|
Article Excerpt
The Earned Income Tax Credit (EITC) is widely recognized to have increased employment, particularly among single mothers with low levels of education. For purposes of calculating the credit, the tax code treats income earned through self-employment in nearly the same way as income earned through wage sector employment. This paper uses tax return data to investigate the effects of the EITC on the reporting of self-employment income to the IRS. The reporting of self-employment income may be particularly responsive to the EITC, as informal self-employment is widespread among the poor, and there is substantial misreporting of self-employment income.
In this paper I document the importance of self-employment income as a source of EITC payments. While only 13.5 percent of EITC recipients report self-employment income, approximately 70 percent of self-employed EITC recipients receive larger credits as a result of their self-employment income. On average, reported self-employment income increases a filer's EITC payment by about $650.
I investigate whether three expansions in the EITC, effective in tax years 1987, 1991, and 1994, are associated with changes in the reporting of self-employment income. I use a difference-in-difference strategy, comparing changes in reported self-employment across those with and without children. This strategy relies on the greater increases in EITC generosity for those with children than for those without children. I also use a more parametric specification that takes advantage of variation in the incentive to report self-employment income provided by state-level EITC programs. I find evidence that the decision to report self-employment income does in fact respond to the incentives created by the EITC. Among the lowest-income filers, the 1994 EITC expansion is associated with a significant increase in the probability of reporting positive self-employment income, equal to 3.2 percentage points for unmarried filers and 4.1 percentage points for married filers. This pattern survives a number of robustness checks. I find somewhat weaker evidence of a response to the high marginal tax rates generated by the phase-out region of the EITC. Unmarried filers with higher levels of income are less likely to report positive self-employment income after EITC expansions.
INTRODUCTION
The Earned Income Tax Credit (EITC) is widely recognized to have increased employment, particularly among single mothers with low levels of education. For purposes of calculating the credit, the tax code treats income earned through self-employment in nearly the same way as income earned through wage sector employment. This paper uses tax return data to investigate the effects of the EITC on the reporting of self--employment income to the IRS.
There are two reasons why the impact of the EITC on self-employment is of interest independent of its impact on wage sector employment. First, while overall self-employment rates have been declining in the long run, sociologists including Edin and Lein (1997) and Venkatesh (2006) have documented widespread informal self-employment among the poor. Edin and Lein interview 379 low-income single mothers, finding that 39 percent of welfare-reliant mothers and 28 percent of wage--earning mothers supplement their income with "off-the-books" employment. This suggests the population most affected by EITC expansions may have relatively good access to informal self-employment opportunities. Second, self-employment is widely recognized to offer substantial opportunities for misreporting. Unlike wage employment, self-employment has no third party withholding tax or reporting information to the IRS. Data from the National Research Program of the IRS (2006) show that 57 percent of self-employment income goes unreported. The discrepancy between reported and actual self-employment income makes it possible for reported self-employment to be particularly responsive to tax incentives. Slemrod (1990) suggests a hierarchy of responses to taxation in which "renaming" behaviors (such as reclassifying some previously unreported income as reported income) are more responsive to a tax change than are "real" changes in behavior (such as committing more time and effort to labor supply).
Figure 1 motivates the analysis of this paper. Using annual cross sections of tax return data, I plot the percentage of low-income returns reporting positive self-employment income, comparing the behavior of EITC recipients and non-recipients. (1) Increases in EITC generosity are marked by vertical lines in the figure. The top panel shows information from returns of unmarried individuals. From 1984-1993, self-employment rates are similar for EITC recipients and non-recipients. After the most recent expansion, the reporting of positive self-employment income falls for non-recipients and rises dramatically for recipients. This is true even when I control for an important change in EITC eligibility. The EITC was first extended to filers without a child in 1994. Looking at taxpayers who claim both the EITC and a dependent exemption for a child living at home, a group that should be fairly stable over time, it is apparent that the reporting of self-employment income increases for EITC recipients relative to non-recipients. The bottom panel shows information from joint returns of married couples. For this group, EITC recipients are always more likely to report self-employment income than are non-recipients. The self-employment gap grows slightly following the most recent expansion.
In this paper I document the importance of self-employment income as a source of EITC payments. While only 13.5 percent of EITC recipients report self-employment income, approximately 70 percent of self-employed EITC recipients receive larger credits as a result of their self--employment income. On average, reported self-employment income increases a filer's EITC payment by about $650. I also investigate whether three expansions in the EITC, effective in tax years 1987, 1991, and 1994, are associated with changes in the reporting of self-employment income. I use a difference-in-difference strategy, comparing changes in reported self-employment across those with and without children. This strategy relies on the greater increases in EITC generosity for recipients with children than for those without children. I also use a more parametric specification that takes advantage of variation in the incentive to report self-employment income provided by state-level EITC programs. I find evidence that the decision to report self-employment income does in fact respond to the incentives created by the EITC. Those with the lowest levels of income are significantly more likely to report self-employment income after the 1994 expansion, consistent with the higher subsidy rate they faced. This pattern is evident for both married and unmarried filers, and survives a number of robustness checks. I find somewhat weaker evidence of a response to the high marginal tax rates generated by the phase-out region of the EITC. Unmarried filers with higher levels of income are less likely to report positive self-employment income after EITC expansions.
[FIGURE 1 OMITTED]
There is a substantial empirical literature investigating the labor supply effects of the EITC among single women, including Dickert, Houser, and Scholz (1995), Eissa and Liebman (1996), Ellwood (2000), and Meyer and Rosenbaum (2001). The consensus from this literature is that the EITC has substantially increased the labor force participation rate of single women, with little effect on hours of work for those already employed. More recently, attention has turned to the effects on labor supply decisions of married individuals. If the primary earner's income is within the EITC phase-out range, the EITC reduces the net wage of the secondary earner and may therefore reduce her labor supply. Eissa and Hoynes (2004) show that the EITC expansions between 1984-1996 increased the participation rate of married men by only 0.2 percentage points, and reduced the participation rate of married women by one percentage point.
Only one previous paper has examined in detail the EITC and self-employment. Joulfaian and Rider (1996) use audit data from 1988 to investigate how the EITC affects tax evasion. They look separately at sole proprietors (the self-employed) and wage earners. They find no evidence that the negative marginal tax rates occurring in the phase-in range of the EITC induce over-reporting of income for either proprietors or wage earners. The positive and large marginal tax rates in the phase-out range do lead to understatement of income among proprietors, but not among wage earners. Even among proprietors, the effect is small. Joulfaian and Rider's simulations suggest that the income understatement of proprietors is about 9 percent higher due to the EITC. Unlike Joulfaian and Rider, I consider whether the EITC affects the decision to report self-employment income rather than treating self-employment status as exogenous. I use more recent data, because EITC expansions have generated benefit amounts much larger than 1988 levels. The use of unaudited tax return data means that I am unable to draw conclusions about compliance.
Although I focus particularly on the EITC, this paper is related to the broader question of how taxes affect self-employment. In the theoretical literature on this question, wage employment and self-employment differ in terms of risk and opportunity for tax evasion. Pestieau and Possen (1991) consider individuals choosing between riskless work in the wage sector and risky entrepreneurship, where tax evasion is possible only for entrepreneurs. In their model an increase in the tax rate has an ambiguous effect on both the fraction of individuals who choose entrepreneurship and the fraction of entrepreneurs who choose to evade. Kesselman (1989) develops a general equilibrium model with a fully compliant "above-ground" sector and a fully noncompliant "belowground" sector. Under certain conditions, including a penalty for evasion that does not change with the tax rate, he shows that an increase in the tax rate can shift workers into the noncompliant sector. Domar and Musgrave (1944) show that a tax system in which losses can be used to reduce taxable income makes a risky investment relatively more attractive, by shifting some risk to the government. Many authors have applied this result to the self-employment decision, arguing that a more progressive tax system can shift workers into the relatively risky self-employment sector. On the other hand, Gentry and Hubbard (2000) show that when loss offsets are imperfect, a more progressive system can discourage self-employment. In short, there is disagreement in the theoretical literature about how a change in the tax rate will affect participation in self-employment.
The empirical literature investigating the relationship between tax rates and self-employment activity, summarized by Schuetze and Bruce (2004), has generated a wide range of estimates. The earliest work in this area, such as the time series evidence of Long (1982) and Blau (1987) and cross-sectional evidence of Moore (1983), indicated a positive relationship between federal marginal tax rates and the level of self-employment participation. Subsequent cross-sectional studies that have addressed the endogeneity of an individual's tax rate have weakened this early consensus. Bruce (2000, 2002) relies on exogenous changes in the payroll tax treatment of self-employment income relative to wage income. He finds that higher relative tax rates on self-employment income are associated with a higher probability of entry into self-employment and a lower probability of exit from self-employment. Using the same source of tax variation, but tax return data rather than survey data, Gurley-Calvez and Bruce (2008) find the opposite pattern; higher relative tax rates on self-employment income are associated with shorter duration of self-employment spells. Moore (2004) relies on the 1986 and 1993 tax reforms. Neither reform is shown to have a consistently significant effect on participation in self-employment. To the best of my knowledge, my paper is the first to take advantage of EITC-related exogenous tax variation to estimate the relationship between marginal tax rates and self-employment activity. (2)
BACKGROUND ON THE EITC
The administrative details of the EITC are described by the U.S. House Ways and Means Committee (2004), and Hotz and Scholz (2003) provide a thorough summary of research on who takes up the EITC, how their behaviors respond to its incentives, and the degree to which they are complying with the tax law. Here I describe the elements of the EITC most relevant to my analysis of reported self-employment income.
The EITC is a refundable credit, with its amount determined by a taxpayer's earned income. Earned income includes wage and salary payments and net income from self-employment. (3) If earned income falls within the phase-in region, the credit amount is equal to the subsidy rate multiplied by earnings. Over the years I analyze, the subsidy rate has ranged from a low of 7.65 percent (for childless claimants) to a high of 40 percent (for claimants with two or more children). The maximum credit is equal to the subsidy rate multiplied by the income cutoff that separates the phase-in from the plateau region. For fliers with children, this cutoff has ranged in nominal dollars from $5,000 in 1984 to $9,390 in 1998. As earnings increase across the plateau region, the credit is constant at the maximum amount. Once earnings rise above the plateau cutoff, the credit is reduced at the phase-out rate. The beginning of the phase--out range has varied from $6,000 in 1984 to $12,260 in 1998, and the end of the phase-out range from $10,000 in 1984 to $30,095 in 1998.
Although the EITC is generally calculated as a function...
|
