Report No. 15296 Report on the World Bank Research Program Fiscal 1994 and 1995 january 1996 Office of the Senior Vice President Development Economics and Chief Economist FOR OFFICIAL USE ONLY ,jt scloed tou oil a on-i 0 nce p-2 The World Bank Washington, D.C. 20433 U.S.A JAMES D. WOLFENSOHN President January 26, 1996 MEMORANDUM TO THE EXECUTIVE DIRECTORS Subject: Bank Group Research Program The knowledge, ideas, and policy advice of the Bank complement the financial resources that it deploys to promote economic development in its member countries. The Bank's research program seeks to expand our knowledge and provide intellectual leadership to the development process, strengthening the quality of the Bank's lending operations and policy advice. The linkages between the Bank's research and operations are very important and are addressed on an ongoing basis. But to be effective, Bank research must also reach beyond the Bank to engage the broader development community and affect the global debate on development issues. These external linkages comprise the theme of this report. The external impacts and linkages of Bank research are examined in Part 1. A variety of evidence is presented concerning the effect of Bank research on policy makers, analysts and researchers, students, and the development community. Part 2 of the report reviews the main trends and findings of Bank research over the past two fiscal years. It reports on the overall size, composition, and coverage of research and summarizes findings for the major thematic areas of the research program. This section also presents the results of evaluations of the Bank journals, the Annual Bank Conference on Development Economics, and completed research projects. Future research themes and directions are outlined in Part 3. Although the broad coverage of research will continue, several new focal points are emerging that are linked by a common concern with institutional capacity and political economy. I believe that a vital and effective research program is crucial to the future success of the Bank. I look forward to your views on the performance of the research program and its suggested future directions. s REPORT ON THE WORLD BANK RESEARCH PROGRAM FISCAL 1994 AND 1995 Research Advisory Staff Washington, DC Bank research-objectives and definition The Bank's research program has four basic goals: * To support all aspects of Bank operations, including the assessment of develop- ment progress in member countries. * To broaden understanding of the development process. * To improve the Bank's capacity to give policy advice to its members. * To assist in developing indigenous research capacity in member countries. This Report's definition of research, like that used in recent years, encompasses analytic work designed to produce results with wide applicability across countries or sectors. Bank research, in contrast to academic research, is directed toward analyzing recognized and emerging policy issues and improving the quality of policy advice. Although motivated by policy problems, Bank research addresses longer-term concerns rather than the immediate needs of a particular Bank lending operation or of a particular country or sector report. Its definition does not, therefore, include the economic and sector work and policy analysis carried out by Bank staff intended only to support specific operations in particular countries. Economic and sector work and policy studies take the product of research and adapt it to particular projects or country settings, whereas Bank research contributes to the broader understanding of development processes that underlies future lending operations and policy advice. Both activities-research and economic and sector work-are critical to the design of successful projects and effective policy. CONTENTS Executive summary External influence of Bank research Bank research during fiscal 1994 and 1995 i New research themes and directions . i Part 1 Bank research in the marketplace of ideas 1 Reaching many audiences 1 Influencing policy 3 Influencing the wider research community 5 Contributing to economics training 7 Reaching the development community 8 Conclusion 8 Part 2 The World Bank research program in fiscal 1994-95 9 Research funding and management 9 Fiscal 1994-95 research projects and findings 11 Evaluation of research programs and projects 28 Part 3 Research themes and directions 33 New challenges in development 33 Focal points for research 34 Boxes 1 Addressing pension problems 4 2 Improving road maintenance in Africa by improving institutions 4 3 Projecting industrial emissions and costs of abatement 5 4 Indonesia's public disclosure program for pollution control 6 5 Learning to measure health and education effects 13 6 Training: Does it work and does it pay? 15 7 A sharp focus on the big question 17 8 Operations questions, research findings, policy changes: The cycle of learning 19 9 Does the invisible hand reach far enough into financial markets? 21 10 Expanding capacity and communication in African institutions 23 11 Can the forests return? 25 12 Watts, water, and waste: Infrastructure services in developing countries 27 Figures I The annual number of Bank publications grew briskly, then stabilized 1 2 The number of Bank-authored journal articles has nearly tripled since 1981 3 3 Bank-authored journal articles receive more citations than the average 6 4 Bank books are the most frequently used Bank-authored outputs on reading lists 7 5 Research has declined as a share of expenditures 9 6 DEC's share of research activity has increased... 10 7 ...and the region's share of RSB funding has grown 10 8 The share of smaller RSB-funded projects has risen 10 Tables 1 Most widely distributed Bank publications in fiscal 1994 and 1995 2 2 The 10 most frequently read Bank publications 2 3 Percentage of magazine references to World Bank research, statistics, and lending 8 Notes 37 Appendix tables 41 1 Bank research in relation to other Bank analytical work and the administrative budget, fiscal 1990-95 43 2A Bank resources devoted to research, by department, fiscal 1994 44 2B Bank resources devoted to research, by department, fiscal 1995 45 3A Bank resources devoted to research, by program objective, fiscal 1994 46 3B Bank resources devoted to research, by program objective, fiscal 1995 47 4 External funding for Bank research, by program objective category, fiscal 1994-95 48 5 External funding for Bank research, by management unit, fiscal 1994-95 49 6 RSB-funded new research projects, by project size, fiscal 1989-95 50 7 The World Bank Visiting Research Fellows Program: Research fellows at the Bank during fiscal 1994-95 51 8A Research Support Budget-funded research, by program objectives, fiscal 1994 52 8B Research Support Budget-funded research, by program objectives, fiscal 1995 56 EXECUTIVE SUMMARY The value-added of the World Bank in many ways research reports among the 10 most widely distrib- derives from its ideas, knowledge, and experience. uted Bank publications in the past two years. A Thus the World Bank is a knowledge-based insti- recent readership survey indicates that the Bank's tution as well as a development institution and a most widely distributed publications are also the financial institution. The Bank's knowledge base most frequently read. Bank research is also dis- comes indirectly from many sources, but it is the seminated through many non-Bank publications; direct product of research, the one Bank activity Bank authors published nearly 300 articles in pro- whose main objective is to expand knowledge fessional journals in each of the past two fiscal about development. Because research and know- years. In a two-way transfer, Bank research also ledge are crucial to the success of the Bank's mis- draws much expertise from the external research sion, the performance, outcome, and likely future community. A third of Bank research projects in- direction of research at the Bank are important volvelocalcounterpartresearchinstitutesandcon- matters for review. This Report reviews these re- sultants. search concerns for fiscal 1994 and 1995. Bank research influences policy in client coun- The previous research report, covering fiscal tries through the Bank's policy papers, economic 1992 and 1993, summarized Bank efforts to in- and sector work, and summaries of best practice, crease research capacity in developing countries as well as through research publications them- throughregionalcapacitybuildingprograms,con- selves. Case studies of policy change in pension ferences, collaborative research, and training-all reform, road maintenance in Africa, and emissions of which continue. This Report attempts to assess reduction programs in several countries provide the influence of the Bank's research program on strong evidence of the many linkages from re- those outside the Bank, using quantitative perfor- search to policymakers, and these are just a few of mance indicators and qualitative approaches, in- many possible examples. The linkages work in cluding case studies. It also reviews changing pat- both directions. The Bank's research agenda res- terns of research funding in the Bank, discusses ponds to new policy issues because researchers are external evaluations of research programs and actively engaged in operational work; indeed, many projects, highlights selected findings of research regional staff carry out research. projects completed in the past two years, and More quantitative measures can also be used as identifies likely future directions of Bank research. indicators of the impact of Bank research. Analyses of the number of citations of Bank work in journal External influence of Bank research articles show that Bank-authored journal articles Assessing the impact of research is inherently dif- are cited from 10 to 50 percent more than the ficult, but much can be learned froLI the effort. averageforeconomicsarticles.Articlesinthe World Available evidence indicates that Bank research is Bank Economic Review and the World Bank Research broad in its coverage of topics, influential among Observer are cited more than articles in any other policymakers, well regarded by other researchers, development economics journal; they rank near heavily used as teaching material in development the thirtieth percentile of all economics journals by economics courses, and widely disseminated in a this measure. The citation rate of papers in the variety of publications produced by the Bank and Proceedings of the World Bank Annual Conference on others. Development Economics exceeds that of the Bank Bank research publications are among the most journals. widely distributed of Bank publications-with the Bank-authored studies, particularly Bankbooks, World Development Report (WDR) continuing as the are well represented on reading lists in university Bank's flagship publication and the first two policy courses in economic development. In a recent pub- ii Report on the World Bank Research Program, Fiscal 1994 and 1995 lished collection of 25 (mostly graduate level) course Bank research covers a range of topics nearly as reading lists, one-sixth of the entries were by Bank wide as Bank operations. Part II of this Report authors. The Bank's Economic Development Insti- summarizes the results of many research projects. tute draws heavily on Bank studies in its courses, In recent years economic management and pov- workshops, and seminars-attended by more than erty and human resource issues have been the 6,600 people in fiscal 1995. Bank research also leading subjects of research, followed by private receives press coverage, which extends its dis- sector development. semination to the general development commu- Research programs and projects financed from nity. The Financial Times and The Economist are theResearchSupportBudgetareperiodicallyevalu- most likely among newspapers and magazines to ated by outside experts in a review process man- carry a story on Bank research and to draw on Bank aged by the Research Advisory Staff. The Annual statistics. Bank Conference on Development Economics, the two Bank journals, and numerous research projects Bank research during fiscal 1994 were evaluated in fiscal 1994 and 1995. and 1995 The evaluators of the conference and its Proceed- Over the two year period reviewed, resources ings volume judged that the conference makes a devoted to research at the Bank have declined, serious contribution to the purposes of the Bank both absolutely and as a share of Bank budgets. and found the Proceedings volume to be well writ- From a peak of $36 million in fiscal 1992, Bank ten. They suggested more papers on policy and research expenditures fell to $26 million in fiscal implementation, institutional economics, and po- 1995. As a share of the budgets of the regional and litical economy; stricter review of draft papers; and central vice presidencies, research declined from presentation of differing views in a point-counter- just over 5 percent in fiscal 1992 to under 3 percent point format. in fiscal 1995. External funding for research has The journal evaluators praised changes made in also declined, from $5.5 million in fiscal 1993 to response to a previous evaluation in 1991 and $2.6 million in fiscal 1995. Accounting for publica- strongly supported the journals overall. They tion delays, these declines in spending will soon be praised the increasing diversity of topics and the reflected in output as well. The reduction in re- relevance of the World Bank Research Observer, but sources is curtailing the breadth and depth of Bank several thought that the World Bank Economic Re- research. Maintaining a critical mass of research view should better explain its published analyses capability in some areas has required compensat- to make them more accessible to a wider range of ing reductions in other areas, with health, infra- readers. Although somewhat improved, the writ- structure, and rural development currently hav- ing style of the journals continued to be dull, ing only modest research efforts. perhaps reflecting too much the style of Bank The location and size distribution of research reports. The differentiation between the journals projects in the Bank has changed, largely in re- was felt to be appropriate. sponse to the creation of the thematic vice presi- The journal evaluations are being reviewed by dencies in fiscal 1993. The share of research activity the journal editor and editorial boards with a view carried out in the Development Economics Vice to improving the coverage and readability of the Presidency increased in fiscal 1994 and then stabi- journals. lized. For the regional vice presidencies, the share Completed research projects were reviewed in of the centrally-funded Research Support Budget international economics, poverty and human re- exceeded their share of overall research, in part sources, macroeconomics, transition, agriculture, because they oversee several research capacity and the environment. Reviewers were impressed building initiatives. Theshare of ResearchSupport by the research and its emphasis on empirical and Budget funds going to smaller projects (less than policy-relevant topics. The success of dissemination $100,000) has increased, probably because review was judged to be more mixed, with some projects criteria increase with grant size and many staff and addressing a variety of different audiences with managers are interested in research projects of appropriateproducts,butothers focusing mainly on shorter duration. academic audiences. Some reviewers thought that Executive summary iii the research reflected too narrow a range of view- tat preservation, and interactions among poverty, points and had too great a sympathy for the "Wash- fertility, and household use of environmental re- ington consensus" view of development. sources. A macrolevel study will explore links The reviews of research projects are sent to among economic growth, environmental condi- project supervisors. The Research Committee and tions, and poverty. the Research Advisory Staff adopt recommenda- Reforming the state will focus on the role of the tions that will improve proposal review and project government in supporting private sector develop- oversight. ment, poverty reduction, and the regulation of economic activity, sofety, and the environment. New research themes and directions Research will analyze how the state can best fulfill Bank research must help meet the development its responsibilities in countries with weak institu- challenge of improving operational results on the tional capacity. ground by strengthening the design, preparation, Managing global integration requires new ap- and implementation of projects and policy reform. proaches to reap the benefits of trade, capital, and Bank operations are increasingly involved in technology flows while reducing the accompany- projects and programs that go well beyond the ing risks and volatility. Research will address the usual confines of economics to include issues in management of private capital flows, trade reform political economy and the behavior of institutions. and regional trading arrangements, and the two- Decentralizing governmental responsibility to way linkages between developing and industrial provinces or municipalities, encouraging private countries. sector provision of services, reducing environ- Aid effect eness is a growing concern as aid mental pollution and conserving fragile habitats, flows diminish. Empirical studies will analyze the targeting programs to the poor; and fostering the impact of aid on hUdget allocations; links among participation of stakeholders in project implemen- aid,savin-;.and investment; project performance; tation all entail a shiftfrom narrow economic analy- -nd the I ;nks between beneficiary participation sis. They point to the growing relevar -e of in'titu- and . J effectiveness. tional economics and political economy to Bank Evaluation of project and policy implementation projects and policies. It is important to devote will have a micro and a macro dimension. At the research resources to developing new tools, iden- micro level the focus will be on the performance of tifying the critical institutional capacities needed social sector projects during implementation; the for success, and learning how to develop those hypotheses underlying project design will be tested. capacities. At the macro level, the challenge is to determine The following topics are likely to be the major why some policies work well in some countries but focal points for research efforts, and many of them not in others; case studies will examine the politi- embody the increased awareness of institutional cal economy of reform. and political economy issues needed in our re- Institutional issues are moving to the forefront search and analytic capacity: of many Bank operations, and it is fitting that they Interactions of poverty, growth, and the environ- be given prominence in the Bank's research pro- ment involve links and tradeoffs that need to be gram. Knowledge about these issues is scarce and quantified. Microlevel analyses will focus on in- the analytics are still weak. Research in these areas dustrial pollution and firm behavior, land use will thereforebe risky, but the payoffs appear tobe conflicts between agriculture expansion and habi- very promising. PART 1 BANK RESEARCH IN THE MARKETPLACE OF IDEAS In addition to being a development institution and a search fairs to reach outside audiences. Modest use financial institution the World Bank is also a knowl- also has been made of video as a medium for dissemi- edge-based institution. The Bank accompanies its nating Bank research. But it is too early to evaluate financial assistance with advice based on its experi- these new approaches. ence working in more than 140 countries and in The Bank's research dissemination efforts focus virtually all sectors of national economies. And in on four main audiences. These include policymakers many ways, the real value added of the Bank is based in developing countries, researchers and analysts on its ideas, knowledge, and experience-increas- interested in economic development, students (espe- ingly true in a world of expanding global markets cially those likely to become members of the first two The Bank's knowledge base comes from many groups), and members of the "development commu- sources. It is an important byproduct of operational nity"-including journalists and nongovernmental experience and analytic work. But it is the direct organizations-who have a general interest in devel- product of research, the one activity carried out at the opment or specific interest in a narrower develop- Bank that has as its main objective the expansion of ment issue. knowledge about development. Research contrib- The Bank's formal publications are a major dis- utes to the solution of new development problems-- semination vehicle for research. Growth in the num- such as those of transition economies and environ- ber of Bank publications stopped in 1985 and in the mental sustainability-and long-term development past 10 years has ranged from 350 to 500 titles a year challenges-such as increasing the productivity of (figure 1). About 70 percent of them are research the poor. Yet it is a modest share of both the Bank's output: 55 percent are facsimile publications such as analytic work (about 14 percent) and its total budget discussion papers and technical papers, 9 percent are (less than 3 percent). books and symposia, 3 percent are statistical publica- Within the Bank, research underpins policy stud- tions, and 3 percent are the Bank journals and the ies and economic and sector work. But to realize its World Development Report. The remaining 30 percent full potential Bank research must also have an exter- of publications do not deal with research. nal -impact. How does Bank research fare in the marketplace of ideas outside the Bank? This section Figure 1 The annual number of Bank attempts to answer that question by applying a vari- publications grew briskly, then ety of measures to assess the use of Bank research by stabilized those outside the Bank. Although evaluating the 500 impact of research is difficult, it is possible to reach Cc o some judgments about the external effectiveness of 4 - Bank research. The links of research tu Bank opera- 2 tions will be addressed in a future report in this series. 300 Reaching many audiences o The Bank's research effort is aimed at many external 200 a audiences and addresses those audiences through a E 2 100- broad range of publications. Publications have been Z the main instrument for disseminating Bank research, 0 and the evaluations reported here focus on them. 75 80 85 90 95 70 7 0 8 0 9 New efforts are under way in electronic dissemina- Year tion using computer networks, such as the Internet, Source: Data from the Office of the Pubisher. and Bank research staff are experimenting with re- 2 Report on the World Bank Research Program, Fiscal 1994 and 1995 Table 1 Most widely distributed Bank respondents had read an average of 8.9 publications publications in fiscal 1994 and 1995 and used the material mainly for research, writing articles or books, and preparing classes or presenta- Number tions. The use of Bank publications for research was Title distributed not limited to academics or consultants: of those responding, nearly four-fifths of government organi- World Development Report 150,000 zations and more than three-fifths of commercial World Bank Atlas 39,000 firms use Bank publications for analytic work. The East Asian Miracle 25,000 From respondents' lists of the Bank publications World Bank and the Environment 22,500 they had read in the past 18 months, the 10 most Proceedings of the World Bank Annual frequently read publications were identified (based Conference on Development Economics 18,000 on percentage of respondents reading them). This list World Bank Economic Review 16,600 (table 2) corresponds well with those shown in table World Bank Research Observer 16,400 1-six titles appear on both lists-suggesting that the Global Economic Prospects 15,000 distribution of Bank publications is a good indicator Adjustment in Africa 15,000 of their actual use by readers. Disbursement Handbook 14,000 Over the past four years one-fifth of the Bank's publications have been translations, predominantly Source: Data compiled by the Office of the Publisher, in French and Spanish. The Bank's research output is most likely to be translated into other languages Distribution varies tremendously, from a few hun- when its distribution is large, or a large share of its dred copies for some of the facsimile publications to target audience speaks another language. The World more than 150,000 a year for the World Development Development Report is translated into more lan- Report, the Bank's most widely distributed publica- guages-nine---than any other Bank publication. tion. Of the 10 most widely distributed publications In addition to distribution in formal World Bank during the past two fiscal years, eight are research or publications, a great deal of Bank research and ana- statistical publications and include the first two policy lytic work is published by Bank staff and consultants research reports (table 1). in professional journals. Over the past 15 years the These distribution numbers are quite high for number of such articles has nearly tripled and is research products, but they include both sales and approaching 300 a year (figure 2). Many of these free distribution. For five of the eight research and articles make their first appearance in the Bank as statistical publications, about one-third of the distri- informal working papers circulated for comment bution is sales and two-thirds is free. For the journals and Proceedings of the World Bank Annual Bank Confer- Table 2 The 10 most frequently read Bank ence on Development Economics, six out of seven copies are distributed free to subscribers in developing coun- tries, who must verify their continuing interest in Share reading receiving the journals every two years. Title (r(percen Distribution makes publications available, but tells little about their use, especially when th y are free. To World Development Report 67 investigate the use of publications, the Bank carried World Bank Economic Review 38 out a survey of readers in April 1995. The survey was World Bank Atlas 30 sent to more than 20,000 customers, reviewers, librar- World Tables 27 ies, bookstores, and recipients of complimentary cop- World Bank Research Observer 27 ies; over 3,400 completed questionnaires were re- Social Indicators of Development 27 turned, one-third from developing countries. Survey Global Economic Prospects 23 respondents were mainly from research and consult- The East Asian Miracle 23 ing firms (32 percent), universities (31 percent), fi- World Debt Tables 23 nancial firms (8 percent), and government agencies (6 Trends in Development Economics 22 percent). Two-fifths of respondents were research- ers, a third were economists or managers, and a Source: New Directions in World Bank Publishing Survey, quarter were teachers. Over the previous 18 months 1995. Bank research in the marketplace of ideas 3 and Promote Growth showed that current systems of Figure 2 The number of Bank-authored providing financial security to old people are ill- journal articles has nearly tripled since equipped to cope with rapidly aging populations the 1981 world over. Over the next 35 years the proportion of 300 the world's population over 60 will nearly double, from 9 to 16 percent. By 2030 nearly 80 percent of the a 250 world's old people will live in today's developing countries. Current systems of old age security are Z200 unsustainable: either pensions will have to be cut or contribution rates raised. 1 so . The report, which sets out a framework for reform, has received wide attention from policymakers, ad- A 100 - vocacy groups, and private sector pension manage- E ment and investment advisory firms. This has con- Z 50 tributed to an upsurge in pension reform efforts among the Bank's client countries. 0 . I I I I I I The report's solution is to meet the three objectives 81 82 83 84 85 86 87 88 89 90 91 92 93 94 of old age security-saving, redistribution to the Year poor, and insurance-with a mix of public and pri- Source: Institute for Scientific Information. vate management, full funding, and pay-as-you-go finance. This would involve a mandatory savings program with attractive returns and a tax-financed, before revision and final publication. For example, pay-as-you-go program to alleviate poverty among nearly two-thirds of the Policy Research Working the elderly. These two programs would be supple- Papers are submitted for formal publication. Forty- mented by voluntary savings. Progress along these eight percent of Bank-authored journal articles deal lines is being made in a number of countries with with economic topics, and the rest are widely distrib- assistance from the Bank (box 1). Conferences have uted among social science and technical fields-with been held in many developing countries at the re- the next five most prominent topics being social quest of governments, and in many OECD countries policy (6 percent), agriculture (5 percent), political at the request of nongovernmental organizations, science (5 percent), education (4 percent), and envi- employer groups, institutional investors, and labor ronment (3 percent). Newsletters also are widely unions, as these groups try to educate themselves used to disseminate research and best practice. One and others about what needs to be done to be done to example, the World Bank Policy Research Bulletin, is avert the looming old age crisis. distributed to 18,000 subscribers outside the Bank, In a very differentarea, the Bank has a long history 16,000 of them in developing countries, of carrying out research on highway maintenance practices, with some of its early analysis leading to Influencing policy the development of the Highway Design Model, a A major goal of Bank research is to improve the base tool used by many highway departments to design of knowledge that supports policymaking, but mea- and manage highway systems. Over the past several suring the impact of Bank research on policymaking years, Bank research on road maintenance has fo- and policymakers is very difficult. There are no data- cused on Africa, where road transport is the domi- bases with information on research inputs to policy. nant form of transport but roads are poorly managed Nevertheless,evidenceisavailablefromspecificcases and badly maintained. The poor state of African showing that Bank research has policy impact. This roads is reflected in the backlog of deferred mainte- section reviews three case studies showing the direct nance, estimated to have a cost of $43 billion. impact of Bank research on policy. Research has shown that poor road maintenance Cross-country studies are particularly compelling policies in Africa are attributable to the institutional to policymakers, and The East Asian Miracle has been frameworkin which roads are managed. Attempts to very influential among Latin American policymakers. improve road maintenance policies need to focus on More recently, the Bank's Policy Research Report the managing and financing of roads as well as on the Averting the Old Age Crisis: Policies to Protect the Old maintenance practices themselves. A combination of 4 Report on the World Bank Research Program, Fiscal 1994 and 1995 Box 1 Addressing pension problems Averting the Old Age Crisis has contributed significantly to cial adjustment loans or credits. In Zambia, for instance, the policy debate on pension reform, providing a blueprint social security reform is a component of the International for Bank staff and policymakers alike. The Bank's involve- Development Association's Recovery and Investment Pro- ment in pension system reform has grown dramatically motion Credit. The Cabinet has approved the reform since the publication of the report. proposal (based on a joint Bank-International Labour The Bank has engaged in extensive discussions on Office report), which establishes an affordable state pen- pension reform in virtually every country in Eastern Eu- sion and promotes the development of voluntary private rope and Central Asia. Latvians, more than a quarter pension schemeswith a regulatory and supervisory frame- of whom are currently pensioners, passed legislation re- work. forming the existing pension system and creating a new, Other countries are revising their pension schemes in funded mandatory savings program. In developing the line with the Bank's recommendations. In Peru, revisions reform, the Latvian government was advised by a consor- were made to the 1992 Pension Reform in response to the tium of advisers from the World Bank and the Swedish guidelines of a Bank policy note. These changes include government. reducing the discrepancy between the public and the In other regions the recommended solutions to pension private pension contributions and harmonizing the re- problems are becoming conditions in structural and finan- tirement age of both regimes. research and field experience has fostered the devel- tries to learn from others' experiences with the opment of institutional reforms that are being imple- reforms. mented in several African countries in a cooperative The Bank has had an active research program on multidonor road maintenance initiative (box 2). environmental issues for nearly a decade, and it has The program's findings are being disseminated begun to influence environmental policy in many through regional seminars attended by representa- developing countries. The underlying theory of pol- tives from government agencies, the business com- lution control is straightforward-emissions should munity, farmers, and academics. Country workshops be reduced from each source to the point where the are also used to build consensus and to allow coun- cost of further reductions equals the benefits from Box 2 Improving road maintenance in Africa by improving institutions Road maintenance reform is being carried out in Africa holder participation; and the road system should be man- undertwoBank-managed,multidonorprograms:theRoad aged according to business-like principles. The RTTP's Maintenance Initiative (RMI) and the Rural Travel and research progressed more slowly but made apparent the Transport Project (RTTP). These long-term programs, now need for a national strategy for rural transport in which in their sixth and fifth years, help participating countries individual projects and programs could be tailored to achieve effective and sustained road maintenance and local physical, institutional, demographic, and economic reduce the transport burden on rural populations. conditions. The programs started with research to define and quan- Ten countries are actively pursuing RMI reforms, and tify transport problems and to identify effective rural trans- six have prepared rural transport or rural roads strategies port strategies. This research formed the basis for discus- through the RTTP. Progress under both programs spans sions with participating countries and potential donors on a continuous spectrum, from research to country and how to proceed with reform. Reaching agreement on core donor dialogue and policy reform to institutional change. issues required several years of research and dialogue. Donors involved in the RMI have agreed to give first By 1992 the RMI had achieved consensus on the funda- priority to the funding of investments related to road mentals: all stakeholders, public and private, must be maintenance, as opposed to new road construction. In involved; financing must come from stablesources that are addition, before making investment decisions, donors insulated from other public demands; the organizational will engage in systematic consultations with other do- structure must be transparent and permit active stake- nors, as well as with the governments concerned. Bank research in the marketplace of ideas 5 that reduction. Many countries set ambient standards strumental in the development of a public disclosure for air or water quality rather than analyze the benefits program in Indonesia (box 4). of emissions reduction in detail. Meeting these ambient standards at least cost requires good information on Influencing the wider research the cost of emissions reduction by source. community Few countries have even approximate informa- The 1995 reader survey mentioned above indicated tion on the level of emissions or the cost of emissions that research was the most frequent reason for con- reduction by industry or plant. But developing coun- sulting Bank publications and that 73 percent of tries frequently have information on the size, sectoral respondents used Bank publications in their research. composition, and location of their industry from Not surprisingly, Bank statistical publications are economic census data. Bank researchers have devel- particularly useful for research, but more general oped a data base that relates industrial emissions to publications such as the World Development Report the data that are typically available in economic also are widely used for research purposes. census files so that industrial emissions can be forecast Quantitative measures of the impact of research for metropolitan areas, provinces, or countries. Com- publications are far from perfect, but one increasingly bining these data with readily available information on applied measure is the number of times that publica- the cost of emissions reduction for particular industrial tions are cited in journal articles. The reasoning is that processes or plant designs can yield rough estimates of the most influential research findings will be those the costs of reducing emissions on a plant-by-plant most widely cited by other researchers. The Institute basis. Box 3 summarizes progress that is being made in for Scientific Information maintains a database on cita- applying this approach in Mexico. tions from over 1,400 professional journals published Besides regulatory approaches to emission reduc- in 35 countries. This database has been used to con- tions, recent research suggests that the public disclo- structmeasures of the citations made to Bank-authored sure of emissions data at the plant level can help to journal articles by other journal articles. reduce emissions by bringing public pressure to bear The citation measure used here is essentially a on polluting firms. Bank research has analyzed the moving-average citation rate. It is the total number of modes and costs of gathering and disclosing infor- citations in a five-year period of Bank-authored ar- mation and the nature of effective public participa- ticles published in the same five-year period, divided tion in pollution control. That research has shown by the number of such articles. This average was that the intensity of pollution generated by plants calculated for Bank-authored articles and, for com- under the same regulatory regime varies widely de- parison, for all published economics articles for each pending on how informed local residents are about a successive five-year interval since 1980 (figure 3). plant's pollution output. This research has been in- Bank-authored articles are cited from 10 to 50 percent Box 3 Projecting industrial emissions and costs of abatement The Bank's Industrial Pollution Projection System (IPPS)is industrial organizations in Mexico, to address key indus- widely used in Bank operations, client countries' environ- trial pollution issues. A similar project is being undertaken mental protection agencies, and internat' nal policy re- in Brazil toestimate theimpactof pollution in Rio de Janeiro search institutes. In Mexico the system's emissions and cost state. parameters are being used in a comprehensive cost-effec- The IPPS has also been used by the World Resources tiveness analysis of industrial pollution. Industries were Institute to produce the first comprehensive cross- ranked according to the environmental damage they cause, country estimates of toxic pollution, and it has provided the and geographic areas were prioritized according to the basis for recent work on trade and environment by the impact of pollution on local populations. The results were OECD. discussed at a government and industry workshop. The IPPS is available through the Bank's Website and Following this workshop a survey was initiated to receives many queries. About 55 percent of users accessing identify environmental practices and problems. The sur- the Website are from overseas (representing 43 countries), vey will be followed by an action plan, formulated in and roughly 10 users a month download the IPPS coeffi- collaboration with several environmental agencies and cients to be applied to country-specific data. 6 Report on the World Bank Research Program, Fiscal 1994 and 1995 Box 4 Indonesia's public disclosure program for pollution control Three factors led to the Bank's involvement in Indonesia's vary from gold (for plants with excellent environmental development and implementation of its public informa- performance), through green (those significantly exceed- tion program-the Program for Pollution Control Evalua- ing standards), blue (those in compliance), red (those tion and Rating. First, the Bank's 1994 Environment Report clearly not in compliance), and black (those very far from on Indonesia recommended public participation ap- compliance). Of the 187 plants initially rated, none re- proaches for pollution control. Second, this recommenda- ceived gold marks, 5 were awarded green, 61 blue, 115 tion coincided with thelndonesianEnvironmentMinistry's red, and 6 black. Officials initially made public the green interest in a program to disclose the environmental perfor- awards, along with numbers but not the names of other mance of polluters to the public. Finally, the Bank's ongo- firms in other categories, giving the plants six months to ing research in Indonesia had extensively analyzed indus- dean up. Full disclosure is now underway. trial wastewater data. BAPEDAL (Indonesia's National Since the program was announced, BAPEDAL has Pollution Control Agency) requested technical assistance received many calls from plant managers anxious to from Bank researchers to develop and implement their know what they can do to improve their rating. Plants not new environmental initiative. included in the program have asked to be included and Indonesia's Program for Pollution Evaluation and Rat- ranked. Overall, the program has sparked considerable ing publicly discloses information on pollution control international interest,and other countries are considering efforts by Indonesian factories. A team of BADEPAL staff adopting it as a model for their public information initia- and Bank researchers devised a way of collapsingcomplex tives. pollution data into a single rating for each plant. Rankings more often than the average economics article over Many researchers outside of the Bank participate the period shown. The most frequently cited Bank in Bank-funded research. In fiscal 1995,50 percent of articles cover a wide range of topics, including analy- the consultant time spent on research tasks in the ses of AIDS, economic growth, agriculture, educa- central vice presidencies was provided by nationals tion, trade, and environmental issues-indicating of developing countries (compared with 41.6 percent that the Bank's research program is producing well- recognized research across a wide range of topics. Indexes of citations have also been used to assess Figure 3 Bank-authored journal articles the impact of professional journals by estimating the receive more citations than the average average number of citations for each journal over 1.40 some time period. One recent analysis (reported in D. N. Laband and M. J. Piette, "The Relative Impact of 4 0 Economics Journals: 1970-90," Journal of Economic Literature, June 1994) covered 130economicsjournals 1.20 -- - - published during 1985-89. It did not include the World Bank World Bank Economic Review and the W-rld Bank Re- o W search Observer, which began publication in 1986. t .......... Applying the same technique employed by Laband and Piette, but for 1993 rather than 1990, puts the Review and the Observer 33rd and 40th out of the 130 economics journals ranked. The Bank's two journals 0.80 ----- - - rank the highest among journals devoted just to development economics. Moreover, the citation in- Economics average dex for the Proceedings of the World Bank Annual Bank 0.60_1_i_i_ i Conference on Development Economics exceeds those 81.85 84-88 87-91 90-94 for the Bank journals. By this measure, the Bank Years journals and conference volume are well regarded by Sourc: Institute for Scientific Information. members of the economic research community. Bank research in the marketplace of ideas 7 of all Bank consultant time). Local research institutes Figure 4 Bank books are the most frequently used and local consultants participated in one out of three Bank-authored outputs on reading lisMs research projects-typically the larger ones. Research on transition economies was most likely to involve Bank facsimde local researchers; research on international econom- ics was least likely to do so. Bank efforts at building economic research capac- Matenal in non-Bank volume ity in developing countries are expanding. Bank sup- port continues for the Africa Economic Research World Development Report Consortium and the Middle East Economic Research Forum, and support has been provided to the new Me n iank jourad China Center for Economic Research. Capacity build- ing initiatives are now being developed in Russia and Aftle i no-hnk joural Ukraine. Contributing to economics training Bankook Research output is not just an input into further I research; it also affects other knowledge-based ac- 0 20 40 60 0 100 120 140 tivities. Education is one of the most important. Re- Number of references on 25 reading Hit search reports, journal articles, and books are key inputs into graduate and undergraduate training in Source: Tower 1995. development economics and into Economic Devel- opment Institute (EDI) seminars and workshops. From compilations of reading lists, which are now reading lists more frequently than do Bank-authored available for many fields, it is possible to compute research outputs produced in any other format. Ar- how often Bank publications are assigned reading in ticles from Bank journals (World Bank Economic Re- courses in development economics. Edward Tower mew, Research Observer, and the Proceedings of the (Development 1: Reading Lists, Eno River Press, World Bank Annual Conference on Development Eco- Chapel Hill, N.C., 1995), has compiled a volume of nomics) appearnearly as frequently as Bank-authored twenty-five reading lists in development economics articles published in all other journals combined- (mostly for graduate courses) that were selected for even though these Bank volumes are relative new- their quality and comprehensiveness. These lists may comers. Material in non-Bank books includes papers be particularly influential because they are likely to by Bank authors in conference volumes or collections be used as guides in preparing other reading lists. that are published outside the Bank. The lists are primarily from U.S. institutions, but they The World Development Report is widely used in also include institutions in Australia, Canada, and teaching. Its seemingly low utilization rate (figure 4) the United Kingdom. derives from the fact that the Bank produces only one One-sixth of the reading list entries are World World Development Report a year, whereas its books Bank publications or articles or papers written under and journal articles appear in ihe tens and hundreds the auspices of Bank research projects. The Bank each year. Because the World Development Report is share of the reading lists varies greatly, from a high of carefully edited for readability to attract a broad 46 percent to a low of 3 percent. This average and audience, it is an attractive tool for teaching and has range is virtually identical to the results of a similar many of the characteristics of a good textbook. examination four years ago, even though there is Within the Bank, EDI also draws heavily on Bank very little overlap between the 1995 reading lists and research, policy papers, and other publications. All of those examined in 1991. As was the case in 1991, Bank EDI's seminars and workshops draw on those works readings are least likely to be used in courses that in some way. Key Bank authors sometimes present have a planning perspective or deal with a narrow their findings in person, often their works are sum- aspect of development. marized by EDI and used as instructional materials, The Bank-authored research that is cited in read- and in almost all EDI activities a number of Bank ing lists comes from a wide variety of sources (figure publications serve as required or recommended read- 4). Bank books and selections from them appear on ing. EDI submits that its comparative advantage and 8 Report on the World Bank Research Program, Fiscal 2994 and 1995 success as an institution is its access to the wealth of Table 3 Percentage of magazine references Bank expertise, experience, and research in the field to World Bank research, statistics, and lending of development. Magazine Research Statistics Lending Reaching the development community Information about the Bank's activities and findings The Economist 13 11 29 from its research and analytical work are dissemi- U.S. News 13 21 21 nated in many secondary sources that draw on origi- Science 38 8 14 nal Bank outputs. Newspapers and magazines pro- Foreign Affairs 20 9 13 vide a great deal of information about the Bank and Fortune 5 15 23 its programs. Forbes 7 5 40 A search of newspaper and magazine databases Source: Magazine Database, Knight-Ridder, Palo Alto, (some containing the complete text of the news California, 1995. sources) found that the Financial Times and The Econo- mist had the greatest number of references to the each year). The remaining reviews cover a broad World Bank, and by a large margin. The Financial range of Bank outputs from the Bank Atlas to reports Times was nearly three times more likely to have a by the Operations Evaluation Department. Reviews reference to the Bank than were other newspapers, are published in newspapers, popular journals and and The Economist was nearly five times more likely magazines, and professional journals. Overall, the to have a reference to the Bank than were other reviews are very favorable, but some common criti- magazines. cisms emerge. Some publications are described as References to the Bank were categorized for the outdated, and reviewers would like the Bank to make Financial Times and a representative newswire ser- more of its data public. Bank documents that evalu- vice, Info-South, and for six magazines from the ate past Bank practices--with successes and fail- twelve most likely to mention the Bank. The majority ures-are well-received by reviewers. Reviews (at of newspaper references to the Bank deal with the least 150 a year) outnumber books published by the lending program; about 15 percent refer to products Bank annually by 3 to 1, meaning that some books are of its research program or to Bank data (references reviewed several times. were about evenly divided between the two). For example, of the 1,055 references to the Bank by the Conclusion Financial Times in 1994, 70 involved Bank research The evidence summarized in this section indicates products, 90 involved Bank statistics, and 576 in- that Bank research is reaching a wide variety of volved Bank lending activities. Magazine articles audiences-mainly through Bank publications, pub- referring to Bank activities are less likely than news- lications produced by other publishers, and through papers to report on lending operations and are more newspapers, magazines, and EDI seminars and work- varied in their coverage. Of the 144 references to the shops. Bank research is widely used and referenced Bank by The Economist in 1994, 19 involved Bank by the international research community. Products research, 22 involved Bank statistics, and 39 involved of Bank research play a prominent role as teaching Bank lending activities. Table 3 summarizes the per- material, readings, and case studies in development centage of Bank references by these th-ee topics for economics courses. And Bank research is surpris- the six magazines. ingly well covered in selected newspapers and maga- Book reviews are another notable source of sec- zines-although it nearly always gets less coverage ondary dissemination of Bank research findings. The than the Bank's lending operations. The direct impact Bank's publications staff often receives copies of of research on policyrnakers has not been quantita- reviews of Bank publications and search for them in tively measured, but its impact is evident from ex- the literature. Data on reviews for the past two fiscal amples that can be readily expanded beyond the few years indicate that an average of 13 reviews of Bank illustrations offered in this section. Dissemination of publications are published each month, but the num- Bank research and analysis must be expanded be- ber is growing-reaching 30 in October 1995. yond conventional printed publications. Electronic About three-quarters of these reviews are of Bank dissemination of papers and data are beginning, and books that present the results of Bank research (from it offers new opportunities for two-way communica- forty to fifty such books are published by the Bank tion and dialogue on criti al development issues. PART 2 THE WORLD BANK RESEARCH PROGRAM, FISCAL 1994-95 This section reviews the World Bank's research pro- gets: 71 percent were financed by departments, 20 gram in fiscal 1994 and 1995-its funding, its find- percent from the RSB and only 9 percent from exter- ings, and its performance, as reported in evaluations nal sponsors. The thematic vice presidencies (TVPs) of past projects and programs financed by the receive most of the external resources (appendix Bankwide Research Support Budget (RSB). The sec- table 5). tion also reports changes in the research program The recent changes in the overall research effort resulting from the Bank's fiscal 1993 reorganization and in the composition of funding result in part from and outlines preliminary findings and conclusions of organizational changes in fiscal 1993-the creation of research projects. the thematic vice presidencies and the delegation of broader responsibility for research to the Develop- Research funding and management ment Economics Vice Presidency (DEC). DEC's share The Bank's analytic work underpins the institution's of Bank research increased in fiscal 1994 and stabi- lending program and policy advice. This work, com- lized in fiscal 1995. posed of research, economic and sector work, and DEC now has a stronger role in sectoral research, policy analysis, has maintained a fairly constant share while the proportion of research carried out by the of about 20 percent of Bank administrative expenses sectoral departments in the thematic vice presiden- over the past six years. But its mix has been changing: cies has declined dramatically (figure 6). Departmen- economic and sector work has grown and research tally funded research in the sectoral departments declined, absorbed nearly 42 staff years in fiscal 1993, but less Bank resources devoted to research have dropped than eight staff years in fiscal 1994 and nine in fiscal by 28 percent since fiscal 1992, from $36.1 million to 1995. Regional research recovered from its nadir in $26.1 million in fiscal 1995. Historically absorbing fiscal 1993, with increases in both RSB-funded and just under 4 percent of the Bank's administrative departmentally funded research (figure 7 and appen- budget, research declined to less than 3 percent in dix table 2). The regional offices also manage research fiscal 1995. Research expenditures have also fallen rapidly as a share of operations and central vice Figure 5 Research has declined as a presidency (CVP) expenditures in the past two fiscal share of expenditures years, to reach historical lows (figure 5 and appendix 6 - table 1). Funding for Bank research comes from three main sources: departmental budgets, the RSB, and external 0 sponsors. Departmental funds and staff time support La research independently or with cofinancing from other sponsors. The RSB cofinances research projects with departments, allocating funds on the basis of a 4 peer reviews of research proposals. Outside agencies also cofinance some research projects. In fiscal 1994, 68 percent of research resources came from depart- O m 3 mental budgets, 18 percent from the RSB, and 14 1 percent from external sources. External research funds have fallen sharply from their high of $5.5 million in 2 fiscal 1993 to $2.6 million in fiscal 1995. This drop in 1990 1991 1992 1993 1994 1995 external funding in fiscal 1995 shifted more of the Fiscal year burden of research expenses to departmental bud- Source: Appendix table 1. 10 Report on the World Bank Research Program Figure 6 DEC's share of research Figure 7 ...and the regions' share of activity has Increased... RSB funding has grown 1001 100- 80 80- 60, 6 06 o40 40 20 -2 CLI 0 I0 88 89 90 91 92 93 94 95 88 89 90 91 92 93 94 95 Fiscal year Fiscal year 0 DEC a Sectoral Depts/TVPs a Regions and CFS U DEC N Sectoral Depts/TVPs 1 Regions and CFS Source Appendix table 2 and RAD data. Source: Appendix table 2 and RAD data. capacity-building initiatives which receive a signifi- During the past three years the TVPs have submitted cant share of RSB funds. about 25 percent of the proposals, the regions about The reduction in resources available to Bank re- 30 percent, and DEC 40 percent. DEC's share of the search is curtailing its breadth and depth. Maintain- largest research projects (those above $300,000) is ing a critical mass of research capability in priority significantly greater than 40 percent, but the distribu- areas has led to reducing research in other areas. For tional effect of this is offset by the regions' manage- example, Bank research efforts are now quite modest ment of large research capacity-building grants (fig- in health, infrastructure, and rural development. Al- ure 7). though it is now increasing, the share of research on The size composition of grants from the RSB has environmental issues was halved in fiscal 1994 and beenchanging(figure8andappendixtable6).Projects again in fiscal 1995 as Bank analytic work turned of shorter duration and smaller budgets have been toward developing policy priorities and identifying increasing, in part because staff time for research in best practices. The wide-ranging topics of Bank research projects Figure 8 The share of smaller RS8- can be categorized by the Bank's program objectives funded projects has risen (appendix table 3). For the past four fiscal years, 100 economic management and poverty and human re- m 00K sources have remained the leading areas of research. -A 80 Fueled by questions about the transition in former :- socialist countries and broader liberalization world- wide, research on private sector development and 2 60 economic management absorbed much time and ef- - fort in fiscal 1994-95. Across the program objectives, C 4o external funds for research are concentrated on envi- ronmental topics, followed by poverty and economic management (appendix table 4). 20 E Proposals for funding from the RSB are submitted from throughout the Bank. Close working relations 0 among DEC, TVPs, and regional staff, and the 88 89 90 91 92 93 94 95 amongDEC. VPs,Fiscal year Bankwide access to RSB support, ensure that re- Source: Appendix table 6. search projects reflect Bank demand for research. The World Bank research program,fiscal 1994-95 ii the TVPs has declined. Smaller grants also have ties with the development community. In fiscal 1994- lower transaction costs and rejection rates: the crite- 95, 11 of 16 fellows were from the Bank's Part II ria and the stages of proposal review at budget ceil- member countries. The topics on which they worked ings of $40,000 and $150,000. Slightly more than half ranged as widely as the Bank's own research pro- of RSB funds have been allocated in grants of less gram (appendix table 7). than $100,000 over the past four years. Since fiscal 1993, almost half of the funding authorized for the Fiscal 1994-95 research projects largest grants (more than $300,000) has been for and findings research capacity-building in the Bank's client The Bank's research program shows both continuity countries. and variety. It continues to examine how poverty is The increase in small projects since fiscal 1992 has reduced, with recent work focusing on shifting pub- generated higher approval and funding rates there- lic financing for social services from universal subsi- after: dies to targeted provision for the poor. Labor market research emphasizes ways to improve education and Approval rates by fiscal year training in developing countries-and ways to re- (percent) train labor, accelerate growth, and generate demand 1992 1993 1994 1995 for output that will fuel employment growth in the formerly socialist countries. Bank researchers also Research proposals 72 86 74 82 are working with officials to design fiscal systems to All funds requested 53 66 45 72 alleviate adjustment problems during recessionary Funds for approved periods. And to contribute to the fiscal decentraliza- projects 85 85 87 82 tion being undertaken in developing countries, the Bank's macroeconomic research is examining how to In addition to research projects, the RSB funds . delegate fiscal authority from central to local govern- longer-term programs of dissemination and outreach. ments. These programs include the Annual Bank Confer- Ways to improve the performance of civil servants ence on Development Economics (ABCDE), a devel- and state agencies are the focus of studies on both opment research journal, and an applied economics public sector management and infrastructure ser- journal, a fellowship program, a working papers vices. In the face of limited administrative capacity in series, and a research newsletter. In 1994 the sixth governments, research on environmental policy also ABCDEfocusedoninstitutionaleconomics,economic has sought solutions outside the public sector. Re- geography, the transition in socialist economies, and search on the role of the state in enhancing private international migration and labor markets. In 1995 sector development has examined social security and the ABCDE examined fiscal decentralization, redis- pension systems, bankruptcy legislation and admin- tribution with growth, demographics (old age sup- istration, the power sector, and agricultural market- port and women's roles), and aid and interactions ing and extension services. And in response to pro- between industrial and developing countries. The posals for regional trade arrangements that arose findings of evaluations of the ABCDE are summa- during the extended Uruguay Round negotiations rized at the end of this section. of the GATT, the Bank has completed studies of The Visiting Research Fellows Program brings several regional trade agreements, finding potential researchers to the Bank to work on crucial develop- benefits where regional markets exploit economies of ment issues and to advise Bank staff. Some fellows scale. are renowned scholars; others are promising junior Summaries of the findings of some of the Bank's researchers, often from developing countries. Both research are provided below; synopses of all World benefit from access to the range of empirical and staff Bank research projects are published in the 1994 and resources at the Bank and the opportunity to develop 1995 Abstracts of Current Studies. 12 Report on the World Bank Research Program Poverty and social welfare The Bank's poverty reduction strategy focuses on in- benefit as the richest. The study identified the most creasing efficient employment opportunities and im- effective components of the marketing approach and proving the health and education of the poor. Equally, institutional strategies to improve implementation.' it emphasizes protecting those in need with the limited Carefully targeted subsidies have also been suc- resources in developing countries. Much of the Banks cessful in a few small-credit programs. Many poor research is directed to identifying which policies will households have needed only a small loan-to start best achieve these ends and whichconditions areneeded poultry farming or handloom weaving, for example- for those policies to succeed. The research is heavily to work their way out of poverty. But effective agen- empirical and draws on the knowledge and experience cies to provide the many small loans required and to of practitioners and researchers in many countries. The collect repayments are rare. A study by the Poverty collaborative nature of this work is illustrated in a and Social Policy Department and South Asia Coun- recent study in which experts from ministries and try Department I staff found that three rural credit government agencies in four Latin American countries programs in Bangladesh have been effective in reach- worked with Bank and other researchers to evaluate ing the poorest households. The most well known, alternative social sector programs (box 5). the Grameen Bank, needs a subsidy of about 20 What are the impediments to the poor in working percent of disbursements for its operations. But its their way out of poverty? Evidence and common repayment rates consistently exceed 90 percent, it sense suggest that inadequate education, training, focuses on the poorest parts of the country, and 94 health, and initial finance are all important con- percent of its clients are women. The study analyzed straints. How best to reduce these obstacles is the the features that have made it successful and sug- subject of ongoing research. A study of the relation gested ways to replicate it.' between income, nutrition, illness, and work in A study on targeting government spending to Rwanda determined threshold levels of nutrition for benefit the poor compiled and analyzed massive working adults, and found few ways that the hungry public expenditure data on health, education, food can compensate in their work for inadequate nutri- subsidies, employment schemes, and cash transfers tion. Higher intakes, perhaps funded through subsi- to identify the approaches most effective under dif- dies for basic foods, are needed to improve the wel- ferent circumstances. The study found that the story fare of the poorest and to increase their ability to work for food subsidies is more complex than for other and thus to raise their income.' transfers. Leakage is less of a problem the poorer the But universal food subsidies have been problem- country (and thusthelarger the target group) and the atic. When the state picks up the tab for rich and poor more dissimilar the consumption patterns of rich and alike, even small per-unit subsidies create huge fiscal poor (when only the "poor" commodities are subsi- burdens. A study of targeting concluded that the dized). But the poorest can be excluded when subsi- more programs are restricted to the poor, the less dies are targeted and distribution centers are few, support they generate among the politically critical registration procedures are complex, and distribu- middle classes-so benefits to the poor can fall as tion is made at schools.' targeting tightens. This study suggests that remov- Whenthefiscalburdenofsubsidizingsocialservice ing subsidies on more attractive proructs may be grows too great, user charges can be adopted or in- politically and fiscally more feasible.2 creased, but at the risk of eliminating access for the In Tunisia in 1985, the food subsidy program cost poorest. A recent project examined families' responses over 4 percent of GNP and 10 percent of government to changes in the price, quality, and availability of expenditures, and the richest quintile received more publicly provided health care. The most compelling than twice the benefit of the poorest. A recent "self- finding was that increases in prices, including for food targeting" program encouraged private distributors and drugs, led to a rise in children's deaths in Ghana to market "superior goods," while nutritionally andC6ted'lvoire.ResultsfromGhanaindicatethatthe equivalent subsidized foods are packaged and sold poor will increase their use if improved quality accom- in less appealing ways. Research found the program paniesuser charges-especially if the quality improve- to be successful: costs had been reduced to 2 percent ments provide local access to basic services and, par- of GNP and 6 percent of public outlays, and the ticularly, to pharmaceuticals, which formerly were poorest quintile received 1.2 times as much of the available only in the regional or provincial capital., The World Bank research program, fiscal 1994--95 13 A recent study commissioned by the Indonesian sive.3 In Hungary research shows that the transfer Ministry of Health illustrates the problems in raising income going to the richest group is twice that going revenues from user fees. Raising fees increased pub- to the poor. While family allowance payments are lic clinics' revenues, but at a significant cost to the progressive, pensions are not. 4 Related research finds poor, who cut their visits and did not go to alternative that transfers are going disproportionately to urban private providers. And large price increases caused recipients and aremissingmany female-headed house- many of the wealthier to shift to private doctors (who holds. This work also identifies implementation diffi- also raised their fees), so the public clinics gained culties and incentive problems in guaranteed income little revenue. The hospitals providing curative care schemes." A broader review of poverty reduction pro- have little leeway to raise prices: increases cause such grams in transition economies suggests that these pro- large declines in use that revenues drop.' grams mustbean integral part of reform, mustbuild on An indirect benefit of access to public health facili- existing programs, and must restrict access to and the ties was revealed in a study of maternal fertility and level of benefits if they are to remain viable.16 child mortality. The research found that women with Market structure and prices differ and change access to health facilities in Tunisia increased the inter- across and within countries, and identifying the poor vals between births, which reduced the risk of mortal- requires analysts to account for these differences. The ity for their children-and showed that it is possible to UN International Comparison Programme (ICP) has break the cycle of high infant mortality and thus a been developing comparable price measures for all larger number of desired births leading to shorter countries-an important improvement over compari- intervals between children and higher death rates.' sons based on exchange rates alone. But a study of Evidence on the risks of high fertility continues to these measures shows that they are biased: the prices mount. In a study of malnutrition in Uganda, the Africa of basic goods are significantly higher than the aver- Technical Department staff found that large household age from which GDP and consumption measures are size and high dependency ratios endanger the nutri- derived, and the poorer the country, the larger the tional status and health of a family's children.9 bias. Thus comparisons based on ICP measures have Broader cross-country patterns have been the sub- underestimated the extent of poverty, especially in ject of research on the relation between economic the poorest countries. Purchasing power parity-based growth, structural adjustment, and income inequality. poverty measures thus need further correction." Empirical research using panel data has found no evidence for the Kuznets curve (which suggests that Box 5 Learning to measure health inequality increases when economies begin acceler- ated growth and decreases in later stages of growth).w Educationeffets Conversely, a Latin America and the Caribbean Region Educating children, training teenagers, supporting study of growth, recession, and adjustment in Latin health and sanitation, providing prenatal car-re all America found that poverty increased significantly important to social welfare, but how do governments durig slups, deceasing or nt increa signg hny measure their effect or select which to fund? A project during slumps, decreasing (or not increasing) when totclthspbemdwtgterxetsfmte economies grew.n1 Both cross-country and household to tackle this problem drew together experts from the survoiey data " s owtht adjustentr p ran h ed Bolivian National Statistical Office, the Youth Training survey data show that adjustment prugrams have adEpomn atesi rga nTiia generally reduced poverty, although the poorest may and Employment Partnership Program in Trinidad not have benefited from later economic recovery. Re and Tobago, the Venezuelan Ministry of the Family no hvebneitd rm atr cnoicreoer..e and Central Statistical Office, the Netherlands Consult- search indicates that investments in human capital and and Trst icad Offic, the andS ns- rural infrastructure are needed to reach these groups.u ant Trust Fund, UDAPSO, Bolivia, and US. universi- The transition of formerly centrally planned econo- ties. An innovative cross