Crime, Punishment and Tax

For example, in Australia section 26–5 of the Tax Assessment Act of 1997 also denies deductions for fines and penalties. In the United Kingdom, although no specific legislation restricts deductions for fines and penalties, the House of Lords in Mcknight v. Sheppard [2 All E.R. 491 (1999)], ruled that fines are non-deductible for tax purposes because deduction would dilute the legislative policy behind the fine. In Canada, Section 67.6 of the Income Tax Act (1985), over turned the Supreme Court of Canada ruling in 65302 British Columbia Ltd. v. The Queen [99 D.T.C. 5799 (2000)], and disallows deductions in respect to any amount that is a fine or penalty.

There are several exceptions to the “maximal sanction” result, see for example Garoupa (1997). In the main analysis of this article, the assumptions required for this result are maintained.

In the article the results are generalized to account for the possibility that illicit gains are disgorged and that potential offenders can engage in legal activities that generate taxable gains.

Png and Zolt (1989) call for the repeal of Section 162 (f) of the Internal Revenue Code. They argue that once income taxation is taken into consideration, efficient deterrence can be maintained either by (1) allowing deductions for monetary sanctions or (2) adjusting the pre-tax fine downward to replicate the effects of a tax-deductibility rule, if deductions for monetary sanctions are not allowed. Moreover, when comparing these two methods for achieving efficient deterrence, Png and Zolt (1989) embrace the first one on the basis of administrative considerations. They argue that under a tax-deductibility rule, there is no need to coordinate income taxation and law enforcement systems. They claim that one level of punishment can, and should, apply to both those who are and who are not subject to income taxation; and that this punishment needs no adjustment when tax rates change.

It is also inconsistent with Png and Zolt’s own recognition that “since enforcement is costly, it will be optimal to [apprehend offenders] randomly” and seems generally unreasonable. Why can punishment be optimally set both before and after the introduction of income taxation, but enforcement efforts cannot be optimally set either after or before the implementation of income taxation?

This article builds and closely follows the works of Bowles et al. (2000) and Tabbach (2009), which analyze a model of crime and law enforcement in which offenders’ gains from harmful acts are monetary or monetary-like in nature and, therefore, can be disgorged. The only difference between the model presented here and the one analyzed by Tabbach (2009) is that here we also account for the opportunity costs of crime. Both Bowles et al. (2000) and Tabbach (2009), however, did not take into consideration income taxation and the role it might play in optimal law enforcement policy. On disgorgement of illegal gains, see also Friehe (2011).

For the purpose of analyzing the effects of income taxation on crime, however, such an assumption is unreasonable, because if gains are taxable then these gains can be disgorged. In any event, note that the qualitative, main results of this article concerning the effects of income taxation on optimal law enforcement policy are not affected by assuming that the gains from the harmful act can be disgorged, and actually these results are easier to demonstrate without this assumption. The example in the Section 1 illustrates this.

Note, however, that Png and Zolt (1989) do not discuss how first-best behavior can be optimally derived. Rather, they derive first-best behavior, assuming that the probability of punishment is given and fixed.

From a legal, tax perspective, there might be a difference between the tax treatment of fines and disgorgement of gains. By and large, however, disgorgement of gains in criminal proceedings is viewed as a “similar penalty” for purposes of Section 162 (f) of the Internal Revenue Code and, accordingly, it is non-deductible for tax purposes. Sometimes, however, voluntary restitutions may be tax deductible. In any event, in the present setting, the difference between a deductibility and a non-deductibility regime is effectively immaterial with respect to disgorgement of gains. The reason is that disallowing deductions for gains disgorgement means that not only the gains are disgorged but nevertheless the offender/taxpayer should pay taxes on these gains. In the present setting the offender/taxpayer will not have enough resources to pay this extra tax.

To see this differently, observe that deductibility of the disgorgement of gains simply means that net gains are zero, so there is no tax. Therefore, if an agent decides to commit the harmful act he expects with probability to obtain and with probability p to pay an effective fine of (since fines are deductible with tax refunds). Therefore the expected after-tax gains from committing the harmful act are . The expected gains from committing the non-harmful act are . The level of deterrence is consequently determined by .

To illustrate this numerically suppose that , and , then deterrence without taxation is determined by . Consider now the effects of a 20% DTR on the marginal offender (i.e., the offender whose before-tax gains are 200). If he commits the non-harmful act he will gain after-tax 64 (800.8). If he commits the harmful act he will not be detected with probability 0.9 and his after-tax gains in that case will be 160 (2000.8). With probability 0.1 he will be detected, his gains will be confiscated and since they are deductible no tax will apply. In addition, he will have to pay a fine of 1,000 but will receive a refund of 200 (1,00020%) from the tax collector, so his after-tax losses are 800. In sum, the expected gains from the harmful act are 64 (0.9160–0.1800). As is evident, the marginal offender without taxation remains the marginal offender with taxation.

Observe that the gains cannot be similarly grossed up, for the simple reason that after punishment, the taxpayer is left with no gains or other resources.

To illustrate this numerically, observe in the previous example (footnote 11) that if law enforcement is not changed, the agent is left with 200 (1,0000.2), reflecting the tax refund he receives. To take this amount away as well, the fine should be grossed up to 1,250. If the probability of punishment is reduced from 0.1 to approximately 0.082, the level of deterrence will not change 200 , but enforcement costs are saved.

In reality, of course, taxes should presumably increase social welfare, because tax revenues are used to finance public goods or to redistribute wealth, but such an increase is assumed away in this article.

Moreover, some may argue that income taxation decreases the future wealth of future potential offenders. This article disregards issues of savings and how taxation affects it.

As explained below for part (3) to be strictly true one should also assume that .

Formally, the effects of introducing income taxation, which is equivalent to increasing offenders’ wealth, on the optimal probability of punishment can be analyzed by examining how the first-order condition, , which determines, , is affected once the level of fines is increased to and, accordingly, the level of deterrence is increased to . Observe first that the marginal costs of enforcement efforts, , are unaffected. Observe next that the marginal benefits, which are the product of the gains from increased deterrence, , and the deterrent value of enforcement efforts, , are affected in opposite ways. On the one hand, since deterrence is increased, the marginal benefits from additional deterrence are reduced, . On the other hand, since fines are increased, the deterrent value of enforcement efforts is increased, . Therefore, whether the marginal benefits of enforcement efforts are increased or decreased, and, consequently, whether the probability of punishment should be increased or decreased will depend on the magnitude of these opposing effects. As explained in the text, the former effect dominates (is dominated by) the latter effect, if the pre-tax level of deterrence is sufficiently high (low).

This analysis is only partial. The deterrent value of enforcement efforts, , is a function of both the level of fines and the probability of punishment. This means that increasing the fine and reducing the probability of punishment generates opposing effects on the deterrent value of enforcement efforts. The increase in the level of fines increases the deterrent effect of enforcement efforts, as explained in the text, but the decrease in the probability of punishment reduces the deterrent effect of enforcement efforts. However, as demonstrated by Tabbach (2009), and revealed in the appendix, as long as the probability of punishment in the no tax world is less than 1/2, the fine effect dominates the probability of punishment effect, so that, in total, the deterrent value of enforcement efforts is increased. Moreover, since by Proposition 1, , the assumption that guarantees that the optimal probability of punishment is less than 1/2.

If , which is possible but not necessary if , then the marginal benefits of enforcement efforts will be actually lower. Then, the optimal probability of punishment may be actually lower.

Observe that if income taxation affects pre-tax level of wealth by the applicable tax rate, then a non-deductibility tax regime will essentially be meaningless, since the feasible fine will be reduced exactly as if the fines were deductible, that is, it will become . Indeed, as pointed out above, in the present set up, disgorgement of gains is effectively treated as if it were tax deductible, since the taxpayer cannot lose more than .

Alternatively, a NDTR can be viewed as being equivalent to the no tax social problem assuming that the feasibility constraint is reduced to , but that the “real” impact of any punishment is inflated by a factor of .

Strictly speaking, there is no optimal tax rate!

For an analysis of optimal law enforcement when wealth varies among potential offenders, see Polinsky and Shavell (1984, 1991). This analysis does not take into account the possibility to confiscate the gains offenders derive from the harmful act or analyze how optimal law enforcement should change as the wealth of a subgroup of offenders is changed.