The World Bank | | S I Fnotes N UNUMBER 49 PUBLIC SECTOR Toward more operaftonally relevant 2 604 indicators of govemance More specific indicators of govemment performance, processes, and institutions can contribute to better govemance-with its attendant benefits for development. A large body of recent research has linked combining information is likely to increase high-quality governance and public institu- accuracy. tioIIs to growth in per capita income, higher These first generation governance indi- How can investment, and other positive development cators have been invaluable in drawing atten- outcomes (see Knack and Keefer 1995 and tion to the crucial role of good governance governance Kaufmann, Kraav, and Zoido-Lobaton 1999). in successful development, and to the need This research has relied almost entirely on for public sector reform. Partly because of indicators be broad indlicators of the quality of governance research findings based on these indicators, and public institutions, produced in most the importance of good governance for developed to cases by sturvevs or expert assessments. development has recently attained the status Iicr-easing awareness of the importance of of conventional wisdom. For example, "good foster practical goverrnance has been accompanied by an and clean government" and "an effective legal increase in the number of commercial firms and judicial system" are two of the pillars of public sector and nongovernmental organization (NGOs) the World Bank's Comprehensive Develop- producing such indicators. ment Framework (Wolfensohn 1999). reforms? The International Country Risk Guide, These indicators can also be used to iden- puiblished by the PRS Group, provides the tify countries with severe corruption and indicators used most widely by researchers. other governance-related problems. The Ailonig the indicator-s in the guide are assess- proliferation in the number and sources of ntients of government bureaucracy, gov- such indicators may improve the accuracy erinnent corruption, anid the rule of law. of such cross-country comparisons. If sev- The guide has manv competitors that also eral independent sources point to coun- assess risk for prospective foreign investors, try X as having severe corruption, we can btut it is unique in offering both broad cov- be more confident in drawing such a con- erage across countries (covering more than clusion than if we were relving on just one 140) and over time (with monthly ratings source. for most nationis dating to 1982). Among development practitioners and Limitations of broad indicators the genertal public Tr-ansparency Interna- Broad indicators of the quality of gover- tional's (Corruption Perceptions Index may nance, such as those provided by the be the best-known governance indicator. International Country Risk Guide and This in(lex is create(d by aggregating cor- Transparency International, are of little rutptionl ratings from the International value in guiding specific public sector (Country Risk Guide and various other reforms. The problem is that they implicate sortircs, otn the reasonable assumption that many institutions and policies at the same FROM THE DEVELOPMENT ECONOMICS VICE PRESIDENCY AND POVERTY REDUCTION AND ECONOMIC MANAGEMENT NETWORK time. For example, a low score on a rule Thus the breadth of first generation gov- of law index points to multiple policy and ernance indicators makes it difficult or impos- institutional culprits. It is undeniably a prob- sible to infer specific policy implications. It lem-but it does not naturally suggest what also creates problems for country ownerslhip, the solution might be, or even who should particularly for governments that are ranked implement it. low. Governments that might be receptivq to Thus there are few examples of gover- being told of a specific deficiency in public nance indicators having a substantial impact sector performance-say, that public employ- on the policies of governments or on the ment levels are above international norms- reforms proposed by donors and interna- might nevertheless resist being told that t vey tional financial institutions. First generation are highly corrupt or that the rule of la, v is governance indicators have drawn attention weak. Most broad indicators are produn ed Most first to the right issues, but identifying practi- by for-profit firms or NGOs from industrial cal ways to reform governments remains a nations, through nontransparent proces.es, generation challenge. Today's public sector reform pro- and unaccompanied by details about he grams tend to be based on plausible but nature or source of the perceived deficien- indicators have largely untested assumptions about the insti- cies in governance. As a result they can eas- tutional mechanisms conducive to good gov- ily be dismissed by governments embarrassed limited relevance ernance and a strong rule of law. by their low rankings. There is thus an obvious need for second Accordingly, second generation gover- for public generation indicators to help determine nance indicators should move in two direc- which institutions are associated with which tions. They should be more specific in sector reform dimensions of public sector performance. measuring performance. And they sho ild Most first generation indicators have two pay more attention to measuring governmr-it properties that limit their relevance for pub- processes and institutional arrangements- lic sector reform. First, they measure gov- notjust performance. More disaggrega-ed ernment performance but do not describe measures of corruption and other dimen- government processes or institutional sions of performance are needed to facilitate arrangements. To argue convincingly that meaningful empirical tests of the effect, of a particular public sector reform will (or various government processes and inst tu- will not) improve government performance, tions. Measures of those processes and ir sti- it is necessary to have measures of both insti- tutions are essential for testing assumptions tutions and performance, to allow tests of on the efficacy of certain public sector refoi [ns the links between them. in improving performance. For example, Second, the best-known governance indi- does increasing civil service pay reduce cators typically measure performance very bureaucratic corruption and increase the c1fi- broadly rather than characterizing specific ciency of public service delivery? dimensions of performance. The broadest indicators are those-such as Transparency Developing second generationi International's Corruption Perceptions indicators Index-that are constructed by aggregating There has been some progress on collect- the ratings provided by other sources. But ing more specific performance indicators. even indicators from single sources can be Examples include: broad. For example, corruption ratings from * Business surveys conducted by the World the International Country Risk Guide and Bank that include questions on the qulal- other sources typically do not differentiate ity of public service delivery, such as i lie among bureaucratic, legislative, or judicial frequency of power outages or disritp- corruption, or between grand and petty cor- tions in water supply. Some of these sur- ruption, or among government agencies. This veys are designed to allow region-al property limits their usefulness in identifying comparisons, making it possible to id . n- reforms that might reduce corruption. tify areas experiencing the most severe PREMNOTE 49 DECEMBER 2COO service disruptions. Such information can bureaucratic performance. Qualitative be far more useful for sector-specific pro- responses to 20 questions were provided jects than broad indicators of bureaucratic by 126 country experts over 1993-96. Indi- quality or government effectiveness. cators constructed from the data include * Business surveys conducted by the Bank an index of meritocratic hiring, an inter- that include questions distinguishing nal promotion and career stability index, between two aspects of corruption (Prad- and a civil service compensation index. han and others 2000). The first aspect, state Rauch and Evans show that subjective rat- capture, refers to corrupt practices that dis- ings of bureaucratic quality and corrup- tort the formulation of laws and regulations. tion from the International Country Risk The second, administrative corruption, dis- Guide and other sources are positively torts the implementation of those laws and related to the meritocratic hiring index, regulations. This distinction has important but are unrelated to the indexes of com- Efforts are being implications for anticorruption strategies pensation and internal promotion and and public sector reforms. These surveys career stability. made to collect also measure the incidence of corruption * In a Bank-IMF project, country econo- in different government agencies. mists are assessing public expenditure more specific * Surveys of public officials conducted by management in 25 heavily indebted poor the Bank that allow the construction of countries. Countries are graded on var- indicators performance measures-accountability, ious aspects of budget formulation, exe- employee morale, focus on results-for cution, and reporting. several government agencies (Manning, * Giulio de Tomasso and Amit Mukher- Mukherjee, and Gokcekus 2000). jee are updating civil service pay and * The International Development Forum employment data for a large grouo of sponsored by the United Nations, Organ- countries (see Schiavo-Campo, d n- isation for Economic Co-operation and maso, and Mukherjee 1997). Empirical Development (OECD), World Bank, and evidence on the impact of civil service International Monetary Fund (IMF) in pay on corruption is mixed-suggesting March 2000 determined that there was not that pay raises do little to reduce bribe a meaningful set of indicators of partici- solicitation by bureaucrats. patory democracy and good governance * The PREM Network's Decentralization that was politically acceptable to govern- Thematic Group is compiling indicators ments. But there was strong sentiment that of fiscal, political, and administrative efforts to identify such indicators should decentralization, including measures of continue, in the recognition that the alter- subnational autonomy in spending and native was the increasing use-at least out- revenues, characteristics of intergovern- side the OECD's Development Assistance mental transfer systems, and regulations Committee (DAC)-of indicators that for subnational borrowing. were nontransparent or unacceptable to * Bank surveys of public officials in 15 coun- governments. DAC plans to post updates tries allow the construction of various mea- on work in this area on the OECD Web sures of the institutional environment, site, to gauge political and practitioner including the credibility of rules con- reactions to possible indicators. straining officials' behavior, the credibility There has also been progress in devel- of policies that officials are asked to imple- oping and testing indicators of government ment, and the adequacy and predictability processes and institutional arrangements. of agency resources (Manning, Mukher- Examples include: jee, and Gokcekus 2000). Coupled with * Using data from 35 developing countries, agency performance measures produced Rauch and Evans (2000) show that mer- from the same surveys, these data permit itocratic civil service systems are associ- analysis of the relationship between insti- ated with lower corruption and better tutional arrangements and performance. PREMNOTE 49 DECEMBER 2000 * The Database of Political Institutions, ters." Policy Research Workitig Paper assembled by Bank staff, contains 113 2196. World Bank, WVashinigton, D.C. variables for 177 countries covering Knack, Stephen, and Philip Keefer. 1 99.5. 1975-98 (Beck and others 2000). The "Institutions and Economic Perfbrmance: variables provide information on elec- Cross-Country Tests Ujsing Alternittive tions, electoral rules. type of political sys- Institutional Measures." Economics atid tem, presence of mnilitary officers in Politics 7: 207-27. government, and party composition of Manning, Nick, Ranjana Muklherjee, amid the opposition, government, and coali- Omer Gokcekus. 2000. "Public Officials tions. From the raw data, indexes of polit- and Their Institutional Enxironment: An ical stability and checks and balances are Analytical Model for Assessing the Impact constructed. The database is designed of Institutional Change on Public Sector Data collection efforts for studying issues such as political coIn- Performance." Policy Research WVorking ditions for economic reform, the politi- Paper 2427. WAorld Bank, Washington, D.C. are expected to cal and institutional roots of corruption, Pradhan, Sanjay, and others. 2000. Anticor- and the appropriate and institutionally ruption in Transition: A Contribution to the greatly enrich sensitive design of economic policv. Effec- PolicDebate. W4ashington, D.C.: Norld Bank. tive reform programs requiire an under- Rauch,James E., and Peter B. Evans. 200(0. the information standing of existing political constraints. "Bureaucratic Structure and Bureaucratic These and other data collection efforts Performance in Less Developed Cotiun- base on governance are expected to greatly enrich the infor- tries."JournalofPublic Economics 75:49-71. mation base on governance and public insti- Schiavo-Campo, Salvatore, Giulio de Tom- and public institutions tutions. Many of these data sets will allow maso, and Amitabha Mukherjee. 1997. for more micro level (such as agency or "An International Statistical Survey of firm) analyses of links between governance Government Employment and Wages.' and economic performance, complement- Policy Research Working Paper 180(6. ing well-known cross-country evidence on World Bank, Washington, D.C. these links. Some of the business surveys Wolfensohn,James D. 1999. "A Proposal for being conducted by the Bank will permit a Comprehensive Development Frame- more in-deptlh analysis of how investment work." http://www.worldbank.org/1cd/ and other decisions by firms are influenced cdf-text.htm by poor service delivery, corrutption, policy This note was written b' Stephen Knack unpredictability, and other manifestations (Senior Research Economist, Public Sector (rouip, of malfunctioning institutiorns. PRE,A' Network and DIevelopment Resear nh Group) and NVick Mllanning (Senior Public Sec- Further reading tor Management Specialist, Public Sector G*rouip, Beck, Thorsten, George Clarke, Alberto Groff, PREM NAetwork). Philip Keefer, and Patrick Walsh. 2000. If you are interested in similar topics, consider "New Tools and New Tests in Comparative joining the Public-Expenditure Managemnent The- Political Economy: The Database of Polit- matic Group (contact TVinaya Swaroop, x37189) ical Institutionis." Policy Research Working or the Administrative and Civil Service Rejorm Paper 2283. World Bank, WVashington, D.C. Thematic Group (contact Nick Manning, Kaufmranni, Daniel, Aart Kraay, and Pablo x31202). For more information, click on The- Zoido-Lobaton. 1999. "'Governance Mat- matic Groups on PRE'Mnet. This note series is inten(led to summarize good practice and kev policy find- hN h M ings on PREM-related topics. The views expressed in these notes are those of the autlors and do not necessarily reflect the views of the World Bank. PREM- notes are distributed widely to Bank staff and are also available on the P'RLM NWVeb site (http://prem). If yoti are interested in writing a PREMniote, cmail vour idea to Sarah Nedolast. For additional copies of this PRENMnotc please Povet RedunonandfvonomeManagemnt contact the PREM Advisory Service at x87736. Prepared for World Bank staff