D

Ja

b

a

JKH

KDCMLTD

1

mlindioatIti

tdFeftd

(

0d

International Review of Law and Economics 29 (2009) 1–7

Contents lists available at ScienceDirect

International Review of Law and Economics

eterrence and tax treatment of monetary sanctions and litigation costs

acob Nussima, Avraham D. Tabbachb,∗

Bar-Ilan University, Faculty of Law, IsraelTel-Aviv University, Faculty of Law, Israel

r t i c l e i n f o

EL classification:42

eywords:

a b s t r a c t

The tax treatment of monetary sanctions and litigation expenditures varies across legal jurisdictionsand time. The effects of these different tax regimes – particularly, on crime deterrence – have not beenfully explored. Instead, legal intuitions in court decisions and legislative reforms are found. This paperexplores the effects of these tax regimes. It shows that our common intuitions are sometimes misguided,since we tend to ignore cross-effects between crime and litigation. For example, contrary to commonly

eterrence

rimeonetary sanctions

itigation expendituresaxationeductibility

held views, it is shown that non-deductibility of monetary sanctions may increase the level of crime,if litigation expenses are deductible. In addition, if deductibility of legal expenses depends only on asuccessful trial outcome, this may also increase amounts spent on litigation and time allocated to crime.As this paper shows, however, a complete deductibility regime, under which both monetary sanctions andlitigation expenditures are deductible, maintains the pre-tax levels of crime and litigation expendituresfor risk-neutral offenders. The paper further explores the effects of different tax reforms.

. Introduction

The tax treatment of crime-related activities and consequencesay affect crime enforcement. Indeed, income derived from

egal and criminal activities is generally taxed. For example, thencome associated with offenses committed in the course of run-ing an otherwise legitimate business, such as speeding taxirivers, double-parked messenger services, or overloading by truck-

ng companies, is regularly reported and taxed. Similarly, manyffenders committing white-collar crimes, such as those violatingnti-trust laws, committing fraud, or trading on insider informa-ion, are also likely to report and pay taxes on their illicit gains.ndeed, as a general principle, tax laws are not concerned withhe legality of the source of income when it comes to taxingt.

On the other hand, the tax treatment of crime-related costs, par-icularly monetary sanctions and litigation expenses, is different inifferent jurisdictions and is also non-coherent within jurisdictions.or example, in Canada, both monetary sanctions and litigation

xpenditures are generally deductible for tax purposes if they ariserom business operation. On the other hand, in the United States,he United Kingdom, and Australia monetary sanctions are non-eductible for tax purposes, because deductibility is considered to

∗ Corresponding author. Tel.: +972 3 640 8108; fax: +972 3 640 5349.E-mail addresses: jnussim@mail.biu.ac.il (J. Nussim), adtabbac@post.tau.ac.il

A.D. Tabbach).

144-8188/$ – see front matter © 2008 Elsevier Inc. All rights reserved.oi:10.1016/j.irle.2008.01.001

© 2008 Elsevier Inc. All rights reserved.

reduce the “sting” of punishment and to frustrate “sharply definedpublic policy”. However, litigation expenditures incurred in defend-ing against criminal charges arising from a trade or business aregenerally deductible in these jurisdictions, because it is deemedto manifest a constitutional right. In Israel, both monetary sanc-tions and litigation expenditures are generally non-deductible fortax purposes, because deductibility would be contrary to publicpolicy. These regimes are only mentioned to serve as examples andcertainly do not exhaust the variety of tax rules that can be foundworldwide.2

This paper explores the positive effects of different taxdeduction regimes. It aims to evaluate (1) the effects of thetax rules governing monetary sanctions and litigation expendi-tures on crime (i.e. deterrence) and litigation; and (2) the legalperceptions and intuitions according to which deductibility ofcrime-related expenses violates sharply defined “public policy”principles.

For example, in the landmark decision of Tellier (1966), the tax-payer, a securities dealer, was prosecuted and convicted for fraudand was sentenced to pay an $18,000 fine and serve four and a halfyears in prison. He also incurred and paid approximately $30,000

in legal expenses. How would the tax treatment of these crime-related costs affect similar behavior? This paper shows that simpleintuitions may take us astray.

2 See Section 2 for legal background and additional tax regimes.

2 Review

cotmetawpwetslowea

rbatCge

gtdt3diallmTaA

lto(itTcar2sit

fiH

ea

payer in defense against criminal charges arising from businessoperations are deductible.9 In addition, Canadian courts tradition-ally limit the deduction of fines via various “public policy” tests.10

However, the Supreme Court of Canada recently changed this inter-

J. Nussim, A.D. Tabbach / International

The key insight of the analysis is based on the observation thatrime and litigation are complements in the sense that the moreffenders-taxpayers engage in crime, the more they are willingo spend on litigation, and the more they spend on litigation, the

ore they are willing to engage in crime. This complementarityffect produces interesting, counter-intuitive results, because taxa-ion may affect not only the relative expected returns from legalnd criminal activity, but also the costs and benefits associatedith litigation. For example, contrary to commonly held views, thisaper shows that disallowing deductions for monetary sanctionshile allowing deduction for litigation expenditures may actually

ncourage crime or discourage litigation (though not both). In addi-ion, the paper shows that allowing the deduction of both monetaryanctions and litigation expenditures does not encourage crime oritigation, but rather maintains their pre-tax levels for risk-neutralffenders. The paper also demonstrates that simple intuition worksell only for a tax regime that disallows deductions for both mon-

tary sanctions and litigation expenditures. In this case, both crimend litigation are discouraged, as would be expected.

Understanding the behavioral effects of different tax deductionules on crime and litigation is important not only for its own sake,ut also for normative analysis, since policy makers such as courtsnd legislators base their legal decisions on their perceptions ofhese effects. For example, following the Supreme Court decision inanada allowing deductions for monetary sanctions, the Canadianovernment reacted recently in a proposed legislation prohibitingxplicitly such deductions.

Moreover, if tax rates are even moderately high, the tax rulesoverning monetary sanctions and litigation expenditures are likelyo have tremendous effects. For example, if the tax rate is 25%, theifference between allowing or disallowing deductions for mone-ary sanctions is equivalent to increasing the magnitudes of fines by3%.3 Similarly, the difference between allowing and disallowingeductions for litigation expenditures is equivalent to penaliz-

ng/taxing the costs of litigation by 33%. This clearly should havelarge impact on the incentives to engage in crime and spend on

itigation, particularly since offenders spend a great amount on theatter. For example, the Rigas family who controlled Adelphia Com-

unications Corp. spent $25,000,000 on their criminal defense.he estimated cost of defending Barnes & Noble and Borders inRobinson-Patman Act suit brought by the American Booksellersssociation was $68,000,000.4

The existing literature on the effects of taxation on crime isimited. The very few works in this area have focused on the taxreatment of fines and similar penalties. The pioneering papersf Png and Zolt (1989) and Zolt (1989) and the paper of Tabbach2003) show that a tax regime, under which both legal and criminalncome are taxable and fines are deductible, maintains the pre-ax level of crime as long as risk-neutral offenders are considered.hey also show that disallowing deductions for fines discouragesrime, compared to allowing deductions or to a world with no tax-tion at all. Polinsky and Shavell (1998) derive a similar result withespect to the tax treatment of punitive damages. Tabbach (2003,

005) also examines the effects of the tax treatment of monetaryanctions on crime if offenders are risk averse or if leisure times variable. Hillman and Katz (1984) and Tabbach (2003) explorehe consequences of taxing legal activity while exempting crimi-

3 Assume, for example, a $100 fine and a tax rate of 25%. The after-tax costs of thene is $75 assuming fines are deductible or $100 assuming fines are non-deductible.ence, a before-tax $133 deductible fine is equivalent to an after-tax fine of $100.4 For other examples see “The Price of Corporate Fraud: Shareholders, Insurers and

ven Executives Foot Bill for Time in Court,” The Washington Post (May 9, 2004), andt http://www.carlperson.com/expensive.php (last visited 15.08.07).

of Law and Economics 29 (2009) 1–7

nal activity from taxation altogether.5 All these works completelyignore the decision of offenders/taxpayers to spend resources on lit-igation to defend against criminal charges, and the tax treatment ofsuch spending. This paper is the first to evaluate the joint effects ofthe tax rules governing monetary sanctions and litigation expendi-tures on crime. This is important not only because of the significantmagnitude of criminal litigation costs, but because the effects ofthe tax treatment of monetary sanctions crucially depend on thetax rule governing the deductibility of litigation expenditures andvice versa. By examining the joint effects of different tax deductionrules, this paper qualifies some of the results derived by Png andZolt (1989), Zolt (1989), and Tabbach (2003) and generalizes others.Therefore, it offers a more complete and comprehensive analysis,which is essential for policy makers.

The paper is organized as follows. Section 2 provides legal back-ground. Section 3 develops the general model of taxation andcrime when offenders can spend resources on litigation. The modelis applied in Section 4 to examine the effects of different taxregimes, including: complete deductibility, no deductibility, par-tial deductibility, and contingent deductibility regimes. Section 5applies the model to examine tax reforms—i.e. a change in tax rules.Section 6 provides concluding remarks.

2. Legal background and relevant literature

2.1. Legal background

Income tax laws generally impose taxation on both legal andcriminal activity. In the United States, for example, income fromcriminal activity is taxed without exception since the landmarkdecision James v. United States.6 In other jurisdictions, the generalprinciple is that the tax system is not concerned with the legalityof the source of the income when it comes to taxing it. Therefore,income derived from criminal activity is regularly taxed.7

While illegal income is typically taxed, the tax treatmentof monetary sanctions and litigation expenditures varies acrosslegal jurisdictions. In general, one can find four different, explicittax regimes: (i) complete deductibility: both fines and litiga-tion expenses are deducted; (ii) partial deductibility: only legalexpenses are deducted; (iii) contingent deductibility: only litiga-tion expenses are deducted and only upon acquittal; and (iv) nodeductibility: both fines and legal expenses are non-deductible.8

2.1.1. Complete deductibilityIn Canada, for example, no specific deduction rule concern-

ing monetary sanctions and legal expenses exists in the tax code.According to the Canadian courts, legal expenses incurred by a tax-

5 The only indirectly related literature is on shifting litigation fees. Indeed, the taxtreatment of litigation expenses functions much like fee shifting in civil actions. Yet,this literature is not informative to our case.

6 366 US 213 [1961].7 See, for example, Tilly (2005, p. 349), Silke, on South African income tax (1989C,

pp. 3–7) “[t]he income tax is not concerned with the legality or illegality of a transac-tion. Receipts and accruals from an unlawful business are taxable if there is a schemeof profit-making involved.”

8 Theoretically, there might be a partial tax regime that would allow deductionsfor monetary sanctions but would deny deductions for litigation expenditures, butits adoption seems unlikely and we are unaware of such a regime in reality.

9 See Minister of National Revenue v. L.D. Caulk Co. of Canada Ltd., 52 D.T.C. 1034(1952) (Ex. Ct.), affirmed by 54 D.T.C. 1011 (1954) (Supreme Court of Canada); RolandPaper Co. Ltd. v. Minister of National Revenue, 60 D.T.C. 1095 (1960) (Ex. Ct.).

10 See, for example, Day & Ross v. The Queen, 76 D.T.C. 6433 (1975) (Fed. T.D.).

Review

prep

2

t1eeltTdtpdw

tdtptlovtfth

2

decce

2

Snid

oipTei

dI

9

F

i

work are given with certainty by the function W(a, t) = (1 − t)w(a),

J. Nussim, A.D. Tabbach / International

retation of the law in 65302 British Columbia Ltd. v. The Queen,uling that fines and penalties are generally deductible.11 The gov-rnment, following the Supreme Court ruling, recently reacted in aroposed legislation explicitly prohibiting such deductions.12

.1.2. Partial deductibilityThis is the contemporary tax regime in the US, the UK, and Aus-

ralia. In Australia, Section 26-5 of the Income Tax Assessment Act997 specifically denies deduction of fines and penalties. Yet, legalxpenses incurred in the course of defense against criminal pros-cution are allowed by Australian courts.13 In the UK, no specificegislation applies to these deductibility issues, and the law is dic-ated by the House of Lords decision in McKnight v. Sheppard.14

he court ruled that fines are not deductible since deduction wouldilute the legislative policy behind the fine.15 Legal expenses, onhe other hand, are allowed by the House of Lord since it finds noolicy infringed by such deduction. Moreover, it argues “that non-eductibility would be in effect an additional fine or penalty forhich the regulatory scheme does not provide.”

In the US, Section 162(f) of the Internal Revenue Code codifieshe landmark decision Tank Truck Rentals Inc. v. Commissioner andisallows the deductibility of monetary sanctions in computingaxable income, even if incurred in a trade or business or in theroduction of legal or illegal taxable income. In Tank Truck Rentals,he Supreme Court held that allowing deductibility of fines vio-ates sharply defined public policy, because it reduces the “sting”f punishment and thus may encourage crime. In Commissioner. Tellier, the Supreme Court held that legal expenses incurred inhe defense against criminal prosecution may qualify for deductionrom taxable income, since no “public policy” is contravened whenhe accused exercises his constitutional right to take a lawyer toelp with the defense.

.1.3. Contingent deductibilityPrior to Tellier, the legal practice in the US conditioned

eductibility of legal expenses on trial outcomes: while legalxpenses incurred in the unsuccessful defense against criminalharges would be barred, those in the successful defense againstriminal charges would not be. (As noted above, the Tellier decisionliminated this distinction in the US.16)

.1.4. No deductibilityIn Israel, no specific legislation on these issues exists. The Israeli

upreme Court, however, held that fines and similar penalties areot deductible due to “public policy” concerns. Legal expenses

ncurred in the course of criminal procedures are also generallyisallowed by Israeli courts for similar reasons.

In addition to these four explicit tax regimes, there are also a fewther implicit tax regimes. Their common trait is the implicit andndirect possibility to partially (or fully) deduct the “real” costs of

unishment regardless of the tax treatment of litigation expenses.his may be possible if punishment is non-monetary. Take, forxample, the suspension or revocation of a trade license. The pun-shment materializes in lost (or reduced) trade profits, so the net

11 99 D.T.C. 5799 (2000). Though the court left a certain limited possibility to denyeductions. Id., paragraph 69. Additionally, deduction of penalties arising under the

ncome Tax Act is explicitly prohibited under Section 18(1)(t).12 Interpretation Bulleting IT-104R3, “Deductibility of Fines and Penalties(̈August, 2002).13 See, e.g., Magana Alloys & Research Pty Ltd v FCT, 49 F.L.R. 183 (1980); Putnin v.CT, 27 FCR 508 (1991).14 McKnight v. Sheppard, 2 All E.R. 491 (1999).15 See also Inland Revenue Commissioners v. Alexander von Lehn, 2 K.B. 553 (1920).16 In McKnight, the House of Lords explicitly rejected this distinction in the Englishncome tax law.

of Law and Economics 29 (2009) 1–7 3

real loss to the offender is the after-tax profits. Since the after-taxprofits are affected by the tax rate, it is as if the real costs of punish-ment are deducted.17 Similarly, imprisonment generates a partiallydeducted punishment. In addition to the disutility of imprison-ment, stemming from its restriction on consumption and leisuretime (freedom), the offender is barred from making profits duringthe period of imprisonment and may also suffer from a reduction inhis future earning capacity. This latter portion of the punishmentis also indirectly deductible since the offender loses its after-taxvalue.

Not only can non-monetary punishment be partially or fullydeductible, but certain other amounts paid as a consequence of acriminal act may not be subject to disallowance. Monetary sanc-tions that are determined “compensatory” rather than “punitive”in nature are generally deducible. For example, deductions aregenerally allowed for amounts paid as damages or as (voluntary)restitution. Similarly, remediation of damages, such as when pol-luters are made to clean up contaminated sites, is generally alsodeductible.

Therefore, although fines and similar penalties are typically non-deductible, non-monetary or “compensatory” punishment may bepartially or fully deductible. Hence, under these scenarios, the US,UK, and Australian partial deductibility regime may become simi-lar to the Canadian complete deductibility regime, and the Israelino deductibility regime may transform into a new, different partialdeductibility regime.

3. A model of taxation and crime

This section presents a general, simple model of taxation andcrime based on Tabbach (2003), modified to include the offenders’decision to spend resources on litigation.18 Like models of prop-erty or income-producing crimes in a world without taxation, thismodel formulates the criminal problem as a labor supply decisionunder uncertainty. Nevertheless, the model may also represent aninvestment portfolio in which resources are allocated between legaland illegal investments—e.g. a business choosing among activitiesor inputs, some of which are illegal.

Assume that individuals can only participate in two marketactivities, work and crime, and that they choose the optimal timeallocation between these two activities at the beginning of a givenperiod. Training or other entry costs are not required in either ofthese activities, and movement between them is costless. A totalamount of time, T, normalized to one, is allocated between thesetwo activities, with a fraction a spent on crime and 1 − a on work,where 0 ≤ a ≤ 1. Time devoted to all other non-market activities(leisure) is fixed.19

Assume that the returns in both activities are a monotonicallyincreasing function of allocated time. The after-tax returns from

where t is the proportional tax rate in place.20 These returns aremarginally decreasing, so w′(a) < 0 and w′′(a) < 0.21 The “real”after-tax returns from crime, on the other hand, depend on three

17 To illustrate, suppose that the before-tax profits are $100,000 and the tax rate is25%. Thus, the real cost of a punishment that denies these profits is $75,000. Takento the extreme, if the tax rate approaches 100%, the real costs of such a punishmentis zero.

18 The model in Tabbach (2003) is based on Ehrlich (1973).19 Interpreting the model in wealth terms (as in Png & Zolt, 1989), instead of fixed

labor (leisure) time, offenders have fixed endowment which they allocate to, orinvest in, legal and criminal inputs.

20 A non-linear tax schedule is more complicated to analyze and adds nothingimportant to the analysis.

21 For ease of notation, the returns from legal activities are defined as a functionof time allocated to crime rather than of time allocated to work. Since w′(1 − a) > 0and w′′(1 − a) < 0, it follows that w′(a) < 0 and w′′(a) < 0.

4 Review

svomaeItepptnbeirtoinataga

c

E

w

X

Y

Z

rtcvwt

E

Tt

−a

−Ooaegeef

eiq

activity, and of how much to invest in litigation.This result generalizes the proposition that the deductibility of

monetary sanctions maintains the pre-tax level of criminal activ-ity for risk-neutral offenders (Png & Zolt, 1989; Tabbach, 2003).

24

J. Nussim, A.D. Tabbach / International

tates of the world: non-apprehension, apprehension without con-iction, and conviction with punishment. If not apprehended,ffenders receive the after-tax returns from crime in a monetary oronetary-like form, as given by the function C(a, t) = (1 − t)c(a), also

ssumed to be concave, c′(a) > 0 and c′′(a) < 0. However, under thexogenous probability q, offenders are not completely successful.nstead, they are apprehended and face criminal charges. Offendershen make a second decision on how much to spend on litigation,. Consequently, they are convicted and punished with probability(e).22 Assume that offenders’ investment in litigation reduces therobability of conviction and punishment with decreasing rates, sohat p′(e) < 0 and p′′(e) > 0. If offenders are successful in their crimi-al defense, they still receive the after-tax returns from their crimesut are worse off by the after-tax amounts spent on litigation:(1 − �t), where � (0 ≤ � ≤ 1) represents the degree of deductibil-ty of litigation costs. However, if offenders are found guilty, theireturns from crime are further reduced by an after-tax fine, given byhe function F(a, ı, t) = (1 − ıt)f(a), where ı (0 ≤ ı ≤ 1) is the degreef deductibility of the fines. As usual, fines are assumed to exhibitncreasing marginal severity, f′(a) > 0 and f′′(a) > 0. To rule out theeed to resort to non-monetary sanctions, namely imprisonments a form of punishment, it is assumed that individuals have ini-ial wealth, W, which is sufficient to pay the fine regardless of themount.23 Finally, assume that tax revenues are used to financeovernment spending that enters individuals’ utility function in andditively separable way.

Under the foregoing assumptions, the individual’s problem is tohoose a and e, subject to 0 ≤ a ≤ 1, to maximize:

[U(·)] = (1 − q)U(X) + q(1 − p(e))U(Y) + qp(e)U(Z), (1)

here:

= W + (1 − t)[w(a) + c(a)]; (2A)

= W + (1 − t)[w(a) + c(a)] − (1 − �t)e; (2B)

= W + (1 − t)[w(a) + c(a)] − (1 − �t)e − (1 − ıt)f (a) (2C)

epresent terminal wealth at the end of the period, under condi-ions of non-apprehension, apprehension without conviction, andonviction with punishment, respectively. U(·) is the individual’son Neumann–Morgenstern utility function defined over terminalealth. If individuals are assumed to be risk-neutral, the maximiza-

ion problem (1) reduces to:

W = W + (1 − t)[w(a) + c(a)] − q[(1 − �t)e + p(e)(1 − ıt)f (a)]. (3)

he first-order conditions of an interior maximum of (3) requirehat:

(1 − t)w′(a) = (1 − t)c′(a) − qp(e)(1 − ıt)f ′(a) (4)

nd

p′(e)(1 − ıt)f (a) + �t = 1 (5)

ffenders choose, in equilibrium, a*(t, �, ı) and e*(t, �, ı) simultane-usly. The optimal level of crime, a*, is chosen where the marginalfter-tax returns from legal activity equal the marginal after-taxxpected returns from criminal activity. The optimal level of liti-

ation – e* – is chosen so that the last dollar spent on litigationquals its marginal benefit in terms of the marginal reduction inxpected punishment plus the marginal tax savings due to litigationee deduction.

22 We use a few simplifying assumptions: exogenous probability of apprehension;does not utilize resources (in this case, time); the probability of punishment is

ndependent of the level of crime. Relaxing any of these assumptions would have noualitative effect on this paper’s results.23 Regarding the effects of taxation on imprisonment, see Tabbach (2003).

of Law and Economics 29 (2009) 1–7

Having set the general model, two kinds of tax effects can beexamined. Section 4 investigates the effects of changes in tax rateson crime and litigation expenditures under various tax regimes.Section 5 examines the effects of tax reforms (comparing differenttax regimes) on these levels. Throughout the analysis it is assumedthat the parameters of the model take on values resulting in aninterior solution.24

4. Tax regimes

This section examines the effects of changes in tax rates on crimeand litigation expenditures under various tax regimes. This analy-sis is important since tax rate changes take place very often, andsince different offenders may be subject to various tax rates. Addi-tionally, the analysis can be interpreted as comparing the differenttax regimes to a no-tax world (at least for small tax rates). Thiscomparison is also important since some offenses – for example,double-parking or speeding – may be committed in a business ora personal capacity and, accordingly, may be subject to or exemptfrom taxation. Furthermore, some offenders may also be effectivelyexempt from taxation.

4.1. Complete deductibility

Examine first the effects of a complete deductibility regimeunder which both legal and criminal incomes are taxable and mon-etary sanctions and litigation expenditures are deductible (withrefund if necessary), all at the same proportional tax rate t (e.g.the Canadian tax regime). Adjusting the model in (3) to capturea complete deductibility regime by setting ı = � = 1, the first orderconditions for optimization require that:

−w′(a) = c′(a) − qp(e)f ′(a) (6)

−p′(e)f (a) = 1 (7)

These conditions do not depend on the tax rate. They are actuallyidentical to the conditions that would prevail with no taxation atall—i.e. conditions (4) and (5) given t = 0. Thus, changes in tax ratesdo not affect the crime level or the amount spent on litigation.25

The intuition behind this simple yet powerful result is straight-forward. A complete deductibility regime implies that expenditureson litigation are cheaper by the percentage of the tax rate, but, at thesame time, less beneficial, because the effective punishment levelis also reduced by the same percentage of the tax rate. In addition,a complete deductibility regime implies that the after-tax marginalreturns from work are reduced by the same proportions as the after-tax marginal expected returns from crime (i.e. the tax rate). Thus,at the margins, a complete deductibility regime does not distortthe decisions of how to divide the time between legal and criminal

The necessary conditions for an interior solution require that the marginalexpected returns from crime are greater than those from work, evaluated at the pointof specialization at work, that is, (1 − t)c′(a) − qp(e)(1 − ıt)f ′(a) > −(1 − t)w′(a),evaluated at a = 0. In addition, the marginal expected returns from crime must beless than those from work, evaluated at the point of specializing in crime, thatis, (1 − t)c′(a) − qp(e)(1 − ıt)f ′(a) < −(1 − t)w′(a) evaluated at a = 1. Furthermore,the marginal reduction in expected punishment, −p′(e)(1 − ıt)f(a), must be greater(smaller) than 1 − �t, evaluated at e = 0 (e = ∞). Note further that the imposition ofeach tax regime may change the conditions of entering the criminal market andspecializing in crime, and thus may affect the participation level of offenders.

25 Put differently, if the optimal choices of a and e in the absence of taxation aredenoted as a* and e*, then a* and e* would be the optimal choices also under acomplete deductibility regime (for any t < 1).

Review

Ir

4

tIs

−−Ttor

w

|

Tatsf

ctnctAaaepdtltletelp

4

te

esotf

J. Nussim, A.D. Tabbach / International

t shows that this proposition holds even if offenders can spendesources on litigation, as long as such spending is also deductible.

.2. No deductibility

Consider next a tax regime under which both monetary sanc-ions and expenditures on litigation are non-deductible (e.g. thesraeli tax regime). In terms of our model, such a regime corre-ponds to ı = � = 0. Adjusting (3) and re-optimizing yields:

(1 − t)w′(a) = (1 − t)c′(a) − qp(e)f ′(a) (8)

p′(e)f (a) = 1. (9)

he behavioral response of offenders to changes in tax rates underhis regime can be investigated by implicitly differentiating theptimal solution, a∗

ND and e∗ND (where ND stands for no deductibility

egime), with respect to t, which yields:

∂a∗ND

∂t= 1

|H| [HeaHet − HeeHat] (10)

∂e∗ND

∂t= 1

|H| [HaeHat − HaaHet], (11)

here

H| =∣∣∣∣

Haa Hea

Hae Hee

∣∣∣∣

and Hij = ∂EW

∂i∂j.

he second-order conditions for a local maximum imply that |H| > 0nd Hee < 0. In addition, direct calculation reveals that Het = 0, andhat Hat = −[w′(a) + c′(a)] < 0, so ∂a∗

ND/∂t < 0. To determine theign of ∂e∗

ND/∂t, observe that Hea = −qp′(e)f′(a) > 0. Since Het = 0, itollows that ∂e∗

ND/∂t < 0.A no deductibility regime leads to a reduction in both the level of

rime and litigation expenditures.26 The economic explanation ofhese results is simple. Increasing (decreasing) the tax rate under ao deductibility regime affects the expected marginal returns fromriminal activities proportionally more (less) than from legal activi-ies. This induces individuals to substitute legal for criminal activity.t the same time, at the optimum, the marginal costs and benefitsssociated with litigation are not directly affected. Indeed, Eqs. (9)nd (7) are in the same form. This follows, of course, since mon-tary sanctions and litigation costs are treated similarly for taxurposes. However, litigation expenditures are affected indirectlyue to changes in the crime level. Since criminal activity and litiga-ion costs are complementary for the taxpayer – in the sense thatower levels of criminal activity induces lower litigation expendi-ures, and vice versa – offenders would reduce their investment initigation costs. The complementarity between crime and litigationxpenditures, which is mathematically given by Hea > 0, is due toheir mutual effect on punishment. Higher crime levels raise thexpected punishment and, hence, increase the marginal benefit ofitigation. Similarly, higher litigation expenditures reduce expectedunishment and, hence, the marginal costs of crime.

.3. Partial deductibility

Analyze now the effects on crime and litigation expenditures ofhe contemporary tax regime in the US, the UK, and Australia: mon-tary sanctions are non-deductible but expenditures on litigation

26 Note also that the imposition of the present regime changes the conditions ofntering the criminal market (and specializing in crime). Under the present regime,uch conditions are less likely to be met. Thus, assuming for example heterogeneityf offenders, a non-deductibility regime not only reduces the level of crime of part-ime offenders (i.e., a < 1), but also the level of participation in criminal activity (bothull- and part-time).

of Law and Economics 29 (2009) 1–7 5

are unconditionally deductible. Accordingly, adjust (3) by settingı = 0 and � = 1, and re-optimize:

−(1 − t)w′(a) = (1 − t)c′(a) − qp(e)f ′(a) (12)

−p′(e)f (a) + t = 1 (13)

Implicitly differentiating the optimal solution, a∗PD and e∗

PD (wherePD stands for partial deductibility regime), with respect to t yields(10) and (11), where now Hat = −[w′(a) + c′(a)] < 0, Het = q > 0, andHea = − qp′(e)f′(a) > 0. The second-order conditions for maximiza-tion imply that |H| > 0, Hee < 0, and Haa < 0, so the signs of ∂a∗

PD/∂tand ∂e∗

PD/∂t cannot be determined unambiguously.These results are very important and interesting. They imply

that under the US, UK, and Australian tax regimes, increasing taxrates will not necessarily reduce crime levels or increase litigationexpenditures, as is commonly thought.27

The economic explanation of these ambiguous effects is as fol-lows. The direct effect of changes in tax rates on crime under the PDregime is similar to the ND regime. Thus, higher tax rates reducecrime, and in turn indirectly lead to lower investment in litigation(since a and e remain complements). At the same time, however,allowing deductions for litigation expenditures subsidizes the costsof defending against criminal charges without affecting the bene-fits of litigation. This creates a direct incentive to spend more onlitigation for any criminal activity level. In turn, increased spendingon litigation indirectly induces higher criminal activity levels (since,again, a and e are complements). Accordingly, when tax rates areincreased, criminal activity and litigation expenditures are directlyaffected in opposing directions: the net expected marginal bene-fit of criminal activity decreases relative to legal activity, on theone hand, while the marginal benefit of litigation for any level ofcriminal activity increases, on the other hand. Since the behav-ioral variables – i.e. criminal activity and litigation expenditures– are complementary, the direct effect on each of them triggers anopposing indirect effect on the other variable.

Accordingly, it is not clear whether offenders would reduce theircriminal activity or increase investment in litigation as a result ofa tax-rate increase. It is possible that offenders would reduce theircriminal activity sufficiently to decrease the total expected punish-ment they face, in turn, also reducing their investment in litigation.Similarly, offenders may be better off increasing their investmentin litigation sufficiently to reduce the total expected punishmentfrom criminal activity enough to lower the marginal costs of crime,thereby increasing their level of criminal activity.28 This result iscounter-intuitive, especially when compared with the two other“extreme” regimes—i.e. complete deductibility and no deductibil-ity. The effect of allowing deductions for litigation costs on crimehas also gone unnoticed by courts in various jurisdictions.

4.4. Contingent deductibility

Lastly, consider the effects of a tax regime under which finesare not deductible but litigation expenses are deductible only uponacquittal (contingent deductibility regime) on crime and litigationexpenditures. In terms of our model, such a regime can be repre-sented by ı = 0 and � = 1 − p. Adjusting (3) and re-optimizing yield

the following first order conditions:

−(1 − t)w′(a) = (1 − t)c′(a) − qp(e)f ′(a) (14)

−p′(e)[et + f (a)] = 1 − t[1 − p(e)] (15)

27 Note again the effect of the tax regime on incentives to participate in the criminalmarket. See supra Notes 24, 26.

28 The only result which is impossible is that offenders will increase their criminalactivity and, at the same time, decrease their litigation expenditures.

6 J. Nussim, A.D. Tabbach / International Review of Law and Economics 29 (2009) 1–7

Table 1The effects of tax rates changes on criminal activity and litigation expenditure under different tax regimes

Tax regimes Complete deductibility No deductibility Partial deductibility/contingent deductibility

Crime Litigation Crime Litigation Crime Litigation

Direct effects – – ↓ – ↓ ↑Cross-effects* – – – ↓ ↑ ↓N

a > 0.

ICy−

t

romeitimetbm

or

5

toTtttaiaCidlliimo

ladios

v

increases the levels of expenditure on litigation and criminal activ-ity (irrespective of the fine deduction rule). The intuition is quite

et effect – – ↓* Because criminal activity and litigation expenditures are complementary, i.e. He# Indeterminate.

mplicitly differentiating the optimal solution, a∗CD and e∗

CD (whereD stands for contingent deductibility regime), with respect to tields (10) and (11), where now Hat = −[w′(a) + c′(a)] < 0, Hea =q(1 − p(e) + ep′(e)) >< 0, and Hea = −qp′(e)f ′(a) > 0.

It can be easily verified, using the second order conditions, thathe signs of ∂a∗

CD/∂t and ∂e∗CD/∂t are ambiguous.

The explanation is similar to that of the partial deductibilityegime, yet not identical. The direct effect of changes in tax ratesn crime is to reduce crime and, in turn, to lower the invest-ent in litigation due to the complementarity effect. The direct

ffect of changes in tax rates on litigation expenditures, however,s different. Under a CD regime, both the marginal benefit andhe marginal cost of litigation are higher. The marginal benefits higher since the expected loss due to conviction is larger. The

arginal costs are higher since the expected deduction for litigationxpenses is smaller. Thus, the direct effects on litigation expendi-ures themselves are ambiguous. Mathematically, this can be seeny observing that: Het = −p′(e)e − p(e) + 1 ≥≤ 0. Due to comple-entarity, the total effect on crime is also unclear.In summary, this section examined the effects of tax rate changes

n crime and litigation expenditures under various tax regimes. Theesults are summarized in Table 1.

. Tax reform

This section demonstrates how the basic model can also be usedo analyze tax reforms. Accordingly, the effects of moving fromne tax regime to another are examined through a few examples.hese comparisons between tax regimes are significant becausehey represent possible tax reforms. When a legislator changes aax rule or when courts decide what the appropriate tax rule is,hey effectively choose between alternative tax regimes, holdingll other variables, including tax rates, constant.29 For example,n Tank Truck Rentals Inc., the Supreme Court decided whether tollow or disallow deductions for monetary sanctions. Similarly, theanadian legislator has been contemplating a partial deductibil-

ty regime instead of the complete deductibility regime that wasetermined by the Supreme Court of Canada. In addition, in Tel-

ier, the decision was whether to allow or disallow deductions foritigation expenses incurred in an unsuccessful defense against crim-nal charges. Although a simple comparison between tax regimess problematic because it disregards differences in tax revenues, it

ay still be important in scrutinizing the reasoning and intuitionf the courts and the legislators.

The effects of the Tank Truck Rentals decision or of the proposedegislation in Canada for denying the deductibility of fines represent

tax reform that replaces a complete deductibility with a partial

eductibility regime. The effects of such reform on crime and lit-

gation expenditure can be illustrated by differentiating offendersptimal choice – i.e. a* and e* – with respect to ı, where ı repre-ents the degree of deductibility of fines. As can be shown, higher

29 While a legislator can simultaneously change tax rules, tax rates, and any otherariable, including enforcement schemes, it generally does not do so.

↓ # #

ı does not necessarily increase crime or litigation expenditures asmight be intuitively expected. In particular, if (p′(e))2 > p(e)p′′(e),then greater deduction of fines actually reinforces deterrence. It fol-lows that the Tank Truck Rentals reform – i.e. disallowing deductionfor fines – does not necessarily increase the “sting” or the “impact”of punishment as the court reasoned. It may actually generatethe opposite result – i.e. stimulating crime – as long as litigationexpenses are deductible.30 The correct intuition is similar to thediscussion in Section 4.3. Put differently, the Canadian’s contem-porary complete deductibility regime does not necessarily dilutecrime deterrence in comparison to the US, UK, or the Australian’spartial deductibility regimes.

This analysis is also applicable to the court’s choice of criminalpunishment. Given a partial deductibility tax regime that disal-lows deduction of fines, imposing a non-monetary punishment(e.g. revocation of trade license, imprisonment, mandated pollu-tion cleanup) may implicitly allow a partial or full deduction ofpunishment.31 Hence, replacing fine punishment with an equiv-alent non-monetary punishment may actually affect the sting ofpunishment under a partial deductibility tax regime.

The behavioral effects of the Tellier decision, which representsa shift from a contingent to a partial deductibility regime, are notobvious and require further analysis. The difference between con-tingent deductibility and partial deductibility regimes is found inthe optimization condition with regard to e (compare (15) with(13)). A contingent deduction of litigation expenses upon acquittalincreases the expected marginal costs of litigation by tp(e) but alsothe expected marginal benefit from litigation by −p′(e)et. It followsthat the effects of the Tellier tax reform depend on the relative mag-nitude of these changes in marginal costs and benefits. This relativemagnitude depends on the elasticity term � = −(p′(e)/1)(e/p(e)),which denotes the elasticity of the probability of punishment (givenapprehension) with respect to investment in litigation. If the elas-ticity is less than unity, then the effect of the Tellier decision, aswould be expected, is to boost incentives to invest in litigation and,in turn, to increase the level of crime. However, if the elasticityis greater than unity, then the Tellier decision, counter-intuitively,diminishes incentives to invest in litigation, and thereby reducescriminal activity. In the case of unit-elasticity – that is, if an increaseof 1% in litigation expenditures leads to a 1% reduction in the prob-ability of conviction and punishment – the Tellier decision has noeffect on litigation expenditures or crime level.

Other tax reforms can be similarly analyzed. For example, it canbe easily shown that changing the deduction rule from disallowingto allowing deductions for litigation expenditures unambiguously

simple. Allowing deductions for litigation cost is equivalent tosubsidizing it by the amount te, which reduces both the costs of

30 Actually, Tank Trunk Rentals Inc. was decided before Tellier (which is explorednext). That is, the existing tax regime when Tank Trunk Rentals Inc. was decidedallowed deduction of legal expenses only of prevailing litigants. Thus, the effect ofTank Trunk Rentals Inc. decision is somewhat different but qualitatively similar.

31 See supra, Section 2.

Review

lade

6

ottasFidirdrtcsas

atdvtFtttfle

aP

aie

J. Nussim, A.D. Tabbach / International

itigation and of crime. These two (direct) effects reinforce onenother, leading to an increase in the levels of litigation expen-itures and crime. Thus, disallowing deductions for litigationxpenditures (even taxing them) would reduce crime.

. Conclusion

The tax treatment of crime and litigation expenditures maybviously affect deterrence and, therefore, is important. However,he resulting effects are not straightforward. This paper exploreshese effects on crime deterrence and litigation effort generated bylternative tax deductibility regimes, given a crime enforcementcheme. It shows that our common intuitions are often misleading.or example, contrary to commonly held views, a partial deductibil-ty regime – such as the US, UK, or Australian tax regime – mayilute, rather than strengthen, crime deterrence. This is true for

ncreasing tax rates or in comparison to a complete deductibilityegime. Similarly, a Canadian-like complete deductibility regimeoes not induce crime or litigation, as commonly believed, butather maintains their pre-tax levels. This paper provides a posi-ive framework for potential tax reforms, such as the one recentlyontemplated by the Canadian legislators. The model and analysisuggested in this paper can also assist in evaluating the effect oflternative criminal sanctions such as monetary vs. non-monetaryanctions.

The qualitative results of this paper, particularly those associ-ted with a complete deductibility regime, are generally robusto certain refinements of the model, specifically those concerningifferent assumptions about the probability of apprehension, con-iction or punishment.32 However, the results may be sensitive tohe risk attitudes of offenders and to the variability of leisure time.or example, if offenders are risk averse, rather than risk neutral,

hen a complete deductibility regime no longer maintains the pre-ax levels of crime and litigation. This is true for two reasons: first,he direct effect of such a regime on crime is ambiguous due to con-icting risk effects, as shown by Tabbach (2003)33; second, it canasily be shown that the direct effect of such a regime on litigation

32 For example, the probability of apprehension can be endogenized, or the prob-bility of conviction and punishment may be a function of the crime level as well.roofs of these cases are available from the authors on demand.33 Complete deductibility reduces the amount of risk associated with criminalctivities, and thus induces offenders to engage in more crime. On the other hand,t may reduce wealth, thereby reducing the attractiveness of crime, for offendersxhibiting decreasing absolute risk aversion.

of Law and Economics 29 (2009) 1–7 7

is also unclear, since taxation may change both the marginal costsand benefits of litigation.34

Similarly, if leisure time can vary, one can expect at leasttwo additional, well-known tax effects to enter the analysis. Onthe one hand, taxation reduces time allocated to both legal andcriminal activities, because it makes leisure relatively cheaperthan market activities (substitution effect). On the other hand, tothe extent that taxation reduces wealth, it increases time allo-cated to market activities as long as leisure is a normal good(income effect). These two additional effects work in oppositedirections and make the analysis more complex and the resultsambiguous.35

Acknowledgements

For their helpful comments, we thank Sharon Hannes, ArielPorat, Omri Yadlin, and the participants in the law and economicsworkshops at the University of California at Berkeley and Tel AvivUniversity, and in the American (2005), Israeli (2005), and Scandi-navian (2006) Associations of Law and Economics conferences.

References

Ehrlich, I. (1973). Participation in illegitimate activities: A theoretical and empiricalinvestigation. Journal of Political Economy, 81, 521–565.

Hillman, A. L., & Katz, E. (1984). Note, excise taxes, import restrictions, and the allo-cation of time to illegal income-earning activity. International Review of Law andEconomics, 4, 213–222.

Png, I. P. L., & Zolt, E. M. (1989). Efficient deterrence and the tax treatment of monetarysanctions. International Review of Law and Economics, 9, 209–217.

Polinsky, A. M., & Shavell, S. (1998). Punitive damages: An economic analysis. HarvardLaw Review, 111, 869–962.

Silke, A.S. (1989C). Silke on South African income tax. Kenwyn: Juta & Co. in associ-ation with Divaris Stein Pub.

Tabbach, A. D. (2003). Criminal behavior, sanctions, and income taxation: An eco-

nomic analysis. Journal of Legal Studies, 32, 383–406.

Tabbach, A. D. (2005). The tax treatment of crime with variable leisure time. Inter-national Review of Law and Economics, 25, 276–289.

Tilley, J. (2005). Revenue law. Oxford and Portland, OR: Hart Publishing.Zolt, E. M. (1989). Deterrence via taxation: A critical analysis of tax penalty provi-

sions. UCLA Law Review, 37, 343–387.

34 Moreover, for risk-averse offenders, crime and litigation need not be comple-ments.

35 See, for example, the analysis in Hillman and Katz (1984) and Tabbach (2005).