53505 ResearchDigest World Bank VOLUME 3 L NUMBER 3 L SPRING 2009 Banking on Politics Why are politicians so much more likely than others to serve on bank ruption and more powerful but less IN THIS ISSUE boards? Probably because of accountable governments. More- Banking on Politics ... page 1 private interests, not public over, while politicians with skills like Former politicians are much more likely than I those needed to run a bank (such as the average citizen to become board members n many countries former cabinet finance ministers) are more likely to at banks members, central bank gover- sit on a bank board than other cabi- nors, and financial regulators net members, those with the most FOCUS What Is the Evidence on Conditional Cash are far more likely than an ordinary expertise but less influence on rent- Transfers? ... page 2 citizen to become board members shifting regulation (such as bank at banks. Indeed, politicians are so supervisors) have the same chance A new report looks at how effective conditional cash transfers have been in much more likely to also be bankers of sitting on a board as the average improving health and education outcomes that the hypothesis that they were politician. randomly selected from a common Under an alternative view, bank- More Capital Not Enough to Grow Female- pool can clearly be rejected, even ers and politicians use regulation to Owned Businesses ... page 3 if it is assumed that the relevant create or shift rents at the expense Why do grants boost profits for population is a very small share of society. A hallmark of this private- microenterprises owned by men but not (say, 1­5 percent) of those in the interest view of financial regulation women? Two factors may explain the country who have a tertiary is the presence of transfers between difference education. bankers and regulators. Campaign A benign, public-interest view contributions, gifts, or bribes are di- No More Cutting Class? ... page 4 would see the prevalence of banker- rect, though hard to document. But When teachers are absent, students learn politicians as a way knowledge and a seat on a bank's board of directors less. What incentives can get teachers to experience are shared between the may be equally effective as a vehicle school? private and public sector. Under this for transfers: it cannot be easily Timing of Evaluations and Duration of view the phenomenon would be differentiated from a legitimate hir- Exposure Affect Estimates of Program positively related to the expertise ing, it is typically well paid, and any Impact ... page 5 of the former politicians and would lack of technical competence is not Timing matters in evaluating the impact produce more efficient banks. There directly damaging to the firm--as of a program. So does the duration of is no particular reason for it to be it would be if the politician were ap- beneficiaries' exposure related to the power of the govern- pointed to an executive position. ment to affect regulation, to its ac- Private-interest explanations can What Drives Firm Productivity countability, or to people's predispo- be grouped under two views--toll- Growth? ... page 6 sition to engage in corruption. booth or capture--depending on Rankings show which aspects of the business As a new paper by Braun and how rents are distributed. Under the climate matter most for firm productivity in Raddatz shows, however, this view is tollbooth view, politicians benefit Eastern Europe not supported by the cross-country from bankers and use the bank to The Challenge of Establishing World-Class evidence. Instead, the prevalence extract rents or for other political Universities ... page 7 of banker-politicians correlates purposes. Under the capture view, Building a world-class university does not negatively with the development of bankers capture the politician to get happen overnight the banking system--and far from more favorable treatment. being randomly distributed, it is higher in countries with greater cor- (continued on page 8) 2 World Bank ResearchDigest FOCUS What Is the Evidence on Conditional Cash Transfers? Cash transfers conditioned on Do these and other claims make tion services by program beneficiaries. households' investing in their sense? Are they supported by the em- Conditions can be justified if house- children are spreading rapidly. pirical evidence? What does all this holds are underinvesting in the human How well do they work? imply for how CCT programs should capital of their children--for example, C be structured or reformed in the coun- if they hold incorrect beliefs about the onditional cash transfer (CCT) tries that have them? And what does returns to these investments, if there is programs transfer cash, gener- it mean for countries that are consid- "incomplete altruism" between parents ally to poor households, on the ering implementing these programs, and their children, or if there are large condition that households make pre- often in circumstances very different externalities to investments in health specified investments in the human from those in which they were first and education. Political economy con- capital of children. Health and nutri- introduced? siderations may also favor conditional tion conditions generally require pe- A new World Bank Policy Research over unconditional transfers: taxpayers riodic checkups, growth monitoring, Report seeks to answer these and re- may be more likely to support transfers and vaccinations for children under lated questions. The report lays out a to the poor if they are linked to efforts age five; perinatal care for mothers; conceptual framework that considers to overcome poverty in the long term, and attendance by mothers at period- the economic and political rationale particularly when the efforts involve ic health information talks. Education for CCTs and reviews the rich evidence actions to improve the welfare of conditions usually include school that has accumulated on them, espe- children. enrollment, attendance on 80­85 per- cially the evidence from impact evalua- CCTs have led poor households to cent of school days, and occasionally tions. It discusses how the conceptual make more use of health and educa- some measure of performance. Most framework and the evidence on im- tion services, a key objective for which CCT programs transfer the money to pacts should inform the design of CCT they were designed. Nevertheless, the the mother of the household or, in programs. And it considers where CCTs evidence on improvements in final some circumstances, to the student. fit in with broader social policies. outcomes in health and education is Countries have been adopting CCT The report shows that there is good mixed. While CCTs have increased the programs--or considering doing so-- evidence that CCTs have improved the likelihood that households will take at an impressive rate. Almost every lives of poor people. Transfers gener- their children for preventive health country in Latin America has such a ally have been well targeted to poor checkups, that has not always led to program. Bangladesh, Indonesia, and households, have raised consumption better child nutritional status. And Turkey have large-scale programs, and levels, and have reduced poverty-- while school enrollment rates have in- Cambodia, Malawi, Morocco, Paki- in some countries by a substantial creased substantially among program stan, and South Africa, among others, amount. Offsetting adjustments that beneficiaries, there is little evidence of have pilot programs. Interest in pro- could have blunted the impact of improvements in learning outcomes. grams that seek to use cash to create transfers, such as reductions in the These findings suggest that to maxi- incentives for households to invest in labor market participation of benefi- mize the potential effects of CCT pro- child schooling has spread from de- ciaries, have been relatively modest. grams on the accumulation of human veloping to developed countries--as Moreover, CCT programs often have capital, they should be combined with reflected most recently in programs in provided an entry point for reforming programs to improve the quality of the New York City and Washington, DC. badly targeted subsidies and improv- supply of health and education servic- In some countries, such as Brazil ing the quality of safety nets. es and should provide other support- and Mexico, CCT programs have be- The report argues that CCTs have ing services. The findings also suggest come the largest social assistance thus been an effective way to redistrib- a need to experiment with conditions program, covering millions of house- ute income to the poor, while recogniz- that focus on outcomes rather than on holds. They have been hailed as a way ing that even the best-designed and use of services alone. of reducing inequality, especially in best-managed program cannot fulfill the very unequal countries in Latin all the needs of a comprehensive so- America; helping households break cial protection system. CCTs need to out of a vicious cycle in which poverty be complemented with other interven- Ariel Fiszbein and Norbert Schady, with Francisco is transmitted from one generation tions, such as workfare or employment Ferreira, Margaret Grosh, Niall Keleher, Pedro to the next; promoting child health, programs and social pensions. Olinto, and Emmanuel Skoufias. 2009. Condi- nutrition, and schooling; and helping The report also considers the ratio- tional Cash Transfers: Reducing Present countries meet the Millennium Devel- nale for conditioning the transfers on and Future Poverty. World Bank Policy Re- opment Goals. the use of specific health and educa- search Report. Washington, DC: World Bank. World Bank Research Digest 3 More Capital Not Enough to Grow Female-Owned Businesses More capital alone fails to help industries such as the making of lace, that more empowered women invest many female-owned businesses, in string hoppers (a typical food), or more of the grant in inventories--and part because they are in industries coir products (from coconut husk). By earn higher profits from it. with poor growth prospects contrast, women who work in more Are these results unique to Sri Lan- mixed industries, such as retail trade ka, or do they apply more generally? A and bamboo, invest The authors present field experiment in Sri Lanka more in their busi- nonexperimental provided grants of $100­200 to nesses and see some In many female- evidence showing randomly selected male and increase in profits. It dominated industries that women also female microenterprise owners. These appears that in many have much lower grants increased the monthly profits female-dominated the optimal size of returns to capital in of male-owned businesses by 9 per- industries the opti- a business is much Brazil and Mexico, cent, but led to no increase in profits mal size of a busi- with the Brazilian for female-owned enterprises. These ness is much smaller smaller and there data also suggest- results were surprising for much of and there is little is little prospect of ing a strong role for the microfinance movement, whose prospect of expand- industry in explain- emphasis on lending to women has ing beyond a certain expanding beyond a ing the differences. been based in part on the idea that subsistence size. But certain subsistence size In ongoing work the women are likely to be poorer and this is not the whole authors are attempt- more credit constrained and thus use story, since even in ing to replicate the resources more efficiently. the relatively mixed industries women experiment in Ghana. In addition, they In a new paper de Mel, McKenzie, still earn lower returns to the grants are using experiments with business and Woodruff explore the reasons than men. training and information in Sri Lanka behind this surprising finding of no A second factor may be differences to see whether women can be helped return to capital for women. They are in bargaining power and capture. An to switch into more profitable indus- able to rule out several important rea- intriguing finding is that when women tries with better prospects for growth. sons, and find two factors that appear received $100 grants, little of the What lessons should the micro- to explain the results. money ended up in the business, yet finance industry take from these re- The authors find that the lower when women received $200 grants, as sults? First, capital alone may not be returns for women are not the result much of it ended up in the business enough to raise the incomes of many of differences in education, entre- as when men received grants--or even poor female microenterprise owners. preneurial ability, risk aversion, or more. But when women do invest the Training and empowerment efforts reasons for going into business. As larger amounts, the returns are zero or may also be needed. These are being would be expected, the data show negative. carried out by some microfinance or- that poor, high-ability individuals The authors argue that a standard ganizations. But little is known about benefit most from additional capi- unitary household model in which the efficacy of such programs, showing tal in their business. But even after the household acts together to maxi- the need for more research. Second, controlling for gender differences in mize income cannot explain these poor, highly able, urban men appear to wealth and a myriad of ability mea- results. They develop a new model of have very high returns to capital, yet sures, the authors find that women household allocation in which some lack access to finance. Developing in- still have much lower returns than of the income and assets of the busi- novative financing tools to help them men. It thus appears that gender is ness are captured by other household should be a priority. not simply masking heterogeneity of members. A key insight of their pa- returns across other key dimensions per is that small, liquid assets such of the investment decision. as inventories might be more easily One factor explaining the dif- captured by a spouse than larger, less ference in returns is differences in liquid assets such as equipment. As industry. Both the amount of the a result, women might underinvest in grant invested in the business and inventories and overinvest in equip- Suresh de Mel, David McKenzie, and Christopher the return to capital decrease as the ment, using the equipment as a way of Woodruff. Forthcoming. "Are Women More Credit share of female owners in an industry protecting assets from capture in ad- Constrained? Experimental Evidence on Gender increases. In particular, returns seem dition to its business use. As support and Microenterprise Returns." American Eco- to be lowest in female-dominated for this speculation, the authors show nomic Journal: Applied Economics. 4 World Bank ResearchDigest No More Cutting Class? Creating the right incentives can for three years, despite interventions One approach is to link teach- reduce high rates of absence that improved other aspects of teacher ers' pay or progression directly to among teachers--and thus improve effort. But in Indonesia and Uganda measures of performance. A recent learning by students preliminary evidence suggests that evaluation of a performance-based I absence fell in the four to five years bonus system for teachers in Israel mproving basic education in de- since the first round of surveys. concluded that it increased student veloping countries requires, at a Attacking the problem effectively achievement, primarily through greater minimum, teachers who are pres- requires knowing who is absent most teacher effort and responsiveness to ent in the classroom and motivated to often and why. In most countries stud- student needs. In North Carolina a teach. But recent research shows that ied, absence is a systemwide problem policy charging teachers for excess teacher absence rates are quite high caused by many teachers, not a small sick leave reduced absence. An experi- in some developing-country primary number of "ghost teachers." Some ment in Rajasthan monitored teacher school systems. Moreover, because frequently cited reasons for absence attendance using cameras and then of poor record keeping and incentives do not hold up to scrutiny. Illness based part of the teachers' salary on to hide absence, administrative re- and noneducational duties appear their attendance rates. Not only did cords often understate the problem. to explain a relatively small share of attendance improve; so did student To monitor teacher attendance absences. Low base salaries have also learning. In Kenya an experiment directly, survey teams carried out been blamed, on the theory that they with incentive pay based on gains in surprise visits in 2002­03 to repre- cause teachers to moonlight at outside student test scores found that the sentative random samples of primary jobs that interfere with teaching. But program increased learning, though schools in eight countries: Bangla- higher-paid teachers are somewhat perhaps only as a result of "teaching desh, Ecuador, India, Indonesia, more likely than others to be absent. to the test." By contrast, a larger-scale Papua New Guinea, Peru, Uganda, and A lack of accountability and incen- experiment in government schools Zambia. Between 11 and 27 percent tives for performance probably ex- in Andhra Pradesh found that when of teachers were absent from schools plains more of the problem. Absence teachers were paid bonuses based at times that they would ordinarily is higher among high-ranking teachers, on their students' learning rates, be teaching. Absences were espe- who are harder to hold accountable; teacher attendance rates did not cially high in poorer and more rural in remote schools and lower-income improve but student learning did. areas: average rates reached about areas, where supervision is weakest; Details of the incentive design 40 percent in the Indian states of and in public schools compared with matter. In some cases teachers have Bihar and Jharkhand, but 15 percent nearby private schools (which are responded adversely to incentives-- in the wealthier state of Maharash- better able to discipline teachers). In by reducing collaboration with one tra. Similarly, Peru's most remote most developing-world school sys- another, excluding low-performing schools averaged 20 percent absence, tems, teacher compensation is largely students from classes, cheating on compared with 7 percent in Lima. unrelated to performance. Not all tests, lessening the academic rigor of High levels of teacher absence teachers take advantage of this lack of classes, or "teaching to the test" to the likely worsen education outcomes. accountability and incentives: most are detriment of other subjects and skills. In many developing countries sub- usually at their post, trying to teach in Other approaches are also prom- stitute teachers are not available, what are often difficult conditions. But ising. Some governments reformed and single-teacher primary schools when teachers lack the professional- the way teachers are monitored and must close when the teacher is ab- ism to meet their responsibilities, they evaluated and based promotions or sent. Studies from India, Indonesia, usually are not held accountable. raises on these evaluations. Finally, the United States, and Zambia have How can policy makers tackle policy makers may be able to improve associated higher absence with teacher absence and, more broadly, teacher performance through nonmon- lower student learning. The most improve the accountability and per- etary incentives. If teachers find their convincing evidence comes from the formance of teachers? First, focus work environment to be more pleasant Indian state of Rajasthan, where a greater policy attention on teacher and more conducive to student learn- randomized experiment that pro- effort, and on measuring attendance ing, they are less likely to avoid it. vided attendance incentives, and cut accurately, perhaps through indepen- teacher absence by half, substan- dent audits like those carried out by tially increased student test scores. researchers. In addition, policy mak- Additional research has begun ers should experiment with promis- to track absence over time, to see ing approaches for improving effort. Halsey Rogers and Emiliana Vegas. 2009. "No whether and how attendance can The best bet is to focus on improving More Cutting Class? Reducing Teacher Absence be improved. In the Indian state of teachers' marginal incentives--the in- and Providing Incentives for Performance." Policy Andhra Pradesh average absence centives they face each morning when Research Working Paper 4847, World Bank, rates stagnated at about 25 percent deciding whether to attend school. Washington, DC. World Bank Research Digest 5 Timing of Evaluations and Duration of Exposure Affect Estimates of Program Impact Timing and duration of exposure puts by the central program office is a significant part of a child's physical are relevant for all evaluations but not necessarily an accurate indicator of and cognitive development occurs be- particularly so for the evaluation of when those inputs reach their intend- fore age three and so the returns to the social programs ed destinations. programs are high- I Impact may be Impact may be est at these ages. mpact evaluations often ignore underestimated if a Besides timing, the the importance of timing and du- program is evaluat- underestimated if a duration of exposure ration: how long after a program ed before it has run program is evaluated during this critical has been launched one should wait its course. For ex- age range also mat- before evaluating it; how long treat- ample, trainees who before it has run its ters--that is, not ment groups should be exposed to a attend only part of course--but even after only whether a child program before they can be expected a training program receives the program to benefit from it, either partially or are likely to benefit its completion estimates before age three but fully; and how to take account of the its less than those could still understate its also whether the heterogeneity in impact that is related who complete it. child does so for to the age of the beneficiary and the Moreover, some full impact much of the interval duration of exposure. The true impact intended beneficia- between birth and of a program may not be immedi- ries of a program age three. ate or constant over time, yet many may be slow to accept the program-- What can be done to incorporate evaluations treat interventions as or may not do so until after they have timing and duration effects in program if they were instantaneous, predict- learned about its benefits, such as by evaluations? able changes in conditions and equal observing the gains to early takers. First, improve the quality of pro- across treatment groups. Whether the In some cases estimates of the gram data, especially the administra- treatment involves immunization or a impact of a social program after its tive records about the design and more process-oriented program such completion could still understate its implementation details of a program. as community organization, there is full impact for reasons that are exter- Second, choose the timing of the often no consideration of the possi- nal to the program design. One such evaluation keeping in mind the time bility that the effects differ depending reason is that the program might yield path of program impacts. The learning on variations in program exposure. additional and unintended outcomes process for program operators or for A new article by King and Behrman in the long run. For example, a micro- beneficiaries could produce a curve argues that timing and duration of finance project may not only provide showing increasing impact over time, exposure are relevant for all evalua- women employment and income but while a Hawthorne effect could show tions but particularly so for the evalu- also improve the future status of their a very early steep rise in program im- ation of social programs. The reason daughters. pact that is not sustainable. Examin- is that these programs often involve Another reason is that spillover ef- ing long-term impacts could point to changes in the behaviors of both ser- fects pertaining more to compliance valuable lessons about the diffusion of vice providers and service users. If one than to timing can appear and intensi- good practices over time. evaluates too early, one risks finding fy with time. Control groups or groups Third, apply an appropriate evalu- only partial or no impact; if one waits other than the intended beneficiaries ation method that takes into account too long, one risks a loss of donor might find a way of obtaining the treat- timing and duration effects. In pilot and public support for the program ment, or they might be affected simply programs it is possible to explore the or the scaling up of a badly designed by learning about the existence of the time path of the impact by allocating program. program, possibly because of expecta- treatment groups to different lengths There are many sources of variation tions that the program will be expand- of exposure in a randomized way, thus in timing and duration of exposure ed to their area. yielding operational lessons about that can affect program impact. Orga- There may be heterogeneous re- program design. nizational factors affect the leads and sponses to programs that are related lags in program implementation. For to the age of the beneficiary. For ex- example, a program requiring material ample, early childhood development Elizabeth M. King and Jere R. Behrman. Forth- inputs (such as textbooks or medi- programs, such as infant and child coming. "Timing and Duration of Exposure in cines) relies on the arrival of those feeding programs, target children soon Evaluations of Social Programs." World Bank inputs in the program areas, and the after birth because epidemiological Research Observer. Also available at http:// timing of the procurement of the in- and nutritional evidence indicates that wbro.oxfordjournals.org/ under "advance access." 6 World Bank ResearchDigest What Drives Firm Productivity Growth? Improving the business environment ness climate indicators and growth in The relationship between the busi- can lead to big gains in the TFP. During 2001­04 a change of one ness climate and firm productivity was productivity of firms standard deviation raised average firm estimated in two steps. First, the de- T TFP by almost 10 percent in the case of tailed AMADEUS firm-level data were he countries of Eastern Europe the infrastructure indicator; 7.8 percent used to obtain TFP as the residual and the former Soviet Union for the financial development indica- of a Cobb-Douglas production func- experienced a productivity surge tor; 5.8 percent for the labor quality tion. Second, change in log TFP was in 1999­2005 that greatly increased indicator; 3.4 percent for the labor regressed on the changes in the six output, drove up living standards, market flexibility indicator; 3.2 percent business climate indicators as well as and lifted 50 million people out of for the governance indicator; and 3 control variables for firm characteris- poverty. The productivity gains came percent for the competition indicator. tics, industry, subnational location, in part from greater capacity utiliza- Using Olley-Pakes decompositions, and country. tion, especially in the Commonwealth the study ranks these six aspects of The estimation equations were of Independent States (CIS). Output the business climate by relative impor- specified to avoid endogeneity and increased as firms began using their tance over the period observed. For the multicollinearity and thus minimize excess labor and capital, idled during sample as a whole, all indicators im- inefficiency and bias in the results. To the deep recession early in the transi- proved on average, and infrastructure mitigate endogeneity in the second tion, and this is captured in total fac- and governance made the largest con- part of the estimation, the study went tor productivity (TFP) estimates. tributions to log TFP growth (27.8 per- beyond the standard practice of al- The productivity gains also came cent and 22.7 percent). Comparatively leviating reverse causality by using in part from structural changes as less important were changes in labor location-industry averages as instru- resources were shifted to higher- quality (7.5 percent), financial develop- ments for firm-level business climate productivity uses in response to ment (5.6 percent), labor market flex- indicators. By first-differencing firm changed economic and institutional ibility (2.6 percent), and competition characteristics and lagging business incentives. Many workers moved out (2.2 percent). environment indicators by one year, it of manufacturing and into services-- There is much variation across the further reduced endogeneity between or, in the low-income CIS countries, countries in the sample. In Poland, unobservable firm heterogeneity and into agriculture. Estonia, and Romania, for example, factor input choices. But most of the productivity surge increases in labor market flexibility, To avoid multicollinearity, prin- was driven by firm dynamics--includ- labor quality, and governance were the cipal component analysis was used ing gains in existing firms, reallocation biggest contributors to TFP growth, to aggregate the detailed and often of resources across firms, entry of new while infrastructure improvement was closely related BEEPS indicators into and more productive firms, and exit of less important. In seven countries at five broad dimensions of the business obsolete firms. Macroeconomic stabil- least one of the business climate in- climate: infrastructure quality, finan- ity, a better governance and business dicators deteriorated over the period, cial development, governance, labor environment, stronger competition, and in Bulgaria most of them did. De- market flexibility, and labor quality. skill development, financial deepen- compositions show that effects also The sixth indicator, competition, was ing, and investments in infrastruc- varied across firm size and industry calculated with AMADEUS data as the ture were key drivers of productivity subsamples. market share of the four largest firms growth. Trade and foreign investment The analysis is based on a unique in the relevant industry. also contributed, by stimulating tech- data set and a carefully designed es- The study's careful methodologi- nology transfers and innovation. timation strategy that addresses the cal design thus makes efficient use A central issue in the microeco- shortcomings of earlier research. The of a unique database to provide new, nomics of growth is the relationship new data set was created by combining detailed, and reliable evidence for a between the business environment the Business Environment and Enter- highly significant link between differ- and firm productivity. A recent pa- prise Performance Survey (BEEPS) da- ent aspects of the business climate per by Anos-Casero and Udomsaph tabase with AMADEUS, a comprehen- and firm productivity. investigates this issue using a newly sive collection of firm-level accounting constructed database of about 22,000 information. The two databases were Paloma Anos-Casero and Charles Udomsaph. 2009. "What Drives Productivity Growth?" Policy manufacturing firms over the period first prepared by eliminating spurious Research Working Paper 4841, World Bank, 2001­04 in eight countries: Bulgaria, observations and biases, then merged Washington, DC. This is a background paper for Croatia, the Czech Republic, Estonia, by matching firm-level data from AMA- World Bank, Unleashing Prosperity: Produc- Poland, Romania, Serbia, and Ukraine. DEUS with averaged BEEPS indicators tivity Growth in Eastern Europe and the The study finds a highly significant corresponding in country, sector, sub- Former Soviet Union (Washington, DC: World link between change in six busi- national location, and firm size. Bank, 2008). World Bank Research Digest 7 The Challenge of Establishing World-Class Universities How to build a world-class however, a world-class university is As the case studies presented in university? Experience points to unlikely to be rapidly created without the book illustrate, universities that three strategies a favorable policy environment and aspire to better results engage in an T direct public initiative and financial objective assessment of their strengths he paradox of the world-class support, if only because of the high and areas for improvement, set new university is that everyone wants costs involved in setting up advanced stretch goals, and design and imple- one, but no one knows what it research facilities and capacities. ment a renewal plan that can lead to is or how to get one. A new book by International experience points to better performance. Salmi makes the case that the supe- three basic strategies for establishing Still, there is no single recipe or rior results achieved by world-class world-class universities: magic formula for "making" a world- universities--highly sought-after · Upgrading a small number of ex- class university. National contexts and graduates, cutting-edge research, and isting universities that have the poten- institutional models vary widely, and dynamic technology transfer--can be tial to excel (picking winners) each country must choose a strategy attributed to three complementary · Encouraging a number of exist- that plays to its strengths, resources, sets of factors: ing institutions to merge into a new and culture. Nor does building a · A high concentration of talent university that could achieve the syner- world-class university happen over- (students and faculty) gies of a world-class institution (hybrid night. Creating a culture of excellence · Abundant resources (financial formula) and achieving high-quality outputs and educational) · Creating new world-class uni- take many years of sustained commit- · Favorable governance (support- versities from scratch (clean-slate ment at the national and institutional ive regulatory framework, autonomy, approach) levels. leadership, academic freedom) Each approach has both positive The dynamic interaction among and negative aspects (table 1). these three sets of factors is the These generic approaches are not distinguishing characteristic of high- mutually incompatible; countries Jamil Salmi. 2009. The Challenge of Estab- ranking universities. A recent survey could pursue a combination of strate- lishing World-Class Universities. Directions of European universities confirms gies to establish world-class universi- in Development series. Washington, DC: World that funding and governance together ties. For example, while Kazakhstan Bank. influence performance. Its findings has opted to create new universities, show that the higher-ranked universi- Vietnam is following a mixed strategy ties tend to have greater management that combines investments to upgrade autonomy, which in turn increases the two existing universities with the es- efficiency of spending and leads to tablishment of two new institutions. higher research productivity. Similarly, simply investing money in an insti- tution or making its student admis- Table 1. Aspects of Three Strategies for Building World-Class Universities sions policy extremely selective is not Condition Upgrading existing institutions Merging existing institutions Creating new institutions enough to build a world-class univer- Ability to attract talent Difficult to renew staff Opportunity to change Opportunity to select sity in the absence of an appropriate and change the brand to the leadership and the best (staff and attract top students to attract new staff; students); difficulties governance framework for the univer- existing staff may resist in recruiting top sity system and strong leadership for students to "unknown" institution; need to the institution. build up research and In the past, government did not teaching traditions play a critical part in nurturing the Costs Less expensive Neutral More expensive growth of world-class universities. The Governance Difficult to change More likely to work with Opportunity to create Ivy League universities in the United mode of operation legal status different appropriate regulatory States grew to prominence largely as within same regulatory from that of existing and incentives framework institutions framework a result of incremental progress rather Institutional culture Difficult to transform May be difficult Opportunity to create than deliberate government interven- from within to create a new culture of excellence tion. Similarly, Oxford and Cambridge identity out of distinct institutional cultures evolved of their own volition over the centuries, with variable levels of pub- Change management Major consultation "Normative" approach "Environmentally and communication to educate all adaptive" approach lic funding but considerable autono- campaign with all stakeholders about to communicate and my in governance and in the definition stakeholders expected norms and socially market the new institutional culture institution of their mission and direction. Today, 8 World Bank ResearchDigest (continued from page 1) These private-interest explana- the prevalence of banker-politicians is Recent Policy Research tions better fit the evidence. They higher. Working Papers imply a greater prevalence of banker- If the direction of causality does politicians where politicians have go from the prevalence of banker- 4799 China's Integration with the World: more power, where the government is politicians to the characteristics of the Development as a Process of Learning and less accountable, and where there are financial system, the evidence sug- Industrial Upgrading Justin Yifu Lin and Yan Wang lower costs of engaging in corruption. gests that these connections result in 4805 Grand Corruption in Utilities Both views predict that the prevalence regulatory barriers to entry and have Charles Kenny and Tina Søreide 4806 Public Finance, Security, and Development: of banker-politicians is unrelated to real consequences for the develop- A Framework and an Application to the financial expertise of the politi- ment of the financial system. Imposing Afghanistan William Byrd and Stéphane Guimbert cians and yields less developed bank- restrictions on board memberships 4816 The Developing World's Bulging (but ing systems. for former politicians could limit the Vulnerable) "Middle Class" Martin Ravallion Moreover, the tollbooth and cap- ability of incumbent bankers to tilt 4817 The Investment Climate in 16 Indian States ture views have different implications regulations in their favor and restrict Giuseppe Iarossi for the distribution of rents and the financial development. 4843 Connecting Lagging and Leading Regions: The Role of Labor Mobility prevalence of banker-politicians that Still, it is important not to take Somik V. Lall, Christopher Timmins, and provide further evidence on the likeli- partial-equilibrium policy conclusions Shouyue Yu 4850 The Demographic and Socio-Economic hood of each explanation. First, the from this discussion. As long as there Distribution of Excess Mortality during the tollbooth view predicts that politi- are rents from restricting entry into the 1994 Genocide in Rwanda Damien de Walque and Philip Verwimp cians would be more likely to serve financial sector, incumbent bankers 4857 Transactional Sex as a Response to Risk in as bank directors when their party is will have incentives to try to ensure Western Kenya in government. Second, banks would that entry barriers are in place. Hiring Jonathan Robinson and Ethan Yeh 4858 Poverty Decline, Agricultural Wages, and be more profitable under the capture politicians to act as board directors Non-Farm Employment in Rural India: view, while leverage, because it allows is only one manifestation of these in- 1983­2004 Peter Lanjouw and Rinku Murgai the shielding of shareholders from centives. Without this avenue, these 4868 Wind Power Development: Economics and expropriation, would be higher under incentives could manifest themselves Policies G. Cornelis van Kooten and Govinda R. the tollbooth view. Third, while under in other ways, such as outright bribes, Timilsina the capture view rents created would that could be even more detrimental 4881 Is Low Coverage of Modern Infrastructure Services in African Cities Due to Lack of accrue primarily to bankers, under the to the stability of the institutional Demand or Lack of Supply? tollbooth view rents would be dissi- framework. Quentin Wodon, Sudeshna Banerjee, Amadou Bassirou Diallo, and Vivien Foster pated through higher costs. 4889 Orphanhood and the Living Arrangements of Overall, the evidence favors the Children in Sub-Saharan Africa Kathleen Beegle, Deon Filmer, Andrew capture view: former politicians are Stokes, and Lucia Tiererova not more likely to be directors when 4890 Enabling Conditions for Second Pillars of their party is in power, connected Pension Systems Heinz Rudolph and Roberto Rocha banks are more profitable without be- ing more leveraged, a larger share of Matias Braun and Claudio Raddatz. 2009. Working Papers can be downloaded at http://econ.worldbank.org rents accrues to bankers, and regula- "Banking on Politics." Policy Research Working To download the World Bank Research E-Newsletter, tion becomes more pro-banker where Paper 4902, World Bank, Washington, DC. 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 disseminating findings of World Bank research. Research Committee and managed by DECRS, the editor), Asli Demirgüç-Kunt, and Alan Gelb. Research The views and interpretations in the articles are those research support unit of the Development Economics assistance: Maximilian Hirn; editor: Alison Strong; pro- of the authors and do not necessarily represent the Vice Presidency (DEC). The Research Digest is not duction: Roula I. Yazigi. Information or free subscrip- views of the World Bank, its Executive Directors, or the copyrighted and may be reproduced with appropriate tions: send email to researchdigest@worldbank.org or countries they represent. source attribution. visit http://econ.worldbank.org/research_digest The World Bank 1818 H Street, NW Washington, DC 20433, USA Printed on Recycled Paper