Money for Nothing: The Dire Straits of Medical Practice in Delhi, India Jishnu Das and Jeffrey Hammer Academic and policy discussions about health care increasingly view the quality of care as a crucial determinant of health outcomes. Policy levers to address quality disparities in health care must cover a variety of underlying causes. If geographic variations in practice quality reflect differences in provider competence, the solution lies in geographic incentives for doctors, greater movement of patients, or both. If variations result from differing provider effort, stronger performance incentives are needed. In a new paper, Das and Hammer explore differences in doctor competence and incentives to explain variations in the quality of care in India. A companion paper was published in 2005 in the Journal of Development Economics. Both papers rely on interviews with 205 doctors using hypothetical medical conditions as well as records of what actually happened, and on measures of effort that reflect what doctors do when they show up for work (abstracting from the number of hours they work). Doctors in the public sector are often absent, though less so in urban than rural areas. The authors find that provider competence and effort both play a role in the quality of care. The data indicate that what doctors actually do is very different from what they know they should do. This gap reflects the cost of effort relative to reward and illustrates a response to incentives: doctors paid for each service they provide are closer to their “knowledge frontier” than those paid fixed salaries. Furthermore, effort and competence are complements, not substitutes. More competent doctors spend more time with patients, ask more questions, and do more examinations. Less competent practitioners deliver worse care because of the direct effects of lower competence and the indirect effects of lower effort. The new paper explains that the differences between what doctors know and what they do reveal disparities across institutional settings and geographic locations. Private sector practitioners do more or less what is expected of them by patients. Poorer patients receive low-quality advice and spend a fair amount of money for nothing—on low-value advice and unnecessary drugs. Wealthier patients get better advice, both because they see more competent providers and because their providers put in more effort. But wealthier patients also spend a fair amount of money for nothing, on unnecessary drugs. Public sector practitioners spend less time and effort on patients and do less than they know they should. But this public-private disparity masks variations in the public sector. Public providers in smaller clinics and dispensaries substantially underperform public providers in hospitals; the latter are comparable to private practitioners. Unfortunately, poor people do not really use public hospitals. Over a two-year period about 55 percent of medical visits by members of low- and middle-income households were to private doctors. Just 31 percent were to clinics and dispensaries—and only 13 percent to public hospitals. The public sector spends more than 80 percent of the government’s health budget on salaries for doctors and heavy subsidies to educate them. This is another form of money for nothing: substantial public spending on cursory, poorly delivered health care. The concentration of more competent providers in richer neighborhoods, combined with the low use of public hospitals, implies that poor urban residents are particularly underserved—for several reasons: • The private practitioners they visit have low competence. • They receive worse medical care both due to the direct effects of lower competence and the indirect effects of lower effort. • Weaker incentives in the public sector offset the protective effects of higher competence. Thus the poor receive low-quality care from the private sector because doctors do not know much—and low-quality care from the public sector because doctors do not do much. Poor households are better off visiting less-qualified private practitioners than more qualified public doctors. The findings of Das and Hammer suggest that weak medical services, at least for poor people, are more a function of a lack of incentives for public providers than a lack of competence among private providers. This would tend to indicate that solutions are more likely to come from improving information for consumers and reducing the demand for extensive inappropriate treatment. The authors suggest changing the incentives of public providers rather than increasing provider competence through training. Jishnu Das and Jeffrey Hammer. Forthcoming. “Money for Nothing: The Dire Straits of Medical Practice in Delhi, India.” Journal of Development Economics. World Bank Research Digest: An Effective Way to Disseminate Research Findings François Bourguignon This new quarterly publication aims to improve dissemination of World Bank research in the development community. Knowledge is an essential input to World Bank operations in general and lending in particular—to the extent that the World Bank is sometimes called the “Knowledge Bank.” This strong complementarity between knowledge and lending has sometimes been described as the “lending-learning-knowledge” cycle. World Bank loans to developing countries for projects, programs, and policy reforms provide a unique opportunity to discover what works and what doesn’t in development depending on the format of the interventions financed by the Bank and their context. This knowledge should make World Bank assistance to developing countries more effective and its financing more attractive. Most of the learning part of this cycle occurs through research—hence the importance of research for the Bank. Relying on actual experience, researchers try to understand the relationship between the design of projects or policies and their development outcomes depending on the characteristics of the area or country where the project or policy was implemented. To do so, they rely on advances in knowledge from the research community, to which they sometimes contribute. They also develop new analytical tools that can be used by practitioners when establishing policy diagnostics or monitoring aspects of development. There are many examples of this tight integration of research and operations within the Bank. The industrial pollution projection system, featured in this Research Digest, relies on close collaboration between environmental agencies in developing countries and World Bank researchers. It has been used in several Bank projects and is being applied by several non-Bank users around the world. Poverty assessments are now common practice in many countries. They were developed by Bank researchers in response to a need from practitioners to base poverty reduction policies on deep knowledge of poor populations. The assessments rely on household surveys pioneered by the Living Standards Measurement Study, launched by the Bank in the 1980s. Impact evaluations are the best example of the close integration between World Bank research and operations. Based on comparisons between agents subject to the interventions being evaluated and control groups, impact evaluations permit the attribution of key development outcomes such as poverty, health status, school enrollment, employment, and business creation to the interventions, rather than other factors. Essentially demand driven, the multiplication of these evaluations will help the development community evaluate particular operations and determine what works and what doesn’t in specific areas. These examples show why applied development researchers, in the World Bank and elsewhere, must be willing to listen and observe. They can do so only through direct contact with Bank task managers as well as government counterparts, analysts, and researchers in partner countries. If researchers get much of their inspiration from actual development interventions that they have been directly or indirectly involved in, the reverse must also be true. For the lending-learning-knowledge cycle to work, practitioners must be aware of the findings of research to improve their diagnostics, monitoring, and decisionmaking. This awareness does not require a perfect match between the findings of research and the questions faced by specific practitioners. Often, ideas for reforming a policy or launching a new development program come from learning about different policies and programs in other countries. This public good aspect of research shows the importance of efficient dissemination of results. Researchers, not practitioners, are the main readers of research work. This is understandable: research is a cumulative process, and researchers must somehow incorporate all relevant past research in their work. Practitioners have other obligations and do not have time to regularly browse the growing number of academic journals publishing the most salient research work. Hence the importance of developing more direct dissemination channels that may inform development practitioners on remarkable results that they should not miss. That is the objective of the World Bank Research Digest. The efficiency of dissemination implies tradeoffs between quantity and precision. The choice is between featuring many research papers with succinct descriptions of their findings and fewer papers with longer accounts of their results and implications. The Research Digest takes the second approach, and each quarter will offer six or seven one-page summaries of remarkable World Bank research. The papers featured will be chosen by Bank research managers based on relevance, originality, and importance. (Readers wishing to access the Policy Research Working Papers and their abstracts, our monthly Research E-Newsletter, the Research Highlights of DEC’s Research Group, and other World Bank research publications should visit http://econ.worldbank.org) Development research is a global public good, and the audience of this publication—if it achieves its goal—is potentially the whole development community. Given the mission of the World Bank—to fight poverty and improve the living standards of people in the developing world—its research activity covers practically every aspect of development in its partner countries. It also bears on issues that go beyond national borders, such as international trade, migration, HIV/AIDS, and climate change. Over time, we intend to reflect all this diversity of themes and geographic focus in the Research Digest. We hope that all those interested in the results of development research will find this new publication both easily accessible and helpful. François Bourguignon is Senior Vice President and Chief Economist of the World Bank. Bank Supervision and Corruption in Lending Thorsten Beck, Asli Demirgüç-Kunt, and Ross Levine Banks provide a large share of external finance to businesses around the globe. The Basel Committee, International Monetary Fund (IMF), and World Bank promote the development of powerful supervision agencies to monitor and discipline bank behavior. Yet there are few studies of international differences in bank supervision and how they influence the obstacles that corporations face in raising external finance. There is also little cross-country evidence on which bank supervision policies facilitate efficient corporate finance and which reduce bank corruption. Beck, Demirgüç-Kunt, and Levine empirically assess the relationship between bank supervision policies and the degree to which corruption in lending prevents firms from raising external finance. Three theories of government regulation provide a framework for their findings. The first, known as the supervisory power view, holds that strong official supervision of banks can improve their corporate governance, because private agents often lack the incentives and capabilities to monitor powerful banks. This theory assumes that governments have the expertise and incentives to ameliorate market imperfections and improve bank governance. An alternative theory, the political/regulatory capture view, argues that politicians and supervisors do not maximize social welfare, but rather their private welfare. Thus, if bank supervision agencies have the power to discipline noncompliant banks, politicians and supervisors may use this power to induce banks to divert credit to politically connected firms. Finally, the private monitoring view argues that bank supervision policies should focus on enhancing the ability and incentives of private agents to overcome information and transaction costs, so that private investors can exert effective governance over banks. Beck, Demirgüç-Kunt, and Levine use data on supervision from Rethinking Bank Regulation: Till Angels Govern by Barth, Caprio, and Levine and firm-level data from the 1999 World Business Environment Survey of more than 2,500 small, medium-size, and large firms in 37 countries. The authors estimate the relationship between the degree of corruption—based on the answer to the question, “Is the corruption of bank officials an obstacle for the operation and growth of your business?”—and measures of supervision power (such as the ability to intervene in banks, replace managers, force provisioning, stop dividends and other payments, or acquire information) and of the degree to which regulations require information disclosure by banks and give private creditors incentives to monitor banks (such as whether bank directors and officials face criminal prosecution for failure to disclose information, whether banks must disclose consolidated accounts, whether international accounting firms audit banks, and whether there is implicit or explicit deposit insurance). The authors control for a range of firm- and country-specific characteristics and use instrumental variables to control for endogeneity. The results strongly refute the view that powerful supervision agencies with the authority to monitor and discipline banks facilitate efficient corporate finance. Firms tend to face greater obstacles in obtaining bank loans as a result of corrupt bank officials in countries with stronger supervision agencies than in countries with weaker agencies. The finding that powerful supervision agencies are associated with lower integrity of bank lending provides support for the political/regulatory capture view, which emphasizes that these agencies are prone to capture and manipulation by politicians, regulators, or both. But this conclusion needs to be tempered. Powerful supervision is so strongly correlated with poor national institutions (ineffective government, absence of the rule of law, high corruption) that it is difficult to identify an independent relationship between supervision power and bank corruption when controlling for these institutional traits. Finally, the authors’ findings are consistent with the private monitoring view: bank supervision strategies that focus on forcing accurate information disclosure—and not distorting the incentives of private creditors to monitor banks—facilitate efficient corporate finance. This view recognizes that private agents face substantive information and enforcement costs when monitoring banks, but also that politicians and regulators act in their own interests and not necessarily to reduce market friction. Private monitoring has an especially beneficial effect on the integrity of bank lending in countries with sound legal and administrative institutions. Bank supervision clearly matters. Policies that redress market failures by forcing accurate disclosure of information reduce the obstacles that firms face in raising external finance, and active bank supervision can help reduce information costs and improve the integrity of bank lending. Powerful supervision agencies too often do not act in the best interests of society. Thorsten Beck, Asli Demirgüç-Kunt, and Ross Levine. Forthcoming. “Bank Supervision and Corruption in Lending.” Journal of Monetary Economics. Who Benefits from Residential Water and Electricity Subsidies? Kristin Komives, Vivien Foster, Jonathan Halpern, and Quentin Wodon Utility subsidies are often seen as a way to expand coverage and make services affordable to poor people. Yet a new study by Komives, Foster, Halpern, and Wodon shows that such subsidies tend to benefit the middle class and well off. The study examines 32 subsidy programs from 13 water utilities and 27 electricity utilities in developing countries. Most of the programs involve quantity-based subsidies (which are equally common for water and electricity)—that is, basing service prices on consumption. The study decomposes determinants of targeting performance to examine varying access, consumption, and targeting performance across rich and poor populations. Explicit utility subsidies come in two forms: those that cut the cost of consumption and those that reduce connection charges. Consumer subsidies are a feature of water and electricity services around the world. Most water utilities charge tariffs far below levels required to recover costs. Nearly 40 percent do not even cover operation and maintenance costs. Average water tariffs in low-income countries are about a tenth of those in high-income countries. For electricity, about 15 percent of utilities have average tariffs that are too low to even cover operation and maintenance costs. The study’s results are sobering. The quantity-based subsidies practiced by four-fifths of water and electricity utilities surveyed are starkly regressive. Poor households capture half as much of the value of the subsidy as they would if the subsidies were distributed randomly across the entire population—while many poor households are excluded altogether because they are not connected to the network. Only 2 of 26 quantity-based subsidy programs come close to achieving even a neutral subsidy distribution. All the rest are regressive. This means that all or nearly all residential customers receive a subsidy—and richer consumers receive more than poorer ones. In most water programs studied, the poorest 40 percent of the population receives only 5–20 percent of subsidy benefits. Yet for poor households that do benefit, utility subsidies represent 3–4 percent of household income. Why do quantity-based utility subsidies perform so poorly? • Poor households are less likely than others to be connected to networks. Households that lack access to services cannot benefit from consumption subsidies. • Differences in consumption between poor and better-off households are less than is often assumed, so using quantity as a basis for targeting is ineffective. • Even when poor households do consume less than better-off ones, tariff structures often fall heaviest on those who consume the least. Households that consume small quantities may face much higher unit prices than do larger consumers. Moreover, failure to apply full cost recovery tariffs to households consuming at higher levels means that even large consumers receive substantial benefits from subsidies. Quantity-based subsidies are not the only type. Many utilities offer connection subsidies, targeted by geographic or proxy means-test criteria. In addition, self-selection is used to target subsidies for lower-quality services such as public standpipes. These schemes perform somewhat better than quantity-based ones. The geographic schemes examined present an almost neutral distribution, while proxy means-testing and self-selection present relatively progressive distributions. Still, these schemes exclude a substantial proportion of poor people—due to low utility coverage for this group. The large difference in access between poor and nonpoor households suggests that connection subsidies are a better way to reach the poor. Assuming that all unconnected households are offered and accept subsidized connections, the distribution of benefits from connection subsidies is nearly always progressive—meaning that lower-income households receive more subsidies than higher-income ones. But this finding assumes that, when introducing connection subsidies, unconnected households at each income level will connect at the same rate. That assumption is unlikely to hold, because utilities often face constraints in expanding their networks into poor areas. And even where networks are already present, many poor households face nonfinancial obstacles to connecting—such as not having legal title to the property they occupy. The study notes that despite large consumption subsidies, only 20–30 percent of poor households in Africa connected to utility networks even when they were available. Unless those rates are dramatically improved, connection subsidies will continue to disproportionately benefit nonpoor households, particularly in Sub-Saharan Africa. Because most water and sanitation subsidies are captured by the nonpoor, most households would see higher bills if subsidies were eliminated or restructured. But the bulk of any price increase would be paid by nonpoor households. The study concludes that there is potential for improving utility subsidies, whether by switching from quantity-based to more sophisticated approaches or by making more use of connection subsidies. Kristin Komives, Vivien Foster, Jonathan Halpern, and Quentin Wodon. 2006. “Water, Electricity, and the Poor: Who Benefits from Utility Subsidies?” World Bank, Washington, D.C. Child Labor and Agriculture Shocks Kathleen Beegle, Rajeev Dehejia, and Roberta Gatti Crop shocks significantly increase child labor and decrease school enrollment—but households with assets can almost entirely offset these effects What are the links between crop shocks, household assets, and child labor? In a new article, Beegle, Dehejia, and Gatti analyze the extent to which income shocks increase child labor and decrease enrollment, and whether household assets mitigate the effects of these shocks. Child labor has traditionally been viewed as a consequence of poverty, but some recent country studies have called that assumption into question. The article examines the role that child labor plays as a buffer against crop losses—and suggests that insurance or access to credit might reduce its extent. The authors also explore how assets affect household responses to income shocks such as crop losses. As such, the article relates to recent development research examining how credit constraints affect child labor. Child labor entails a tradeoff for households between immediate benefits and, to the extent it interferes with the development of a child’s human capital, potential long-run costs. When faced with a transitory shock, households would ideally use asset holdings—either as a buffer or as collateral to obtain credit—to offset it. The article also relates to the broader development literature on the permanent income hypothesis and consumption smoothing. If households succeed in smoothing their consumption but lack buffer stocks or are credit constrained, they must use other mechanisms to cope with income shocks. Using four rounds of household panel data from the Kagera region of Tanzania, the authors find that income shocks increase child labor by 50 percent. Households in Kagera use almost no purchased inputs and rudimentary technology, and wage labor is limited. When hit by a shock, households tend to increase their use of child labor—typically by having children substitute for adults in household activities such as gathering firewood and water. The data also show that shocks cause a 20 percentage point drop in school enrollment (compared with an average enrollment rate of 70 percent). Households with assets can offset income shocks to a substantial degree: at the mean level of asset holdings, households can offset more than 80 percent of the effect of income shocks. The article finds that household assets decrease in response to shocks but that wealthier households draw down assets to a lesser extent—suggesting that they may be borrowing in response to shocks. The results also suggest that poorer households use assets as a buffer stock, drawing them down in times of need, whereas wealthier households’ behavior is consistent with access to credit. Beegle, Dehejia, and Gatti present a range of evidence to corroborate this interpretation of their results. They show that households are more likely to take loans when they experience a shock and when they hold durable assets. Moreover, the probability of taking a loan in response to a shock is higher for households with durable assets. But the authors acknowledge that there are different possible interpretations of their results. The effect of shocks on households could be explained by myopia or by an extremely high discount rate relative to the interest rate. Still, the authors believe that they offer the most plausible interpretation of their findings. Regardless of how the empirical results are interpreted, the fact remains that households increase child labor in response to crop losses. Child labor is a major policy problem—not only for moral reasons but also because it slows the accumulation of human capital and is inimical to development. The findings of this article imply that child labor could be reduced with policies that insure agricultural households against crop losses and other shocks. Increasing household access to credit in response to crop shocks would also reduce child labor and raise household welfare. Kathleen Beegle, Rajeev Dehejia, and Roberta Gatti. Forthcoming. “Child Labor and Agriculture Shocks.” Journal of Development Economics. Market Access for Sale Hiau Looi Kee, Marcelo Olarreaga, and Peri Silva Has foreign lobbying by exporters from the Western Hemisphere improved their access to U.S. markets? For many countries foreign lobbying is likely to be associated with preferential rather than nondiscriminatory access to markets—for two reasons. First, there are stronger incentives to lobby for preferential access. Exporters expect to grab market share away from exporters from other countries, not just from domestic producers in the importing country—as under a most favored nation (MFN) tariff reduction. Second, developing countries have played little role in multilateral trade negotiations but have benefited from several bilateral and unilateral preference schemes when exporting to developed markets. The recent proliferation of preferential agreements between the United States and its trade partners in the Americas indicates that their foreign lobbying is likely to focus on preferential access. In the early 1990s only about 15 percent of Latin American exports entered the United States under preferential schemes. By the early 2000s that share was 50 percent. Kee, Olarreaga, and Silva explore the role of foreign lobbying in obtaining tariff preferences. They first develop a simple framework, based on a model developed by Grossman and Helpman, to explain the extent to which lobbying by foreign firms affects their preferential access to U.S. markets. They then estimate an equilibrium contribution schedule for 34 countries in the Western Hemisphere using data on foreign lobbying expenditures from the U.S. Department of Justice and on tariff preferences at the industry level from the International Trade Commission. The authors find that foreign lobbying contributions explain variations in U.S. tariff preferences across products and countries at a statistically significant level. When setting preferences, the U.S. government puts five times more weight on foreign lobbying contributions than on forgone tariff revenues. These estimates suggest that market access is for sale—and foreign firms are buying it. These striking results contrast with evidence on domestic lobbying. In a 2004 review of empirical approaches to the political economy of trade policy, Gawande and Krishna conclude that the weight granted to domestic producers and their lobbying in the U.S. government’s objective function is close to the weight granted to social welfare. Kee, Olarreaga, and Silva offer two explanations for the difference between their estimates and those in the earlier literature. Their article assumes that contribution functions are not differentiable, so the Nash equilibrium solution to the lobbying game differs from the one in the Grossman and Helpman article. In this setup the government is left on its participation constraint, whereas in Grossman and Helpman’s work the government grabs part of the lobbying rent. This additional restriction implies that smaller amounts of foreign lobbying contributions will be sufficient to move the government away from welfare maximization, which may explain larger tariff preferences. The authors also assume that the government grants different weights to tariff revenues than to other components of social welfare in its objective function. If the weight put on tariff revenues is lower than the weight put on other components of social welfare, that could explain the difference between the authors’ estimates and those in the rest of the literature. Hiau Looi Kee, Marcelo Olarreaga, and Peri Silva. Forthcoming. “Market Access for Sale,” Journal of Development Economics. Public Disclosure —A Tool for Controlling Pollution Susmita Dasgupta, Hua Wang, and David Wheeler Though the research in this article is not new, it has had a major impact on policy in several Asian countries in recent years—offering lessons from a policy experience with a long time horizon Most pollution regulation is still based on the quantity of emissions. Market instruments are rarely used, but public disclosure of pollution has shot up since the 1980s. Public disclosure has particular appeal in developing countries, where corruption and weak enforcement often make it difficult for regulators to control pollution by more conventional means. Although weaknesses in conventional regulation open the door for community pressure through public disclosure, public interpretations of emissions data can be hampered by low education levels and a scarcity of technically informed nongovernmental organizations (NGOs) to help with interpretation. The World Bank has worked on this issue with environmental agencies from several developing countries, with the goal of developing and evaluating environmental performance ratings. Indonesia began its disclosure program in 1994, followed by the Philippines’s EcoWatch in 1997, China’s GreenWatch in 2000, Vietnam’s Environmental Information Disclosure System for Hanoi in 2002, and two programs in Uttar Pradesh, India, in 2002 and 2004. After 10 years of collaborative work, the Bank has learned several lessons about this kind of policy experimentation and evaluation. First, counterpart institutions value the Bank’s contributions—particularly its strong policy orientation, extensive field experience, knowledge of related initiatives in other countries, analytical skill, and, perhaps most important, capacity to sustain an innovative vision while engaged in the complex institutional, technical, and political activities that always surround new policy initiatives. Second, a successful transition from pilot experimentation to large-scale implementation depends on several factors: credible demonstration of impact; widespread dissemination of information about the pilot program, its rationale, and results; and sustained technical assistance delivered to collaborating institutions, both during the pilot phase and in scaling up. Third, long time horizons are critical for this kind of work. The full development cycle, from catalytic research to national scale-up, may take 5–10 years. Finally, flexibility is essential. Successful collaboration to promote policy reform does not happen smoothly, predictably, and on a fixed schedule. It requires constant attention to relationships, operational details, and emerging opportunities, as well as consistent application of high technical standards and analytical skills. The environmental performance rating programs adopted by Indonesia, the Philippines, China, Vietnam, and India include the following elements: • They focus on conventionally regulated pollutants rather than toxic chemicals. • They use locally recognized benchmarks to grade performance. • They compare audited factory emissions reports with benchmark standards and grade the results in a few overall categories. • They assign locally understood color codes or symbols to performance grades. • Performance rating agencies maintain close communications with audited facilities. Table 1 summarizes evidence on the impacts of these collaborative disclosure programs in Indonesia, the Philippines, China, and Vietnam. In all four countries the programs strengthened conventional regulation by significantly increasing compliance. Susmita Dasgupta, Hua Wang, and David Wheeler. 2005. “Disclosure Strategies for Pollution Control.” In T. Tietenberg and H. Folmer, eds., The International Yearbook of Environmental and Resource Economics 2005/2006: A Survey of Current Issues. Cheltenham, U.K.: Edward Elgar. Table 1: Performance Rating Programs: Changes in Polluters’ Compliance Status Number Share of factories of factories (percent) Increase in compliance Country, program Noncompliant Compliant Noncompliant Compliant (percentage points) Indonesia, PROPER 1995 92 54 63 37 1997 57 89 39 61 24 Philippines, EcoWatch 1997 48 4 92 8 1998 19 26 42 58 50 China, GreenWatch Zhenjiang 1999 23 68 25 75 2000 14 77 15 85 10 China, Greenwatch Hohhot 1999 43 13 77 23 2000 21 35 38 62 39 Vietnam, Hanoi 2001 45 5 90 10 2002 38 12 76 24 14