48385 ResearchDigest WorldBank Volume 2 l NumbeR 2 l WiNteR 2008 Infrastructure and Growth in Developing Countries l Stéphane Straub A survey of the literature suggests with results in steady-state growth In ThIs Issue what type of research is most depending on the assumptions infrastructure and Growth in Developing likely to be useful for policy and made on aggregate returns to scale. Countries ... page 1 what data are needed More interesting are the potential in- D How best to analyze the links between direct channels--maintenance costs, infrastructure and growth? A survey of the rawing practical lessons for adjustment costs, labor productivity, literature points in useful directions infrastructure policy in devel- economies of scale and scope--that oping countries requires high- reveal the possibility of a growth ef- FoCuS quality empirical research and more fect resulting from investments in Finance for All? Policies and Pitfalls in and better data. In a new paper infrastructure above and beyond the expanding Access ... page 2 Straub surveys the theoretical and simple factor accumulation effect. A pro-poor policy for expanding access to empirical research on infrastructure The second set of issues in- finance should also focus on the excluded sectors in developing countries-- cludes composition (new invest- nonpoor energy, transport, telecommunica- ments or maintenance, operational exports and Productivity: Comparable tions, and water and sanitation--to or capital expenditures, private or evidence for 14 Countries ... page 3 highlight what works and what public investments), the sequenc- Exporting firms have a bigger edge in doesn't when trying to understand ing of reforms, and the importance productivity in some countries than others. the causal pathways between in- of different infrastructure sectors. Understanding why is key frastructure investment and devel- The theory on these topics is far opment outcomes. Straub draws patchier. It consists mostly of some local inequality and Project conclusions on two questions: what insights indicating that the balance Choice ... page 4 type of research is most likely to between new investment and main- Even programs targeted to the poor--and be useful in the search for practical tenance is likely to depart from what even when carefully designed--are policy recommendations, and what is optimal--and that maintenance vulnerable to capture by local elites type of data are needed to carry out spending that is lower than optimal that research. will lead to higher operational costs, What matters to African Firms? ... page 5 The literature review is organized both for infrastructure facilities and Do firms simply complain about everything? around two main sets of issues. The for private capital goods that rely on Or can their perceptions help guide policy reform? first, where most contributions have them. focused, involves understanding On the empirical side Straub Work-Related migration and Poverty the links between infrastructure and surveys 64 recent papers, analyzing Reduction ... page 6 economic growth at the level of the some 140 specifications in detail. Migration for work has helped reduce poverty economy, region, and sector. Growth Two-thirds of these specifications in Nepal. Policy choices can help it continue theory provides theoretical founda- support a positive and significant to do so tions, and so does the new eco- link between infrastructure invest- nomic geography literature, which ment and growth. But because of the economic effects during outbreaks of has the added virtue of allowing diversity of techniques, indicators, infectious Disease ... page 7 the prediction of joint outcomes in samples, and time frames, the litera- Governments may try to hide an outbreak growth and spatial inequality. ture provides few clear messages on of infectious disease. But honesty is the best Direct channels from infrastruc- questions of interest to policymak- policy ture to growth capture the impact ers--what is the proper timing of of an increase in this form of capital on the productivity of other factors, (continued on page 8) 2 WorldBankResearch Digest FOCUS Finance for All? Policies and Pitfalls in worthy and not everyone will or should Expanding Access qualify for credit. There are many ex- amples of overly relaxed credit policies having reduced national welfare. which can exclude large shares of the Government policies to expand ac- Recent advances in measuring population. cess to financial services should focus access to finance help in evaluating The report's conclusions confirm on reforming institutions, developing its impact on households and firms some traditional views and challenge infrastructure to take advantage of others. For example, recent research technological advances, encouraging and devising policies to broaden it provides additional evidence to sup- competition, and providing the right A port the widely held belief that finan- incentives through prudential regula- cross the developing world cial development promotes growth. tions. In contrast, direct interventions less than half the population The research shows how better access by governments, such as through has an account with a financial to finance helps by creating an envi- credit subsidies or government-owned institution, and in most of Africa less ronment conducive to the entry of new financial institutions, can be coun- than a fifth of households do. Modern firms and to innovation and growth. terproductive, reducing incentives for development theory emphasizes lack Research also shows that small firms the private sector to deliver services of access to finance as playing a big benefit the most from financial de- to the poor. Despite the best efforts of part in persistent income inequality velopment and greater access--both microfinance institutions, delivering and slower growth. Without inclu- through easier entry and through the financial services to the poorest with- sive financial systems that provide relaxing of constraints to growth. Thus out subsidies remains difficult. But broad access, poor people and small inclusive financial systems also have credit is not the only financial service enterprises have to rely on their own consequences for the composition of needed by the poorest--and in many resources to invest in education or the enterprise sector and the competi- cases not even the main one. The re- take advantage of promising growth tion in it. port suggests that subsidies may be opportunities. Recent evidence suggests that fi- better spent on overcoming barriers to Despite the emphasis in theory, nance is not only pro-growth but also savings and payment services, essen- promoting broader access to financial pro-poor. Beyond benefiting directly tial to participate in a modern market services has received much less from access to financial services, small economy. attention in development practice, firms and poor households can also By reviewing and highlighting a mostly because of serious data gaps benefit indirectly from the effects of large body of research, the report also in this area. A new World Bank Policy financial development. For example, identifies many gaps in our knowledge. Research Report, prepared by Asli the poor may benefit from having jobs Much more research is needed to mea- Demirgüç-Kunt, Thorsten Beck, and and higher wages as better developed sure and track access to financial ser- Patrick Honohan, provides indicators financial systems improve overall vices, to evaluate its impact on devel- of access for the first time. The efficiency and promote growth and opment outcomes, and to design and report shows that financial exclusion employment. Similarly, small firms evaluate policy interventions. Building affects not only the poor but also a may see their business opportunities data sets on access to finance that large share of the nonpoor in many expand with financial development, benchmark countries annually would developing countries. even if the financial sector still mostly help focus policymakers' attention and Beyond measuring access to serves large firms. Thus pro-poor finan- allow us to track and evaluate reforms finance, it is important to identify cial sector policy requires a broader aimed at broadening access. Better the barriers that prevent poor people focus of attention than access for the data at the firm and household level and small firms, especially those in poor: extending access to the excluded are important for improving our under- rural areas or in the informal sector, nonpoor micro and small entrepre- standing of the impact of access and from using financial services. The neurs can have a strongly favorable designing better policies for building many barriers include prohibitive indirect effect on the poor. more inclusive financial systems. costs, distance from services, and the Market failures relating to informa- inability to produce formal documents tion gaps, the need for coordination on when needed. Ethiopia has less than collective action, and concentration of one bank branch per 100,000 people, power mean that governments across and in Cameroon it costs $700--more the world have an important part to than GDP per capita--to open a play in building inclusive financial sys- checking account. Indeed, in many tems. But not all government action is developing countries banks impose equally effective. Thus it is important World Bank. 2007. Finance for All? Policies high minimum account balances and to have realistic goals. For example, and Pitfalls in Expanding Access. World annual fees to open and maintain even in the most developed economies Bank Policy Research Report. Washington, D.C.: checking and savings accounts, not all would-be borrowers are credit- World Bank. WorldBankResearch Digest 3 Exports and Productivity: Comparable Evidence for 14 Countries countries and countries with a more business-friendly environment and l International Study Group on Exports and Productivity a more effective government exhibit larger estimated exporter premiums. Exporting firms have strong study comparisons for a given country The findings on exporter premiums productivity premiums. are difficult to make because the stud- suggest a strong correlation between Understanding why these differ ies differ in the specifics of the ap- productivity and exports but do not across countries is key for policy proach followed. establish causality. To test the self-se- In a recent study the International lection hypothesis, the study examines T Study Group on Exports and Productiv- preentry differences in productivity be- hat exporting firms outperform ity uses comparable micro-level panel tween export starters and nonexport- counterparts selling only in the data for 14 countries and a set of iden- ers. If better firms become exporters, domestic market is well estab- tically specified empirical models to one should find significant differences lished through anecdotal evidence investigate the relationship between in productivity between future export and microeconometric research. Two exports and productivity. The results starters and future nonstarters even alternative (but not mutually exclu- are internationally comparable, both before some of them begin to export. sive) hypotheses offer explanations qualitatively and quantitatively. The Indeed, the results show preentry pro- for this difference in performance. study covers countries in Asia (China), ductivity premiums for export starters. The first hypothesis points to self- Latin America (Chile, Colombia), and The preentry premiums are larger in selection into export markets by the the European Union (Austria, Belgium, smaller or less open countries and in more productive firms. Selling goods Denmark, France, Germany, Ireland, countries with less business-friendly in foreign markets requires firms to Italy, Slovenia, Spain, Sweden, the regulations. Thus, in such environ- incur additional transport and distri- United Kingdom), with contributions ments future export starters must have bution costs, the expenses of skilled by economists in all these countries. a strong productivity advantage to personnel to manage foreign net- The study's main results are in line cover the extra costs associated with works, and production costs to cus- with the big picture from the litera- becoming an exporter. tomize domestic products for foreign ture. The study finds that exporters To test the learning-by-exporting markets. These costs constitute an are significantly more productive than hypothesis, the study investigates the entry barrier that less productive firms nonexporters when observed and postentry differences in productivity cannot overcome. Moreover, firms may unobserved firm heterogeneity is con- growth between export starters and be forward-looking, with their desire trolled for, and whether performance is nonexporters. The results provide only to export tomorrow leading them to measured by labor productivity or by weak evidence of learning by export- improve their performance today. total factor productivity. It shows that ing. But the authors caution that their The second hypothesis points to the productivity premiums of export- methodology does not control for the the role of "learning by exporting." ers increase with the share of exports potential bias caused by self-selection Flows of knowledge from foreign buy- in total sales but do not differ by firm of the most productive firms into ers and competitors help improve the size. exporting and thus that this weak evi- postentry performance of export start- Interestingly, the exporter pre- dence deserves further scrutiny using ers. In addition, firms selling in foreign miums differ considerably across more sophisticated techniques. markets are exposed to more intense countries, with Colombia exhibiting The study's key findings are the competition and must improve their the largest estimated premium and cross-country differences in exporter performance faster than firms that sell Sweden the smallest. To understand premiums that are shown not to only domestically. the cross-country heterogeneity in ex- be related to the use of different Cross-sectional differences be- porter premiums, the study conducts types of data or to the application tween exporters and nonexporters a meta-analysis that econometrically of differently specified econometric may therefore be explained by ex ante relates the estimated premiums to fea- models. The crucial next step in the differences between firms (the more tures of the estimation (for example, research agenda is to develop a solid productive firms become exporters) whether ordinary least squares is used understanding of the nature and but also by ex post differences (ex- and how many observations are in- causes of the differences in exporter porting makes firms more productive). cluded) and to country characteristics premiums across countries--a The big picture emerging from the (for example, size proxied by GDP, level prerequisite for sound policy literature on exports and productivity of development proxied by GDP per prescriptions that can help foster is this: the more productive firms self- capita, and indicators of institutional export-driven growth. select into export markets, but ex- quality). The meta-analysis shows Ana M. Fernandes and coauthors from the Inter- porting does not necessarily improve that the level of development cannot national Study Group on Exports and Productiv- productivity. But this big picture hides explain the observed cross-country ity. 2007. "Exports and Productivity: Comparable substantial heterogeneity. Cross- differences in exporter premiums. But Evidence for 14 Countries." Policy Research Work- country comparisons and even cross- it also shows that larger or more open ing Paper 4418. World Bank, Washington, D.C. 4 WorldBankResearch Digest Local Inequality and Project Choice choose latrine projects less often, as a result of a concentration of power in l Caridad Araujo, Francisco Ferreira, uador to investigate whether there is the hands of richer people, who do not Peter Lanjouw, and Berk Özler elite capture in the pattern of proj- need latrines. ect choice by communities awarded Both predictions are borne out Elites can significantly affect grants from the Social Investment empirically, as can be seen in figures 1 the outcomes of participatory Fund. The data include reliable pov- and 2. The results presented in these processes, even those that are erty and inequality estimates at the figures are robust to a large number of carefully designed community level, obtained by com- controls and alternative specifications. bining high-quality household survey Controlling for infrastructure need and D data with census data. They also in- a set of geographic and demographic ecentralization of spending clude administrative data on project variables, the poverty headcount is authority and community par- approvals and expenditures by type associated with a greater probability ticipation in decisionmaking and by community for almost three of the community receiving a latrine have become common elements in years. project. Inequality (measured by the policy and program design in many Social investment funds typically expenditure share of the top 1 percent, developing countries. Since the offer communities a menu of projects top 3 percent, top 5 percent, and so 1980s a growing number of antipov- from which to choose, most of which on) reduces the likelihood of a latrine erty programs have been designed produce public goods and all of which project being chosen. This effect tends as decentralized, demand-driven are intended to benefit the poor. The to fade as the elite is defined as a pro- initiatives, where the local beneficia- paper exploits the fact that the menu gressively larger group and vanishes ries decide--through a participatory offered by Ecuador's fund included when the more commonly used Gini process--what type of expenditures basically two types of projects: lo- coefficient is used. That the expendi- will be made. cal public goods (accessible to all, ture share of the richest 1 percent in Yet the theoretical literature re- though their valuation may vary across a community is so strongly correlated mains ambiguous in its assessment of individuals) and excludable (private) with the type of project selected may the rationale for decentralization. This goods. In Ecuador by far the most indicate that project choice is heavily ambiguity arises from the tradeoff be- important private good provided was influenced by a few dominant actors. tween the local government's advan- latrines, built on plots belonging to These results are consistent with tage in access to superior information poor community members with no the paper's model of project choice at lower cost and the risk that capture previous access to toilet facilities. under political inequality. They sug- of decisionmaking by special interest The paper constructs a theoreti- gest that even programs targeted to groups may be greater at the local cal model of project choice between the poor, and offering only projects level than at the national. Thus better public and private goods, under the designed to reduce poverty, are vulner- empirical analysis for specific pro- assumption that political power is able to capture by local elites. Elites grams is needed, with particular at- positively correlated with socioeco- can affect the outcomes of participa- tention to whether programs might be nomic status. The model predicts tory processes even when they are diverted from their objectives through that, controlling for inequality, poorer reasonably carefully designed. the influence of local elites. communities will select latrine proj- As evidence of this kind mounts, A new paper by Araujo, Ferreira, ects more often than richer ones. It there may be more general implica- Lanjouw, and Özler uses a unique also predicts that, controlling for pov- combination of data sets from Ec- erty, more unequal communities will (continued on page 7) Figure 1. Headcount Index and Probability of Receiving a Latrine Project Figure 2. Local Inequality and Probability of Receiving a Latrine Project .35 tcejorpenirtalagniviecerfoytilibaborP .30 .25 .20 .15 tcejorpenirtalagniviecerfoytilibaborP .25 .15 .05 .10 0 0.25 0.35 0.45 0.55 0.65 0.75 0.85 0.05 0.07 0.09 0.11 0.13 0.15 Headcount index Expenditure share of top 1 percent Note: Partial linear regressions, linearly controlling for the expenditure share of the top 1 percent. Note: Partial linear regressions, linearly controlling for the headcount index. WorldBankResearch Digest 5 What Matters to African Firms? thing to look at is therefore the under- lying industrial structure (that is, the l Alan Gelb, Vijaya Ramachandran, Moving up the scale to lower mid- ratio of camels to hippos), from which Manju Kedia Shah, and Ginger Turner dle income, firms must confront prob- you can infer what the real problem is lems caused by weak governance and (no water). African firms discriminate low administrative and bureaucratic It is certainly true that the mix of reasonably between constraints. capacity. These include corruption, the firms surveyed will reflect a degree Their perceptions can help prioritize tax system (rates and administration), of self-selection, whether for regula- policy reforms and the control of crime and violence. tory and governance issues or other Some aspects of regulation will be less country characteristics. One would C troubling to firms in these settings-- not expect to find many high-tech an perceptions data help us even if labor laws are stringent, the computer firms in Burundi or a vibrant understand investment climate state's weak capacity to enforce them shipbuilding industry in Botswana. constraints facing the private means that they are less likely to be But the survey data suggest that sort- sector? Or do firms simply complain perceived as a serious problem. ing effects do not dominate firms' about everything? In a new paper Labor policies are most responses: Gelb, Ramachandran, Shah, and Turn- problematic for firms in upper- ˇ Within countries responses are er analyze firm perceptions using en- middle-income countries. Unless relatively uniform across types of terprise survey data from 28 countries higher income is due to "exogenous" firms, including foreign-owned firms, in Sub-Saharan Africa. They show that factors such as large hydrocarbon which presumably are able to compare reported constraints reflect country deposits, the state tends to become conditions across countries. Devia- characteristics and vary systematically more capable. But business will not tions in firms' responses are apprecia- by income level. Reported constraints be the only constituency--labor will ble only where expected (for example, are also correlated with objective also exercise its voice. Shortage of foreign firms are less constrained by measures of the investment climate. skilled labor is also seen as a more finance). Countries are divided into three serious constraint at higher income ˇ Across countries the intensity income groups--low income, with a levels, where firms engage in more of complaints correlates with other GDP per capita of less than $400 (Bu- sophisticated operations. country-level indicators. For example, rundi to Burkina Faso in the figures), But are we missing the real story complaints about finance are far more lower middle income, $400­2,000 by not considering information from prevalent in countries with low finan- (Zambia to Cape Verde), and upper firms that have chosen not to locate cial depth. At the level of the firm, middle income, $2,000 and higher in Africa or that simply do not exist? It perceptions are well correlated with (Swaziland to Mauritius). Figure 1 is worth considering the "camels and objective measures of the business shows how perceptions vary with hippos" anecdote often mentioned by environment. income (selected data are reported Ricardo Hausmann. He argues that if Perhaps the most convincing evi- here). In countries at the low end of you are in a desert and interview cam- dence comes from looking at firms the income scale manufacturing firms els, you will get a very different idea that have adjusted to a constraint. In are most likely to be concerned about about the main problems of living and many countries firms with generators the most elemental constraints to do- working in the desert than if you could complain slightly more about electric- ing business: Is there a reliable power interview hippos, who don't live there. ity than those without them (figure 2). supply? Can finance and premises be The authors argue that in considering secured? Can the firm plan ahead? the investment climate, the interesting (continued on page 6) Figure 1. Firm Perceptions of Constraints across Income Levels Figure 2. Perceptions of the Electricity Constraint among Firms With and Without Generators )%(erevesrorojamsatniartsnocgniknarsmriF100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 )%(erevesrorojamsatniartsnocgniknarsmriF 10 0 Mali BeninKenya uritania SenegalLesothCameroon o e bia a 0 i wi Niger a he car Guinea LesothCameroon o AngolapeVerdeSwazilandNamiSouth AfricaBotswanaMauritius bia Burkina Mali FasoZambiaBeninKenya ritaniaSenegal Congo,De Burundm.Rep.ofMalawnea-BissauRwanda i i NigerUganda Gui Gambia,MadagasMozambiquTanzania The car e Guinea BurkinaFaso Zambia Ma AngolapeVerdSwazilandNamiSouth AfricBotswanaMauritius Ca Burundm.Rep.ofMalanea-Bissau Rwanda Gui Ugandmbia,TMadagasMozambiqueTanzania Ga Mau Ca Electricity Access to finance Corruption Skilled labor availability Congo,De Firms with generators Firms without generators 6 WorldBankResearch Digest Work-Related Migration and Poverty vested. Local remittances are treated Reduction as a mix of transitory and permanent income and are more often used for l Michael Lokshin, Mikhail Bontch- the effects of work-related migration both investment and consumption. Osmolovski, and Elena Glinskaya and remittances on rates of poverty The findings on the role of work- and inequality in Nepal. Relying on related migration and remittances in Migration and remittances are data from a 2004 national survey, the Nepal have important implications for key factors in reducing poverty in authors estimate a model of house- public policy. They suggest that strate- Nepal--and are likely to remain so hold migration decisions jointly with gies for promoting economic growth consumption equations instrumenting should take into account both domes- W for the endogeneity of the migration tic and international migration. While ork-related migration and decision. They compare the observed no panacea for the development chal- remittances are among the rates of poverty and inequality in Ne- lenges in Nepal, such policies could key factors in the decline in pal with those simulated under coun- have an important effect on poverty poverty in Nepal since 1996. Indeed, terfactual scenarios. reduction. Since Nepal has a plenti- more than a million prime-age men The results show that almost 20 ful supply of labor, migration for work from Nepal are working outside the percent of the decline in the head- could provide employment and higher country today. Remittances from Nep- count poverty rate in Nepal between earnings for a significant segment of alese expatriates grew by 30 percent 1995 and 2004 can be attributed to the labor force. a year in 1995­2004, from less than 3 the increase in work-related migration Indeed, unless local labor markets percent of GDP to about 15 percent, and the resulting remittances. Without and the global economic environ- exceeding the combined shares of migration, the aggregate poverty rate ment change dramatically, growing tourism, foreign aid, and exports. in Nepal would have been 33.6 percent numbers of Nepalese men and women About $1 billion comes into Nepal rather than the 30.0 percent observed will seek job opportunities outside as remittances each year, and inflows in 2004, and mean per capita spending Nepal. Migration and remittances are through private and unofficial chan- 14,000 Nepalese rupees rather than therefore likely to remain important nels could be even larger. 15,000. More than half the increase contributors to poverty reduction in Remittances sent home are the in poverty would have been due to the country. most tangible benefit of work-related a higher number of would-be poor migration for Nepalese households. Michael Lokshin, Mikhail Bontch-Osmolovski, among households with a migrant liv- and Elena Glinskaya. Forthcoming. "Work- Remittances enable households to ing abroad. Migration and remittances Related Migration and Poverty Reduction in Ne- overcome the constraints of credit had only a marginal impact on income pal." Journal of Development Economics. and risk and help them move into inequality in Nepal. modern and more productive activi- Migration and remittances have a (continued from page 5) ties. Remittances are typically spent strong impact on the living conditions The reason may be that these firms on housing and schooling, and a of households with a migrant. Among tend to be more dependent on elec- significant share directly supports households with a member who mi- tricity and that generator power costs households' consumption. grates within Nepal, the poverty rate some three times as much as grid But remittances are only one con- would be twice as high as the present power. If the self-selection process for sequence of migration. When a young, level if the migrant had stayed home. firms is incomplete, the constraints productive male leaves home, mul- Among households with a member identified by those present will likely tiple adjustments need to be made in working abroad, the poverty rate be seen as even more serious by those the labor supply of those left behind. would also be substantially higher if that have chosen not to enter. Thus in Migration changes the relative pro- that member had not migrated. Uganda, for example, easing the power ductivity of the remaining household One important conclusion emerg- constraint could bring in many new members; it affects households' pref- ing from the analysis is that the elas- firms as well as improve conditions for erences in risk aversion and in deal- ticity of poverty reduction in Nepal established ones. ing with uncertainty; and it provides over the past decade is higher for Overall, firms appear to discrimi- access to new, superior technologies domestic migration than it is for in- nate reasonably between constraints. (such as entrepreneurship and hy- ternational migration. One plausible Their views are a useful first step in giene). Migration also has implica- explanation for the difference is that the consultative process between tions for investments in the health remittances derived from work in for- business and government and can and education of the migrant's chil- eign countries are more likely to be help prioritize more specific behav- dren. The cumulative effects of these invested in productive assets and real ioral analysis and policy reforms. changes determine the observed pov- estate. This is often attributed to the Alan Gelb, Vijaya Ramachandran, Manju Kedia erty status of a household receiving preferences of the sender and also Shah, and Ginger Turner. 2007. "What Matters remittances. to the notion that households tend to African Firms? The Relevance of Perceptions In a recent paper Lokshin, Bontch- to treat international remittances as Data." Policy Research Working Paper 4446. Osmolovski, and Glinskaya investigate transitory income that should be in- World Bank, Washington, D.C. WorldBankResearch Digest 7 Economic Effects during Outbreaks of that people did indeed at times hold Infectious Disease excessively high perceptions of the risk of becoming infected with SARS or, l Milan Brahmbhatt and Arindam Dutta another rather than in the standard if infected, of dying from the disease. costs due to illness and death. The But some of the survey evidence also main economic effects arise from the suggests that perceptions of the risk of Outbreaks of infectious disease can uncoordinated and sometimes panicky disease are sensitive to new informa- lead to severe economic disruptions efforts by large numbers of individuals tion and that people are constantly even when little illness or death to avoid infection. These efforts, such trying to update and improve their ultimately occurs as fleeing from the area of an outbreak subjective probability assessments. or reducing contact with other people, Finally, the authors consider wheth- R lead to sharply reduced demand for er public information strategies can ecent years have seen a grow- many service sector activities, as help reduce unwarranted panic. A pre- ing interest in the economics of occurred during the SARS outbreak. liminary question is why governments infectious diseases. This interest Brahmbhatt and Dutta observe that often seem to have strong incentives has come with the emergence of many recent work in economic epidemiology to conceal information about out- new infectious diseases--more than may provide a natural framework for breaks of infectious disease. The au- 30 in the past 25 years, including HIV/ analyzing SARS-type events. The key thors review recent game theory analy- AIDS and severe acute respiratory syn- intuition of this approach is that self- sis that clarifies government incentives drome (SARS)--as well as the return interested, forward-looking individuals in the context of infectious disease. of old threats such as pandemic hu- adapt their behavior to take account In the early stages of an outbreak, man influenza. of the prevalence of a disease and the when there is much uncertainty about Interest has also increased with threat it poses to them. These changes whether it will turn into an epidemic or fresh experience of the enormous in behavior will differ according to the fizzle out, the government has an in- human and economic costs that can disease. But they will generally have centive to simply "wait and see," espe- arise from illness and death, as with both economic and epidemiological cially if an announcement might itself HIV/AIDS in developing countries. But consequences, leading, for example, start a panic or provoke severe trade as with SARS in East Asia in 2003 and to both negative demand shocks in the and travel restrictions. the plague in Surat, India, in 1994, economy and lower rates of new infec- An important finding is that the outbreaks of infectious disease can tions and disease prevalence. more numerous are nonofficial sources create severe economic disruptions A critical question then is how peo- of information about a possible out- even when relatively little illness or ple form their subjective probability break of disease, the weaker are the death ultimately occurs. In a new judgments about the risk of disease. government's incentives to conceal its paper Brahmbhatt and Dutta seek to While the assumption of rational ex- own information. Thus honesty may improve the understanding of these pectations provides a useful bench- indeed be the best public policy under "SARS type" effects and look at public mark, research in psychology and be- modern conditions, where such tech- information strategies that could help havioral economics in recent decades nological innovations as cell phones mitigate the costs they entail. reveals substantial biases in prob- and the Internet allow easy global Brahmbhatt and Dutta begin with ability judgments in many contexts. mass communications. a narrative account of the Surat and Recent theoretical work on information Milan Brahmbhatt and Arindam Dutta. 2008. SARS events. In Surat official esti- cascades and herding behavior sug- "On SARS-Type Economic Effects during Infec- mates ultimately counted only 52 gests the possibility that in situations tious Disease Outbreaks." Policy Research Working suspected plague-related deaths. But of imperfect information people may Paper 4466. World Bank, Washington, D.C. economic losses were estimated at rationally look to the behavior of oth- up to $2 billion, a result of a mas- ers as a source of information and yet (continued from page 4) sive and sudden flight of population arrive at erroneous conclusions. Under from Surat, panic in other cities, a fall conditions of high uncertainty, poor tions for the design of social funds in tourism, and trade embargoes on information, and emotional stress dur- and other community-driven develop- Indian exports. In the four East Asian ing an outbreak of infectious disease, ment programs. There may be a need economies most affected by SARS-- individuals could well arrive at signifi- to develop clearer rules on how deci- China, Hong Kong (China), Singapore, cantly biased subjective assessments, sions must be made in each commu- and Taiwan (China)--economic losses at least for a time. That would lead to nity with a view to making it harder for amounted to an estimated 0.5­1 per- less than optimal decisions, resulting the more powerful to exercise undue cent of GDP in 2003, despite these in the aggregate in an excessively high influence. economies' experiencing only about cost of private preventive actions. 700 SARS-related deaths. Brahmbhatt and Dutta look at pub- Caridad Araujo, Francisco Ferreira, Peter Lanjouw, The economic essence of such lic opinion surveys taken during the and Berk Özler. Forthcoming. "Local Inequality events thus appears to reside in SARS outbreak that provide sugges- and Project Choice: Theory and Evidence from Ec- costs of prevention in one form or tive (though not conclusive) evidence uador." Journal of Public Economics. 8 WorldBankResearch Digest (continued from page 1) reforms, how much spending should different infrastructure sectors. This is Recent Policy Research be allocated to infrastructure at dif- an area where the World Bank has a Working Papers ferent stages of development, and comparative advantage in developing what impact infrastructure investment cross-country databases. 4458 An Alternative Framework for Foreign has on gaps in development between The main efforts to develop data exchange Risk management of Sovereign countries or regions or between rural should concentrate on microeconomic Debt Martin Melecky and urban areas. household- and firm-level survey data, 4461 Services trade and Growth On the sequencing of reforms to taking into account ownership of sur- Bernard Hoekman and Aaditya Mattoo support infrastructure investments, for veys and limitations in their current 4474 indigenous and Colonial origins of Comparative economic Development: the example, studies suggest that restruc- design. This work should be undertak- Case of Colonial india and Africa C. A. Bayly turing should be undertaken before en with a view toward upward aggre- 4476 informality among Formal Firms: Firm-level, privatization and that a regulatory gation to generate village- or district- Cross-Country evidence on tax Compliance framework should be put into place level statistics on average access and and Access to Credit Roberta Gatti and Maddalena Honorati and supported by institutional ar- additional data on quality, costs, and 4484 on measuring the benefits of lower rangements such as transparency and institutional aspects. It should then transport Costs Hanan G. Jacoby and Bart Minten autonomy. Studies also suggest when address econometric issues linked to 4486 on the Welfarist Rationale for Relative feasible competition should be intro- the use of this type of data, such as Poverty lines Martin Ravallion duced. But the knowledge in this area the endogeneity of firms' technologi- 4487 Fiscal Redistribution and income inequality in needs to be refined, for example, to cal choices. latin America Edwin Gońi, J. Humberto López, and Luis account for finer aspects of regulation. The economic geography approach Servén Moreover, it is relevant only when the is very promising, with its ability to in- 4491 Risk-based Supervision of Pension Funds: A Review of international experience and private sector is involved, and private sert micro-level data in a global frame- Preliminary Assessment of the First outcomes investment still accounts for a rather work that accounts for the spatial, Gregory Brunner, Richard Hinz, and Roberto Rocha small share of the total and is likely to sectoral, and macroeconomic links 4493 Where in the World Are You? Assessing the do so for some time to come. of investments in infrastructure. But importance of Circumstance and effort in a From these gaps in knowledge a such infrastructure-related work, both World of Different mean Country incomes and (Almost) No migration research agenda emerges. The main theoretical and empirical, is still in its Branko Milanovic limitation of the macroeconomic ap- infancy. Efforts should be stepped up 4499 upgrading the investment Policy Framework of Public Pension Funds proach is that most interesting ques- to develop the relevant theory, to in- Dimitri Vittas, Gregorio Impavido, and tions cannot be addressed with very clude infrastructure sectors other than Ronan O'Connor 4500 Health and Civil War in Rural burundi aggregate data. Straub therefore urges transport, and to develop suitable em- Tom Bundervoet, Philip Verwimp, and Richard moving away from trying to estimate pirical strategies to test the models. Akresh 4501 the Quality of medical Advice in low-income the link between output and aggregate Countries indicators of infrastructure. Efforts Jishnu Das, Jeffrey Hammer, and Kenneth Leonard involving macroeconomic data should be geared toward understanding how Stéphane Straub. 2008. "Infrastructure and the political, institutional, and regu- Growth in Developing Countries: Recent Advances latory environment has affected the Working Papers can be downloaded at http://econ.worldbank.org and Research Challenges." Policy Research Work- to download the World Bank Research E-Newsletter, delivery and efficiency of services in ing Paper 4460. World Bank, Washington, D.C. go to Data & Research at http://www.worldbank.org the World bank Research Digest is a quarterly publica- the Research Digest is financed by the bank's editorial Committee: Jean-Jacques Dethier (managing tion aimed at disseminating findings of World bank Research Committee and managed by DeCRS, the editor), Asli Demirgüç-Kunt, and Alan Gelb. editor: research. the views and interpretations in the articles research support unit of the Development economics Alison Strong; research assistance: thi trang linh Phu; are those of the authors and do not necessarily repre- Vice Presidency (DeC). the Research Digest is not production: Roula i. Yazigi. information or free sub- sent the views of the World bank, its executive Direc- copyrighted and may be reproduced with appropriate scriptions: send email to researchdigest@worldbank tors, or the countries they represent. source attribution. .org or visit http://econ.worldbank.org/research_digest The World Bank 1818 H Street, NW Washington, DC 20433, USA Printed on Recycled Paper