsas the average of, respectively, rh and r∗h for those households falling in the lefthand tail of the distribution of the equivalent total family expenditure for thearea of residence. Table 3 reports these estimates, which can be read as follows:a family living in the cold area is in “utility poverty” if the ratio between itsactual expenditure for water, electricity and heating and its total expenditureis above 9.26%, or it is in “utility poverty” if the ratio between the minimumexpenditure for utilities — see Table 2 - and its total expenditure is above 6.12%.As we suggest a new estimate of the minimum expenditure for water, elec-

tricity and heating, we are in fact suggesting a new overall absolute poverty line.In order to assess to what extent this variation is crucial for the overall measureof the poverty in Italy, in Table 4 we compare the percentage of householdsfalling below the official and the revised poverty line. The last column show thefraction of families that are not able to finance the consumption of the minimum

13

quantities of goods other than the utilities, once they have paid the utility bills,that is, those in public utility induced poverty according to the Stone’s (1993)principle. The figures show that i) our regional specific revision of the povertyline does not alter the overall picture of the absolute poverty; ii) the numberof households in public utility induced poverty is higher than the number ofpoor, and the difference is particularly remarkable for the cold regions, whereexpenditure for heating is higher. In terms of number of households, in 2004there were about 804,000 families below the absolute poverty line (of which only91,000 in the cold area), and about 172,000 (51,000 in the cold area) whose totalexpenditure was above the poverty line, but that could devote too little to theexpenditure of other goods.The estimates of the headcount indexes based on actual and potential budget

shares are shown in Table 5. The two procedures give two dramatically differentpictures of the affordability issue. Although the critical thresholds consideredare close each other (see Table 3), the distribution of the actual budget sharesis such that HI signals an utility affordability problem for about 20% of thefamilies, with a peak of over 50% for water. If we consider the distribution of thepotential budget shares r∗h, only about 6.5% of the families faced an affordabilityproblem in 2002, and this fraction dropped to 5.3% in 2004 (the difference isstatistically different from zero).The picture changes again when we look at the headcount index HI∗∗ based

on the more sensible residual income approach (Table 6). Now the fractionof households facing affordability problems goes from 23% in 2002 to 20% in2004, and about 80% of them are classified as in public utility induced povertybecause of the low level of expenditure in utilities, despite of the fact that theirexpenditure for other goods is above the minimum expenditure for these goods.Finally, the poverty gap indexes PGI∗∗ computed for α = 1 and α = 2

provide evidence against the conclusion, based on the decomposition of HI∗∗ intable 6, that the excessively low expenditure in utilities is the most importantreason of the public utility induced poverty. When the intensity of the povertyis taken into consideration (see Table 7) it is clear that the main problem isdue to the excess of expenditure for the utilities, with the warmer and poorerregions with by far the most relevant problems of affordability.

5 ConclusionsIn this paper we have investigated a new affordability measure to capture theeffects of price reforms in public utility markets. This new measure considersthe affordability problem looking at the “residual” income of the household onceit has financed the purchase of a standard quantity of utility services. If thisresidual income falls above the absolute poverty line (net of public utilities),then, we claim the household has no affordability problems.Among the relevant advantages of this new approach we have that: a) it

requires to define a consistent standard level of consumption; b) it can relyon the results obtained by the well established literature on general poverty

14

indexes; c) it does not require the estimation of the demand functions for utilityservices; d) it helps to identify those households which are more likely to faceaffordability problems due to inefficiency of their technological endowment.We empirically test this new affordability measure on the Italian household’s

consumption of water, electricity and gas in the period 2002-2004: these sectorshave undergone a deep institutional and regulatory (tariff) reform through thelast decade. We then compare empirical results from our new affordabilitymeasure with outcomes obtained with standard affordability measures. We findthat all the measures adopted record a reduction in the affordability issue in theperiod between 2002 and 2004 for the Italian electricity, gas and water supply.However, our results highlight dramatically different pictures of the affordabilityissue belonging to the different measure adopted.Moreover, the new affordability measure we propose makes it possible to

distinguish different sources of the affordability for different households. This,in turn, allows one to better target policies to reduce affordability in publicutilities and - possibly - to coordinate more efficiently with other anti-povertyactions.

15

References[1] Balestrino, A. (1999), "The desirability of in-kind transfers in the pres-

ence of distortionary taxes", Journal of Economic Surveys, 13/4, pp.333-354.

[2] Bramley, G., (2004), “Affordability, Need and the Intermediate Mar-ket: measurement, change and significance”, paper presented at the ENHRConference July 2nd-6th 2004, Cambridge, UK.

[3] Barr, N., (1992): "Economic theory and the welfare state: a survey andinterpretation", Journal of Economic Literature, XXX, 741-803.

[4] Bergstrom, Blume and Varian, (1986), "On the private provision ofpublic goods, Journal of Public Economics 29, 25—49.

[5] Besley, T. (1998): "A simple model for merit good arguments", Journalof Public Economics, 35, pp. 371-383.

[6] Chaplin, R. and A. Freeman (1999), “Towards an Accurate Descrip-tion of Affordability”, Urban Studies, 36 (11),. 1949-57.

[7] DEFRA (2001). The UK Fuel Poverty Strat-egy - 1st Annual Progress Report 2003,http://www.dti.gov.uk/energy/consumers/fuel_poverty/index.shtml

[8] DEFRA (2004): "Fuel poverty in England: the government’s plan ofaction"

[9] Foster, J. E., J. Greer, and E. Thorbecke (1984), “A class of decom-posable poverty measures”, Econometrica, 52, pp. 761-766.

[10] Frankhauser, S. and S. Tepic (2006), “ Can poor consumers pay forenergy and water? An affordability analysis for transition countries”, forth-coming in Energy Policy

[11] Hancock, K. E. (1993), “‘Can Pay? Won’t Pay?’ on Economic Princi-ples of ‘Affordability”’, Urban Studies, 30(1), pp. 127-145.

[12] Kutty N.K. (2005), “A New Measure of Housing Affordability: Esti-mates and Analytical Results”, Housing Policy Debate, 16 (1), Fannie MaeFoundation

[13] Miniaci, R., C. Scarpa and P. Valbonesi (2005) “Restructuring Ital-ian Utility Markets: Households Distributional Effects”, FEEM Nota diLavoro 134.2005

[14] Robinson, M., G. M.Scobie and B. Hallinan (2006), “Affordabilityof Housing: Concepts, Measurement and Evidence”, New Zealand TreasuryWP 06/03

16

[15] Sandmo A. (1983), "Ex post welfare economics and theory of meritgoods", Economica, 50/197, pp. 19-33.

[16] Sefton T. (2002), “Targeting Fuel Poverty in England: Is the GovernmentGetting Warm?”, Fiscal Studies, 23(3), pp. 369-99

[17] Sefton T. and J. Chesshire (2005), “Peer Review of the Methodologyfor Calculating the Number of Households in Fuel Poverty in England”,www.dti.gov.uk/files/file16566.pdf

[18] Stone M.E. (1993), Shelter Poverty: New Ideas on Housing Affordability,Philadelphia, Temple University Press.

[19] Thalmann P. (2003), “House poor or simply poor?”, Journal of HousingEconomics, 12, pp. 219-317.

[20] Ugaz C. and C. Waddams, (2003), "Utility privatization and regulation:a fair deal for consumers", Edward Elgar Publishing.

17

Table 1: Average monthly family expenditure by area and family size, year 2002. Euro

Total expenditure

# family members Warm Tepid Cool Cold Italy

1 991.84 1282.71 1337.16 1459.16 1325.77

2 1434.71 1776.13 1877.75 2238.24 1962.38

3 1936.31 2254.23 2613.65 2797.54 2510.60

4 2305.41 2505.00 2939.86 3113.04 2740.45

5 or more 2339.30 2493.34 3172.04 3238.58 2730.07

Total 1761.33 1965.17 2203.17 2340.77 2126.27

Water

# family members Warm Tepid Cool Cold Italy

1 13.47 12.59 13.03 11.06 12.06

2 16.68 15.66 18.05 14.74 15.80

3 20.03 18.98 20.05 18.77 19.23

4 22.61 21.82 23.78 21.63 22.22

5 or more 25.14 24.05 26.23 25.10 25.01

Total 18.98 17.89 18.94 16.32 17.55

Electricity

# family members Warm Tepid Cool Cold Italy

1 27.08 24.39 22.17 20.04 22.47

2 34.88 31.80 28.76 29.71 30.80

3 45.23 42.42 44.58 43.83 43.92

4 43.26 37.28 38.21 37.93 38.75

5 or more 53.10 47.10 50.80 50.82 50.64

Total 39.55 34.58 33.78 32.17 34.36

Natural gas and other fuels

# family members Warm Tepid Cool Cold Italy

1 22.82 39.38 54.45 73.71 55.70

2 29.22 54.28 74.72 90.50 71.55

3 36.27 58.14 82.84 96.73 76.74

4 42.95 64.06 89.25 109.53 79.62

5 or more 44.38 66.84 99.47 113.76 76.47

Total 34.73 54.36 76.33 91.56 70.69

Table 2: Reference expenditure estimates, year 2002. Euro.

Official reference expenditure ( and up cpx x )

# Family members Heating Electricity Other goods

1 18.97 9.68 354.73

2 23.77 13.01 538.08

3 28.31 17.44 770.66

4 28.99 21.86 982.86

5 or more 31.72 27.93 1293.26

Our estimates of the reference expenditure upx

Water Electricity Heating

# Family members Warm Tepid Cool Cold Warm Tepid Cool Cold Warm Tepid Cool Cold

1 7.23 7.23 7.03 6.64 16.07 13.81 12.40 11.29 10.13 11.13 19.98 19.38

2 9.65 8.84 10.45 8.31 20.38 18.75 16.61 16.61 12.73 17.23 27.10 27.65

3 11.25 11.25 12.06 10.85 24.90 21.42 22.51 22.01 16.87 21.68 31.79 33.20

4 12.06 12.05 13.49 12.37 27.66 25.18 26.80 26.78 19.64 22.26 35.43 41.51

5 or more 12.86 12.25 16.07 15.67 32.15 29.65 34.62 29.47 19.71 23.99 38.02 43.00

Table 3: Threshold budget shares

Threshold ratios based on actual expenditure ( /u

h h hr x x= )

Water Electricity Heating Utilities

Warm 2.31% 4.83% 2.35% 8.32%

Tepid 1.89% 3.78% 3.69% 8.45%

Cool 2.07% 3.11% 5.17% 9.34%

Cold 1.38% 2.64% 5.91% 9.26%

Threshold ratios based on reference expenditure ( * /up

h hr x x= )

Water Electricity Heating Utilities

Warm 1.60% 3.57% 2.33% 7.50%

Tepid 1.31% 2.66% 2.36% 6.33%

Cool 1.37% 2.43% 3.64% 7.44%

Cold 1.01% 1.95% 3.15% 6.12%

Table 4: Absolute poverty vs. Stone's (1993) public utility induced poverty

2002

Official poverty line Revised poverty line Other goods only ( c cp

hx x< )

Warm 8.68% 8.96% 9.76%

Tepid 5.46% 5.26% 6.15%

Cool 3.58% 3.57% 4.34%

Cold 1.75% 1.74% 2.28%

Italy 4.17% 4.17% 4.87%

2003

Official poverty line Revised poverty line Other goods only ( c cp

hx x< )

Warm 7.99% 7.97% 8.65%

Tepid 4.68% 4.38% 5.34%

Cool 2.75% 2.55% 3.35%

Cold 1.38% 1.42% 1.86%

Italy 3.58% 3.50% 4.15%

2004

Official poverty line Revised poverty line Other goods only ( c cp

hx x< )

Warm 8.70% 8.72% 9.59%

Tepid 4.98% 4.87% 5.79%

Cool 2.67% 2.73% 3.46%

Cold 0.87% 0.89% 1.34%

Italy 3.53% 3.53% 4.20%

Table 5: Headcount indexes based on budget shares

Based on actual expenditure ( HI ) Based on reference expenditure (*HI )

2002 2002

Water Electricity Heating Utilities Water Electricity Heating Utilities

Warm 50.69% 14.09% 26.23% 17.54% 6.33% 6.15% 6.00% 6.10%

Tepid 55.72% 13.44% 25.94% 19.19% 7.56% 7.63% 7.48% 7.44%

Cool 41.92% 14.52% 26.45% 22.13% 6.45% 6.27% 6.34% 6.47%

Cold 50.20% 15.99% 26.44% 22.45% 6.74% 6.23% 6.38% 6.35%

Italy 50.36% 14.85% 26.29% 20.73% 6.80% 6.53% 6.54% 6.56%

2003 2003

Water Electricity Heating Utilities Water Electricity Heating Utilities

Warm 51.52% 13.02% 28.52% 19.03% 6.07% 5.84% 5.72% 5.80%

Tepid 54.79% 13.76% 27.73% 19.38% 7.03% 6.75% 6.62% 6.81%

Cool 44.02% 14.29% 27.40% 21.78% 5.01% 5.21% 5.39% 5.20%

Cold 49.53% 16.73% 23.74% 21.20% 6.11% 5.94% 6.37% 6.20%

Italy 50.30% 15.02% 26.06% 20.46% 6.15% 6.00% 6.16% 6.12%

2004 2004

Water Electricity Heating Utilities Water Electricity Heating Utilities

Warm 52.93% 13.78% 29.61% 18.82% 7.36% 4.99% 6.56% 6.01%

Tepid 54.42% 13.58% 29.21% 21.48% 7.90% 5.59% 6.94% 6.48%

Cool 45.16% 14.15% 30.63% 23.34% 6.41% 4.34% 6.13% 5.60%

Cold 48.61% 14.04% 24.09% 20.38% 5.45% 3.45% 4.70% 4.35%

Italy 50.23% 13.90% 27.19% 20.73% 6.49% 4.34% 5.75% 5.31%

Table 6: Headcount indexes based on residual income

2002

c cp

hx x< u up c cp

h hx x and x x< ≥ **HI

Warm 9.76% 21.89% 31.65%

Tepid 6.15% 17.63% 23.77%

Cool 4.34% 18.86% 23.20%

Cold 2.28% 16.97% 19.25%

Italy 4.87% 18.34% 23.21%

2003

c cp

hx x< u up c cp

h hx x and x x< ≥ **HI

Warm 8.65% 20.52% 29.17%

Tepid 5.34% 14.70% 20.03%

Cool 3.35% 16.34% 19.69%

Cold 1.86% 16.41% 18.27%

Italy 4.15% 16.82% 20.98%

2004

c cp

hx x< u up c cp

h hx x and x x< ≥ **HI

Warm 9.59% 17.71% 27.31%

Tepid 5.79% 12.97% 18.77%

Cool 3.46% 14.54% 18.00%

Cold 1.34% 17.08% 18.42%

Italy 4.20% 15.95% 20.15%

Table 7: Poverty gap indexes.

2002

( ) /u cp c

h hI x x n−∑ ( ) /c up u

h hI x x n−∑ ( )

2

/u cp c

h hI x x n−∑ ( )

2

/c up u

h hI x x n−∑

Warm 17.50 2.87 6221.78 57.95

Tepid 11.65 2.27 4118.84 44.63

Cool 5.73 3.25 1534.62 86.90

Cold 2.96 2.97 754.65 82.08

Italy 8.08 2.84 2666.78 69.82

2003

( ) /u cp c

h hI x x n−∑ ( ) /c up u

h hI x x n−∑ ( )

2

/u cp c

h hI x x n−∑ ( )

2

/c up u

h hI x x n−∑

Warm 14.93 2.60 5151.72 52.86

Tepid 8.74 1.78 2924.99 34.30

Cool 5.56 2.76 1714.69 73.68

Cold 2.60 3.03 750.84 86.81

Italy 6.76 2.63 2219.42 66.83

2004

( ) /u cp c

h hI x x n−∑ ( ) /c up u

h hI x x n−∑ ( )

2

/u cp c

h hI x x n−∑ ( )

2

/c up u

h hI x x n−∑

Warm 18.10 2.06 6281.91 38.29

Tepid 9.82 1.48 3177.54 26.51

Cool 4.50 2.43 1090.69 65.18

Cold 1.42 3.15 361.20 88.86

Italy 6.90 2.47 2221.18 62.19