Profi t Tax Evasion under Oligopoly with Endogenous Market Structure

Transcription

1 Profi t Tax Evasion under Oligopol with Endogenous Market Structure Profi t Tax Evasion under Oligopol with Endogenous Market Structure Abstract - This note investigates the impact of profit tax evasion on firms output decisions in a Cournot oligopol setting in which the market structure is determined endogenousl. It is shown that tax evasion intensifies market entr and raises aggregate output, while production of each incumbent firm decreases. Therefore, tax evasion choices affect activit decisions and an evadable profit tax distorts the market outcome. Laszlo Goerke Department of Economics, Eberhard Karls Universit Tübingen, D Tübingen, German and Institute for the Stud of Labor (IZA), D Bonn German and CESifo, Munich, German Marco Runkel Department of Economics, Universit of Munich, D Munich, German and CESifo, Munich, German National Tax Journal Vol. LIX, No. 4 December 2006 INTRODUCTION firm s decision to evade taxes ma be separable from its A activit choice. This separabilit feature has important implications since, for example, the neutralit of a profit tax is preserved in the presence of tax evasion opportunities. Snthesizing previous analses, Yaniv (1995) establishes the conditions that guarantee that a firm s input or output choices are independent of its evasion decision. There are a number of settings, however, in which the separabilit feature ma not hold. First, this will be the case if the firm cannot attain its preferred extent of tax evasion and its choice variable is the fraction of the tax base declared, instead of the absolute amount (Kreutzer and Lee, 1986; Wang and Conant, 1988; Yaniv, 1995, 1996). Activit and evasion choices ma not be separable either if, second, the extent of tax evasion exerts a direct negative effect on gross profits (Kreutzer and Lee, 1988; Virmani, 1989) or, third, the detection probabilit and/or the penalt rate are determined endogenousl b, e.g., the firm s reported revenues or costs (Marelli, 1984; Marelli and Martina, 1988; Virmani, 1989; Lee, 1998). Finall, separabilit will not hold if the uncertaint caused b tax evasion is complemented b a second source of uncertaint regarding either the outcome of the firm s activit (Yaniv, 1995) or the investment decision (Panteghini, 2000). All of the above contributions take the market structure as given. However, tax evasion occurs because firms are better off when misreporting than when telling the truth to tax authorities. Hence, tax evasion affects the firms paoffs and the gains from entering the market. The innovation of our analsis is to endogenize the market structure. We rule out all causes for a non separabilit of output and evasion choices established thus far, and develop a Cournot oligopol model 851

2 NATIONAL TAX JOURNAL with profit tax evasion and an endogenous number of firms. 1 Prior to the decisions about the activit level and the amount of profit tax evasion, firms choose whether to enter the market at positive costs. We can show that tax evasion influences the production level of each firm and alters aggregate output. The reason is that, under a given market structure, tax evasion enhances the expected paoff of each firm. This increase will attract new firms. As a consequence, output of the incumbent firms falls. The additional output of the new firms outweighs the decline in the production of the firms alread in the market so that aggregate output increases. In sum, we identif the endogeneit of the market structure as a further reason wh profit tax evasion affects individual and aggregate production. Hence, under oligopol with endogenous market structure, profit taxes will not be neutral and we shall argue below that tax evasion will tend to render the market structure less efficient. MODEL AND RESULTS Consider an industr with a continuum of potential firms that will produce a homogeneous good once the have entered the market. To ensure that non marginal firms have positive gross profits in equilibrium and an incentive to evade a tax on profits, firms are assumed to differ in production costs. A firm of tpe has costs C(x, ), given a production quantit of x units, with C x > 0 and C xx 0. Moreover, we assume C(0, 1 ) C(0, 2 ) for 1 > 2 and C x 0, with strict inequalit in at least one of these two conditions. Hence, a firm with a large has higher fixed and/or higher marginal production costs than a firm with a small. Note that these conditions impl C > 0. The cost parameter is continuousl distributed over the interval [, ]with <, according to the distribution function F() and the densit function f() = df()/d. Without loss of generalit, the mass of firms is normalized to unit. A firm makes three decisions at most. First, it decides whether to enter the market at costs k. If it enters, it will subsequentl select output and the extent of tax evasion. To ensure a subgame perfect solution, we start with the latter decisions and take market entr as given, so that entr costs k are sunk. Consider a firm of tpe and assume that this firm produces an output of x in the second stage. Let x be the output of all other firms operating in the market. Aggregate output is := x + x, and P() with P < 0 defines the inverse demand function. We focus on the most relevant case of strategic substitutes defined in Bulow, Geanakoplos and Klemperer (1985). Marginal profits of an one firm will then decrease with the output of the firm s rivals. Formall we have P + xp < 0. This implies 2P + xp C xx < 0, which is a well known necessar condition for the stabilit of the equilibrium in oligopol models (Dixit, 1986). Each firm has to pa a profit tax at rate τ. The tax base of firm is [1] Π= xp( x + x) C( x, ) k. The firm can evade taxes b understating profits. Let s be the absolute amount of understatement. The tax authorit audits firms with an exogenous probabilit q, and if a firm is audited, tax evasion will be detected with certaint. In this case, the firm has to pa the full amount of taxes due and a penalt that is proportional to the amount of taxes evaded. The penalt rate is exogenous and denoted b δ > 0. After tax profits of the firm will equal 1 Virmani (1989) also considers a model with (free) entr of firms. However, since the output market is alwas perfectl competitive, the market structure is effectivel exogenous. 852

3 Profi t Tax Evasion under Oligopol with Endogenous Market Structure n [2] Π = ( 1 τ) Π+ τs if tax evasion remains undetected, and d [3] Π = ( 1 τ) Π δτs if the tax authorit catches the firm evading taxes. In line with most of the literature referred to in the introduction, a firm is assumed to maximize the expected utilit of profits. It has an increasing and strictl concave utilit function U. 2 Without loss of generalit, we normalize the reservation utilit, i.e., the utilit resulting from not entering the market and producing a quantit of zero, to U(0) = 0. Expected utilit of the firm, taken entr as given, reads [4] EU q U n d = ( 1 ) ( Π ) + qu( Π ). For a given market structure, the firm maximizes [4] with respect to output and evasion, taking into account [1] [3]. Supposing an interior solution, which implies q < 1/(1 + δ), and denoting optimal values b an asterisk, the first order conditions can be written as [5] ( 1 qu ) '[( 1 τ)[ x* P ( ) Cx (*, ) k] + τs*] δqu '[( 1 τ)[ x* P( ) Cx (*, ) k] δτs*] = 0, [6] P ( ) + x* P'( ) C( x*, ) = 0. Equation [5] states that expected marginal benefits from tax evasion equal expected marginal costs in terms of the utilit loss caused b the penalt. According to [6], marginal revenues just offset marginal x production costs. This condition for the firm s optimal output is the same as the respective requirement in the absence of tax evasion. More specificall, output does not depend on the evasion variable s and the tax enforcement parameters q and δ. For a given market structure, therefore, the separabilit propert derived in earlier contributions holds. However, this propert does not necessaril impl that the firms output is not affected b tax evasion. As we will now show, endogenizing the market structure creates a link between output and the extent of tax evasion. From equations [5] and [6], the optimal amount of tax evasion s* can be written as a function of the cost parameter, aggregate output and the detection probabilit q, i.e., s* = S(,, q), while, due to the separabilit propert, the optimal output level x* depends solel on the cost parameter and aggregate output, i.e., x* = X(, ). 3 Totall differentiating [6] ields [7] X Cx = P' C X xx < 0, P' + xp'' = < 0. P' C xx According to [7], low costs firms have a higher output than high costs firms and an increase in aggregate production reduces the firm s output. The latter propert is due to our assumption of strategic substitutes. Inserting x* = X(, ) and s* = S(,, q) into [4] defines the maximum expected utilit EU*(,, q) of the firm, provided it enters the market. Differentiating EU*(,, q) with respect to the cost parameter, for a given aggregate output level, and taking into account [5] [7] ields 2 See Lee (1998), Marelli (1984), Marelli and Martina (1988), and Yaniv (1995, 1996), inter alia. The subsequent results can also be obtained if a strictl convex cost of evasion function is assumed and firms maximize profits less tax paments and evasion costs. 3 These are implicit representations of the firm s optimal choices since = x* + x still contains x*. We, therefore, appl the fitting in function approach first used b Selten (1973). Note that both X and S actuall depend on other parameters of the model. But these parameters are irrelevant and, thus, are suppressed throughout. 853

4 NATIONAL TAX JOURNAL [8] EU n = ( 1 τ)[( 1 q) U '( Π ) d + qu '( Π )][ x* P' X + C ] < 0. Hence, the maximum expected utilit is a decreasing function of so that, for an given aggregate output, high costs firms have lower expected utilit than low costs firms. This propert allows us to formalize the market entr decision in the first stage and, thus, to determine the overall equilibrium of the oligopol game. A firm of tpe will enter if the maximum expected utilit resulting from entr, EU*(,, q), is not less than the reservation utilit U(0) = 0 of abstaining from such an action. Let * be the cost parameter of the marginal firm, i.e., of the firm that obtains an expected utilit equal to the reservation utilit and that, accordingl, is just indifferent between entering and not entering. Suppose, further, that all firms with a cost parameter [,*] enter the market, while the remaining firms do not. Expected utilit of the marginal firm and aggregate output can then be written as [9] EU* ( *, *, q) = 0, * [10] * = X(, *) df( ). Equations [9] and [10] determine the equilibrium of our oligopol game. To see this, note that, according to [8] and [9], all firms [,*] have a non negative paoff from entering the market, while the expected paoff of all firms [*, ] is negative. Hence, none of the firms in the market has an incentive to exit and none of the firms outside the market could increase its paoff b entering. The entr stage attains an equilibrium. This equilibrium is unique. To demonstrate the claim, suppose we have an equilibrium with aggregate output o. Due to [8], we know that EU*( 1, o, q) > 0 implies EU*(, o, q) > 0 for all [, 1 ]. This means that if, in the equilibrium, firm 1 is in the market, all firms with a lower cost parameter have to be in the market as well. Similar, EU*( 2, o, q) < 0 implies EU*(, o, q) < 0 for all [ 2, ], i.e., if firm 2 is not in the market, then no firm with a higher cost parameter will be in the market. From these two properties it follows that, in ever equilibrium, there exists a o [, ], which divides the whole set of firms into a subset of firms that are in the market, i.e., all [, o ], and a subset of firms not in the market, i.e., all ] o, ]. To complete the proof of uniqueness, we finall show that firm o is the one with zero expected utilit. This is proven b contradiction. Suppose EU*( o, o, q) > 0. Then there is a small ε > 0 such that EU*( o + ε, o, q) 0, i.e., firm o + ε is also in the market. But this contradicts the propert that, in equilibrium, all firms [ o, ] cannot be in the market. An analogous argument excludes EU*( o, o, q) < 0. Hence, we unambiguousl have o = * and o = * as determined b equations [9] and [10]. Having established a unique equilibrium of the oligopol game, a comparative static analsis clarifies the impact of tax evasion on aggregate output * and output per firm x* = X(, *). Note that [9] and [10] define the marginal firm * and aggregate output * as functions of the detection probabilit q, and that our model includes a world without evasion as a special case for q 1/(1 + δ). Thus, a decrease in q, starting from a value of q = 1/(1 + δ), simulates the transition from a world without tax evasion to a world in which evasion becomes profitable. Totall differentiating [9] and [10] ields [11] [12] d* 1 = 1 dq Δ * X(, *) df( ) d* 1 = X( *, *) f( *) EU q < 0, dq Δ EU q < 0, 854

5 Profi t Tax Evasion under Oligopol with Endogenous Market Structure since Π d < Π n, 2P' + xp'' C xx < 0 and [8] implies 1 X = (2P' + xp'' C xx )/ (P' C xx ) > 0, and * 1 X (, *) df( ) = 1 F( *) * + [ 1 X (, *)] df( ) > 0, EU d n = U( Π ) U( Π ) < 0, q EU = ( 1 τ) x* P'( 1 X ) Δ= X( *, *) f( *) EU n d [( 1 qu ) '( Π ) + qu'( Π )] < 0, * (, *) ( ) 1 >. X df EU 0 With respect to the output of a non marginal firm [ o, *[, we use [7] and [12] and obtain dx* [13] X d * = >0. dq dq Equations [11] [13] contain our main results. The show that profit tax evasion will have an impact on output if the market structure is endogenous. According to equation [12], aggregate output is increased b tax evasion, while [13] implies that the output of each firm operating in the market will be reduced. The intuition is as follows: for a given market structure, tax evasion increases expected utilit of all firms. As a consequence, the marginal firm in the absence of tax evasion will have a positive expected utilit level if evasion becomes feasible. Moreover, there are firms whose expected utilit from entr into the market has been negative without tax evasion but becomes at least zero in the presence of evasion activities. These firms enter the market. Formall, * rises and the (new) marginal firm is characterized b a higher cost parameter according to [11]. Hence, tax evasion attracts new firms. The output of these new firms reduces output of the incumbents due to our assumption of strategic substitutes. But the additional production of the new firms more than compensates the reduction in the incumbents output so that aggregate output is enhanced b tax evasion. These effects of tax evasion on the firms output are indirect. A change in the probabilit of being detected evading taxes does not affect the first order condition [6] of the firm with respect to output. Instead, tax evasion changes the market structure and so induces the firms to adjust their output levels. This is the reason wh, for a given market structure, we obtain the same separabilit result as the analses referred to in the Introduction. But none of the previous contributions considers the case of an endogenous market structure. In contrast, we show that with an endogenous market structure, tax evasion influences output b changes in the number of firms and, in this sense, evasion and output decisions are no longer separable. The non separabilit result has important implications for the neutralit of profit taxes. One of the main motivations of previous studies to investigate the relation between tax evasion and activit choices was the question whether a profit tax can be used to induce a monopolist to increase its inefficientl low output. Those authors who obtain the separabilit propert conclude that the conventional neutralit of a profit tax is true even in the presence of tax evasion, while authors disproving separabilit argue that under tax evasion a profit tax is no longer neutral. In our framework, we can raise a similar question. It is well known from the analsis b Mankiw and Whinston (1986) that unrestricted entr in a Cournot oligopol with positive entr costs induces an inefficientl large number of firms and aggregate output, relativel to a second best world in which the number of firms, but not the output per firm, can 855

6 NATIONAL TAX JOURNAL be regulated. 4 The reason is that the marginal entrant ignores that the incumbents output and profit levels are reduced b her entr. Thus, we ma ask whether a profit tax is suitable to internalize this externalit and which role tax evasion plas. To answer this question, initiall consider the case without tax evasion (q 1/(1 + δ)). The profit tax is then neutral because the market entr condition [9] reduces to x*p( ) C( ) k = 0. Hence, the profit tax alters neither the firms output decisions, as characterized b [6], nor the marginal firm. In the presence of tax evasion, b contrast, the non separabilit result derived above invalidates the neutralit of the profit tax because the firms incentives to enter the market are changed. The number of firms and aggregate output are increased further since evasion raises profits and leads to additional entr. Hence, an evadable profit tax is likel to aggravate the inefficienc derived b Mankiw and Whinston (1986) instead of removing it. 5 CONCLUDING REMARKS Extending previous studies to the case of an endogenous market structure, this paper proves that profit tax evasion ma influence the activit levels of firms. In our setting, tax evasion increases the expected paoff from production so that more firms find it profitable to enter the market. Aggregate suppl in the whole market rises since the increase in the number of firms more than compensates the reduction in output per firm. Finall, we would like to emphasize that the main rationale of our argument is also true in a more general setting comprising other tpes of imperfect competition and taxes. To see this, we can use a similar approach as Yaniv (1995). Analogous to his equation [1], define firm s legitimate net profits (i.e., net profits if this firm full complies to the tax law) as G(x 1, x 2,,x,, x n,, k, τ), where x i is the activit level of firm i (for heuristic reasons we now assume a discrete firm space), τ ma be an kind of tax (not necessaril a profit tax) and all other variables have the same meaning as in our Cournot model. If the term (1 τ)π in [2] and [3] is now replaced b G, firm s optimal activit level is determined b G(x 1, x 2,, x,, x n,, k, τ)/ x = 0. This is the same condition as in the absence of tax evasion and, thus, for a given market structure, we would obtain the same separabilit result as Yaniv (1995). But with an endogenous market structure, tax evasion influences market entr and the dimension of the vector (x 1, x 2,, x,, x n ). This, in turn, changes the individual firm s activit level and invalidates the separabilit result. It will be an interesting task for future research to investigate the economic consequences of this non separabilit for other taxes, such as value added or emission taxes, and other tpes of competition such as Bertrand competition or product differentiation. Acknowledgments We thank an anonmous referee for ver helpful and constructive comments. All errors remain ours. 4 This inefficient entr result also holds in a framework with heterogeneous firms, as is the case in our model. A proof of this assertion is available from the authors upon request. 5 In the case of strategic complements, i.e., P' + xp'' > 0, tax evasion exerts almost the same effects on the market equilibrium. The onl exception is that the individual output of incumbent firms increases. Interestingl, however, evasion then tends to render the market structure more efficient as, without tax evasion, the number of firms is inefficientl small. Proofs of these assertions can be obtained from the authors upon request. 856

7 Profi t Tax Evasion under Oligopol with Endogenous Market Structure REFERENCES Bulow, Jerem I., John D. Geanakoplos, and Paul D. Klemperer. Multimarket Oligopol: Strategic Substitutes and Complements. Journal of Political Econom 93 No. 3 (June, 1985): Dixit, Avinash K. Comparative Statics for Oligopol. International Economic Review 27 No. 1 (Februar, 1986): Kreutzer, David, and Dwight R. Lee. On Taxation and Understated Monopol Profits. National Tax Jour nal 39 No. 2 (June, 1986): Kreutzer, David, and Dwight R. Lee. Tax Evasion and Monopol Output Decisions: A Repl. National Tax Jour nal 41 No. 4 (December, 1988): Lee, Kangoh. Tax Evasion, Monopol, and Nonneutral Profit Taxes. National Tax Jour nal 51 No. 2 (June, 1998): Mankiw, N. Gregor, and Michael D. Whinston. Free Entr and Social Inefficienc. Rand Journal of Economics 17 No. 1 (Spring, 1986): Marelli, Massimo. On Indirect Tax Evasion. Journal of Public Economics 25 No. 1 2 (November, 1984): Marelli, Massimo, and Ricardo Martina. Tax Evasion and Strategic Behaviour of the Firms. Journal of Public Economics 37 No. 1 (October, 1988): Panteghini, Paolo M. Tax Evasion and Entrepreneurial Flexibilit. Public Finance Re view 28 No. 3 (Ma, 2000): Selten, Reinhard. A Simple Model of Imperfect Competition, Where 4 Are Few and 6 Are Man. International Journal of Game Theor 2 No. 3 (1973): Virmani, Arvind. Indirect Tax Evasion and Production Efficienc. Journal of Public Economics 39 No. 2 (Jul, 1989): Wang, Leonard F.S., and John L. Conant. Corporate Tax Evasion and Output Decisions of the Uncertain Monopolist. National Tax Journal 41 No. 4 (December, 1988): Yaniv, Gideon. A Note on the Tax Evading Firm. National Tax Journal 48 No. 1 (March, 1995): Yaniv, Gideon. Tax Evasion and Monopol Output Decisions: Note. Public Finance Quarterl 24 (October, 1996):

8