Civil Liberties, Democracy, and the Performance of Government Projects

LANT PRITCHETT AND DANIEL KAUFMANN

politics or "democracy" and the perfor- —and the performance of the set of govern- How does the extent of civil mance of government investments. ment investment projects financed by the World Bank. Two indicators of a project's liberties and democracy in What, how, and how well? success were used: the project's economic a country affect the perfor- The interrelationship between govern- rate of return, which is determined after mance of its government's ments and economic development and project completion, and a simple rating poverty reduction is enormously complex. indicating whether the project accom- investment projects and, more Three questions are relevant to politicians, plished its developmental objectives. generally, the government's policymakers, and their advisors: what One might think that Bank researchers should governments do; how should gov- would investigate the success or failure of effectiveness? ernment decisions be made; and how well Bank projects to examine the Bank's perfor- will governments be able to carry out their mance. Although evaluators at the Bank do choices? In a wide variety of situations, it use such indicators for that purpose, data MPIRICAL analysis demonstrates may be that the'how well question is much on the success of Bank-financed projects that the extent of civil liberties in a more important than the what, since how also serve as indicators of borrower country country affects the performance of well a government spends its resources may performance, for three reasons. a government's investment pro- be more important than how much or what First, the projects financed by the World jectsE. This finding contributes to accumu- it spends its money on. Bank are undertaken by the borrowing lating evidence that the degree to which This article reports on new evidence that country's government, which is responsible citizens' "voices" in the public sphere are links how government decisions are made for project implementation. Governments repressed or are allowed to be "heard" has and how well they perform. This is chose to undertake these projects (although an important influence on whether the obviously a broad set of questions, which all of them must meet the Bank's internal accountability necessary for government we narrow to the relationship between criteria for financing). Hence comparisons efficacy will be created. In what is perhaps overall rankings of countries' political of ex post success indicate how well govern- a surprising contrast, there is no clear rela- characteristics—in particular, the degree of ments carry out projects they chose, rather tionship between indicators of electoral civil liberties and the extent of democracy than judging what projects they chose.

Lant Pritchett, Daniel Kaufmann, a U.S. national, is a Senior Economist in the World Bank's Development a Chilean national, is a Lead Economist in the World Bank's Development Research Group. Research Group.

26 Finance & Development / March 1998

©International Monetary Fund. Not for Redistribution Second, the data give a unique ranking of link between measures of the extent of civil econometric model and when allowances project success on a comparable basis for a liberties in a country and the performance are made for sector and regional differences large number of countries. Most countries of World Bank-supported projects. Table 1 in returns. Other, more recent investiga- rarely evaluate systematically their own gives the estimated relationship between tions have also found positive relationships investment projects, much less do so in a civil liberties and both the economic rate of between Bank project success and other, way that would permit their findings to be return and project success. The average broader indicators of "governance" quality compared with those of evaluations done by economic rate of return in the sample was such as the "index of credibility" reported other countries. While there is a great deal 16 percent. The estimates obtained imply in the Bank's World Development Report of imprecision in the art of project assess- that if a country were to move from having 1997, and an indicator of "institutional ment, the rankings obtained are fairly reli- the worst civil liberties observed to the quality" used by the Bank's Operations able indicators of success or failure. best, the economic rate of return on its Evaluation Department. We prefer mea- Third, since the World Bank is a multi- projects would improve by between 8 and sures of civil liberties, since we found lateral institution with written guidelines, 20 percentage points. This would mean rankings of civil liberties had greater procedures, and consistent methodologies increasing the return on projects by at least power to explain the performance of gov- that apply to all member country borrow- 50 percent and perhaps more than doubling ernment projects than did other indicators ers, it is unlikely that differences between the average return. Similar results are and because civil liberties are likely to have countries in the success of Bank-supported found for the summary indicator of project an effect precisely by affecting intermediate projects are mainly a consequence of incon- success: improving the degree of civil liber- measures of government performance like sistent actions of the Bank. ties in a country from the worst civil liber- credibility and corruption. ties in the sample considered by the current In addition to investigating the effects of Results study to the best would reduce the rate of civil liberties on the success of Bank- The study on which this article is drawn project failure by almost half. supported projects, we examined the data (see reference) did not construct a new mea- The effects of civil liberties show up in in more detail to see what effects the extent sure of civil liberties but instead relied on both the social and economic sectors. The of political liberties (or of the existence of four different cross-national rankings. One, economic rate of return is calculated for democracy) had. Again, to guard against undertaken by Freedom House, ranks coun- only those sectors in which the calculation peculiarities or anomalies in any particular tries annually based on a checklist of 14 of financial returns is possible (for exam- measure, we used a variety of indicators of criteria, which include media free of ple, industry, infrastructure, and agricul- democracy. Freedom House provides a censorship, open public discussion, free- ture). These are the "hard" sectors where ranking of political liberties based on crite- dom of assembly and demonstration, and one might have believed governance factors ria such as whether the chief authority or personal social rights (for example, the would be less relevant than in the social, or legislature was elected by a meaningful right to travel and the right to own prop- "soft," sectors, such as education and' process, there are multiple political parties, erty). Another rating, undertaken by health. The results obtained in the study and there is freedom from domination by Charles Humana, ranked countries in 1985 show an interesting symmetry with previ- the military. Another researcher, Alberto on the extent to which they complied with ous empirical work, which has shown that Alesina, created a subjective index of the human rights enumerated in the poor macroeconomic policies affect not democracy that ranks countries from 1 International Covenant on Civil and just returns in the "economic" sectors but (most democratic) to 3. The Institutional Political Rights adopted in 1966 by the also project performance in the "soft" sec- Reform and Informal Sector (IRIS) Center United Nations General Assembly. These tors like education and health (Kaufmann at the University of Maryland has also cre- included, among others, the right of peace- and Wang, 1995). This study shows that ated an indicator for democracy. ful assembly, freedom of opinion and poor project performance, even in the Interestingly, none of these indicators of expression, and the right and opportunity economic and "hard" sectors, is related to democracy or of political liberties had any to take part in public affairs. Two other poor governance. impact on project performance. narrower measures—of "media pluralism" The relationship between project success Moreover, the issue of whether civil liber- and of "freedom to organize"—were also and civil liberties is not only reflected in a ties or democracy was the more important used and gave results similar to the more correlation between two variables but also influence on project performance was general indicators. holds when indicators of macroeconomic investigated. Although the two are quite The most important finding of the policy, external shocks, and countries' highly correlated, the data indicate that the current study was a strong and consistent economic features are included in an estimated impact of civil liberties on project

Relationship between a country's civil liberties and its investment project performance

Indicator of project success Percentage point increase in economic rate of return Percentage point decrease in probability of prefect resulting from an improvement to civil liberties failure resulting from an improvement in civil liberties

By one standard deviation From worst to best By one standard deviation From worst to best Humana 4,5 20 2.1 9.4 Freedom House 1.9 3.2 13.2

Source: Jonathan Isham, Daniel Kaufmann, and Lant H. Pritchett, 1997, "Civil Liberties, Democracy, and the Performance of Government Projects," World Bank Economic Review, Vol. 11 (May), pp. 219-42. Note: The reported effects are estimated from a large sample of projects, controlling for other country-specific structural and policy characteristics, as well as for variations in the individual sectors.

Finance & Development / March 1998 27

©International Monetary Fund. Not for Redistribution Table 2 can change rapidly from being unsuccess- Indicators of expressions of citizen discontent for countries with ful to being successful by capturing high, medium, and low economic rates of return the synergies among dynamic leaders, motivated cadres of workers, and local Deviations of each activity from involvement. population-adjusted averages More broadly, there is also evidence that Economic rate of Average economic Regional (average Is zero for each indicator) return category 1 rats of return distribution Riots , Demonstrations Strikes countries with more extensive civil liberties (controlling for other variables) have per- High 22.2 South Asia: 3; 2.48 0.30 3.19 East Asia: 3 formed better in providing schooling to V girls and reducing infant mortality (as Medium 17.0 Latin America and 0.00 0.16 -0.02 noted in the Bank's World Development the Caribbean: 5; Sub-Saharan Report 1991). Ongoing research also sug- Africa: 2; gests that improved civil liberties in a coun- Middle East and try also reduce the incidence of corruption. North Africa: 3; South Asia: 1 The connection of civil liberties and the expression of the citizens' voice explains an Low 11.2 Sub-Saharan -0.19 -0.04 -0.23 otherwise deeply puzzling finding. If one Africa: 9; Latin America and examines the data for connections between the Caribbean: 2; a variety of indicators of sociopolitical South Asia: 1 instability and project returns, one finds that, surprisingly, countries with higher Source: Authors' calculations. 1 Determined by average rates of return classified by country for all countries with at least 1 0 projects during 1 974-87. economic rates of return had a much greater incidence of riots, strikes, and demonstrations (adjusted for population size) than countries with low rates of return success is not merely a consequence of the with the findings of interrelated research (Table 2). Given that expressions of social fact that democracies have extensive civil that emphasizes the "citizens' voice" as a unrest of this type are often thought to have liberties. Civil liberties seem to be a power- key mechanism. A centralized, top-down, an entirely negative influence on invest- ful force for improving performance even technocratic approach to governance has ment and performance, this was puzzling. in less than fully democratic settings, while often failed to deliver the hoped-for results. The current study discovered, however, if there were to be improvement in democ- The research on the impact of the citi- that countries with limited civil liberties racy without increases in civil liberties zens' voice comes from a variety of settings. had very little expression of social discon- (which would admittedly be unusual), this Accumulated empirical work has shown tent, while much more expression occurred would have no impact on performance. that the success of irrigation and rural in countries with more extensive civil liber- water supply projects is sharply affected by ties. Indeed, once the study controlled for Implications the degree of citizen participation in their civil liberties, the apparent positive impact We feel the best way to understand this design, construction, and operation. In one of riots and strikes disappeared completely. new evidence on the relationship between study of 121 donor-financed water projects, There is, however, an indication that coun- civil liberties and government performance those that were highly participatory were tries in which all modes of expressing dis- is to view it within the context of the successful 68 percent of the time, while the content are repressed experience worse debates on the micromechanics of govern- projects with little or no participation by performance on investment projects than ment efficacy. No government works per- the intended beneficiaries of the projects countries in which public expression of fectly, of course. Examples of government were highly successful only 10 percent of social tensions is tolerated. misbehavior are abundant, ranging from the time (Isham, Narayan, and Pritchett, At the same time, narrowly defined polit- severe political repression and kleptocracy 1995). Research vinto the relative perfor- ical democracy is neither necessary nor suf- on a massive scale to more mundane cor- mance of irrigation systems in two Asian ficient to allow full expression of the ruption, malfeasance, and indifference. countries by Robert Wade also emphasized citizens' voice. Some of the worst, most While paying attention to "government the degree of local control over the irriga- ineffectual, and corrupt governments in the failure" is a useful antidote to an abstract tion officers as a key element in how well world are functioning democracies (while and naive characterization of governments, the irrigation systems performed. Robert many others are not, of course), and some focusing on government failure is unpro- Putnam's investigation into the relative effi- well-functioning states are arguably not ductive, since simply having no govern- cacy of Italian regional governments settled fully democratic. The issues of who exer- ment at all is neither possible nor desirable, on the social relationships among the citi- cises power in the day-to-day operations of and rapid economic development requires zens as the key determinant of government a country's government are too complex to certain crucial functions to be carried out performance—a finding that is consistent be fully resolved merely by finding out by governments. The pressing question is with the conclusions of recent research whether or not elections have been held. not government size per se but rather what work done on schools in Tanzania and on Although the current study was primar- are the social, political, and institutional public spending in the United States. Judith ily designed to use the data on World Bank- conditions that create incentives for gov- Tendler's recent examination of the enor- financed projects to examine the relative ernments to do well. This question obvi- mous gains in health achieved in the efficacy of governments using a comparable ously has no one answer, but the current Brazilian state of Ceara, Good Governance measure, the fact that these were donor- study's results on civil liberties are in line in the Tropics, suggests that governments financed projects does raise questions about

28 Finance & Development /March 1998

©International Monetary Fund. Not for Redistribution the implications for donors. The first and types of donor selectivity are appropriate nomic performance and for designing the obvious implication is that the Bank has have revolved around questions of what mechanisms for delivering government not been able to "cocoon" its own the government should do. Alternatively, services of all types, from roads to schools. projects from the overall domestic environ- donors could place further emphasis on the These linkages need to be better under- ment. Earlier work on the relationship how well dimension of government perfor- stood and incorporated—in ways that will between the returns on Bank-supported mance. The efficacy of the government vary, perhaps enormously, from sector to projects and the macroeconomic environ- activities that development agencies fi- sector—into development activities. IF&DJ ment played a role in changing the once- nance and encourage is obviously central to prevalent view that projects and policies the agencies' mission and hence is a natural could be neatly separated and that the per- concern. To the extent that project perfor- formance of projects could be buffered from mance is determined by whether the This article draws on Jonathan Isham, Daniel national policies. The interrelationship be- expression of citizens' voice and their par- Kaufmann, and Lant Pritchett, 1997, "Civil tween a country's macroeconomic policies ticipation in governance are encouraged— Liberties, Democracy, and the Performance of and the success of aid in increasing its eco- whether broadly through open and free dis- Government Projects," World Bank Economic nomic growth rate has been reconfirmed. cussions (and criticisms) and the existence Review, Vol. 11 (May), pp. 219-42. (See "Aid Spurs Growth—in a Sound Policy of a free press, or narrowly through partici- Environment" by Craig Burnside and David pation in the design and implementation of Suggestions for further reading: Jonathan Isham, Deepa Narayan, and Lant Finance & Development, Dollar, December individual projects—this becomes a legiti- H. Pritchett, 1995, "Does Participation Improve 1997.) In the context of the current study, a mate area for dialogue in the design of the Performance? Establishing Causality with Sub- similar logic applies to project success and projects to be financed. jective Data," World Bank Economic Review, government efficacy. Vol. 9 (May), pp. 175-200. There are very good arguments, however, Conclusion Jonathan Isham and Daniel Kaufmann, for development agencies to have policies As a final caveat on a sensitive and diffi- forthcoming, "The Forgotten Rationale for that keep them out of domestic politics. The cult area of discourse, the argument should Policy Reform," Quarterly Journal of Economics. present results do not suggest otherwise. In not be made that civil liberties are impor- Daniel Kaufmann and Yan Wang, 1995, this context, two aspects of the current tant only because they are instrumental to "Macroeconomic Policies in the Social Sectors, study's findings should be kept in mind. greater government efficacy, since the rea- World Development, Vol. 23, pp. 751-65. World Bank, World Development Report First, the most hotly contested, narrowly sons for this go much, much deeper. 1991: The Challenge of Development (New York: political dimensions of democracy were not Nonetheless, civil liberties, along with other Oxford University Press for the World Bank). found to determine project performance. forms of expression and incorporation of World Bank, World Development Report Second, most of the difficult questions citizens' voice, do appear to have an instru- 1997: The State in a Changing World (New York: about sovereignty, conditionality, and what mental value for improving a country's eco- Oxford University Press for the World Bank).

l l:; //from the International Monetary Fund Banking Soundness and Monetary Policy: Issues and Experiences in the Global Economy Edited by Charles Enoch and John H. Green -

Banking sector problems have affected many IMF member countries, and measures to remedy these problems as well as to prevent their recurrence deeply concern central bankers and policymakers. The papers and comments published in this volume were presented at the Seventh Seminar on Central Banking, sponsored by the IMF. The topics include banking soundness and the macroeconomy, prudential standards, the role of the central bank during banking crises, and bank restructuring and supervision. US$29.50. Available in English, (paper). ISBN 1-55775-645-7. xii + 542 pp. 1997. Stock #BSMWEA TO ORDER, PLEASE WRITE OR CALL:

INTERNATIONAL MONETARY FUND .<£^o, Publication Services • Box FD-198 • 700 19th Street, N.W. • Washington, D.C. 20431 U.S.A. ^S'f!^' Telephone: (202) 623-7430 • Telefax: (202) 623-7201 • E-mail: [email protected] • Internet: http://www.imf.org *^V.fGr^ All orders must be prepaid. American Express, MasterCard, and VISA credit cards accepted. ^Pf^ «*KV

Finance & Development / March 1998 29

©International Monetary Fund. Not for Redistribution