40832 pproaches A u g u s t 2 0 0 7 N o t e N u m b e r 1 6 Output-Based Aid in Infrastructure A Tool for Reducing the Impact of Corruption Yogita Mumssen and Charles Kenny C orruption in infrastructure leads to big losses. Estimates of the share of construction spend- ing lost to bribe payments around the world range from 5 percent to more than 20 percent. In an Indonesian community-driven road construction project, a fairly extreme example, about 24 percent of expenditures was "lost," largely due to pilfered money and supplies (Olken 2004). But the financial losses from bribe payments are just part of the economic damage from corruption. In the Indonesian road project much of the corruption took the form of theft of construction materials--an act that causes economic damage mostly through its effect on access to improved infrastructure services. The theft of a marginal dollar of materials reduced the discounted benefits of the road to consumers by an estimated $3.41, because building a road with insufficient ma- terials shortens its life. Beyond poor-quality construc- tion, corruption can encourage the construction of the wrong infrastructure--white elephant projects for which there is no demand, constructed solely to extract rents. Solar homes are provided on an OBA basis to the rural poor in Bolivia It is important to reduce the financial cost of corruption by limiting bribe payments. But even more important is to ensure that corruption does not reduce the quantity and quality of paid for their investments and related services only infrastructure provision. Output-based aid (OBA) is a tool that after making the connections and collecting bills for can help achieve these goals. a period of months to the satisfaction of an indepen- dent monitoring agent. Similarly, in isolated parts of rural Bolivia local How does output-based aid work? dealers that are providing poor residents with solar OBA approaches use explicit, performance-based subsidies to help deliver basic services where policy concerns justify public funding to complement or replace user fees. Subsidies are targeted to the poor 1This and other OBA projects discussed here are among and channeled through service providers that are paid the around 100 World Bank­financed OBA projects in largely only after delivering agreed services, or outputs. infrastructure, health, and education that were identified by The aim is to improve the efficiency of public funding an International Development Association discussion paper (IDA 2006). Since the paper's release in October 2006, about through better targeting and greater accountability. 20 more such projects have been developed or identified. In the coastal areas of Colombia, for example, gas utilities are connecting pre-identified poor house- Yogita Mumssen is an infrastructure economist and holds to their networks and providing them with Charles Kenny is a senior economist with the World Bank's appropriate cooking appliances.1 The utilities will be Finance, Economics, and Urban Department. Supporting the delivery of basic services in developing countries pproaches 40832 home systems and maintenance services are paid only projects. Similar examples exist in other sectors. An after delivering the solar panel--and furthermore, with OBA project in Dakar, Senegal, involves the provision of some of the payments withheld until only after the on-site sanitation facilities to poor households. Pay- service provider has demonstrated the development of ments to local artisans constructing the facilities (and sustainable maintenance programs. to the local nongovernmental agency implementing the By linking payment to the delivery of specified ser- scheme) are withheld until after construction and in vices, OBA approaches transfer performance risk to the part until after a period of use. The project also involves service provider. In a "pure" OBA project the provider monitoring demand and use for several years. In rural self-finances service rollout, receiving reimbursement Paraguay several water supply projects award 10-year only after successful delivery is verified. In reality this service contracts to local companies, compensating condition might need to be relaxed somewhat, while them for systems built and connections made through still linking the profitability of provision to outputs. subsidies tied mostly to the delivery of connections. For example, combining output-based payments with Road projects have also used an OBA approach: some up-front support for construction not only would output--and performance-based road contracts and, reduce the risk to providers, but also would increase before them, similar arrangements (such as CREMA a project's affordability and likely success given loan contracts in Latin America). These involve payments tenors and interest rates in many of the environments for road maintenance (and now increasingly for con- where OBA is being piloted. As long as up-front pay- struction) based on kilometers of road maintained to a ments remain sufficiently below the cost of construc- specified standard over a given period. The payments, tion, OBA would still encourage quality construction often monthly, are based on road availability, speed and sustained maintenance of needed infrastructure. attainable, and other parameters that can be easily monitored, including by civil society. Why could OBA limit corruption? OBA arrangements like these cannot entirely pre- clude all acts of corruption (such as collusion). But Why is output-based aid a useful tool for reducing linking payments to the delivery of outputs can have corruption's adverse impact on development? The a powerful impact. For example, an OBA contract for main reason is that by prespecifying the output at a roads that includes construction gives service providers given price and disbursing payment only after delivery a strong incentive to ensure that materials are not sto- is verified, OBA approaches help ensure that corrup- len: if the road they build is of poor quality and cannot tion does not derail delivery of expected results. More- meet service standards, payments will be held back. over, in contrast with inputs, outputs are by definition By providing payment only on the delivery of relatively easy for everyone to monitor--and results are service, OBA approaches can help limit the impact of therefore transparent and measurable. corruption in another way as well: by allowing auditors Another important reason is that OBA approaches and stakeholders time to uncover malfeasance before often use competition. Traditional approaches also funds are disbursed. In a recent World Bank­financed use a competitive process to select construction OBA project in Southeast Asia, for example, allegations contractors and in some cases service providers. But of corruption (vigorously denied by the government) OBA approaches base the competition on the lowest resulted in the project being put on hold and eventually subsidy required to deliver prespecified outputs that canceled. The corruption allegations unfolded before are relatively easy to measure and of tangible benefit to outputs were fully delivered and verified, so the World the user. That helps minimize the corruption that can Bank had not disbursed all related funding. This ex- fester where monopoly power is combined with consid- ample provides an important message about OBA and erable discretion and accountability is lacking. corruption: the risk to service providers from behaving Transparency and accountability corruptly remains high for a longer period, possibly resulting in nonpayment for sunk investments. OBA projects center on outputs that are relatively easy to monitor. The presence or absence of a working Competition electricity connection, for example, is directly experi- Where OBA schemes involve direct competition for enced by households and easily monitored by donors, subsidies, they limit the opportunity for corrupt civil society, or other interested parties--as can be seen agents to drive a wedge between prices and costs so in the Colombian gas and Bolivian rural electrification as to pocket the difference. All OBA schemes in rural Supporting the delivery of basic services in developing countries pproaches 40832 telecommunications, for example, involve international being paid. These services have been repeatedly shown competitive tenders to determine who requires the low- to provide significant benefits--in better health and est subsidy to provide rural communities with access to lower mortality (especially among children)--as well as telecommunications and information technology ser- high economic rates of return (Banerjee and He 2003). vices. Efficiency gains from these competitive tenders The risk of financing white elephants--infrastructure have sometimes been much higher than expected; in with no economic value--is considerably lower than in one case in Latin America competition was so success- traditional aid models. ful that no public subsidy was needed. Similarly, in the rural water scheme in Paraguay the What challenges remain? government subsidy required for the OBA projects was While output-based aid can do much to help reduce in some cases 25 percent less than in a similar non- the development impact of corruption, further gains OBA project. And in the Bolivian example competition may require tackling some remaining challenges. will result in the delivery of substantially more solar home systems than originally expected. Quality Even where subsidies are given to a monopoly pro- A big advantage of the OBA process is the comparative vider, "yardstick competition" based on benchmarks ease of monitoring to ensure that outputs are delivered. is used to ensure that the subsidies are not out of line OBA projects should make full use of this advantage, with national or international norms. The payment for through physical audit, surveys of beneficiaries, and outputs (and related subsidy) for the natural gas proj- oversight by civil society. And to support broad moni- ect in Colombia was determined through benchmarked toring, OBA projects should include an active com- cost allowances provided by the regulator. munication strategy that advertises what services are to be delivered to whom and at what price. While proj- Limits on discretion ect-specific outputs may be easy to monitor, however, Where decisionmakers have considerable and poorly compliance with general regulations or laws governing specified discretion to choose among projects, designs, the sector may be less transparent. Detecting poor con- or bidders, they can use this discretion to channel struction or below-standard delivery can be difficult. financing to corrupt bidders in return for a kickback. The OBA process can limit this discretionary power be- Pricing cause it is less susceptible to corrupt agents designing Pricing exercises are a potential entry point for cor- project specifications to favor a particular bidder. ruption, especially where a project is to involve mo- In some OBA projects, for example, the defined nopoly provision. This is another argument in favor outputs are "technology neutral" (such as the deliv- of competitive bidding, where the subsidy level is set ery of electricity services of a given quantity to a given by a market mechanism. But even with competitive community) rather than technology specific (the selection, there will be an important role for tools to delivery of services using particular equipment). One detect collusion (such as close analysis of the variation such project is a rural electrification scheme in Senegal between bids) and price fixing (including independent that permitted bidders to use whatever technology they monitoring of the bid process and publication of proj- deemed appropriate to deliver electricity services to ru- ect details, bids, and contracts). ral customers. (An additional grant element was used Benchmarking is a particularly powerful tool, both to "level the playing field" for renewable technologies, to help detect collusion in competitive bidding and however, to allow cleaner solutions.) to help set subsidy rates for OBA contracts negoti- ated with incumbent providers. But benchmarking is Development impact a complex and data-intensive process. Use of this tool Output-based aid pays for the delivery of services, at will require much more work to improve databases for competitive prices, that directly benefit consumers and benchmarking, and even then it will need to be comple- are themselves indicators of development progress. So mented by carefully designed project-specific costing even if corruption does occur in an OBA project, the exercises. risk that it will significantly reduce the project's devel- opment impact is lowered. In a water project, for ex- Targeting ample, even if the winning bidder is selected as part of To ensure that services reach the intended beneficia- a corrupt deal, it still must deliver water services before ries, OBA projects may use a mix of precise targeting Supporting the delivery of basic services in developing countries pproaches 40832 (such as through household surveys combined with Measuring success community participation to minimize errors of inclu- How can the effectiveness of output-based aid in reduc- sion or exclusion) and broader geographic targeting ing corruption or its impact be measured? We have (where large shares of the population are poor). But only limited knowledge about the extent of corruption OBA projects still face a risk of mistargeting, which in infrastructure--and even less about its impact on may be exacerbated by corruption. Using other target- development. It is unlikely that we will ever be able to ing approaches--such as self-selection, by subsidizing say with certainty whether OBA schemes involve smaller technologies that only the poor are likely to use (water bribe payments on average than similar schemes using kiosks, for example)--may also minimize the chances traditional approaches, nor could we say with precision for corruption in targeting. what the broader impact on development would be. Ultimately, the targeting exercise needs to take into A far more tractable and useful analysis is one that account not only the costs and benefits of the ap- would allow us to say that OBA projects have a higher proaches used, but also the ability to limit discretionary economic rate of return on average than do traditional powers of officials that may be prone to corruption. approaches, given that payments are only made after Procurement results are delivered. Such an analysis might not result in hard evidence that corruption did not exist--but it Output-based aid cannot completely ring-fence a project would show that the agreed outputs were delivered as from broader governance failures. But as with other expected and for the agreed price, suggesting the devel- types of project financing, a range of transparency and opment impact of corruption is reduced. accountability tools can be used to reduce the potential for corruption in the procurement process. These include References external monitoring, and e-procurement approaches that maximize dissemination of procurement documents Banerjee, Abhijit, and Ruimin He. 2003. "Making Aid and the transparency of the evaluation process. Work." Massachusetts Institute of Technology, But attempts to ring-fence OBA projects should Cambridge, MA. be evaluated with care. One advantage of OBA is its IDA (International Development Association). 2006. clear and direct focus on the delivery of agreed outputs "A Review of the Use of Output-Based Aid Ap- rather than on the procurement of related inputs--and proaches." IDA14 midterm discussion paper. Wash- excessive procurement control would remove that ad- ington, DC. vantage. In short, there may be an efficient level of ef- Olken, Benjamin A. 2004. Monitoring Corruption: Evidence fort that should not be exceeded in trying to eliminate from a Field Experiment in Indonesia. NBER Working corruption in OBA projects. The overriding priority Paper 11753. Cambridge, MA: National Bureau of should be to ensure the greatest consumer benefit from Economic Research. government and donor financing. About OBApproaches OBApproaches is a forum for discussing and dis- The case studies have been chosen and presented seminating recent experiences and innovations by the authors in agreement with the GPOBA for supporting the delivery of basic services to the management team, and are not to be attributed to poor. The series will focus on the provision of water, GPOBA's donors, the World Bank or any other af- energy, telecommunications, transport, health and filiated organizations. Nor do any of the conclusions education in developing countries, in particular represent official policy of the GPOBA, World Bank, through output, or performance,-based approaches. or the countries they represent. To find out more, visit www.gpoba.org The Global Partnership on Output-Based Aid Supporting the delivery of basic services in developing countries