* 2 1467 Viewpoint Note No. 211 June mIs " Designing Direct Subsidies for the Poor-A Water and Sanitation Case Study Vivien Foster, Direct subsidies are an increasingly popular means of making infrastructure services more Andres Gomez- affordable to the poor. Under the direct subsidy approach, governments pay part of the water bill Lobo, and Jon athan of poor households that meet certain eligibility criteria. This approach was first used in water Halpern sector reforms in Chile in the early 1990s and is an alternative to the traditional method in which governments pay subsidies directly to utilities, often allowing the price of water to fall below economic costs indiscriminately. This Note illustrates how simulation techniques can be used to inform the design of direct subsidy schemes, ensuring that they are both cost-effective and accurate in reaching the target population. Universal access to adequate water and sanitation mize distortions in the behavior of water utilities services has long been recognized as essential to and their custorners. The main drawbacks of public health and individual welfare. In most direct subsidies are higher administrative costs countries government policy has traditionally and the difficulty of designing suitable eligibility been to keep water companies in public owner- criteria. ship, and to keep tariffs artificially low through a range of more or less explicit subsidy measures. Despite the growing interest in direct subsidies, The results have often been unsatisfactory. Service their administrative cost and accuracy of targeting quality and coverage remain inadequate in many have, as yel, been little studied. This Note attemnpts countries, and subsidies directed at public water to fill the gap by presenting the results of a simu- companies have often benefited the middle lation performed in Panama in 1998. The simula- classes rather than the poor, who remain uncon- tion is based on a willingness-to-pay survey and nected to the public network. This has led to a complementary informnation from the 1997 Living search for alternative subsidies that would guar- Standards Measurement Study (LSMS) survey and antee access to basic services for the poorest. the customer database for Panama's national water utility IDAAN (Instituto de Acueductos y In the direct subsidy system introduced during Alcantarillados Nacional). The work was done in the Chilean reforms in the early 1990s, prices are the context of efforts to reform IDAAN, which still allowed to signal their true economic scarcity operates as a traditional utility. It considers the rel- costs while subsidies are paid to consumers who ative merits of a water consumption subsidy ver- cannot afford their bills rather than to water util- sus a sewerage connection subsidy. ities. Government funds are used to cover part of the cost of subsistence consumption for Assessing need for a subsidy households that meet certain poverty-related cri- teria. The main advantages of direct subsidies are As a general rule, the case for a water subsidy is that they are transparent and explicit, and mini- something that needs to be assessed rather than The World Bank Group * Private Sector and Infrastructure Network Designing Direct Subsidies for the Poor-A Water and Sanitation Case Study TABLE 1 ZONAL AND INDIVIDUAL CRITERIA IN PANAMA, 1997 (percent) ,__ sutners would be willing to pay up to US$0.46. The imnplication is that the price of potable water would have to rise sul)stantially before it became Errors oa exclusion 94 6 unaffordable to low-income households in Errors of inclusion 31 93 urban Panama. For sewcrage conncctions the current cost is US$1,000 (including cost of Source: Encuesta Nacional de Niveles de Vida, Panama, 1997. huse nt inobt willing t p i household plumbDing) but willingness to pav is only LS$270. Thus sewerage connections are already well beyond the means of low-income assumed. It is not axiomatic that water services households. Subsidies might therefore be justi- are unaffordable to low-income households, nor fiecl to the extent that the wider social benefits that subsidizing water consumption is the best of expanding the sewerage network- were at way of promoting access to sanitation or improv- least as high as the associated subsidy. ing public health among the poorest. Designing eligibility criteria An important starting point is to review existing subsidy arrangements and consider how effec- The intended beneficiaries of subsidy schemes tively they reach the poor. A review of basic data are invariably those living below the poverty in Panama revealed a number of important line, usually expressed as an annual threslhold points. The first was that poverty in Panama is for per capita income or expenditture. In practice primarily rural, but IDAAN operates almost it is very difficult to measure income levels exclusively in towns. Only 16 percent of directly and, therefore, to determine whether a IDA-AN's customers live below the poverty line. particular household should benefit from the Nevertheless, at least two-thirds of IDAANs CUS subsidy. To get around this problem, it is neces- tomers are subsidized. With the exception of a sary to develop eligibility criteria that show a pensioner discount. the determination of eligi- high degree of correlation with the uniderlying hility for these subsidies has been largely at the poverty variable of interest, can be measured discretion of IDAAN, and has often been based objectively and observed easily. and are difficult on payment history. The sheer number of ben- to falsify or misrepresent. eficiaries relative to the number of poor house holds served by IDAAN indicates that the vast Eligibility criteria can either he zonal (based on majority of subsidy beneficiaries are not poor, by the characteristics of the area where the house- the strict standard provided above. hold lives) or individual (based on the charac- teristics of the household itself). The key issue is To find out how many people are excluded from the extent to xw hich the criteria succeed in iden- water and sewerage services because they gen- tifying the target group. There are two forms of uinelv cannot afford them, a contingent valua- targeting errors. Perhaps the most serious are tion survey was conducted. It sought to establish errors of exclusion, when members of the target the maximum amount that households were group are not captured by the eligibility criteria willing to pay to consume a typical volume of and hence fail to receive tlie subsidy. A subsidy piped water and to obtain a connection to the scheme with a high exclusion error clearly fails sewerage network. Willingness to pay is com- on its own terms. Errors of inclusion occur wlhen pared with current prices to provide an indica- people outside the target group fortuitously tion of affordability. The premise is that a comply with the eligibility criteria and conse subsidy is only justified when willingness to pay quently receive thie subsidy. Such leakage of falls short of true economic cost. The results of funds to unintended beneficiaries redtuces the the survey show that while the current cost for efficiency of die subsidy. to particular, it inflates water is t-S 0.21 per cubic meter; poorer con- the cost of the subsidy to taxpayers. TABLE 2 HOUSEHOLDS WITH PROPOSED INDIVIDUAL CRITERIA OF ELIGIBILITY FOR SUBSIDIES IN PANAMA, 1997 Extreme poverty Poverty Non-poor Total Thou- Per- Thou- Thou- Per- Thou- Per- Criterion sands cent sands sands cent sands cent Poor-quality floor materials 10.6 88.6 24.3 78.3 187.6 56.0 222.6 58.8 Lack of telephone connection 9.6 79.8 21.7 69.8 113.2 33.8 144.4 38.2 Primary educated head of household 7.8 65.3 15.2 49.1 85.2 25.4 108.3 28.6 Toilet facilities located outside house 7.2 60.4 18.3 58.9 82.3 24.5 107.8 28.5 Lack of sanitation beyond most basic 7.1 58.9 15.0 52.7 52.7 15.7 74.8 19.8 None of the above 0.0 0.0 1.0 3.3 98.8 29.5 99.8 26.4 One or more of the above 12.0 100.0 30.0 96.7 236.3 70.5 278.4 73.6 Two or more of the above 11.3 93,9 25.9 83.5 140.7 42.0 178.8 47.3 Three or more of the above 9.5 78.8 20.2 65.2 80.1 23.9 110.5 29.2 Four or more of the above 6.8 56.1 13.2 42.6 47.6 14.2 68.3 18.1 All five of the above 2.9 24.1 5.2 16.9 16.2 4.8 24.3 6.4 Source: Encuesta Nacional de Niveles de Vida, Panama, 1997. As a criterion, zone of residence has the advan- proxies and poverty levels, using data from the tage of being easy to observe and relatively hard LSMS survey. The results for the most successful to falsify. How well it correlates with the under- proxy variables are repolted in table 2. As the lying poverty measure of interest depends on the table illustrates, it is difficult to find a single vari- size of the zones used and the extent to which able that performs well in minimizing both errors poverty is geographically concentrated. In of exclusion and errors of inclusion. For exam- Panama, simulations were performed using zones ple, poor quality floor materials have a very low defined by the 1997 LSMS survey, comprising error of exclusion, being found in 88.6 percent approximately 15 households eaclh. The simula- of extremely poor households, but a high error tions explored the effect of giving a subsidy to all of inclusion, also being found in 56.0 percent of households in zones that met a particular set of non-poor households. Clearly, the most power- socioeconomic criteria; for example, zones where ful eligibility criterion is one that combines infor- more than 50 percent of households were living mation on all of these characteristics. For the in extreme poverty. Such zonal eligibility criteria purposes of the Panama simulation, the eligibil- were found to have Iiigh errors of exclusion, with ity criterion was taken to be that the household as much as 94 percent of the target population should meet two or more of the criteria identi- being excluded from the scheme (table 1). fied in table 2. The main benefit of moving from However, errors of inclusion were comparatively zonal to individual criteria is a dramatic reduc low; only 31 percent of the subsidies would go to tion in the errors of exclusion, which fall from 94 households above the extreme poverty line. percent to 6 percent. This comes at the cost of much higher errors of inclusion, with a very high While reliable information about poverty levels proportion of the subsidy funds (93 percent) at the zonal level may be obtained based on cen- leaking to households outside the target group. sus or survey data, it is much harder to obtain a reliable estimate of individual household income Estimating administrative costs and expenditure in orcler to assess eligibility for individual subsidies. In Panama, an extensive Using individual eligibility criteria requires an search for proxy variables was conducted by administrative apparatus for the screening of making cross-tabulations between candidate potential candidates. Typically, a standardized Designing Direct Subsidies for the Poor-A Water and Sanitation Case Study household interview is conducted by a social households. The subsidy should be capped at worker to collect the information needed to apply some pre-determined subsistence consumption the eligibility filters. This selection process can be level. expensive. In Panama, some initial estimates sug- gested that the interviews could cost as much as Third, the shorter the duration of the subsidy and US$10 per household. To this must be added the the higher its absolute level, the greater the prob- fixed costs of running the subsidy program. ability that the potential loss of the subsidy may act as a disincentive for households to attempt to Since administrative costs do not vary signifi- improve their economic circumstances. This cantly with respect to the size of the subsidy effect, known as the "poverty trap," suggests that given, schemes that pay out very little to each eligibility for subsidies should not be reassessed beneficiary tend to be highly inefficient from an too frequently. In Chile, for example, eligibility administrative point of view. A simulation of is reassessed every three years. administrative costs for different types and illus- trative levels of subsidies in Panama shows that Conclusion for a water consumption subsidy of US$1.50 per month, the administration costs absorb 40 per- The case of Panama shows how useful simula- cent of the total value of the subsidy, whereas for tion techniques can be in informing choices a one-off sewerage connection subsidy of between alternative policy instruments. The US$750 the costs fall to 7 percent. These simula- main lessons from the experience in Panama are: Viewpoint is an open tions illustrate that low value subsidies are hard * Before jumping to the conclusion that a sub- forum intended to to justify in administrative terms, unless the selec- sidy is required. policymakers should examine enicourage tion procedures can be shared and jointly admin- the poverty profile of water utility customers dissemination of and debate onideas, istered across a number of stubsidy schemes. and collect evidence on willingness to pay in innovations, and best relation to the true costs of service provision. practices for expanding Preserving economic incentives * It is extremely difficult to find sound individ- the private sector The views published in this ual or zonal eligibility criteria. However, cal- series are those of the It is important, in determining how high a sub- culating the errors of inclusion and exclusion authors and shoald not sidy to pay, to avoid the perverse incentives that is a helpful way of assessing the options. be attributed to the World Bank or any of its can be created by subsidies. First, a subsidy that * Given significant fixed administrative costs in affiliated organizations. covers the full cost of the service will eliminate subsidy programs, direct subsidies of relatively Nor do any of the con- incentives for the efficient use of water, and is low value are unlikely to be cost-effective. clusions represent ulkl official policy of the likely to create a non-payment habit that may be World Bank or of its difficult to break at a later date. Therefore, full- References Executive Directors or scale subsidies should be avoided. Pavment the coLintries they Foster, vivien, Andres G6mez-Loho, and Jonathan 11a1pern. represent. should also be encouraged by making the sub- Fortlcoming. Designing Direct SLhsidies for Water and Sanitation sidv conditional on the household1 paying its Services: lessons from Panama" Policy Research Working Paper. Ta order additionalWolBak asign,DC copies please call own share. In Chile, water subsidies are capped \\orcLl Bank, Washington, e C 202 458 1111 orcontact at 85 perceint of the bill and are onily disbursed lorthcoming information and Nodeling Issues in )esigning Suzanne Smith, editor, upon proof that the household has paid its share. 'o ntec and Sanitation Subsidv Schemes." Policv Research Working Room FtlK-208, The Piper. World Bank, Washington, D.C. World Bank, 1818 H Street, NW, Washington, Second, a subsidy that applies to all levels of con- D.C. 20433, or Internet sumption may encourage excessive use of the Vivien Foster (ffoster@worldbank.org) and address ssmith7@ worldbank.org. The service. In Argentina a subsidy of utility bills for Jonathan Halpern (jhalperngzworldbank.org), series is also available pensioner households had to be discontinued Latin America and the Caribbean Region, and org/html/fpd/notes/). because consumption levels rose as family, Andres Gomez-Lobo (agomezlo& @ Printed on recycled friends, and neighbors took advantage of the deconfacea.uchilecl). Department of paper lower cost of using utility services in pensioner Economics, University of Chile