11Finance & Development / March 1998

Corruption: Causes, Consequences, and Agenda for Further Research

PA O L O M A U R O

What do we know about corruption, how do we knowit, and what steps do we needto take to improve our under-standing of corruption and enhance governments’effectiveness in combating it?

VER THE LAST few years, theissue of corruption—the abuse ofpublic office for private gain—hasattracted renewed interest, both

among academics and policymakers. Thereare a number of reasons why this topic hascome under fresh scrutiny. Corruption scan-dals have toppled governments in bothmajor industrial countries and developingcountries. In the transition countries, theshift from command economies to free mar-ket economies has created massive opportu-nities for the appropriation of rents (that is,excessive profits) and has often been accom-panied by a change from a well-organizedsystem of corruption to a more chaotic anddeleterious one. With the end of the coldwar, donor countries have placed lessemphasis on political considerations in allocating foreign aid among developingcountries and have paid more attention tocases in which aid funds have been misusedand have not reached the poor. And sloweconomic growth has persisted in many

countries with malfunctioning institutions.This renewed interest has led to a newflurry of empirical research on the causesand consequences of corruption.

Economists know quite a bit about thecauses and consequences of corruption. Animportant body of knowledge was acquiredthrough theoretical research done in the1970s by Jagdish Bhagwati, Anne Krueger,and Susan Rose-Ackerman, among others(Mauro, 1996). A key principle is that corruption can occur where rents exist—typically, as a result of government regulation—and public officials have discre-tion in allocating them. The classic exampleof a government restriction resulting inrents and rent-seeking behavior is that of an import quota and the associated licensesthat civil servants give to those entre-preneurs willing to pay bribes.

More recently, researchers have begun totest some of these long-established theoreti-cal hypotheses using new cross-countrydata. Indices produced by private ratingagencies grade countries on their levels ofcorruption, typically using the replies tostandardized questionnaires by consultantsliving in those countries. The replies aresubjective, but the correlation betweenindices produced by different rating agen-cies is very high, suggesting that mostobservers more or less agree on how cor-rupt countries seem to be. The high pricespaid to the rating agencies by their cus-tomers (usually multinational companiesand international banks) constitute indirectevidence that the information is valuable.

These indices are obviously imperfectowing to their subjective nature, but canyield useful insights.

Causes of corruption Since the ultimate source of rent-seeking

behavior is the availability of rents, corrup-tion is likely to occur where restrictions andgovernment intervention lead to the pres-ence of such excessive profits. Examplesinclude trade restrictions (such as tariffsand import quotas), favoritist industrialpolicies (such as subsidies and tax deduc-tions), price controls, multiple exchange ratepractices and foreign exchange allocationschemes, and government-controlled provi-sion of credit. Some rents may arise in theabsence of government intervention, as inthe case of natural resources, such as oil,whose supply is limited by nature andwhose extraction cost is far lower than itsmarket price. Since abnormal profits areavailable to those who extract oil, officialswho allocate extraction rights are likely tobe offered bribes. Finally, one would expectthat corruption is more likely to take placewhen civil servants are paid very low wagesand often must resort to collecting bribes inorder to feed their families.

While all of the hypotheses describedabove are empirically testable, in the sensethat data are available for that purpose,only a few have actually been tested. Whatempirical studies have been done supportcertain hypotheses: namely, that there isless corruption where there are fewer traderestrictions; where governments do not

Paolo Mauro, an Italian national, is an Economist in the IMF’s European I Department.

O

engage in favoritist industrial policies; and perhaps where natural resources aremore abundant; and that there is somewhatless corruption where civil servants arepaid better, compared with similarly quali-fied workers in the private sector (VanRijckeghem and Weder, 1997).

Consequences of corruptionFrom economic theory, one would expect

corruption to reduce economic growth bylowering incentives to invest (for bothdomestic and foreign entrepreneurs). Incases where entrepreneurs are asked forbribes before enterprises can be started, orcorrupt officials later request shares in theproceeds of their investments, corruptionacts as a tax, though one of a particularlypernicious nature, given the need forsecrecy and the uncertainty as to whetherbribe takers will live up to their part of thebargain. Corruption could also be expectedto reduce growth by lowering the quality ofpublic infrastructure and services, decreas-ing tax revenue, causing talented people toengage in rent-seeking rather than produc-tive activities, and distorting the composi-tion of government expenditure (discussedbelow). At the same time, there are sometheoretical counterarguments. For example,it has been suggested that governmentemployees who are allowed to exact bribesmight work harder and that corruptionmight help entrepreneurs get aroundbureaucratic impediments.

One specific channel through which cor-ruption may harm economic performance isby distorting the composition of govern-ment expenditure. Corrupt politicians maybe expected to spend more public resourceson those items on which it is easier to exactlarge bribes and keep them secret—forexample, items produced in markets wherethe degree of competition is low and itemswhose value is difficult to monitor. Corruptpoliticians might therefore be more inclinedto spend on fighter aircraft and large-scaleinvestment projects than on textbooks andteachers’ salaries, even though the lattermay promote economic growth to a greaterextent than the former.

Empirical evidence based on cross-country comparisons does indeed suggestthat corruption has large, adverse effects onprivate investment and economic growth.Regression analysis shows that a countrythat improves its standing on the corruptionindex from, say, 6 to 8 (0 being the most cor-rupt, 10 the least) will experience a 4 per-centage point increase in its investment rateand a 0.5 percentage point increase in its annual per capita GDP growth rate

(Mauro, 1996). These large effects suggestthat policies to curb corruption could have significant payoffs. The associationbetween corruption and low economicgrowth remains broadly unchanged whenestimated for a group of countries withextensive red tape. Therefore, there is nosupport for the claim that corruption mightbe beneficial in the presence of a slowbureaucracy. The most important channelthrough which corruption reduces economicgrowth is by lowering private investment,which accounts for at least one-third of cor-ruption’s overall negative effects. At thesame time, the remaining two-thirds of theoverall negative effects of corruption on eco-nomic growth must be felt through otherchannels, including those mentioned above.While it is difficult to disentangle thoseother channels, there is some evidence thatone of them—the distortion of governmentexpenditure—plays a significant role.

Based on cross-country comparisons, itseems that corruption alters the composi-tion of government expenditure: specifi-cally, corrupt governments spend less oneducation and perhaps health, and proba-bly more on public investment. Regressionanalysis shows that a country that im-proves its standing on the corruption indexfrom 6 to 8 will typically raise its spendingon education by 1⁄2 of 1 percent of GDP, aconsiderable impact. This result is a matterfor concern, because there is increasing evidence that educational attainment fos-ters economic growth.

Of course, empirical results related to aphenomenon that is, by its very nature, dif-ficult to measure must be treated with ahigh degree of caution. Two issues thatmerit special attention in this context arethose of causality and the possible role ofother forms of institutional inefficiency.

Why do countries judged to be corruptexperience slow economic growth? Is it thatcorruption harms growth or simply thatlow growth leads consultants to give badcorruption grades to a country? To dealwith this issue, one can take variables (suchas a country’s colonial history or the extentto which its population is divided alongethnolinguistic lines) that happen to be correlated with corruption but have noeffect on economic growth or governmentspending other than through their impacton the efficiency of institutions, and usethem as instrumental variables in theregression analysis. Through this statisti-cal trick, it is possible to get around prob-lems relating to the subjectivity of thecorruption indices, and it can be shownthat corruption—together with other forms

of institutional inefficiency—causes loweconomic growth.

Corruption is most prevalent where thereare other forms of institutional inefficiency,such as political instability, bureaucratic redtape, and weak legislative and judicial sys-tems. This raises the question of whether itcan be established that corruption, ratherthan other factors correlated with it, is thecause of low economic growth. Regressionanalysis provides some evidence that if onecontrols for other forms of institutional inef-ficiency, such as political instability, corrup-tion can still be shown to reduce growth.Nevertheless, it is hard to show conclusivelythat the cause of the problem is corruptionalone, rather than the institutional weak-nesses that are closely associated with it.The truth is that probably all of these weak-nesses are intrinsically linked, in the sensethat they feed upon each other (for example,red tape makes corruption possible, andcorrupt bureaucrats may increase theextent of red tape so they can extract addi-tional bribes) and that getting rid of corrup-tion helps a country overcome otherinstitutional weaknesses, just as reducingother institutional weaknesses helps it curbcorruption.

Agenda for further research While there is a well-established body of

theoretical knowledge, as well as some ten-tative results on the causes and conse-quences of corruption, several morequestions need to be answered to enablegovernments to design effective policiesaimed at curbing corruption.

If the costs of corruption are sohigh, why don’t governments get ridof it? A possible answer is that once a cor-rupt system is in place, and a majority ofpeople operate within that system, individu-als have no incentive to try to change it or torefrain from taking part in it, even if every-body would be better off if corruption wereto be eliminated. Consider the followingexamples:

• You live in a society where everybodysteals. Do you choose to steal? The probabil-ity that you will be caught is low, becausethe police are very busy chasing otherthieves, and, even if you do get caught, thechances of your being punished severely fora crime that is so common are low.Therefore, you too steal. By contrast, if youlive in a society where theft is rare, thechances of your being caught and punishedare high, so you choose not to steal.

• You are a new junior civil servant in anadministration where everybody, includingyour superiors, is very corrupt. Somebody

Finance & Development / March 199812

offers you a bribe to help him avoid payingtaxes. You decline the offer. A few hourslater, you receive a telephone call from yourboss, who would have liked a cut of yourbribe. Your boss suggests that if you treat afriend nicely (by accepting the bribe), youmay be promoted, while if you don’t, youwill be transferred to a remote provincialoffice. You then take the bribe and share itwith your boss and colleagues. If, instead,the administration in which you work isvery honest, you are likely to behave hon-estly to avoid the risk of being fired.

• Individuals A and B are members ofthe same government. Suppose, on the onehand, that A is very corrupt and has estab-lished a private bribe-collection system for her own gain. The need to pay substan-tial bribes reduces entrepreneurs’ incen-tives to invest and imposes asignificant burden on economicgrowth. Citizens realize that eco-nomic growth is being harmed bythe corrupt government, thoughthey may not know exactly who issoliciting bribes. Therefore, theydecide not to reelect the govern-ment. This shortens B’s horizon, makinghim more inclined to extract a large propor-tion of current output and to disregard anyensuing adverse effects on future output. Inother words, B will seek to obtain a largeslice of the cake today since he knows thatthe government that he participates in willsoon be ousted. On the other hand, follow-ing a similar line of reasoning, if A does notcollect bribes, then B will also refrain fromdoing so.

The last example may provide an expla-nation not only for the persistence of corruption but also for the empirical obser-vation that, on average, countries that aremore corrupt tend to be more politicallyunstable. It also suggests that both corrup-tion and political instability may resultfrom the failure of members of the samegovernment or ruling elite to coordinatetheir actions. In that sense, corruption andpolitical instability may be two sides of thesame coin. This example may fit the casesof countries that are bedeviled by frequentcoups whereby corrupt regimes succeedone another. At the same time, it does notexplain a number of other relevant cases,such as those of dictators who haveremained in power for many years byallowing their supporters to collect largebribes, or those of governments formed by groups of individuals who have beenable to agree on bribe levels that are high, but not so high as to cause them to be ousted.

All of the above examples show that oncecorruption has become ingrained, it is verydifficult to get rid of. As a result, corruptiontends to persist, together with its adverseconsequences. This leads to an importantpolicy conclusion, which is consistent withinternational experience over the past fewdecades. Attempts to eliminate corruptiontend to succeed when reforms are under-taken in a very sudden and forceful mannerand are supported at the highest levels ofgovernment. However, an equally relevantquestion is what characteristics make coun-tries more likely to fall into a high-corrup-tion, low-growth trap.

Corruption breeds poverty, butdoes poverty breed corruption? Onestriking empirical finding is that poorercountries are usually considered to be more

corrupt. This result must be treated withcaution, since it may well be driven by theobservers’ perceptions. However, if oneassumes for a moment that this findingreflects a genuine correlation, it may beuseful to explore its sources. We have seenthat there is evidence that corruption lowerseconomic growth, thereby breeding povertyover time. At the same time, poverty itselfmight cause corruption, perhaps becausepoor countries cannot devote sufficientresources to setting up and enforcing aneffective legal framework, or because peoplein need are more likely to abandon theirmoral principles. Researchers have begun toanalyze the link between civil servants’wages and the extent of corruption. It hasbeen suggested that reasonable wages are anecessary condition for avoiding corrup-tion, though not a sufficient one.

Which forms of corruption areworse? Available indices of corruption aregeneral and do not distinguish betweenhigh-level corruption (such as kickbackspaid to a defense minister in exchange forhis country’s purchase of expensive jetfighter aircraft) and low-level corruption(such as petty bribes paid to a junior civilservant for expediting the issuance of adriver’s license). Nor do they distinguishbetween well-organized corruption andchaotic corruption. (When corruption iswell organized, the required amount andappropriate recipient of a bribe are wellknown, and payment guarantees that the

desired favor will be obtained.) Therefore,we still do not know which kinds of corrup-tion are more deleterious and should betackled first. Country-specific studies andanecdotal evidence suggest that high-leveland low-level corruption tend to coexist andreinforce each other. Thus, this distinctionmay not be relevant. On the other hand, thedistinction between well-organized corrup-tion and chaotic corruption may be morerelevant, since a fairly convincing theoreti-cal case can be made that the latter hasworse effects than the former.

Under a well-organized system of cor-ruption, entrepreneurs know whom theyneed to bribe and how much to offer them,and are confident that they will obtain thenecessary permits for their firms. It hasalso been argued that well-organized cor-

ruption is less harmful because,under such a system, a corruptbureaucrat will take a clearlydefined share of a firm’s profits,which gives him an interest in thesuccess of the firm. In contrast,under chaotic corruption, entre-preneurs may need to bribe several

officials, with no guarantee either that theywill not face further demands for bribes orthat the permits they seek will actually bedelivered. In addition, if multiple agentsrequest bribes from the same entrepreneurwithout coordinating bribe levels amongthemselves, they are likely to make exces-sive demands, with the result that entre-preneurial activity comes to a halt. Whilechaotic corruption seems a priori to bemore deleterious than well-organized cor-ruption, there is currently little empiricaldata available to test this hypothesis.

What is being done, and what elsecould be done? Many countries and insti-tutions have paid increasing attention to theproblem of corruption, and the debate onpossible policy options is still ongoing. Indeciding how to allocate aid funds, somedonor countries have begun to give moreimportance to recipient countries’ actions tocurb corruption. Member countries of theOrganization for Economic Cooperationand Development have acted to criminalizethe bribery of foreign public officials.International institutions, which havealways played an important role in reduc-ing the scope for corruption, are now givingmore prominence to the issue. For example,the IMF has always encouraged countriesto liberalize their economies (for example,by eliminating trade restrictions), terminateoff-budget operations, and ensure budgettransparency. The guidelines on gover-nance, which were approved by the IMF’s

13Finance & Development / March 1998

“Corruption and politicalinstability may be two sides

of the same coin.”

Executive Board in August 1997, formalizethe IMF staff’s involvement in such tasks.

An example of the remainingchallenges. One of the most difficult pol-icy issues is how to prevent corruptionfrom distorting government expenditure.This important issue goes to the heart ofdonors’ concerns about the possible misuseof aid funds. Donors find it difficult toensure that aid funds are spent wisely,because resources are fungible. For exam-ple, a donor may give aid funds to enablethe recipient country to build a school,which the recipient may indeed use for thatpurpose; however, the availability of aidfunds to build the school makes it possiblefor the recipient to use the resulting savingsto buy sophisticated weapons, whose purchase may provide more scope for ille-gally diverting funds into individuals’pockets than school construction could.How should this problem be addressed?Clearly, donor countries should pay atten-tion to the overall composition of govern-ment spending and not focus narrowly onhow their own funds have been spent, butmany donors may not have sufficientresources to do this. One possible approachthat has been suggested is to have an inter-national institution monitor the overallcomposition of government expenditure, as

a service to both the recipient country’s citi-zens and the donor community. Thisapproach would not, however, be easy toimplement. Recipient countries would prob-ably resist attempts by the rest of the worldto play a role in determining the composi-tion of their public spending. In addition, asa practical matter, it may be difficult toensure that spending items are not simplyrelabeled, with no real improvement in thecomposition of government expenditure.

How should policy effectiveness beassessed over the next decade? Wehave a reasonable theoretical understandingof the causes and consequences of corrup-tion, and have begun to get a sense of theextent of these relationships through empir-ical research. A consensus is emerging thatcorruption is a serious problem, and severalbodies in the international arena havebegun to take policy measures to curb it. Atthe policy level, although we may still be atthe stage of learning by doing, action isbeing taken. It is important to ensure thatten years from now, we can look back ontoday’s focus on corruption and observethat some concrete results were attained inthis domain. To that end, those bodies thatare taking action against corruption oughtto establish criteria to evaluate their poli-cies. Each entity would need to devise its

own evaluation criteria, and it should dothat now, so that the effectiveness of its poli-cies can be assessed accurately and fairlyover the next decade.

The need to define “concrete results”might appear to be a tall order in an areawhere quantification is difficult. A place to begin, however, could be the well-established body of knowledge on thecauses of corruption. For example, effortsto curb corruption could be assessed on the basis of how effective they were inbringing about the implementation of poli-cies known to reduce corruption, such asthe elimination of government restrictionsthat create rents.

Finance & Development / March 199814

Suggestions for further reading: Paolo Mauro, 1996, “The Effects of Cor-

ruption on Growth, Investment, and Govern-ment Expenditure,” IMF Working Paper 96/98(Washington: International Monetary Fund).

Vito Tanzi and Hamid Davoodi, 1997, “Cor-ruption, Public Investment, and Growth,” IMFWorking Paper 97/139 (Washington: Inter-national Monetary Fund).

Caroline Van Rijckeghem and BeatriceWeder, 1997, “Corruption and the Rate ofTemptation: Do Low Wages in the Civil ServiceCause Corruption?” IMF Working Paper 97/73(Washington: International Monetary Fund).

F&D

World BankPubl icat ions

For US customers, contact The World Bank, P.O. Box 7247-8619, Philadelphia, PA 19170-8619.

Phone: (703) 661-1580, Fax: (703) 661-1501. Shipping and handling: US$5.00. Airmail delivery outside

the US is US$13.00 for one item plus US$6.00 for each additional item. Payment by US$ check drawn on

a US bank payable to the World Bank or by VISA, MasterCard, or American Express. Customers outside

the US, please contact your World Bank distributor.

World BankPubl icat ions

V i s i t o u r W e b s i t e : h t t p : / / w w w . w o r l d b a n k . o r g

World Bank Spurs Fight Against AIDS—The Daily Telegraph

AIDS Surge is Forecast for China, India, and Eastern Europe—New York Times

These are just two of the headlines that appeared worldwide upon release of the

World Bank’s largest report to date on the growing global AIDS crisis. The

report, Confronting AIDS: Public Priorities in a Global Epidemic, addresses

this devastating disease from the perspective of government policymakers in

developing countries—policymakers outside the health sector—who shape and

finance national efforts to combat the epidemic.

Confronting AIDS addresses an overarching question: Given the limited public resources in developing coun-

tries, which measures for confronting the epidemic should be public priorities? The report is the product of 18

months of research—with extensive support from UNAIDS and the European Commission—and contains a wealth

of information and statistics that you cannot afford to ignore. Confronting AIDS can now be yours for only $30.00

plus shipping and handling by ordering directly from the World Bank. Please see below for ordering information.