The World Bank OCTOBER NUMBER 7 PU BLIC SECTO R New frontiers in diagnosing and combating corruption Corruption flourishes where policies provide incentives for it and restraining institutions are weak. Diagnosing corruption helps a country understand the short- comings in its policies and institutions and design a strategy to strengthen the state's performance. Over the past year the World Bank has and social costs of corruption. And it estab- helped Albania, Georgia, and Latvia mea- lishes a baseline against which the successes Empirical surveys sure corruption and design strategies to and failures of reform can later be mea- combatitandimprovegovernance.Allthree sured. Repeated surveys, starting 18 to 24 can provide the countries are now refining and imple- months after a reform program begins and menting these strategies. This note explains at least once a year thereafter, are key to information needed how empirical surveys can inform-and giving the government the information it transform-the policy dialogue, so that a needs and refocusing reform efforts. to develop an workable anticorruption agenda can be established. It also highlights challenges What empirical approaches anticorruption in performing these surveys, and in trans- should be used? lating surveyresults into prioritiesfor insti- Until recently it was considered impossi- agenda tutional reform. ble to systematically measure corruption in government institutions and assess its eco- Why measure corruption? nomic and social costs. Data consisted of Implementing reforms to improve gover- general measurements of public and expert nance is inherently difficult. Because such perceptions of aggregate corruption in a reforms dramatically diminish the rents country. But recent advances include cross- from corruption, they are often resisted country analysis of data on perceptions of by senior officials, other politicians, and corruption against institutional and other bureaucrats. Yet such resistance can often correlates, to better understand its causes becloakedbythelackofconcreteevidence and consequences. These studies have on corruption and by the assumption-now improved our understanding of corruption disproven-that corruption cannot be mea- and helped identify potential problems in sured. When such evidence is available, the countries' institutional arrangements. For debate on corruption can be depoliticized example, cross-country analysis shows that and its focus shifted to substantive issues. corruption is higher in countries that Measuring corruption offers other ben- repress civil liberties. Such "flags" do not, efits as well. It can help establish priorities however, provide the country-specific detail forreformbyidentifyingactivitiesandagen- that is needed to depoliticize the policy cies where corruption is concentrated. It debate and design rigorous anticorruption educates the public about the economic agendas. FROM THE DEVELOPMENT ECONOMICS VICE PRESIDENCY AND POVERTY REDUCTION AND ECONOMIC MANAGEMENT NETWORK The newest frontier in the fight against lic funds, theft of state property, bribery to corruption is to survey the parties to cor- shorten processing time, bribery to obtain ruption directly and simultaneously-includ- monopoly power, and bribery in procure- ing household members, enterprise ment. In Georgia the most common form managers, and public officials-and ask them of corruption (from this short list) is embez- about the costs and private returns of paying zlement of public funds. In Albania and bribes to obtain public services, special priv- Latvia the most common form is theft of ileges, and governmentjobs. Until recently state property. In addition, bribery in pro- skeptics believed that parties to corruption curement is common in all three countries. had an incentive to underreport it. But with Respondents are appropriate survey instruments and inter- Institutional causes of corruption differ, viewing techniques, respondents are willing suggesting different priorities for reform willing to discuss to discuss agency-specific corruption with In Albania a weak judiciary is one of the remarkable candor. Even with underre- main causes of corruption; regulatory fail- agency-specific porting and nonresponses to some sensitive ures are much less important. Regulatory questions, the results offer telling lower- failures are more serious in Georgia and corruption with bound estimates of corruption. Latvia, both in terms of excessive regula- The limits of different empirical measures tions and the discretion granted to regula- remarkable candor ofcorruptionpointtothedesirabilityofusing tors enforcing them. The data provide multiple approaches and data from differ- information that can help establish priori- ent sources. Consistent findings across these ties in each of these areas. For example, approaches and sources significantly enhance detailed statistics were collected on the bribes the reliability and ease of acceptance of the paid by enterprises to regulators in differ- empirical evidence-as well as the credibil- ent agencies. This information can be used ity of the actions it suggests. to establish which agencies are receiving the largest share of side payments (figure 1). What are the early results? Detailed surveys of corruption were con- Enterprises would pay higher taxes if ducted in Albania, Georgia, and Latvia. corruption were eliminated Preliminary results provide a startling pic- Corruption has serious implications for pub- ture of systemic corruption that hurts pub- lic finance (table 1). A large number of small lic welfare, taxes private sector activity, and bribes are paid to officials to avoid paying is deeply institutionalized. taxes, customs duties, and other liabilities to the state. Moreover, other types of bribes (such There are many types of corruption, and as unofficial payments to public officials for each country 's pattern is distinct special privileges, such as a favorable judi- Respondents reported many types of cor- cial decision, that do not have direct fiscal ruption, including embezzlement of pub- implications) may crowd out payment of tax Table i Corruption and lost tax revenue (percent) Indicator Albania Georgia Latvia Enterprises willing to pay higher taxes if corruption were eliminated 53 71 30 Additional taxes as a share of revenue of those enterprises willing to pay higher taxes if corruption were eliminated 11 22 15 Additional taxes as a share of revenue all enterprises are willing to pay if corruption were eliminated 6 16 4 Source: 1998 World Bank survey of 483 enterprise managers in Latvia (Latvia Facts), 350 managers in Georgia (GORBI), and 356 managers in Albania (ACER). Figure 1 Where does corruption occur in Georgia? Contact with Phone line Enterprise road police installations registration Leases of State banking Building state-owned services permits commercial W and x _ g I _-: -o / \ ~Water and real estate Property electricity registration _ 1I-/\ s I - - I \ se~~~~~~rvices The burden of Border crossing Weights and at customs l -measures corruption is much -inspections Fire and greater for poorer ~1~YJ ~ sanitary Exportaandialinspections households import permits Clearances for procurement or privatization financial inspections Source: World Bank-GORBI survey of 350 enterprises, May 1998. and other liabilities to the state. Indeed, cor- corrupt, suggesting that corrupt officials ratio- ruption is quite costly for firms: in Albania nally "invest"when buying their public office. and Latvia bribes account for 7 percent of The pattern of these payments differs, revenue in firms that admit to paying them. however. In Latvia ministerial positions are In Georgia bribes account for 15 percent of purchased more often than in Albania and firms' revenue. Lost fiscal revenues are high Georgia, and lower-level positions are pur- in all three. countries, especially Georgia. chased less often. This pattern suggests that grand corruption may be more of a prob- Corruption disproportionately hurts the lem in Latvia, while petty corruption is more poor serious in Albania and Georgia. In Georgia 14 percent of households admit to paying bribes and in Latvia, 12 percent. What role did the data play? Although richer households are more likely The anticorruption programs in Albania, to pay bribes, the burden ofcorruption-mea- Georgia, and Latvia share several features. All sured as the fraction ofincome paid in bribes- three countries first sought assistance from is much greater for poorer households. the Bank in designing reforms to improve governance. Given the inherent difficulties Bureaucrats pay for lucrative positions of public sector reform, strong client own- In Albania, Georgia, and Latvia the price of ership was essential. Next the countries com- obtaining "high rent" positions iswell known mitted to open and transparent among public officials and the general pub- policymaking-including collecting detailed lic, suggesting that corruption is deeply insti- data on corruption and sponsoring public tutionalized (figure 2). Higher prices are paid workshops to discuss the data and the policy forjobs in agencies and activities that house- agenda. Finally, the govern ments initiated holds and enterprises report to be the most policy processes that should culminate in anti- Figure 2 Purchasing public positions Customs @gMU L g@.l&t5@2,@ @° inspectors Tax inspectors Natural resource @ licensers Judges X Albania 3@g3 lilL2IA; .W. m Georgia Ordinary police m Latvia It will be essential Inesigtos ,-1 Investigators to continue to andprosecutors Local officials refine the Ministers methodology for 0 10 20 30 40 50 60 Percentage of public officials believed to have purchased their positions transforming survey Source: 1998 World Bank surveys of 218 public officials in Latvia (with Latvia Facts), 350 public officials in Georgia (with GORBI), and 97 public officials in Albania (with ACER). evidence into corruption programs for regulatory reform, achievement in the policy dialogue on cor- reform priorities civil service and public administration reform, ruption. Still, challenges remain. public finance reform, and judicial reform. In addition, efforts were made to promote Refining the methodology the participation of civil society. It will be essential to continue to refine the Next the data collection and dissemina- methodology for transforming survey evi- fion began. In Albania and Georgia the data dence into reform priorities. The new diag- were presented in workshops that were open nostic tools are at the frontier in measuring to members of all branches of government, corruption in specific agencies, but expe- the business community, and civil society. In rience with measuring corruption's eco- Albania the data dramatically altered the pol- nomic and social costs is at an infant stage. icy debate, moving it from vague, unsub- The direct financial cost of corruption in stantiated accusations to a process focused different agencies is a key indicator of cor- on empirical evidence and systemic weak- ruption's impact on welfare and private sec- nesses. In addition, the survey results were tor development. The willingness of firms printed on the front page of every major news- to pay additional taxes if corruption were paper. In Georgia the immediate effect was eliminated also provides an important mea- less dramatic but still significant. In Latvia the sure of its full costs, as do measures of pub- government opened the debate on corrup- lic preferences for reducing corruption tion with a workshop in early 1997. In June relative to other policy objectives. Still, other 1998 the new government presented the basic approaches are needed. And while current program at a public conference, prior to the instruments diagnose the prevalence of completion of the survey. Presentation and many forms of corruption, innovative tools analysis of the data are expected later in the are needed to measure grand corruption. year, and will be used to refine the program Finally, the Bank needs to strengthen its and establish priorities. capacity to help policymakers integrate the lessons from these empirical diagnostics What challenges lie ahead? with the political feasibility of different pol- The collection, analysis, and dissemination icy recommendations. It is worth empha- of country-specific data mark a special sizing that such feasibility is not a constant: the broad empowerment that conies from ness community to participate in the policy this empirical, technocratic approach can process. "Watchdogs" outside government tilt the balance toward reformists, in alliance can be established to monitor the state's com- with civil society. mitment to the anticorruption agenda. In Several strategies can be used to establish addition, the government may be able to cred- reform priorities. One strategy being used ibly commit to reform by allowing private in Albania, Georgia, and Latvia is to conduct competition with some public services-for focus groups in which different constituen- example, allowing private forms of dispute cies discuss, among other issues, petty and resolution as an alternative to the judiciary. grand corruption and the feasibilityof poten- Finally, data collection needs to be institu- tial reforms. Another strategy to address the tionalized, so that statistics on corruption can Anticorruption political feasibility of reform is to assess a coun- be updated at least once a year. Broad dis- try'sreadinessto reform, analyzingwhateffect semination of these statistics can further efforts should institutional and policy reforms will have on empower stakeholders to continue reforms. key stakeholders. In Albania, Georgia, and Latvia, NGOs focus on reforming helped develop policies. One strategy that Implementing reforns was used to encourage ongoing participation public policies The most difficult stage of an anticorruption was to hire a surveyor who could potentially program is after the survey data have been serve as a watchdog against future corrup- and institutions collected, analyzed, and disseminated-when tion. This surveyor's reputation as an inde- the government must start introducing pendent professional was crucial in the face reforms that tackle fundamental sources of of intense public scrutiny. corruption. A natural temptation for a coun- try's leader is to launch the program by ask- Designing agency-speciJfic surveys and other ing for the resignations of senior officials who tools manage the most corrupt agencies. But in The Bank has already begun to design sur- many countries corruption is so pervasive vey instruments whose purpose is to col- andsystemic thatitcannotbeaddressed solely lect detailed information on behavior in by individualizing the problem. some of the most dysfunctional government Ultimately, anticorruption efforts should agencies. In addition, the Bank is advising focus on reforming public policies and insti- countries to gather hard data on corrup- tutions, with explicit high-level leadership tion in the delivery of specific services. For and commnitment. Survey data provide a pic- example, corruption in the supply of saline ture of the most dysfunctional activities and by public hospitals can be established by hence priorities for reform. Based on the comparing the price-after accounting for country-specific priorities that have emerged, transport and other idiosyncratic costs-of the challenge is to implement credible reforms saline purchased by different hospitals. in each area. Such action has already begun in some countries. Latvia, for example, has Emerging conclusions initiated reforms to reduce corruption in cus- Diagnostic surveys are a useful and power- toms and tax administration. But challenges ful tool for unbundling corruption and iden- remain-for instance, what can be done to tifying specific correlates, its costs, and reform Albania's thoroughly corruptjudicial problem areas. Surveys can focus the polit- system, and how can deregulation be imple- ical dialogue on concrete areas for reform. mented in Georgia and Latvia when vested The public transparency generated by hard interests in government ministrieswill devise data and a technocratic approach can fuel ways of continuing to extract rents? a participatory process that mobilizes and energizes civil society and generates pres- Sustaining reforms sures for reform. Reforms can be sustained by encouraging But rigorous surveys and in-depth analy- all branches of the state, civil society, and busi- sis of their findings are just one input into a multifaceted process for combating cor- "Diagnosing and Combating Corruption: ruption. And they can be misused if they A Framework with Application to are not carried out and placed in the Transition Economies." World Bank, proper context. Surveyors need to be inde- Washington, D.C. pendent and methodologies rigorous; oth- Transparency International and the World erwise the results can be misleading. Bank's Economic Development Institute. Moreover, there is a risk of scapegoating 1998. "New Perspectives on Combating unless the focus shifts quickly to credible Corruption." Paper prepared for reforms of underlying public sector dys- Transparency International's annual functions. meeting, Kuala Lumpur, Malaysia, Data are powerful Data are powerful in mobilizing support September 11-1 6, and for the IMF-World for reforms, but so are the vested interests Bank annual meetings, Washington, D.C., in mobilizing resisting such reforms. The challenge for October 6-8. political leaders, civil society, and donors is support for to capitalize on the insights and momen- This note was written by Daniel Kaufmann tum generated by the diagnostics, move (DivisionManager;RegulatoiyReformandPrivate reforms, but so from diagnostics to action, and make mean- Enteiptise), SanjayPradhan (SectorLeader, Europe ingful progress on the ground. and Central Asia), and Randi Ryterman (Senior are the vested Public SectorManagement Specialist, Europe and Further reading Central Asia). The authors are grateful to Jim interests resisting Kaufmann, Daniel, SanjayPradhan, and Randi Anderson for data analysis and Bill Moore for Ryterman, with James Anderson. 1998. graphics support for this note. such reforms This note series is intended to summarize good practice and key policy findings on PREM-related topics. PREMnotes are distributed widely to Bank staff and are also P D | available on the PREM website (http://prern). If you are interested in writing a PREMnote, email your idea to Asieh Kehyari. For additional copies of this PREMnote please contact the PREM Advisory Service at 87736. Prepared for World Bank staff