A Vice or a Virtue?

FJELDSTADtute, Bergen, Norway

nd

atu

tioer

One

due to corruption and tax evasion (Alm,

mestic revenue mobilization is considered the

reforms in recent years have focused on re-

1988;

mpor-

increase corruption. Actually, as we elaborate

World Development Vol. 31, No. 8, pp. 14591467, 2003 2003 Elsevier Ltd. All rights reserved

Printed in Great Britain0305-750X/03/$ - see front matter

-750X(03)00089-5

programme Taxation, aid and democracy and NUFU

Lambsdor, Lise Rakner, Michael Schinke, Alice

Sindzingre, Ole Therkildsen and the referees for valuable

comments. Points of view and possible errors are entirelydesigning the tax structures and improvingtax administration. Addressing scal corrup-root of the adjustment and growth problemsfaced by many poor countries (Chand &Moene, 1999). To alleviate this problem, tax

through the programme Administrative authority and

trust between state and society. We would like to thank

Jens Andvig, Wendy Belcher, William Ellis, JohannKrugman, Alm, Collins, & Remolina, 1992;Richupan, 1984). This tax-base erosion isparticularly damaging since insucient do-

*This paper was prepared with nancial support fromthe Research Council of Norway through the researchBahl, & Murray, 1991; Bird, 1990, 1992;impacts of corruption loom largest is in theassessment and collection of taxes (Galtung,1995). Studies in various developing countriesindicate that it is not uncommon that half ormore of the taxes that should be collectedcannot be traced by government treasuries

tance of incentive schemes in motivating taxocers to work harder and in accordance withthe overruling objective of improving revenueperformance (Chand & Moene, 1999; Das-Gupta & Mookherjee, 1998; Mookherjee,1997). Such incentive schemes may, however,tion athree arguments to support our view. First, while an increase in corruption may raise revenues inthe short run, in general, the opposite will be the case in the longer run. Second, the instrumentalvalue of reducing corruption goes far beyond its eects on tax evasion and tax revenues. Acceptingcorruption as a policy strategy to increase tax revenues may undermine values of democracy andgood governance. Third, eliminating corruption should be considered an end in itself. Thus,contrary to recent suggestions on incentive reforms in tax administration, the reasonable startingpoint for policy debates in this area should still be that an increase in scal corruption is not anappropriate instrument for raising tax revenues. Sustained development cannot occur in aninstitutional framework that fosters corruption and extra-legal tax enforcement. 2003 Elsevier Ltd. All rights reserved.

Key words corruption, tax evasion, tax administration, incentives

1. INTRODUCTION

of the areas of government where the

grated part of this strategy (Klitgaard,Toye & Moore, 1998).A growing literature emphasizes the iparadox justify policies that stimulate corruption? Our answer is no, and this note puts forwardFiscal Corruption:

ODD-HELGEChr. Michelsen Insti

a

BERTIL TUNNorwegian School of Economics

Chr. Michelsen Insti

Summary. Recent literature on tax administraallowing scal corruption. By strengthening thargued that tax evasion may be reduced and tax

doi:10.1016/S0305www.elsevier.com/locate/worlddevnd tax evasion has become an inte-

145GODDEN *nd Business Administration andte, Bergen, Norway

n in poor countries suggests there are virtues inbargaining power of corrupt tax ocers, it isevenues increased. But does such an intriguingour responsibility. Final revision accepted: 1 June 2002.

9

nating corruption is an end in itself. In ourview, any reasonable conception of a good so-

WORLD DEVELOPMENT1460ciety should count corruptionthat is, theabuse of public oces and rules for personalgainas intrinsically bad (Section 5).

2. THE VIRTUE OF FISCALCORRUPTION

How may corruption contribute to reducingtax evasion and thereby increasing tax reve-nues? The essential link, studied by Mookherjee(1997) among others, is based on the idea thatthe possibility of negotiating bribes from eva-sive taxpayers motivates corrupt tax ocers towork harder in order to detect evasion. 1 This isanticipated by the taxpayers, and hence taxevasion is less attractive because it is morelikely to be detected.Since corruption works to make tax evasion

less appealing and thereby increases tax reve-nues, it might be a good idea to design a bonussystem for tax collectors that mimics or com-petes with the bribery system already in place inmany tax administrations. Actually, this hasbeen attempted in Ghana (Chand & Moene,1999) and has been suggested in several othercountries. The intention behind a bonus systemis to initiate more work eort among tax col-lectors by promising them a share of the taxin Section 2, a standard way of justifying in-centive schemes is by showing that suchschemes strengthen the position of corrupt taxocers and thereby make tax evasion less at-tractive. Nevertheless, it is argued, in caseswhere the eect on taxpayers compliance andgovernment revenues is positive, incentiveschemes are still justied: Eliminating cor-ruption is. . . not an end in itself; eects on taxevasion and revenues are more fundamental(Mookherjee, 1997, p. 16).In this note we put forward three arguments

that question this way of defending incentiveschemes. First, while an increase in scal cor-ruption may contribute to an increase in taxrevenues in the short run, it is highly implau-sible that such an increase is sustainable (Sec-tion 3). The dynamics of corruption suggestthat policies of this kind will decrease tax rev-enues in the longer run. Second, a muchbroader view of this problem is needed in thedevelopment debate, including the eects ofscal corruption on accountability and gov-ernment legitimacy (Section 4). Third, elimi-revenues. This is the way it works for noncor-rupt tax collectors, who within a bonus system,aim at detecting evasion because this increasestax revenues and thereby their income. Butwhat about corrupt tax collectors?Consider a bribe as the outcome of a negoti-

ation between an evasive taxpayer and a corrupttax collector. The introduction of a governmentbonus certainly makes the bribe less attrac-tive for the corrupt tax collector, because hehas to give up the bonus when accepting thebribe. But, this does not necessarily insure thatthe tax collector becomes less corrupt. Actu-ally, it makes him stronger in his negotiationswith the taxpayer, and as a result he receivesa larger part of the pie not reported to the taxauthorities. Thus, the bonus system providesincentives for the corrupt tax collector aswell (by increasing the negotiated bribe), andmay thereby contribute to increase tax reve-nues. This happens because the bonus systemstrengthens the position of the corrupt tax col-lector and therefore may increase overall cor-ruption. 2

Generally, the implications of a bonus systemdepend on whether the tax administration con-sists of corrupt or noncorrupt tax collectors. Inboth cases, we might experience an increase inoverall tax revenues, but in the case of corrupttax collectors, the bonus system may also leadto increased corruption. Hence, in a situationwhere there is a mixture of corrupt and non-corrupt tax collectors, it seems straightforwardto say that we have to make a tradeo betweenthe gain of more revenues and the problem ofmore corruption when evaluating a bonus sys-tem. But, this is not how bonus systems arejustied in the theoretical literature on corrup-tion and incentives.Mookherjee (1997), for example, considers

bonus systems in the context of corrupt taxcollectors only, and then argues for the need togo beyond the question of what levels of cor-ruption arise and examine induced eects on taxcompliance and audit incentives (p. 13). Hence,when evaluating bonus systems, Mookherjeesolely considers the possible gain in tax revenuesfollowing from the fact that the position ofcorrupt tax ocers is strengthened. In our view,this way of justifying bonus systems should berejected because it does not capture the long-term eects of an increase in corruption on taxrevenues and government legitimacy. We nd ithighly implausible that sustained developmentcan occur in an institutional framework thatfosters corruption and extra-legal tax enforce-

ment.

FISCAL CORRUPTION 1461Mookherjee is, of course, aware of the vicesof corruption, and stresses the important pointthat when considering incentive reforms wealso need to take into account the possibilityof wider administrative reforms, includingchanges in supervision systems, informationand monitoring procedures. More precisely, hesuggests that if incentive reform causes vari-ous undesired side eects, the range of policyinstruments must be expanded to moderatetheir eects (p. 8). This is however, a prob-lematic position within the present mode ofreasoning. If one considers an increase in cor-ruption an undesirable side eect to be mod-erated, then an incentive reform cannot bejustied by showing that it increases tax reve-nues by (possibly) inducing more corruption.Such a justication would be undermined bythe policies aiming at reducing corruption.Let us examine another example of how the

strengthening of the position of corrupt taxcollectors has been considered part of a vir-tuous circle in reforming tax administrations.Chand and Moene (1999) are concerned withthe need for noncorrupt higher-level bureau-crats in tax administration when introducingbonus systems, and illustrate this by the fol-lowing story. Consider the case of a corrupt taxcollector who tries to negotiate a bribe from anevasive taxpayer in return for underreportinghis tax liabilities. If they do not reach anagreementthat is, if the taxpayer refuses topay the bribe and the collector reports theevasiona higher-level bureaucrat is informedabout the true tax liability of the taxpayer andsettles the case. If the higher-level bureaucrat iscorrupt, the evasive taxpayer pays him a bribeand provides taxes only on the underreportedtax liability. In contrast, a noncorrupt higher-level bureaucrat collects the taxes on the truetax liabilities. Therefore, the presence of anoncorrupt higher-level bureaucrat strengthensthe position of the corrupt tax collector in thenegotiations with the taxpayer. Why? Becauseit becomes less important for the corrupt col-lector to reach an agreement with the taxpayer.The collector knows that as long as the higher-level bureaucrat is not corrupt, he will receivethe bonus on the whole tax liability if he doesnot reach an agreement with the taxpayer. Thiswould not be the case if the higher-level bu-reaucrat was corrupt. The tax collector wouldthen not receive any bonus. Hence, in order tohave an eective bonus system, noncorrupthigher-level bureaucrats are required. They

make it possible for the corrupt tax collector toget a higher bribe by strengthening the collec-tors bargaining position in relation to taxpay-ers and thereby also stimulate his work eort.Consequently, this will contribute to an in-crease in tax revenues in the short run. But isthis really the virtue of having noncorrupthigher-level bureaucrats in tax administrationwhen introducing a bonus system?Let us close this section by briey pointing at

some of the mechanisms we expect to reect thereal virtues of an incentive reform. First, asalready stressed, an eective bonus system in-duces more eort among noncorrupt tax col-lectors. Second, and maybe more important, abonus system, within an administration con-taining noncorrupt higher-level bureaucrats,may cause less corruption among tax collec-tors. 3 Let us provide a simple illustration ofthis point. Assume that a company reports theprot R, whereas the true prot is P. The taxrate is t and the bonus rate is c. All tax collec-tors assign a certain disvalue 1=m to accepting abribe, mP 1, where m 1 would imply that thetax collector is indierent between receiving acertain amount of money as a bribe or as aregular bonus. 4 If the tax collector does notaccept the bribe and reports the evasion to anoncorrupt higher-level bureaucrat, then hereceives a bonus on the true prot. In this case,a collector would only accept a bribe b if: 5

ctR b=m > ctP 1Obviously, the bribe will not exceed the taxessaved on the underreported amount tP R. 6Hence, on the basis of (1), we nd a cut-ovalue m such that no tax collector having avalue above m would choose to be corrupt. 7

m 1=c 2From (2), we can see that an increase in thebonus c decreases m which indicates that thenumber of corrupt tax collectors should de-crease in an eective bonus system. 8

To summarize, there are important positiveeects from incentive reforms in the tax ad-ministration. The reforms make noncorrupt taxcollectors work harder, and may reduce thenumber of corrupt tax collectors in the ad-ministration. Hence, possible tradeos must bemade between reducing corruption and in-creasing tax revenues when considering incen-tive reforms. We doubt however, that it is areasonable strategy to improve revenue collec-tion by strengthening the bargaining power ofcorrupt tax ocers vis-aa-vis taxpayers. Thus,

we question the claim that one of the positive

do not operate on their own, but are inuencedby the behavior of their reference group, suchas colleagues and friends. 10 As stressed byFehr and Gaachter (2000, p. 167), [s]ocial

WORLD DEVELOPMENT1462sanctions by peer members are probably a veryimportant determinant of eort behavior inwork relations. Therefore, if a tax ocereects of such reforms is that increased taxrevenues can be achieved by stimulating cor-ruption among corrupt tax collectors.

3. LONG-TERM IMPACTS OF FISCALCORRUPTION

Poor taxpayer compliance is particularlydamaging in situations with substantial budgetdecits, as is the case in many poor countries(Tanzi, 1991). But, accepting scal corruptionas an instrument for raising revenues in theshort run may undermine tax collection in thelonger run, for several reasons. Let us herepoint at some of the most important ones.First, implicit in the discussion of the positive

link between scal corruption and tax revenuesis the assumption that the willingness to paytaxes is independent of the way taxes are col-lected. This assumption is in contrast to theliterature on reciprocity considerations in taxcollection. For instance, Smith (1992, p. 227)argues that tax authorities unresponsive, cor-rupt and unfair treatment of taxpayers fostersdisrespect for and resistance against tax au-thorities and tax laws. 9 In a study of Tanzania,Fjeldstad and Semboja (2001) nd that theunresponsive way taxes are enforced appears tohave fuelled tax resistance. Accordingly, theyargue, tax evasion may to some extent be in-terpreted as a strategy of public resistance andopposition against the authorities. Hence, anincrease in corruption may establish a negativepublic perception that causes citizens to beunwilling for a long period to enter into recip-rocal relationships with the government. Thus,accepting scal corruption as an instrument toraise revenues may contribute to underminingthe legitimacy of the tax administration, andthereby increase tax evasion and decrease taxrevenues over time (Tanzi, 1995, 2000). Webelieve this to be an important issue, becausepublic opinion is not easily restored over time.Second, the relationship among tax collec-

tors also needs to be considered. Tax collectorsknows that colleagues are becoming more cor-rupt, his commitment to honest behavior maybe weakened. There are at least three argu-ments supporting this view (Sah, 1991; Baner-jee, 1992):

internalized feelings of guilt from fraudu-lent behavior become weaker as the numberof corrupt tax ocers increases;when many others are involved in corrup-tion, the loss of reputation (stigma) for eachcollector when discovered decreases; andwhen many others are corrupt, this lowersthe probability of corruption being revealeddue to the fact that the capacity of internaland external investigation units are con-strained.In other words, corruption may corrupt

(Andvig & Moene, 1990). Thus, an increase inscal corruption may initiate a vicious circle inthe long run in the tax department.Third, this vicious circle may have impacts on

the recruitment process of the tax administra-tion. It is reasonable to assume that more scalcorruption among tax collectors will attractpotentially more corrupt employees (Besley &McLaren, 1993). Furthermore, an atmosphereof corruption can result in a recruitment processbased on the wrong premises (Huther & Shah,2000). Signicant above-market rate wages inspecic public institutions in order to reduceshirking and corruption may imply that onegets two prices for the same type of service.This may in general create fertile ground forcorruption and rent-seeking where attractivejobs are likely to be sold, and the sale pricehas built in the capital value of the salary sur-plus. Andvig (1999), for instance, reports fromAzerbaijan that a regular customs ocial ata fat site has to pay US$100,000 to get hisposition. A position is normally nanced bythe incumbent borrowing from family andfriends. The customs ocial is assumed to haveearned enough to repay the investment aftersix months. Thereafter he is supposed to senda percentage (85% is indicated by Andvig) ofwhat he gains on corruption upward to his su-periors.Fourth, accepting corruption may have

negative impacts on the future possibility ofreforming the tax system. For instance, impor-tant stakeholders, including bureaucrats andpoliticians, as well as powerful taxpayers, mayresist changes in an attempt to protect their in-uence and control of the tax system. Accordingto Winters (1996, p. 166), the strongest resis-tance to tax reforms in Indonesia came from the

tax ocials themselves, since they had the most

FISCAL CORRUPTION 1463to lose from the depersonalization and simpli-cation of the tax system. Flatters and Macleod(1995, p. 409), also referring to Indonesia, assertthat tax collectors actively opposed simplica-tions in property tax administration, income taxlaws and tari structures. Moreover, some ob-servers argue that the extensive public sectorregulations and complicated tax systems ob-served in many poor countries are the result of adeliberate strategy by civil servants, includingsenior tax ocials, to facilitate corruption(Tanzi, 2000; Myrdal, 1968).Developments in the tax administration in

Ghana, which is the crosscutting case to whichChand and Moene (1999) refer, may supportour general point. Following the reforms, taxrevenues in Ghana increased from 4.6% ofGDP in 1983 to 17% in 1994 (Chand & Moene1999, p. 1135, Table 2), despite reductions intax rates. But if the increase in tax revenues inGhana was due to a strengthening of the posi-tions of corrupt tax ocers, then we wouldexpect a dicult future for the tax administra-tion in the longer run. In fact, this seems to bewhat Ghana has experienced; the initial successhas not been sustained (Devas, Delay, &Hubbard, 2001, p. 213). According to Hadler(2000, p. 40), the tax administration in Ghana,the rst country in Africa to establish an au-tonomous revenue agency, is reputedly now indisarray. 11

In summary, increasing scal corruption bystrengthening the position of corrupt tax o-cers may initiate two vicious circles in thelonger run. On the one hand, it may reducepeoples willingness to pay taxes; on the otherhand, it may weaken a commitment to honestbehavior in the tax administration. Both theseeects are closely related to the importance ofvalues in tax collection and tax compliance.Our general point is that inducing scal cor-ruption in the long run undermines the valuesessential to an ecient tax administration. Asobserved by Amartya Sen:

Indeed, in societies in which corrupt behaviour of thestandard type is quite unusual, the reliance is, to agreat extent, on compliance with codes of behaviourrather than on nancial incentives to be corrupt. Thisforces attention on the norms and modes of behaviourthat respectively prevail in dierent societies (Sen,1999, p. 276).

Of course, this does not imply that incentivesare of no importance. But we question the ideaof fostering scal corruption in order to gain

12short-term increases in tax revenues.4. GOVERNMENT TRUSTWORTHINESS

Fiscal corruption is likely to undermine gov-ernment trustworthiness and, thus, the legiti-macy of the government, where legitimacyrefers to citizens approval of the government,which in turn justies citizens obedience. 13

When the institutions are legitimate, citizenshave a predisposition to consider obedience tothem as reasonable and appropriate (Fauvelle-Aymar, 1999). A governments lack of legiti-macy, on the other hand, diminishes almost bydenition the perceived moral justication forobeying its laws (we return to this point inSection 5). Furthermore, of particular impor-tance in this context is that citizens disrespectfor the tax laws may initiate disrespect for otherlaws, and, thus contribute to further under-mining the legitimacy of government (Graetz,Reinganum, & Wilde, 1986). This suggests avicious circle where distrust breeds distrust. Incontrast, government trustworthiness and wide-spread public support tends to legitimize thepublic sector, and may therefore impose somesocial norm to pay taxes. Hence, it is importantto take a broader view of the societal eects ofcorruption in tax administration. 14

The need for a broader view on taxationderives from the fact that taxation is essentialfor shaping state-citizen relations (Levi, 1988;Moore, 1998). For instance, in Europe over thepast two centuries, taxation and disputes overthe use of revenues have stimulated the devel-opment of greater citizen rights and privileges,with democratic institutions enforcing ac-countability and greater transparency in ex-penditures (Tilly, 1992). Moreover, it almostgoes without saying that scal corruption, asan integral part of tax collection, does notcontribute to establish productive state-societyrelations. Survey research from a number ofcountries concludes that citizens in general viewcorruption negatively even where it is wide-spread. Miller, Grdeland, and Koshechkina(1998), for instance, in a study of bribery inthe Czech Republic, Slovakia, Bulgaria andUkraine, nd that public opinion in all fourcountries is against corruption. The morality ofpublic oceholders is therefore most likely animportant source of government trustworthi-ness (Brennan, 1998; Hardin, 1996).Recent research also indicates that citizens

trust in their fellow citizens is strongly inu-enced by whether they have condence in thegovernment that they share (Brehm & Rahn,

1997). This observation strengthens the need

government trustworthiness contributes to so-cial behavior in general and a productive set of

benet in an appropriate way as assessed by a suitablebenchmark of comparison. Since the primary subject

WORLD DEVELOPMENT1464common norms in society. These norms areimportant for establishing the more informalsocial networks and associations of civic en-gagement that eects the productivity of thecommunity (Putnam, 1993). Moreover, theyare also crucial for strengthening the formal-ized institutional relationships such as the po-litical regime, the rule of law, the court system,as well as the tax system, that may have im-portant eects on the rate and pattern of eco-nomic development (North, 1990; Olson, 1982).

5. ELIMINATION OF CORRUPTIONAS AN END IN ITSELF

Mookherjee (1997, p. 6), claims that theelimination of corruption should not be con-sidered an end in itself, and he substantiatesthis point by arguing that complete eliminationof corruption may be impossible. We reject thisline of reasoning. A noncorrupt society may bea utopian ideal, but this does not undermine thepossibility of assigning intrinsic disvalue tocorruption. To consider reduced corruption anend in itself is simply to say that it is importantin its own right, and does not have to be jus-tied (as a value) on the basis of its eects onthe economy and society in general. We believethis to be a reasonable position to take. Ofcourse, there are other ends to consider, andhence we have to make tradeos. But this onlyshows that there is a plurality of constitutivefor a broader view of the societal eects of scalcorruption. We know that a functioning socialorder requires social behavior (Coleman, 2000;Putnam, 1993; Serageldin & Grootaert, 2000)and a productive set of common norms (Bard-han, 1995; Oe, 1999), which will only evolve ina society of trustworthiness (Dasgupta, 1988;Sztompka, 1999).To summarize, there are two main reasons

for taking a broader view on scal corruption.First, when government is perceived to betrustworthy, citizens are more likely to complywith its demands in general (Levi & Stoker,2000). From this perspective, governmenttrustworthiness is closely linked to citizensperceptions of the capacity of the governmentto make credible commitments about the use oftheir taxes, as well as the governments pro-cedures for designing and implementing pol-icy nonarbitrarily (Levi, 1997, 1988). Second,elements in the process of development.of justice is the basic structure of society, these fairterms are expressed by principles that specify basicrights and duties within its main institutions. . .(Rawls, 1993, p. 16).

Moreover, Rawls (1993, p. 19) argues thatanyone with a sense of justice should apply andact from the public conception of justice whichcharacterizes the fair terms of social coopera-tion, and hence ought not to be involved in anykind of corruption.Considerations of this kind may be perceived

as having little relevance to policy debates inpoor countries that are far from any ethicalequilibrium of fair cooperation. We doubt,however, the validity of such a point of view. Inparticular, we believe that Rawls line of rea-soning may contribute to establish an under-standing of the main institutions in society ingeneraland tax administration in particularas ways of specifying fair terms of cooperation,where violation of these terms is consideredwrong in itself. By recognizing this, we also seethe plausibility of considering the eliminationof corruption as an end in itself.

6. CONCLUDING REMARKS

The point of departure for this note is theliterature showing that increased scal corrup-tion in some contexts may contribute to in-creased tax revenues. We do not argue againstthe relevance of this argument. On the con-trary, it is important to clarify this relationship.There are obviously cases where tradeosmust be made between reducing corruption andincreasing tax revenues by using incentive re-forms. We doubt however, that it is a reason-able strategy to improve revenue collection byCorruption is the violation of establishedrules for personal gain, and the disvalue ofcorruption depends on the legitimacy of theserules. But within a fair system of cooperation,the elimination of corruption should be con-sidered an end in itself. This has been arguedforcefully by Rawls among others, who views afair system of cooperation as involving:

terms that each participant may reasonably accept,provided that everyone else likewise accepts them.Fair terms of cooperation specify an idea of recipro-city: all who are engaged in cooperation and who dotheir part as the rules and procedures require, are tostrengthening the bargaining power of corrupt

tion the claim that one of the positive eects ofsuch reforms is that increased tax revenues can

normative reasoning, we conclude that the

propriate instrument for raising tax revenues.

T

problems likely to be encountered in designing and

implementing incentive reforms, and on evaluating the

2. The total amount of bribes received by corrupt tax

collectors does not necessarily increase in equilibrium.

are, of course, other ways of measuring corruption than

by the bribe rate. For instance, corruption can be

4. This way of modeling moral costs is quite standard

within a static framework, and pursued by among others

can, however, easily be extended to include monitoring,

but this would not add anything to our story.

m is not the critical value of m dening the partitioning

13. Following Lipset (1959, p. 86), legitimacy can be

FISCAL CORRUPTION 14656. The possibility of extortion is not considered in this

simple example (see Hindriks, Keen, & Muthoo, 1999

and Klitgaard, 1988).

7. Eqn. (2) is not well dened for the exact case wherethere is no bonus, but the implication of the equation isdened as the capacity of a political system to engender

and maintain the belief that existing political institutions

are the most appropriate or proper ones for the society.

14. Daunton (1998) provides an excellent historical

account of the role of trust and trust formation in the

British scal administration from the Napoleonic warshence the tax collector knows for certain that the

acceptance of a bribe will not be punished. The example 12. See also Elster (1989, p. 158).Laont and Tirole (1991). Within a dynamic setting

however, more elaborate modeling is needed.

5. We do not consider the issue of monitoring, andmeasured as the proportion of corrupt tax ocers in the

tax administration. Although the precise conclusion on

the eect of incentive reforms on corruption may depend

on how corruption is measured, this is not essential for

our argument.

3. See also Besley and McLaren (1993) for a related

discussion.This depends on the reaction of the taxpayer to the fact

that the work eort of the tax collector increases. Thereeects of pay-for-performance schemes for tax collectors

on corruption and tax revenues.NO

1. More generally, Mookherjee (1997) focuses on thebe achieved by stimulating corruption amongcorrupt tax collectors. Based on existing liter-ature on corruption, incentives, compliance andtax ocers vis-aa-vis taxpayers. Hence, we ques-11. To explain this development in revenue perfor-

mance, we also have to look at factors other than

corruption, including general economic trends and

changes in tax policy.mechanism in question.

9. This proposition can also be stated in positive terms:

Tax authorities responsive, honest, respectful and fair

treatment of taxpayers tends to foster respect for and co-

operation with the tax system.

10. For a more general analysis of these mechanisms,

see Hessing, Elfers, and Weigel (1988) and Snavely

(1990).of the set of tax collectors into corrupt and noncorrupt,

and that we implicitly assume a continuous distribution

of the value of m among tax collectors. Thus, we cannotdraw any denite conclusions from (2), but for our

purpose this should give a reasonable indication of the8. Here, we ignore the equilibrium response of the

companies, and assume that they do not increase

underreporting when the bonus increases. Notice thatthat if the bonus approaches zero then all tax ocers will

be corrupt (except for tax ocers having a deontological

approach to corruption; represented by m equal toinnity).ESSustained development cannot occur in an in-stitutional framework that fosters corruptionand extra-legal tax enforcement.reasonable starting point for policy debates inthis area should be the straightforward one thatan increase in scal corruption is not an ap-to WW II.

EWORLD DEVELOPMENT1466tax administration? Public Administration and Devel-opment, 21, 211222.

views.org/cgi/content/full/3/47.Lipset, S. M. (1959). Some social requisites of democ-REFER

Alm, J., Bahl, R., & Murray, M. N. (1991). Tax baseerosion in developing countries. Economic Develop-ment and Cultural Change, 39(4), 849872.

Andvig, J. (1999). Corruption in the former USSRcountries and international oil business in Azerbai-jan. Report 253:99 (December). Oslo: NorwegianInstitute of International Aairs.

Andvig, J., & Moene, K. O. (1990). How corruptionmay corrupt. Journal of Economic Behaviour andOrganization, 3(1), 6376.

Banerjee, A. (1992). A simple model of herd behaviour.Quarterly Journal of Economics, 107(3), 797817.

Bardhan, P. (1995). Research on poverty and develop-menttwenty years after redistribution with growth.Paper prepared for the Annual conference on devel-opment economics, May 12. The World Bank,Washington, DC.

Besley, T., & McLaren, J. (1993). Taxes and bribery: therole of wage incentives. Economic Journal, 103, 119141.

Bird, R. (1990). Expenditures, administration and taxreform in developing countries. International Bureauof Fiscal Documentation: Bulletin (June), 263267.

Bird, R. (1992). Tax policy and economic development.Baltimore/London: The Johns Hopkins UniversityPress.

Brehm, J., & Rahn, W. (1997). Individual-level evidencefor the causes and consequences of social capital.American Journal of Political Science, 41(3), 9991023.

Brennan, G. (1998). Democratic trust: a rational choicetheory view. In V. Braithwaite, & M. Levi (Eds.),Trust and governance (pp. 197217). New York: TheRussel Sage Foundation.

Chand, S. K., & Moene, K. O. (1999). Controlling scalcorruption. World Development, 27(7), 11291140.

Coleman, J. C. (2000). Social capital in the creation ofhuman capital. In P. Dasgupta, & I. Serageldin(Eds.), Social capital. A multifaceted perspective (pp.1339). Washington, DC: The World Bank.

Dasgupta, P. (1988). Trust as a commodity. In D.Gambetta (Ed.), Trust: Making and breaking coop-erative relations (pp. 4972). Oxford: Blackwell.

Das-Gupta, A., & Mookherjee, D. (1998). Incentive andinstitutional reform in tax enforcement: An analysis ofdeveloping country experience. New York/Oxford:Oxford University Press.

Daunton, M. (1998). Trusting the Leviathan: Britishscal administration from the Napoleonioc wars tothe second World War. In V. Braithwaite, & M. Levi(Eds.), Trust & Governance (pp. 102134). NewYork: The Russel Sage Foundation.

Devas, N., Delay, S., & Hubbard, M. (2001). Revenueauthorities: are they the right vehicle for improvedElster, J. (1989). Nuts and bolts for the social sciences.Cambridge: Cambridge University Press.NCES

Fauvelle-Aymar, C. (1999). The political and taxcapacity of government in developing countries.Kyklos, 52(Fasc. 3), 391413.

Fehr, E., & Gaachter, S. (2000). Fairness and retaliation:The economics of reciprocity. Journal of EconomicPerspectives, 14(3), 159181.

Fjeldstad, O.-H., & Semboja, J. (2001). Why people paytaxes: The case of the development levy in Tanzania.World Development, 29(12), 20592074.

Flatters, F., & Macleod, W. B. (1995). Administrativecorruption and taxation. International Tax andPublic Finance, 2, 397417.

Galtung, F. (1995). Current strategies for combatingcorruption: A study of corruption in the tax admin-istration. Occasional Working Paper 8:95. Berlin:Transparency International.

Graetz, M. J., Reinganum, J. F., & Wilde, L. L. (1986).The tax compliance game: towards an interactivetheory of law enforcement. Journal of Law, Econom-ics and Organization, 38, 132.

Hadler, S.C. (2000). Best practice in tax administrationin sub-Saharan Africa. A handbook for ocials.Draft (June), International Tax Program, HarvardLaw School, Boston.

Hardin, R. (1996). Trustworthiness. Ethics (107), 2642.Hessing, D., Elfers, H., & Weigel, R. (1988). Tax evasion

research: measurement strategies and theoreticalmodels. In W. F. van Raaij, G. M. van Veldhoven,& K. E. W~aarneryd (Eds.), Handbook of economicpsychology (pp. 516537). Boston/London: KluwerAcademic Publishers.

Hindriks, J., Keen, M., & Muthoo, A. (1999). Corrup-tion, extortion and evasion. Journal of Public Eco-nomics, 74, 395430.

Huther, J., & Shah, A. (2000). Anti-corruption policiesand programs: a framework for evaluation. WorldBank Policy Research Working Paper No. 2501.Washington, DC: The World Bank.

Klitgaard, R. (1988). Controlling corruption. Berkeley/London: University of California Press.

Krugman, P. R., Alm, J., Collins, S. M., & Remolina, E.M. (1992). Transforming the Philippine economy.Quezon City: APO Production Unit, Inc.

Laont, J.-J., & Tirole, J. (1991). The politics ofgovernment decision-making: a theory of regulatorycapture. Quarterly Journal of Economics, 106, 10891127.

Levi, M. (1988). Of rule and revenue. Los Angeles/London: University of California Press.

Levi, M. (1997). Consent, dissent, and patriotism. Cam-bridge: Cambridge University Press.

Levi, M., & Stoker, L. (2000). Political trust andtrustworthiness. Annual Reviews Political Science(3), 475507, Available: http://polisci.Annual Re-racy, economic development and political legiti-macy. American Political Science Review, 53, 69105.

Miller, W., Grdeland, AA., & Koshechkina, T. (1998).What is to be done about corrupt ocials? Publicopinion in Ukraine, Bulgaria, Slovakia and theCzech Republic. Paper prepared for the Networkof Institutes and Schools of Public Administrationin Central and Eastern Europe (NISPAcee) 6thannual conference, Prague, March 1820.

Mookherjee, D. (1997). Incentive reforms in developingcountry bureaucracies. Lessons from tax administra-tion. Paper prepared for the Annual Bank conferenceon development economics, Washington, DC: TheWorld Bank.

Moore, M. (1998). Death without taxes: democracy,state capacity, and aid dependency in the fourthWorld. In G. White, & M. Robinson (Eds.), Towardsa democratic developmental state (pp. 84121). Ox-ford: Oxford University Press.

Myrdal, G. (1968). CorruptionIts causes and eects.In G. Myrdal (Ed.), Asian drama: An enquiry into thepoverty of nations (Vol. II, pp. 937951). New York:The Twentieth Century Fund (Re-published in A.

for the developing countries. Departmental Memo-randum No. DM/84/46. The International Mone-tary Fund, Fiscal Aairs Department, Washington,DC.

Sah, R. K. (1991). Social osmosis and patterns of crime.Journal of Political Economy, 99(6), 12721295.

Sen, A. (1999). Development as freedom. Oxford: OxfordUniversity Press.

Serageldin, I., & Grootaert, C. (2000). Dening socialcapital: an integrating view. In P. Dasgupta, & I.Serageldin (Eds.), Social capital. A multifacetedperspective (pp. 4058). Washington, DC: The WorldBank.

Smith, K. W. (1992). Reciprocity and fairness: positiveincentives. In J. Slemrod (Ed.), Why people paytaxes. Tax compliance and tax enforcement (pp. 223250). Ann Arbor: The University of Michigan Press.

Snavely, K. (1990). Governmental policies to reduce taxevasion: coerced behaviour versus services and valuesdevelopment. Policy Sciences, 23(1), 5772.

Sztompka, P. (1999). Trust. A sociological theory.Cambridge: Cambridge University Press.

FISCAL CORRUPTION 1467Heidenheimer & M. Johnston (Eds.). (2002), Polit-ical corruption. Concepts & contexts (3rd. ed., pp.265279). New Brunswick/London: TransactionPublishers).

North, D. (1990). Institutions, institutional change andeconomic performance. New York: Cambridge Uni-versity Press.

Oe, K. (1999). How can we trust our fellow citizens? InM. E. Warren (Ed.), Democracy & trust (pp. 4287).Cambridge: Cambridge University Press.

Olson, M. (1982). The rise and decline of nations:Economic growth, stagation and social rigidities.New Haven: Yale University Press.

Putnam, R. (1993). Making democracy work: Civictraditions in modern Italy. Princeton, NJ: PrincetonUniversity Press.

Rawls, J. (1993). Political liberalism. New York: Co-lumbia University Press.

Richupan, S. (1984). Income tax evasion: a review ofthe measurement of techniques and some estimatesTanzi, V. (1991). Public nance in developing countries.Aldershot: Edward Elgar Publishing.

Tanzi, V. (1995). Corruption, arms-length relationshipsand markets. In G. Fiorentini, & S. Peltzman (Eds.),The economics of organised crime (pp. 161180).Cambridge: Cambridge University Press.

Tanzi, V. (2000). Corruption, governmental activities,and markets. In V. Tanzi (Ed.), Policies: institutionsand the dark side of economics (pp. 88106). Chel-tenham: Edward Elgar.

Tilly, C. (1992). Coercion, capital and European states,AD 9901992. Cambridge, MA: Blackwell.

Toye, J., & Moore, M. (1998). Taxation, corruption andreform. The European Journal of Development Re-search, 10(1), 6084.

Winters, J. A. (1996). Power in motion: Capital mobilityand the Indonesian state. Ithaca, NY: Cornell Uni-versity Press.

Fiscal Corruption: A Vice or a Virtue?IntroductionThe virtue of fiscal corruptionLong-term impacts of fiscal corruptionGovernment trustworthinessElimination of corruption as an end in itselfConcluding remarksReferences