Marginal and average tax rates and the incentive for self-employment.

• Author: Robson, Martin T.

1. Introduction

   How do income taxes influence the incentives for individuals to engage in entrepreneurial activity? For the United Kingdom, the government view has been that high rates of income tax are inimical to enterprise, prompting a series of reductions in tax rates since the late 1970s. The U.K. government position is that "a more beneficial tax regime has been an important factor in stimulating enterprise" (Employment Department 1989, p. 16), and this approach has been mirrored in many other countries.

   In fact, the economic literature provides only weak support for this rationale for tax cuts. Models involving linear income tax systems, such as those developed by Kanbur (1981) and Kihlstrom and Laffont (1983), find that changes in the income tax rate have an ambiguous effect on the number of entrepreneurs, whereas empirical evidence, such as that of Blau (1987) and Evans and Leighton (1989) in the United States and Parker (1996) and Robson (1998) in the United Kingdom, typically find a positive relationship between income tax and the self-employment rate as a measure of "entrepreneurship."(1) If anything, the evidence is that lower tax rates lead to a reduction in self-employment.

   The explanation that is usually put forward to account for these findings is that higher tax rates lead individuals to take up self-employment to take advantage of the greater opportunities for tax avoidance and evasion that self-employment is presumed to afford relative to paid employment (i.e., wage earning). However, in the specification of the relationship between self-employment and income tax, no distinction is typically made between the marginal and the average tax rate. For example, Blau (1987) uses measures of the marginal tax rate in his study of the determinants of U.S. self-employment, and Evans and Leighton (1989) use the average tax rate. Although the prevailing view seems to be that such a distinction is largely irrelevant, there is a growing literature that suggests that changes in the marginal and average tax rates have differing and possibly opposing effects on economic activity, so that the difference in specification is in fact likely to be important.(2)

   In this paper we explore the relationship between self-employment and changes in the marginal and average income tax rates. We pick up on the suggestion of Robson (1998) that these tax rates will have different effects on the incentive to be self-employed. A high marginal tax rate may discourage entrepreneurial activity by reducing the returns from effort and thereby provide a disincentive for self-employment. In contrast, a high average tax rate may encourage evasion and thereby provide an incentive for individuals to take up self-employment because the opportunities for evasion are usually considered much greater in this mode.

   We develop a model in which individuals choose the level of effort to supply and the proportion of tax to evade. To distinguish between the employment modes, we suppose that at the optimal position the marginal returns from effort are greater in self-employment than in paid employment and that the probability that tax evasion is detected is lower. We analyze the effect of pure changes in the marginal and average tax rates on the individual's decision at this position. The theoretical work is supported by an empirical investigation of the relationship between self-employment and income tax using pooled cross-section data for 15 OECD countries. The conclusion is that marginal and average tax rates work in opposite directions on the choice between self-employment and paid employment. Lower marginal rates act mainly on effort and increase the incentive for self-employment, but lower average tax rates reduce the gains from evasion and are mainly an incentive for paid employment.

   In the next section we set out our basic theoretical framework and derive the results on the effects of the tax system. Section 3 carries out the empirical investigation for OECD countries. Section 4 draws conclusions.

2. Theoretical Framework

   We develop a model in which an individual is considering entering employment but in which, in the initial position, the benefits and costs of entering either self-employment or paid employment are exactly balanced, so that the individual is indifferent between these. We then consider the implications of changes in the marginal and average income tax rates in making one of these employment modes more attractive and thus on the individual's choice. The results can be straightforwardly generalized to individuals who are already in one of these employment modes.(3) We begin by setting out our two main assumptions, by which we distinguish self-employment from paid employment.

   Characteristics of Self-Employment

   Underlying our modeling are two assumptions about the characteristics of self-employment compared with paid employment or wage earning. We believe that these are reasonable and potentially fruitful generalizations of the literature as follows:

   ASSUMPTION A1. Pretax remuneration in self-employment is much more closely related to the level of individual effort than is the case in paid employment.

   We believe that this is reasonable because, unlike paid employment, the self-employed individual is the owner, manager, and employee of the business and so (subject to running costs) can retain all the earnings from an increase in effort. In this case, remuneration is directly related to effort. We use "effort" to capture the quality and quantity dimension of labor supply and note that remuneration is poorly related to effort in paid employment because the terms of the employment contract are not only fixed over the short term but may also be imperfectly adjusted and tend to reflect the completion of physical tasks (see Brown et al. 1986). By contrast, these relationships are stronger in self-employment, which for this reason is often taken as synonymous with "entrepreneurship." We model Assumption A1 by supposing that gross pretax income is more responsive to effort in self-employment than in paid employment in the region of the individual's optimal position.

   ASSUMPTION A2. The opportunities for tax evasion are much greater in self-employment than in paid employment.

   We believe that this is reasonable because whereas the tax on income in paid employment is collected at source by the employer (through the pay-as-you-earn system in the United Kingdom), an individual in self-employment self-assesses income for tax purposes and does this only retrospectively. Further, the self-employed negotiate contracts directly with suppliers and customers, and there exists much greater scope for informal arrangements that cannot be separately verified from either within or outside the firm. An assumption such as A2 often characterizes the modeling of the "underground" economy in studies of evasion (e.g., Jung, Snow, and Trandel 1994), and we model it by supposing that the probability that evasion is detected by the tax authorities is relatively smaller in self-employment, so that the greater opportunities for tax evasion are reflected in a smaller likelihood of being prosecuted.

   Both assumptions point to the advantage of entering self-employment, but there are various startup costs (e.g., initial capital, establishing a market, sourcing of inputs, and risk) that give rise to a "barrier to entry" into self-employment. We do not model this entry barrier explicitly, but the startup costs mean that we can potentially observe individuals entering either employment mode.

   The Model

   We set up our model and notation as follows. We suppose that the individual's pretax income [Y.sub.i] in self-employment (i = s) or paid employment (i = p) depends on effort [e.sub.i] ([e.sub.i] [greater than] 0), according to a well-defined income function [Y.sub.1] = Y([b.sub.i][e.sub.i]), where Y[prime] [greater than] 0. The [b.sub.i] ([greater than] 0) is a shift parameter that by Assumption A1 varies between employment modes, so that [b.sub.s] [greater than] [b.sub.p]. Although income may be higher in paid employment than in self-employment at low levels of effort, we suppose that our income function holds in the region of the optimal effort level and thus that income is more responsive to effort in self-employment.(4)

   The tax liability [T.sub.i] of the individual (whether declared or undeclared) is given by an exogenously determined tax schedule that is laid down by the government. Following Malcomson and Sartor (1987), we write this tax schedule in general form as [T.sub.i] = T([Y.sub.i], z), where z is a vector of parameters defining the tax system (e.g., allowances and tax bands). It is assumed that the schedule is the same between employment modes. Because the broad direction of our results are unaffected by this assumption, we think that it is reasonable for modeling purposes, although in practice the self-employed may have greater scope for deductions of business expenses.(5)

   We denote [m.sub.i] as the marginal tax rate (i.e., d[T.sub.i] = [m.sub.i]d[Y.sub.i]) and [a.sub.i] as the average tax rate (i.e., [T.sub.i] = [a.sub.i][Y.sub.i]). As an empirical regularity, we suppose that 0 [less than] [a.sub.i] [less than] [m.sub.i] [less than] 1, which includes not only progressive tax schedules but also those that are linear with an allowance on initial income, in which case it is assumed that taxable income always exists. The schedule allows us to consider pure changes in taxation as follows. A pure increase in the marginal tax rate is a change in z such that [Delta][m.sub.i] [greater than] 0 and [Delta][a.sub.i] = 0, and a pure increase in the average rate is a change in z such that [Delta][a.sub.i] [greater than] 0 and [Delta][m.sub.i] = 0. We can give intuition to these pure tax changes by considering the linear income tax schedule with an allowance. Here, a pure increase in the average tax rate results from a reduction in the allowance, whereas a pure...