4 Controlling Fiscal Corruption

Sanjeev Gupta, and George Abed
Published Date:
September 2002
  • ShareShare
Show Summary Details
Sheetal K. Chand and Karl O. Moene 

1. Introduction

It has been widely observed that insufficient domestic resource mobilization and wasteful public expenditures are at the root of the adjustment and growth problems faced by many countries, including especially those in transition to a market economy. To remedy this situation the basic strategy has relied on reforming the design of taxes and their administration, accompanied by efforts to control expenditure waste and fraud. The often less-than-satisfactory outcome of this strategy has led to a growing perception that corruption and its corrosive effects on fiscal performance should also be addressed.1

It is difficult, however, to obtain adequate information on corruption because the involved parties naturally strive for concealment. Broadly, on the side of revenue, some indication of the extent of corruption can be obtained by comparing the theoretical yield of taxes and actual collections, but the shortfall could also be caused by other factors such as the poor organization and administration of tax departments. It is also difficult to establish how much corruption adds to expenditures as standards for comparison are not readily available, especially for public goods, while with respect to procurement and contract specifications, practices may vary between the private and public sectors. Detailed information on behavioral responses is not readily available and therefore researchers have often had to rely on anecdotal accounts.2 Such a situation does not facilitate the construction of an analytical framework to explain corruption and to develop ways to control it.

In the literature, three basic factors have been identified as contributing to corruption.3 The first is the overall level of potential benefits from corrupt behavior. Benefits refer to the payoffs from evasion in the case of taxes or excessive billing of government expenditures. The second relates to the costs of bribery, or the penalties and sanctions applicable to briber and bribee. The third concerns the bargaining power of officials, or the extent of exclusive discretionary powers in the hands of individual bureaucrats. High tax rates imply high potential benefits from evasion. A poor record of applying sanctions lowers the costs of being corrupt. Considerable discretionary powers vested in the hands of officers confer the means for extracting bribes. An environment exhibiting all three features is likely to exhibit high levels of corruption. Suggested traditional remedies follow readily: lower tax rates; apply adequate penalties more rigorously; and reduce the discretionary powers of fiscal officers, with the last two also applicable in the area of government outlays.

These remedies however, may not be adequate. It may also be necessary to address issues concerning the conditions of service and more generally the motivation of fiscal officers. Obviously, if officers are poorly paid, or become so as a consequence of an adjustment strategy, for example, wage control in a context of massive price increases, they might be more inclined to engage in corruption.4 Should the top echelons of government be corrupt, there is likely to be a contagion effect on lower levels.5 This effect could become more pronounced if it is generally perceived that the administrative structures are being used to promote the self-interest of their managements rather than the mandated objectives. The contagion effect may be further stimulated if, as is often the case in such environments, officers view their own career progression to be dependant on the extent to which they compromise with higher levels. Establishing an organizational structure that is less conducive to compromising behavior, combined with the right incentive structure and the appropriate esprit de corps, can help contain contagion effects. It is likely that the payment of a properly designed bonus, especially if emoluments are inadequate, could be a powerful incentive. But, for such a measure to be effective it is essential to reduce corruption at higher levels.

Most of the attempts at formal modeling in the literature have focused on the behavior of the briber, for example, the taxpayer.6 Less attention has been paid to the behavior of the fiscal officer, their conditions of service, and their motivation, which is the focus of this paper.7 To address these issues, a two-pronged approach is adopted. First, a framework is set up in Section 2 to analyze how bonus payments to tax officers can promote less corrupt outcomes. Second, a case study of how rampant fiscal corruption was brought under control, and revenue performance markedly improved, is presented in Section 3.

2. Modeling Incentive Effects

There are three categories of players in the model that are set out here: taxpayers, tax collectors, and their managers. Taxpayers try to reduce their tax liability through evasion. Should they be caught evading they try to bribe the tax collector. It is assumed that all tax collectors are not well paid and that they supplement their incomes through bribes. It is further assumed that their ability to extort ever higher bribes is limited by the scope for the taxpayer to appeal to the tax collectors’ managers, some of whom may be dishonest. On appeal, the latter uncover the true extent of evasion and either levy a penalty, or, if they are corrupt, they ask for a bribe. The tax collector loses in the process.

To capture these interactions we study a four-stage sequential game. The game involves weighing various risks in the course of optimizing the net benefits to be obtained. Into this framework, a bonus scheme is introduced with a view to promoting less corrupt behavior on the part of the tax collector. In order to abstract from the diverse characteristics of taxpayers, the analysis refers to a “firm” that engages in the quintessential activity of attempting to increase its net profits which may include recourses to tax evasion. Mutatis mutandis, the same principle applies to other categories of taxpayers. The sequencing of the game is as follows.

(a) Stage (1)—The Firm’s Behavior

The true profit is П, but the firm may find it advantageous to report lower profits R than the true level. According to the books, the reported profit is R≤ П. The main idea below is that the firm chooses how much of its profits to report. To disguise the true profit, however, involves costs. The firm cannot run activities that are easily monitored by tax collectors; it must keep double accounts and so on. The costs of trying to evade taxes is captured by assuming that the true profit П and the reported level R are related as П(R). It is assumed that П′(R) ≥ 0 for R ≤ П and П′(R) = 0 for R = П. Moreover, П″(·) < 0.

(b) Stage (2)—The Tax Collectors’ Behavior

The tax collector checks the books with an intensity μ implying that tax evasion is discovered with a probability μ. The tax collector chooses μ by the choice of how much work effort he is willing to put in. The cost of effort is denoted c and μ is increasing in c at a decreasing rate, hence μ = μ(c) with μ′(·) > 0 and μ″(·) < 0. With a bonus scheme, the tax collector receives a fraction γ of all taxes he collects. We derive the impact of a bonus for the amount of tax collected by modeling the case with a bonus parameter γ, and then comparing the cases γ = 0 with γ > 0. The tax rate on profits is t.

(c) Stage (3)—Threat Points and the Bribe

A tax collector who has found evidence of tax evasion may take a bribe b1 for not reporting the firm. The level of the bribe b1 is determined by bargaining between the firm and the tax collector. The agreement is constrained by the possibilities of both the firm and the tax collector to appeal the case to higher-level tax authorities.

(d) Stage (4)—Behavior of Higher-Level Bureaucrats

Higher-level tax authorities handle the case if there is a disagreement between the firm and the tax collector who claims to have evidence of tax evasion. By inspecting the books in the light of the evidence provided by the tax collector, higher-level tax authorities can always find out what the true profit is. But not all bureaucrats employed at this level are honest. A fraction θ of the higher-level bureaucrats is corrupt and willing to take a bribe b2 for reporting a taxable profit equal to R. Honest bureaucrats report what they find. The presence of corrupt officials implies that when a tax case is appealed, it is handled by a corrupt official with probability θ and by a noncorrupt official with a probability (1 − θ).

We start backward at stage (4), when the case is considered by higher-level tax authorities. If the case is handled by a noncorrupt official, he collects the tax tП. If the case is handled by a corrupt bureaucrat, he reports tR and receives a bribe b2, determined by bargaining between the firm and the bureaucrat. The profit of the firm above the disagreement level (1 − t)П, is

The corresponding surplus to the corrupt bureaucrat is simply the bribe b2. The equilibrium bribe is just a share of the tax saved

where α is the bargaining power of the bureaucrat. Eqn. (2) can be justified by applying the asymmetric Nash bargaining approach, or by exploring the bargaining conflict in an extensive form game such as in Rubinstein (1982). The bargaining power coefficient α captures how harmful a postponement of the deal is to the taxpayer relative to the tax collector and may incorporate both specific institutional factors such as the expected future treatment of the tax authorities and general aspects such as the relative impatience of the two.

At stage (3), if tax evasion is identified, the bribe to a nonreporting tax collector b1 is determined by bargaining. Both can use an appeal to higher-level tax authorities as a threat against the other. Both sides perceive what will happen in that case: the firm obtains π1 and the tax collector u1 (in expected terms), given by

where, in both expressions, the probabilities of being treated by a higher-level corrupt official is θ. Letting the bargaining power of the tax collector be β and that of the firm be (1 - β), the equilibrium level of the bribe is the value of b1 that maximizes

which by inserting for π1, u1 and b2 can be expressed as


Observe that φ is less than unity and that φ. and, therefore, the bribe b1 is increasing in the bonus parameter γ. The implication is straightforward. The higher the bonus the more income the tax collector forgoes by not reporting the true level of profits, thus the higher the bribe has to be. It is also clear from Eqns. (6) and (7) that φ and the bribe b1 is decreasing in the incidence of corruption θ among higher-level tax authorities. The reason is that an appeal from the tax collector is a less severe threat to the firm as long as it can bribe its way also at this level, while an appeal by the firm would lead to a lower expected bonus to the tax collector, the more corruption there is at this higher level. Thus a high level of corruption among higher-level bureaucrats makes the tax collector weaker and the firm stronger. Finally, it might also be noted that increases in the tax rate raise the level of the bribe.

At stage (2) the tax collector decides his work intensity μ. Consider the case where all tax collectors are dishonest. Thus the tax collector maximizes

which give us the first-order condition

It is clear from Eqns. (7) and (9) that the μ is positive even without a bonus, that is with (γ = 0). Without a bonus, φ in Eqn. (9) is just φ(0,θ) ≡ β(1 - θ) + βθα. Thus the possibility of obtaining bribes provides work incentives for the tax collector. Observe, however, that work incentives become weaker, the more higher-level corruption there is, as φ declines when θ goes up. Moreover, since φ is increasing in the bonus parameter γ, the chosen work intensity μ goes up with the bonus for each level of tax evasion. But, the impact of an increase in the bonus on work effort becomes smaller when corruption of higher-level bureaucrats goes up. In the limit, the impact of γ on work effort is zero when all higher-level bureaucrats are corrupt and θ = 1. The tax collector then obtains no expected gain in bonus income by appealing the case to higher-level authorities. There is no bonus income forgone by accepting the bribe for not reporting. Accordingly, there are no incentives to the tax collector to work harder with a bonus than without when all higher-level bureaucrats are corrupt. Finally, Eqn. (9) shows that for each level of φ, tax collectors work harder the more tax evasion they expect to find, i.e., the lower they expect R to be and the higher is t. How work intensity μ is affected by the other variables is summarized in the table.

At stage (1) the firm determines R in order to maximize expected profits net of taxes and bribes. The maximand is

The first-order condition for maximizing V with respect to R is

First, observe that reported profit R is higher the harder tax collectors are supposed to work. This is so since П”(·) < 0 and the RHS of Eqn. (11) is decreasing in the work intensity μ. It is also readily seen that higher taxes make tax evasion more tempting. As shown by Eqn. (11), reported profit R declines as t goes up. Next, since φ is decreasing in the incidence of higher-level corruption θ (and the RHS of Eqn. (11) is decreasing in φ in which case it would increase when φ decreases), reported profits are lower and tax evasion higher, the more higher-level corruption there is. Similarly, since φ is increasing in the bonus parameter, the level of reported profits R is higher, the higher the bonus. With a bonus scheme tax collectors have to be paid higher bribes to be willing not to report any acts of tax evasion that they identify. The firm would know this, which is captured by the impact on the RHS of Eqn. (11) via the dependence of φ on γ in Eqn. (7). (In addition, firms can infer that with a bonus, tax collectors would work harder in order to identify tax evasion. This is captured by the impact on the RHS of Eqn. (11) of the dependence of μ on γ from Eqn. (9), to be incorporated when we solve for the equilibrium behavior of both taxpayers and collectors.) The impacts on reported profits of increases in the variables of the model are summarized in Table 1.

Table 1.Work Intensity Affected by Other Variables
Increase in:Rμθγt
Effects on: R++

The equilibrium levels of μ and R are those that solve Eqns. (9) and (11) simultaneously as illustrated in Figure 1. The downward sloping curve shows the tax collector’s choice of μ for each level of R (since for each R there is a unique [П(R) - R]). The upward sloping curve shows the firm’s choice of reported profits R for each level of μ. The (Nash) equilibrium of the model is where the two curves intersect.

Figure 1.Incentive Policies and Revenue Yield

As illustrated in the figure, an increase in the bonus parameter shifts both curves to the right, implying that the equilibrium level of reported profits goes up. An increase in corruption among higher-level bureaucrats, however, shifts both curves to the left, implying that the equilibrium level of reported profits R goes down. A decline in the tax rate shifts the curve describing the firm’s behavior to the right as the direct incentives for tax evasion decline. Yet, in equilibrium, lower taxes do not necessarily imply less tax evasion since a lower tax rate, in addition, shifts the curve describing tax collectors’ behavior to the left as tax collectors work less hard. Thus the net effect on reported profits depends on which is the strongest of the direct incentive effect on taxpayers’ behavior and the indirect effect via a less efficient tax collection.

Returning to the bonus, it may seem as if a higher bonus parameter implies increasing corruption among tax collectors since the bribe b1, all else being given, goes up with the bonus. All else, however, is not given. Tax collectors work harder and firms have fewer incentives to hide their true profits both because they have to pay higher bribes when detected and because for each level of tax evasion the probability of being detected goes up with the bonus. As a consequence, the gap between П - R narrows and even though tax collectors obtain a higher share of the gap with a bonus than without, the equilibrium bribes may very well decline with the introduction of a bonus.

As demonstrated, the efficiency of the bonus scheme depends on the incidence of higher-level corruption. It is, therefore, important to incorporate how the incidence of higher-level corruption θ may be affected by the bonus to tax collectors. A simple and robust way to make θ endogenous is to assume that the incidence is increasing in the potential gain from acting corruptly captured by b2. Thus let

where F(·) > 0. Now we have an even more interesting closed loop of the impact of a bonus to tax collectors on the functioning of the tax system; A bonus implies that tax collectors work harder and that firms voluntarily report higher profits. This implies, once more, that the gap between the true and reported profits (П - R) narrows which again reduces the incidence of corruption among higher-level bureaucrats according to Eqn. (12). As θ goes down, work effort of tax collectors increases and reported profits of firms go up both because μ is higher for each level of tax evasion and because there is less to be gained by an appeal to higher-level authorities. Thus introducing a bonus scheme leads to more honesty in parts of the administration that are not directly affected by the bonus. Moreover, a more honest administration makes the bonus scheme more efficient. Thus, there is an honesty multiplier in the sense that a bonus makes firms act more honestly, which also induces higher-level bureaucrats to act more honestly as there is less to gain from corruption. In turn, this outcome induces firms to engage in less tax evasion and so on.

The honesty multiplier does not require that tax collectors work harder with the bonus than without. As seen from Figure 1, an increase in the bonus may have an ambiguous impact on work effort μ. To see how the multiplier works let us for now assume that μ is given. We then have that Eqn. (11) determines R as a function of φ ≡ φ(γ. θ), written R = R(φ) where R′(φ) > 0, while Eqn. (12) determines the fraction of higher-level authorities who are corrupt as a function of R, expressed as θ = θ(R) where θ′(R) < 0. Let us in addition define the following elasticities

Using these short-hand expressions, the total effect of an increase in the bonus parameter can be written as


As stated before, an increase in the bonus has no effect when θ = 1. When θ < 1 every initial stimulus from a higher bonus Ry > 0 is multiplied by m > 1 to obtain the overall effect. A higher γ makes tax evasion less profitable simply because the higher bonus implies that tax collectors demand higher bribes in order to report the firm. When firms take that into account, they evade less taxes which again makes it less tempting for higher-level officials to be corrupt. If we assume symmetric bargaining power α = β = ½ and that half of the higher-level officials are initially corrupt, θ = ½ and that σρ = ½, we get m > 1.2. An increase in the response elasticities to σρ = 1 increases the multiplier to m = 1.5.

The tax collected is proportional to

since in equilibrium the same amount is reported whether the tax collector detects tax evasion or not. In both cases, he only reports tR. The value of T is of course a gross tax and one may be concerned that even though R goes up, the net tax income after the bonus to tax collectors is paid, may go down. A bonus scheme, however, does not have to be excessively generous and adjustments can always be made to initial salary levels. The salary could be reduced by an amount not far from that of expected bonus incomes. If a target level of emoluments including bonuses is maintained, then the relevant tax income is rendered proportional to T in Eqn. (14) and tax incomes unambiguously go up with the introduction of the bonus scheme.

3. A Case Study

It is difficult to find well-documented case studies of successful attacks on fiscal corruption, perhaps because there may not be many successful cases. Although it remains to be fully documented, Ghana provides an interesting case of a country that suffered badly from widespread fiscal corruption, but eventually took effective steps to contain it, at least for a while. From the mid-1960s up to the early 1980s, Ghana’s economy underwent a sustained decline, owing to adverse terms of trade, and domestic policies that became increasingly more interventionist. During this period there was widespread recourse to price and income controls, comprehensive regulations, and an elaborate licensing system, which had the effect of driving much of the economy underground.8 This, of course, contributed to a decline in taxable capacity.

(a) Growth in Corruption

Initially, attempts to recoup revenue relied largely on raising tax rates. This stimulated a further erosion of tax bases, however, and the revenue response was inadequate. As indicated in Table 2, by 1983 the tax ratio had progressively collapsed to 4.5% of GDP from around 13% in 1973, and even higher in earlier years. To supplement the dwindling resources of the state, a variety of quasi-fiscal measures, which implicitly tax some in favor of others, were also introduced. In the main, these took the form of multiple exchange rates, foreign exchange rationing, negative real interest rates, and credit rationing, with state controlled agencies and favored individuals being the principal beneficiaries. Political and bureaucratic discretion were relied upon in applying the quasi-fiscal measures. Such conditions stimulated rent seeking which expanded to the point where the productive economy collapsed. Eventually, declining state resources forced a curtailment of expenditures, and deep cuts were made in the salaries of civil servants, and in essential infrastructure and social outlays. These moves contributed to a series of major breakdowns that affected critical areas such as telecommunications, energy generation, the transport network, and ports, thereby hampering exports and economic growth. The resulting supply shortages were made worse by droughts and the severe rationing of foreign exchange in the face of export collapse, growing external debt arrears, and exclusion from the international capital markets.9

Table 2.Ghana: Fiscal Collapse and Recovery (Ratios to GDP)a
Total revenue and grants15.09.15.614.623.4
Tax revenue12.88.14.612.317.0
Direct taxes2.
Taxes on domestic goods and services2.
Taxes on international trade7.
Total expenditure and net lending19.
Current expenditure14.68.97.410.616.1
Capital expenditure3.
Overall deficit−4.1−5.4−
Sources: Statistical Service of Ghana; and Fund staff estimates.

Refers to Central Government.

Refers to fiscal year on a July/June basis. From 1983 onward, the fiscal year was converted to a calendar year basis.

Sources: Statistical Service of Ghana; and Fund staff estimates.

Refers to Central Government.

Refers to fiscal year on a July/June basis. From 1983 onward, the fiscal year was converted to a calendar year basis.

The pronounced and persistent decline in revenue was associated with a tax and customs administration that became increasingly disorganized. This manifested itself in several ways including the haphazard storage of taxpayer files, which were in any case poorly organized; desultory examination of returns; the mixing up of assessment and collection functions in the hands of the same officers; and the virtual cessation of properly conducted audits. Such practices of course made it easier for taxpayers to evade taxes and for corrupt tax officers to collect rents.10 At the same time, the tax system became riddled with all sorts of devices that were used by fiscal officers to supplement their incomes. A prominent instrument was the use of tax clearance certificates (TCCs). These were required for a wide and growing range of transactions, for example, acquiring trading licenses, a passport, and so forth. The fiscal officer enjoyed considerable discretion in granting TCCs and, according to all accounts, exploited the opportunity to extract bribes.

Both anecdotal accounts and the charges subsequently leveled by the future President Rawlings pointed to widespread corruption at all levels, and most damaging, at higher levels. This is likely to have contributed to widespread demoralization of the tax and customs service, which is no doubt reflected in the poor revenue performance. Civil servants, who had been among the most highly paid in the country, had by 1983 experienced a real erosion in their salaries to one-sixth of what they had been in earlier times. At the same time, because of the loss of job opportunities elsewhere in the economy, employment in the public service increased sharply, further preempting the scope for restoring real wages. Aside from suffering from badly eroded emoluments, what would be the point of applying extra effort to uncover tax evasion if collusion between the taxpayer and higher-level officials would deny the fruits of one’s labor?11

(b) The Reform

The takeover by the Rawlings regime at end 1981, on an anti-corruption platform, was the start of a far-reaching reform of the economy. At first, in an attempt to contain corruption and to raise revenue, extreme measures were taken. Some officers were charged with corruption and were executed; untrained revolutionary cadres were assigned to identify potential taxpayers and to collect taxes (in the so-called operation tax harvest); general exhortations to pay taxes were made; and sanctions were threatened in the event of nonpayment. An ominous sounding “National Investigations Committee (NIC)” paralleled by another body called the “Office of Revenue Commissioners (ORC)” were set up to enforce fiscal obligations, if necessary through arrest and seizure. But, these measures appeared to have had only temporary effects. A frightened population or terrified tax officers may turn in more revenue for a time. But, if the incentive structure that concerns taxpayers and tax collectors is not reformed, and reinforced by appropriate organizational and procedural changes, revenue will subside. The likelihood for such an outcome increases when tax collectors are paid well below subsistence levels, and the population remains harassed and subject to very high compliance costs.

What can be done to improve the fiscal situation in a sustainable manner and, in particular, the revenue position? Three essential elements that concern the structure of incentives are involved: first, tax bases need to be coaxed back from the underground; second, taxpayers must be induced to pay taxes; and, third, tax officers must be motivated to collect them. A piecemeal solution involving only one or other of the preceding elements may not work. Thus simply improving the motivation of tax officers, but not reforming a tax system that had become, at least on paper, punitive because of the high rates, could wreak havoc on the private sector. They would now be subject to the full brunt of the higher tax rates, and no doubt would redouble their efforts at evasion. On the other hand, not doing anything to improve the effectiveness of tax administration but merely lowering tax rates may not increase revenue.12

The failure of the initial, piecemeal, attempts at reform led the Rawlings Government, late in 1984, to initiate an integrated approach that was pursued over several years involving all of the three requirements mentioned above. Actions had already been taken in the context of the reform program to reduce key distortions such as the overvalued exchange rate, and its concomitant of rationing, which encouraged tax bases to emerge from the underground. These were supplemented by the staged lowering in tax rates to more reasonable levels, for example, the top marginal rate of tax on income was reduced in steps from 55% (65% on personal income) to 25%. The large number of excise duties were reduced and merged into a revamped manufacturer’s sales tax at substantially lower rates. The tax mix was changed in favor of indirect taxes on domestic production and consumption. In particular, reliance on the harder to evade excise duties on gasoline products was increased sharply. Generally, taxes were simplified and steps taken to document them better so as to render them more transparent. There was more effective enforcement of withholding and the application of presumptive methods, while the use of TCCs was circumscribed.

A central element of the strategy, which was vigorously pursued from 1985 onward, was to reorganize the revenue service and, inter alia, restrain the exercise of discretionary powers by tax officers. Considerable emphasis was placed on improving the conditions of service of tax and customs officers. Owing to the legal restrictions imposed by civil service rules, their pay could not be raised without parallel increases for the rest of the civil service. But a general pay increase could not be provided because of the lack of fiscal resources. Part of the solution that could be adopted to break this version of the chicken and egg problem would be to introduce a bonus scheme. But, this could not be granted to each tax officer without providing similar bonuses to other ministry of finance officials. More generally, the required restructuring of the revenue collection function could not be undertaken if the needed recruitment, training, promotion and salary policies differed significantly from those of the civil service. There was also the question of how to ensure that the revenue departments would have sufficient resources to procure the various inputs that would enable them to fulfill their functions. As long as the revenue departments were an integral part of the central government structure, they could not be exempt from any across-the-board cuts that were frequently applied to ensure compliance with overall expenditure targets in the context of adjustment programs.

It was therefore decided to move the two revenue departments—Customs, Excise and Preventive Service (CEPS) and the Internal Revenue Service (IRS)—out of the Ministry of Finance and Economic Planning (MFEP). A new authority, designated the National Revenue Service (NRS), was established to coordinate closely the work of the revenue agencies. The agencies were given considerable operational autonomy, with their own boards, and entrusted with the tasks of assessing and collecting revenue under the overall supervision of the National Revenue Secretariat (NRS). The latter, armed with the clear mandate of “…supervising the activities of the revenue institutions and recommending revenue policy directly to the government” (Terkper, 1994a, p. 1394), was headed by a minister of state of known integrity and drive.

An important aspect involved strengthening the surveillance functions of the NRS over the autonomous income tax and customs services, thereby making credible the minister’s threat to keep an eye on the tax officers. A key measure was to weed out at the outset those officers who were regarded as irredeemably corrupt, and several officers left. Selective recruitment was initiated to replace incompetent and unmotivated staff. Facilities were also set up to enable the general public to complain about corrupt officers. Significant reforms were undertaken in the organizational structure and work procedures of the IRS and CEPS. These included better separation between different revenue functions such as assessment and collection; offices, whose work was organized more efficiently; examinations that were more assiduously conducted, and more frequent taxpayer audits.13

Two innovative schemes were introduced in 1986 for motivating staff and ensuring adequate resources for the revenue departments. The first, involved sizable, across-the-board, annual bonuses to tax and customs officers that would be paid if pre-set revenue targets for the two agencies were exceeded. The bonus rate is 15% of base pay, with a lower floor of 10%. The scheme is nondiscriminatory, applying to all full time employees, which is viewed as contributing to better morale and esprit de corps.14 At the same time, it is compatible with using the regular pay scale and schedule for merit increases to reward superior performance of individual officers.

Second, and parallel with the bonus system, an incentive-based mechanism was employed for some years to provide resources to the revenue agencies to cover their routine operations. The two agencies were allowed to keep a small percentage of the revenue collected, to pay for both the bonus schemes. This procedure conferred the incentive that if they needed more resources they would aim at a higher level of collections. The proceeds could, at the discretion of the heads of the two tax departments, be used for the improvement of facilities.15 The use of such funds was subjected to ex post auditing. This agency bonus system lasted until the end of 1992, by which time it was presumably felt that sufficient improvements in the conditions of service had occurred.16

(c) Some Outcomes

Table 2 shows that the results from pursuing the integrated reform strategy were highly favorable. Despite the major reductions in tax rates, the tax ratio rose sharply from 4.5% in 1983 to over 12% of GDP by 1988, reaching 17% by 1994.

It is difficult to disentangle the effects of the three elements mentioned above with respect to the reemergence of tax bases, improved tax compliance, and more effective administration. To some extent, the revenue increases reflect the success of the strategy implemented to reduce the size of the unofficial sector. Some part of the improvement must be attributed however, to the reform of the tax administration and its better motivated officers. This can be established by noting that simply bringing into the open the unofficial sector and subjecting it to taxation need not raise the tax ratio. While revenue will rise, the GDP denominator also increases. Furthermore, with an entrenched behavior of noncompliance, even if the increase in the taxable base is readily observable, for example, goods being brought out into the open for sale, it need not lead to increases in revenue. Ignoring for the moment tax administration, the tax ratio could rise if new taxes are introduced, or the tax system is rendered more progressive. But, the reforms involved reducing the nominal progressivity of the tax system. Furthermore, although there were significant increases in revenue from the introduction of the essentially new tax on petroleum products, losses were sustained from phasing out export duties. It must follow, therefore, that improvements in tax administration made a significant contribution to the observed rise in tax ratios. Doubtless, revamped administrative procedures is a positive factor, but in the end, surely, it is the superior, well-motivated, performance of the fiscal officer that makes the difference.

The revenue improvement enabled the government to increase sharply its outlays. Together with a policy of civil service retrenchment, major wage increases were granted to civil servants. This resulted in big improvements in average real emoluments and was accompanied by an incentive-oriented decompression of the salary scale. Aside from contributing to greater efficiency of the government machine, the revenue increases made possible higher maintenance, infrastructural, and social outlays, which had for long been neglected, and paved the way for the rehabilitation of the economy. On the whole, the expenditure increases were undertaken responsibly as is indicated by the reduction in the overall deficit and its eventual conversion into sizable surpluses.

The success of the Ghanaian experiment has been emulated in several countries both in Africa and elsewhere. For example, Uganda, drawing on technical assistance from the NRS, set up a comparable structure, which helped promote a superior revenue performance, as have Kenya and Malawi.

In 1991, the NRS, whose independence was apparently a source of bureaucratic friction with the Ministry of Finance, was effectively downgraded. It was brought under the direct control of the MFEP, with its head now reporting to one of the deputy ministers in the MFEP, instead of being headed as it had been by a full minister. The downgrading helps explain some of the difficulties that have since been experienced in coordinating and monitoring the work of the two revenue agencies. The autonomy of the two revenue agencies has increased relative to that of the emasculated NRS. There is reason to believe that bribery and corruption, which had abated considerably during the heyday of the NRS has become more pronounced. In particular, the performance of the CEPS has deteriorated.17 With a view to restoring earlier levels of compliance, the creation of a central revenue board outside the MFEP is now planned. It is intended to confer sufficient powers on this body so that it can once again effectively monitor and supervise the assessment, collection, and audit activities of the revenue agencies.

(d) Relationship with the Model

For a full empirical evaluation of the model, more information than was generated by the case study is required, for example, the size of the bribes for senior officials, the varying intensity of work effort and the conditions responsible, and so on. Nevertheless, there are some suggestive confirmations. A prediction of the model is that bonuses are more effective in stimulating revenue collection if corruption at the more senior levels is reduced. From its inception, and during its period of autonomy, the NRS exercised firm surveillance over the revenue agencies, and undertook active measures to reduce if not eliminate corruption especially at the higher levels. At the same time, the bonus schemes described above were introduced. It is suggestive that marked increases in the tax ratio should have occurred over this period. After 1991 however, during which the NRS was emasculated, the continued operation of the bonus scheme for emoluments (but not for departmental inputs) was not associated with a pronounced increase in tax ratios. Moreover, it appears that on occasion, the CEPS staff did not achieve the bonus, unlike the IRS. While deficiencies with both agencies were noted by the Minister of Finance, on balance, the CEPS appeared to be worse off.18 Some confirmation is therefore provided of the model’s prediction that a bonus scheme may lose effectiveness if management practices deteriorate.

4. Conclusion

The model set out above provides analytical support to the basic strategy that was pursued in Ghana to restrain corruption and to improve revenue performance. It brings out clearly the importance of attending, in an integrated manner, both to the conditions of service of fiscal officers and the organizational setup. Simply providing bonuses is not enough: corruption at higher levels of management has to be contained to allow bonuses to become more effective. The analysis of the model shows that once this process is initiated a virtuous circle can result that involves the progressive shrinkage of the gap between reported and true tax liabilities thereby reducing the incentive for corruption. Obviously, conclusions based on a single case study should be viewed with caution, but if they can be shown to accord with optimization behavior they gain in plausibility.

An extensive bonus scheme can be criticized as leading to an overly zealous tax collection authority. Such criticisms are appropriate when dealing with a country that has a settled, well-run tax system, decently paid tax officers, taxpayers who understand their obligations, proper accounting and reporting systems, and so on. But in a situation where tax officers are not paid a living wage and where the general climate is one of pervasive rent-seeking—unfortunately, a widespread condition in many transition and other developing countries—a less orthodox solution that has been successfully tried in some countries such as Ghana may be much more effective in the interim. Most fiscal officers, if properly trained, have a sense of professional pride and would not condone corruption. They become corrupt partly out of necessity or because of peer group pressure.


    Allingham, M. andSandmo, A. (1972) Income tax evasion: A theoretical analysis. Journal of Public Economics1 (November), 323–338.

    Andvig, J. C. andMoene, K. O. (1990) How corruption may corrupt. Journal of Economic Behavior and Organization13 (January), 63–76.

    Besley, T. andMcLaren, J. (1993) Taxes and bribery: The role of wage incentives. Economic Journal103 (January), 119–141.

    Chand, S. K. (1993) From controls to sustainable liberalization: Ghanaian lessons. In Transition to Market, ed.V.Tanzi.International Monetary Fund, Washington, DC.

    • Search Google Scholar
    • Export Citation

    Flatters, F. andMacLeod, W. B. (1995) Administrative corruption and taxation. International Tax and Public Finance2 (November), 397–417.

    • Search Google Scholar
    • Export Citation

    Haque, N. U. andSahay, R. (1996) Do government wage cuts close budget deficits? Costs of corruption. IMF Staff Papers43 (December), 754–778.

    • Search Google Scholar
    • Export Citation

    International Monetary Fund (1997) Governance guidance note. IMF Survey, August5, 234–238.

    Klitgaard, R. (1988) Controlling Corruption. University of California Press, Berkeley.

    Mookherjee, D. andPng, I.P.L. (1995) Corruptible law enforcers: How should they be compensated?Economic Journal105 (January), 145–159.

    • Search Google Scholar
    • Export Citation

    Roe, A. andSchneider, H. andPyatt, G. (1992) Adjustment and Equity in Ghana. OECD Development Centre, Paris.

    Rose-Ackerman, S. (1978) Corruption: A Study in Political Economy. Academic Press, New York.

    Rose-Ackerman, S. (1997) Corruption and development. Paper presented at the Annual Bank Conference on Development Economics, Washington, DC, April.

    • Search Google Scholar
    • Export Citation

    Rubinstein, A. (1982) Perfect equilibrium in a bargaining model. Econometrica50, 97–109.

    Tanzi, V. (1998) Corruption around the world: Causes, consequences, scope, and cures. IMF Working Paper/98/63, International Monetary Fund, Washington, DC. [Revised version reproduced as Chapter 2 in this volume—Ed.]

    • Search Google Scholar
    • Export Citation

    Tettey, W. J. (1997) Transforming tax administration in Ghana: The tension between computers and human agency. Development Policy Review15, 339–356.

    • Search Google Scholar
    • Export Citation

    Terkper, S. E. (1994a) Ghana’s tax administration reforms (1985–93). Tax Notes InternationalMay23, 1393–1400.

    Terkper, S. E. (1994b) Ghana’s budget focuses on strengthening fiscal administration. Tax Notes InternationalMarch7, 633–637.

    Van Rijckeghem, C. andWeder, B. S. (1997) Corruption and the rate of temptation: Do low wages in the civil service cause corruption?IMF Working Paper/97/73, International Monetary Fund, Washington, DC. [Revised version reproduced as Chapter 3 in this volume—Ed.]

    • Search Google Scholar
    • Export Citation

Reprinted from World Development, Vol. 27, Sheetal K. chand and Karl O. Moene, “Controlling Fiscal Corruption,” pp. 1129–1140 (1999), with permission from Elsevier Science.

The authors are indebted to Sveinung Skjesol for assistance in the preparation of this paper. They also wish to thank an anonymous referee for very helpful comments.


See especially Klitgaard (1988) and Rose-Ackerman (1978, 1997). For an example of institutional recognition in a multilateral context see the Governance Guidance Note of the IMF (1997).


Klitgaard (1988) presents some intriguing cases. See also Tanzi (1998).


Thus see Rose-Ackerman (1978).


Van Rijckeghem and Weder (1997). The experience of Sierra Leone is instructive. In the last quarter of 1986, an adjustment program was adopted that froze wages of public servants so as to contain inflation. At the same time, the exchange rate was floated and restrictions on external transactions removed. The exchange rate depreciated rapidly from about US$ 1 = Leones 5 to US$1 = Leones 35 by the end of the year. While the former rate was overvalued in comparison to a parallel market rate of US$1 = Leones 18, the new rate had widely overshot. Despite a small reduction in import tariff rates, importers confronted an increase in customs duties of severalfold. Removal of price restrictions resulted in full pass-through of the exchange rate effect and the price of imported staples such as rice rose by a factor of about seven. Given the resulting severe erosion in the real wages of customs officers, the stage was set for mutually self-serving deals with importers, and officially recorded imports plummeted. The fiscal deficit jumped from 2.3% of GDP in 1986 to 14.8% in 1987.


The classic reference is Allingham and Sandmo (1972).


See, however, Besley and McLaren (1993), Flatters and MacLeod (1995), and Haque and Sahay (1996), who examine incentive effects but without the explicit treatment of bonuses as is done here. Our approach is closer to that of Mookherjee and Png (1995).


Even external donor support dwindled. As Roe and Schneider (1992, p. 15) note “… the disenchantment of the aid community with Ghana’s economic policies in general, its reputation for poor management and corruption, and its extremely poor record in honoring debt commitments … ruled out any prospect that external grants and loans could compensate for the collapse of tax revenues.”


Indeed, anecdotal accounts referred to the tax system as one of more or less voluntary payments of tax, where the amount to be paid bore little relation to the true tax liability but was the outcome of a self-serving bargain between the taxpayer and the tax collector.


According to Roe and Schneider (1992, p. 49) “The respect and influence accorded to the civil service was substantially damaged by the corruption of the later Nkrumah years … and chronically impaired by the heightened corruption and mismanagement presided over by the governments of most of the 1970s....extremely tiny differentials of less than 2:1 between the salaries of top civil servants and unskilled workers was guaranteed to make the recruitment and retention of skilled and committed professionals extremely difficult....the failure…to match the costs of rising consumption needs inevitably prompted absenteeism, moonlighting, corruption and woefully poor morale.”


A finding of the model set out earlier is instructive in this regard. Although lower tax rates reduce the costs of complying which would raise revenue, they may also reduce the diligence with which tax officers undertake their work. The supply-side contention that lower tax rates increase revenue needs to be qualified.


More steps however, could have been taken. For example, Ghana still has to introduce a common taxpayers’ identification number, which is essential for fully effective control.


A disadvantage of this scheme is that once the revenue target is in sight the incentive to continue working hard may diminish. The disincentive effect can be avoided by setting the bonus as a percentage of the amount by which actual revenues exceed the target. But, the size of the bonus is then less assured. With wage costs of tax officers roughly 2% of total collections, a 15% annual bonus to be paid out of excess revenue would require annual revenue excesses of 10% if 3% of the excess were set aside for this purpose. Clearly, it would be difficult to ensure a 10% revenue excess each year.


See Terkper (1994a) for a more detailed description.


Terkper (1994a), mentions additional reasons why this bonus scheme for supplementing departmental resources was discontinued, especially that the practice was never backed by legislation. Retaining part of revenue would violate the constitutional requirement that all revenue collected should be assigned to the consolidated fund.


Terkper (1994b, p. 633) notes that “the year (1993) ended with the minister (Dr. Botchwey) himself criticizing the customs administration before Parliament for a series of valuation malpractices…”


Tettey (1997, p. 351) provides an interesting account of how computerization of customs operations, including the random selection of customs officers to handle transactions, failed to curb malfeasance, “…the reality is that large-scale corruption continues…An officer can arrange with a colleague, selected by the computer, to ‘help’ an importer circumvent the system…It is not uncommon to find…importers giving perquisites to officials with whom they may not be dealing in the present in the hope that their good turn will deserve another in the future.”

    Other Resources Citing This Publication