WATER P-NOTES ISSUE 5 JUNE 2008 44723 Understanding Demand When Reforming Water Supply and Sanitation: A Case Study from Sri Lanka M any countries are weighing urgent reforms Such deferrals are likely complicate matters to bring safe water supply and sanitation later; indeed, neglecting the interests of consumers (WSS) services to hundreds of millions of has often derailed the reform process. poor city dwellers. Past reforms, unfortunately, have often ignored consumer preferences and percep- tions, resulting in overly optimistic projections of Integrating transaction design the revenue potential of reform projects. When and consumer research-- revenues fall short, private partners may seek to re- negotiate their contract, resulting in tariff increases a cautionary tale and other changes that increase project costs across the board. Such situations can undermine Two proposed PSP transactions in Sri Lanka dem- political commitment to reforms in general and to onstrated that understanding demand and design- private sector participation (PSP) in particular. ing a transaction can proceed in parallel. In 2000, the government of Sri Lanka proposed that private Understanding consumers can help avoid such operators manage WSS services for the town of Ne- situations. Different groups of consumers have dis- gombo, north of Colombo, and along the strip of tinct preferences and perceptions that may influence the coast stretching between the towns of Kalutara their decisions about new water systems. Unfortu- and Galle. nately, studies of consumers' water-related prefer- ences are often deferred because collecting data The intention of the government at the time takes time and costs money. Often there is pressure was to attract local private investors, preferably in to complete reforms quickly--sometimes to take ad- a consortium with international water companies. vantage of a political opportunity--so the necessary The operator would receive a fixed amount for ev- research is not done. In other cases, the challenge ery cubic meter of water sold to customers under a of increasing efficiency and improving governance tariff structure set by the government. The 15-year may seem so daunting that the specific interventions contract contained provisions related to coverage required to make reform beneficial to the poor may targets, tariff structures, service levels, and subsi- be overlooked or consciously deferred. dies. Most of those provisions were based on as- sumptions about demand. This note reports the key messages and findings of "Getting the Assumptions Right: Private Sector Partici- pation, Transaction Design, and the Poor in Southwest Sri Lanka," by Caroline van den Berg, Subhrendu Pattanayak, Jui-Chen Yang, and Herath Gunatilake, published in October 2006 by the World Bank (Water Supply and Sanitation Sector Board Discussion Paper Series 7). Readers may download the com- plete paper from www.worldbank.org/water WATER P-NOTES The plan was to provide 95 percent of resi- dents, including most of the poor, with a 24-hour Table 1. Predicted uptake rates by supply of in-home piped water of a quality that subgroup under private sector met national standards. The use of wells and provision, with connection charge of public taps would be phased out. Considerable US$95 (percent) 44723 investment, as well as tariff increases of up to 100 percent, would be required to meet the target and District Poor Nonpoor to deliver anticipated returns to the operators. At Greater Negombo the time, the monthly tariff for 20 cubic meters Connected 49 64 of water was only a fraction of the cost of supply. Unconnected 32 47 However, fearing political fallout from large tariff Kalutara-Galle increases, the government decided to subsidize the Connected 44 59 operators. The size of the subsidy was to be deter- Unconnected 27 42 mined by the investment required, less any future increases in the tariff. Households were to pay a connection charge of US$95 for their metered connection to the network. households was even lower: 27 percent in Negom- In the Kalutara­Galle coastal strip, the connection bo and 32 percent in the Kalutara­Galle coastal charge was to be paid by consumers; in Greater strip (see table 1). Negombo, it was to be partially subsidized. Several factors affected residents' willingness to connect and pay for piped water. While poverty The clash between assumptions and cost were key, access to alternatives, location, and reality and perceptions also mattered. Households with greater self-provision options--primarily through private wells--showed lower demand, particularly As the project advanced, a team working with the if they were satisfied with their non-network source. government and funded by the World Bank­Neth- The likelihood that residents who were satisfied erlands Water Partnership began an investigation with their current source would not connect to the into the preferences and perceptions of residents new network undermined the goal of near-uni- in the service area. We were part of that team. In versal coverage on which the government's plans 2003, we conducted surveys to shed light on the depended. Uptake simulations showed that remov- characteristics, service levels, preferences, and ing the connection fee would significantly affect willingness to pay of all households in the two households' willingness to connect. Even so, one service areas. The results suggested that rethinking simulation in which poor households would pay the PSP transactions would be necessary. Unfortu- no connection fee and no monthly consumption nately, our findings arrived too late to salvage the charge suggested that some 30 percent of poor, proposed reform. The government abandoned it unconnected households would choose not to be in 2004. We had found that many of the assump- connected. tions underlying the PSP transactions were flawed, jeopardizing the economic fundamentals of the The low potential uptake rates raised questions proposed transactions. Those assumptions are ex- about the financial viability of the PSP transactions. plored below. The government asked for new calculations based on lower demand assumptions. Those calculations suggested that higher operator subsidies and con- The assumption that universal sumer tariffs would be needed. service could be reached Nevertheless, slack demand for the proposed piped water services posed a problem for the gov- Our survey research indicated that consumer ernment because, to the extent the poor opted out uptake would be less than 50 percent at the pro- of the plan, a growing share of the subsidies need- posed connection fee--significantly lower than the ed to sustain the transactions would go to house- anticipated 95 percent. The uptake rate for poor holds that did not need them. ISSUE 5 · JUNE 2008 The assumed preference for who chose to remain connected. The availability house connections of alternative water sources kept the price of piped water in check, reducing the scope for increasing Subsequent research showed that the initial assump- tariffs. tion--that households favor house connections--did Nor was the reform structure effective at target- 44723 not hold. Nonpoor households were more disposed ing the poor. The tariff structure during the study toward house connections, whereas poor house- period significantly subsidized users of less than 20 holds preferred lower service levels, with price likely cubic meters per month. But with only 28 percent a factor. of poor households having a house connection, Our data also showed that creating small net- 72 percent were excluded from subsidies. Mean- works, or "mini-grids," might be a good alterna- while, as 43 percent of nonpoor households had tive, especially for poor households. Allowing the house connections and consumed an average of utility to use household storage tanks could reduce only 19 cubic meters per month, significant subsidy investment costs, especially in areas farther from resources were being captured by the nonpoor the current piped network. We also found that without generating revenues that could be used for 24-hour piped in water was not a perceived neces- cross-subsidies. sity in homes already receiving less than 24-hour The households that received subsidies through piped supply, and that--except for the safest level the very low tariffs in the lower blocks received a of water (drinkable from the tap)--incremental benefit equal to 2.2 percent of monthly income--a improvements in drinking water quality were not significant amount. But altering the tariff structure highly valued. would be unlikely to improve targeting performance, It was clear that different consumers valued because the patterns of piped water consumption of piped water service differently. House connections the poor and nonpoor were similar, and because so and mini-grids were preferred to public taps, but the many of the poor were not connected--and would household's current water supply sources were pre- not choose to connect. ferred to all alternatives. The poor were no different We explored other targeting mechanisms and from the nonpoor in their preferences for consump- learned that the poor were likely to live farther from tion volumes, water quality, safety, and hours of urban centers and from piped water networks. We supply. also found little explicit clustering of the poor--many areas showed a mix of poor and nonpoor. This meant that spatial or geographic factors to iden- Assumptions concerning tariffs, tify poor and nonpoor customer groups could be subsidies, and connection fees only partially successful. Combining geographic and household characteristics to design a subsidy Water tariffs in 2003­04 were low, even for the scheme would likely have been effective, but design- poor, who did not limit consumption to the two ing and implementing such a scheme would have subsidized blocks of the tariff structure. The average increased program costs. reported monthly water bill of the poor represented less than 1 percent of their household spending, much less than the common rule of thumb of 5 Conclusions percent. The government's models had assumed that Several months after we submitted our conclusions the tariff increases required to fund the expansion and recommendations, and quite apart from the programs would not affect consumption of piped details of our findings, the general concept of pri- water. We found, however, that only households vate sector participation in infrastructure ran into that already depended exclusively on piped wa- political opposition within the central government, ter were likely to accept the increases, whereas and the transactions analyzed here were aban- households that had access to other reliable doned. sources of good water were very unlikely to do The fact remains that unsubstantiated assump- so, thereby forcing the tariff still higher for those tions undermined the government's policy. Our WATER P-NOTES research cast doubt on the practicality of universal investment--accordingly becomes more complex, coverage, revealed complexity in consumers' per- because rules of thumb and extrapolations from ceptions of piped water supply (in an environment one place to another cannot be applied without risk of attractive alternatives), and questioned the pov- of error. erty impact of a project that would likely benefit 44723 largely the nonpoor. Given the disparity between Baseline surveys of demand for services can im- assumptions and reality, it would have been very prove the design of WSS reform projects. Findings difficult to ensure the financial viability of the two can improve the design of subsidies, service levels, contracts. and technical standards, strengthening their pro- poor impact and enhancing their acceptability and Although the districts in the proposed service sustainability. Such studies also help set and moni- areas are geographically close, they showed sub- tor progress toward program targets. Often their stantial diversity in consumption preferences--for costs are cited as a disincentive, but they tend to be example, in willingness to pay for piped water low in the context of the overall investment and typi- connections. Water supply and sanitation are lo- cally are less than the services of the various other cal services; hence, one-size-fits-all solutions may advisors required. Overall, they are a small price to fail if local differences are not accommodated. pay for ensuring the viability of large, visible invest- Designing a PSP transaction--or any public sector ments vital to millions of people. 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. THE WORLD BANK | 1818 H Street, NW | Washington, DC 20433 www.worldbank.org/water | whelpdesk@worldbank.org