WATER P-NOTES ISSUE 20 OCTOBER 2008 46472 Water, Electricity, and the Poor: Who Benefits from Utility Subsidies? U tility subsidies to consumers of water and elec- countries are a tenth of the level applied in high- tricity services are often justified as a mecha- income countries, while average electricity tariffs in nism for making services affordable for the low-income countries are half that level. poor. After all, an estimated 1.1 billion people in the developing world lack access to safe water; 2 Subsidies provided to utility customers can be billion are without electricity, and 2.4 billion without a significant drain on the public treasury. The most sanitation. But critics object that such subsidies can striking examples come from the countries of the work against improving quality of service to existing former Soviet Union, and electricity subsidies exceed consumers and extending access to unconnected 1 percent of GDP in India, Mexico and Argentina. households. Financially strapped utilities are often Public subsidies for drinking water and sanitation inefficient, provide low-quality services, and lag be- tend to be smaller as a percentage of GDP because hind in expanding networks. total system costs are lower. In India, drinking wa- ter subsidies have been estimated at 0.5 percent During the 1990s, experts urged that water and of GDP. In contrast, proportion of costs recovered electricity services should charge enough to fully through user charges tend to be much lower for wa- cover costs. Households could spend 10­50 per- ter than for electricity. cent more on water and electricity without major ef- fects on poverty levels, but in many countries much There are two broad means by which utilities larger price increases are needed to recover costs. subsidize their customers--consumption subsidies A substantial proportion of the population of lower- and connection subsidies. Consumption subsidies income countries may find it difficult to pay the full reduce charges for consuming water or electricity, cost of services. while connection subsidies are one-time reductions in connection charges. A survey of tariffs in Latin America, Africa, and Asia found that three-quarters How prevalent are utility of utilities in both sectors include some form of con- sumption subsidy. subsidies? The most common form of targeted subsidy is Consumer utility subsidies for water and electric- based on volume of water or electricity consumed. ity services are common in the developing world. Increasing block tariffs (IBT)--stepped tariffs that Global surveys indicate the majority of electrical charge an increasingly higher price per unit to all utilities and water utilities charge tariffs that often consumers as their consumption increases into don't even cover operating and maintenance costs subsequent blocks--are common. Some utilities (see table 1). Average water tariffs in low-income (primarily electricity) apply a different kind of quan- Excerpted from Water, Electricity, and the Poor: Who Benefits from Utility Subsidies? by Kristin Komives, Vivien Foster, Jonathan Halpern, and Quentin Wodon, and Roohi Abdullah (Washington, DC: World Bank, 2005). The publication is available in PDF format from www.worldbank.org/water WATER P-NOTES Table 1. Most water utilities in the developing world do not recover their full costs Probable degree of cost recovery by water utilities, by country income level and region Percentage of utilities whose average tariffs appear to be...a Grouping of Too low to Enough to cover Enough O&M and water utilities cover basic O&M most O&M partial capital Global 39 30 30 By country income level High-income 8 42 50 Upper-middle-income 39 22 39 Lower-middle-income 37 41 22 Low-income 89 9 3 By region OECD 6 43 51 Latin America and Caribbean 13 39 48 Middle East and North Africa 58 25 17 East Asia and Pacific 53 32 16 Europe and Central Asia 100 0 0 South Asia 100 0 0 Sources: ADB 2004; ADERASA 2005; GWI 2004; NIUA 1999. Note: Data are drawn from utilities serving 132 major cities worldwide, broken down geographically as follows: OECD, 47; South Asia, 24; Latin America and Caribbean, 23; East Asia and Pacific; 19; Middle East and North Africa, 12; Europe and Central Asia, 6. The same group of countries is broken down by income group as follows: high-income, 52; upper-middle-income, 18; lower-middle-income, 27; low-income, 35. O&M = operation and maintenance. a. Based on GWI 2004 (box 2.4). tity-based subsidy, termed volume-differentiated tar- How well do subsidies perform? iffs (VDT), composed of two or more different tariffs, the first highly subsidized and the second less or not Recent individual case studies which sought to assess at all, to which consumers are assigned based on the extent to which subsidies benefit poor households consumption volume. are not directly comparable (differing coverage, Another form of targeting uses administrative methods and metrics). This study addresses this gap decisions to determine subsidy eligibility, choosing by systematically examining targeting performance for example all customers in a particular demo- of a large number of consumer utility subsidies. graphic or employment category (e.g. pensioners or The sample includes 45 subsidy programs from 27 war veterans), those living in a particular neighbor- electrical utilities, and 32 programs from 13 water hood or region (geographical targeting), or those utilities. Most of the programs involve quantity-based whose means are determined to fall below a certain subsidies, which are equally common in water and level. Geographic targeting is sometimes combined electricity sectors. There are 12 cases from Latin with an IBT or VDT. America, 7 from Eastern Europe, 4 from South Asia, and 4 from Sub-Saharan Africa. No case study ma- The Chilean water subsidy program is a widely terial was available for East Asia or the Middle East. cited example of a means-tested consumer utility subsidy, subsidizing between 40 and 70 percent Quantity-based subsidies (ie., through the of up to 15 cubic meters of water for poor house- tariff structure) employed by 80 percent of water holds. Utilities apply this discount to the water bills and electricity utilities are starkly regressive. Poor of eligible households and are reimbursed by the households capture only half as much value of the government. subsidy as they would if subsidies were randomly distributed across the entire population. Many poor 2 ISSUE 20 · OCTOBER 2008 households are excluded from subsidy programs altogether, because they are not connected to the Figure 1. Quantity-based subsidies network. are not effective in targeting the poor A targeting performance indicator of 1.0 indicates an income-neutral subsidy Measuring subsidy performance 120 The study constructed a targeting performance in- (%) 100 Hungary (S) dicator () representing the share of subsidy ben- 80 India, State IBTs households efits received by the poor divided by the proportion subsidy 60 Guatemala Kathmandu Bangalore, India of the population in poverty. A value of 1.00 for poor 40 Honduras São Tamé and Peru implies that distribution of the subsidy across of Sri Lanka 20 Principe Cape Verde Cape income classes is neutral, with the benefit share receiving Rwanda (S) Share 0 Verde going to poor households equal to their share of 0 0.2 0.4 0.6 0.8 1.0 1.2 the population. Neutral targeting means that the Benefit targeting performance indicator ( ) subsidy performs no better than a random subsidy Electricity Water that delivers equal benefits to all. A value greater Sources: Appendices D.1, D.6, E.1, and E.6. than 1.0 implies that the subsidy distribution is Note: S=simulated progressive, with the poor receiving a larger share of the total benefits than their share of the total population. A W value below 1.00 indicates a re- tion between the poor and the nonpoor--particular- gressive subsidy. ly between the poor and the middle class--are less than they are often assumed to be. The "poor" in this study are defined as the poorest 40 percent of households. The targeting Also, various common features of tariff struc- performance indicator is the product of five fac- tures fall heaviest on those who consume the least. tors: rates of access to the utility network, uptake Fixed charges mean that households that consume rates among those having access, targeting rates small quantities face higher unit prices than larger among those with connections, the rate of subsi- consumers and that large consumers benefit sub- dization enjoyed by those targeted to receive the stantially from subsidization. subsidy, and the quantity of subsidy consumed by An appealing solution to this problem is to recipients. modify tariff structures so that non-poor households are transformed from subsidy recipients into net cross-subsidizers, for example by reducing the size Quantity-based subsidies: of the subsidized first block of an IBT or reducing nearly always regressive the subsidy threshold of a VDT, while raising the rate charged in the unsubsidized portions of the tar- Almost invariably, the nonpoor benefit dispropor- iff to more than average cost. tionately from quantity-based subsidy programs. Only 2 of 26 quantity-based subsidy cases come close to achieving even a neutral subsidy distribu- Other consumption subsidies: tion; the rest are all regressive or highly regressive less regressive (see figure 1). In one water program, the poorest 40 percent of the population received just 10 per- Geographical targeting raises on average to cent of the benefits. An electricity program in Cen- 0.99--a substantial improvement over quantity- tral America distributed just 8 percent of benefits to based targeting but still no better than a random the poorest 40 percent of the population. distribution of subsidies.1 The effective application of IBTs and VDTs re- quires households to have functioning meters to 1 measure water and electricity use, which the poor The only exception is categorical targeting (such as discounts for pensioners or veterans), which proved in this study to be regres- are less likely to have. And, differences in consump- sive in all cases considered. 3 WATER P-NOTES Means testing has the greatest effect on improv- face nonfinancial obstacles to connecting (such as ing targeting performance, with taking a strongly not having legal title to the property they occupy). progressive average value of 1.31. However, the gains in targeting accuracy can also exclude a high When only 50 percent of the unconnected poor percentage of intended beneficiaries (low participa- households connect, benefit targeting becomes re- tion rates). Moreover, means-testing systems can be gressive. Even so, the targeting performance of the costly to implement and administer if done solely on simulated connection subsidies is better than the a sectoral basis. In contrast, in countries with exist- consumption subsidies. ing means testing systems, the incremental cost of Also, despite sizable consumption subsidies utilizing them to allocate of utility subsidies is very only 20­30 percent of poor households in the Afri- modest. can case studies connected to utility networks even Service-level targeting is another alternative. when networks were available, making connection Our sample of subsidy cases includes two examples subsidies not progressive there. in the water sector--public tap subsidies in Ban- galore and in Kathmandu. But subsidy value is relatively small and errors of exclusion are high, be- Overcoming the limits of cause more poor households (in those cities) have subsidies: an action agenda private taps than use public taps. Because utility subsidies address only price they are best seen as one part of a package of measures to Connection subsidies: a better ensure access to utility services for the poor. Other way to reach the poor? measures may be more promising: · Reducing cost of service, through efficiency Connection subsidies may be more effective in gains in operating and capital expenditures and reaching the poor. Our simulations assumed all by improving revenue collection. unconnected households were both offered and ac- cepted subsidized connections. Distribution of ben- · More frequent billing and elimination of efits from these universal subsidies is progressive. minimum consumption requirements and fixed If connection subsidies are further combined with charges. Other options include prepayment, fi- geographic targeting or means testing, simulations nancing connection costs, and providing devic- suggest that could be raised to 1.30 or 1.71, on es that help households control consumption. average respectively. · Easing legal restrictions that work against ex- However, the simulations results are based on pansion of services to the poor, such as techni- the assumption that with subsidies, unconnected cal norms that oblige utilities to use high-cost households at each income level will connect at the technologies, legal-tenure requirements that same rate. In practice, that is unlikely, because utili- prevent services from being extended to periur- ties often face constraints in expanding their networks ban neighborhoods, and regulations restricting into all geographic areas and many poor households services provided by small-scale providers. The Water Sector Board Practitioner Notes (P-Notes) series is published by the Water Sector Board of the Sustainable Development Network of the World Bank Group. P-Notes are available online at www.worldbank.org/water. P-Notes are a synopsis of larger World Bank documents in the water sector. 4 THE WORLD BANK | 1818 H Street, NW | Washington, DC 20433 www.worldbank.org/water | whelpdesk@worldbank.org