|
Article Excerpt
Abstract--Existing experimental literature on tax compliance and crowding-out examines either the incentive to evade tax or the incentive to give, but not both. This paper provides an experimental examination of the behavior of tax evasion and voluntary contributions when both publicly and privately provided public goods are present. The experimental evidence suggests that the privately provided public good is a substitute for the publicly provided public good, but the converse does not hold, and that the level of compliance may be underestimated, ceteris paribus, if private contributions are not taken into account.
INTRODUCTION
The issue of tax compliance has attracted extensive attention since Allingham and Sandmo's (1972) pioneering theoretical paper. Due to the advantage of quarantining unwanted disturbances outside the laboratories, in recent years more and more researchers have relied on experimental methods to explore the behavior of tax evasion. Among them, Alm and his coauthors have conducted a series of inspiring tax compliance experiments (Alm, McKee, and Beck, 1990; Alto, Jackson, and McKee, 1992a, 1992b, 1993; Aim, McClelland, and Schulze, 1992; Alm, Sanchez, and de Juan, 1995; and Alm and McKee, 1998). Their experiments are of particular interest since they consider the tax-funded public good that is incorporated in the theoretical model posed by Cowell and Gordon (1988), while earlier experimental studies, for instance, Spicer and Becker (1980), Spicer and Thomas (1982), Spicer and Hero (1985), Robben, Webley, Elffers, and Hessing (1990), and Collins and Plumlee (1991), were generally developed from the theoretical work by Allingham and Sandmo (1972) and a later extension by Yitzhaki (1974), in which the expenditure in relation to tax revenue is neglected.
Alm and his coauthors apply the voluntary contribution mechanism, which is frequently used in experiments related to the voluntary provision of public goods, to tax compliance experiments. In this mechanism, subjects voluntarily contribute some or all of their endowments to a privately provided public good. The extent to which they benefit from the public good depends on the total contributions made by the group. Instead of contributing to the public good directly, in Alm and his coauthors' tax compliance experiments, subjects decide on the amount of income reported to the tax authority. Then the declared income is taxed and the taxes are used to fund the public good. The level of the public good is determined based on the total amount of taxes paid by the group. The rest of their story is similar to the theoretical settings and other experimental designs in the literature: subjects' true incomes are audited by a certain or an uncertain probability. The subject who is audited and caught underreporting will pay the tax evaded and a fine, which is some multiple of the evaded tax. A major finding of their studies is that the level of compliance is higher when the taxes collected are used to fund the public good than when the taxes are discarded.
Although neglecting the use side of tax revenue is unrealistic in exploring the behavior of tax evasion, the tax-funded public good is not the only category of public good in reality. As observed in the real world, many public goods are provided by the private sector alone and receive no support from the government. Churches, the Red Cross, private shelters, private dog pounds, and private lighthouses are examples. Since public goods are not restricted to being provided by the public sector, the employment of the voluntary contribution mechanism in Alm et al.'s experiments actually gives rise to some interesting questions. Besides paying taxes to fund the publicly provided public good, if individuals also contribute voluntarily to the public good provided by the private sector, how will they allocate their income between the two public goods? Furthermore, will compliance be lower if the privately provided public good is valued relatively more highly than the publicly provided public good? Will taxes crowd out private contributions and will the crowding-out become more severe if the publicly provided public good is valued relatively more highly than the privately provided public good?
The effect of taxes on private contributions has been extensively discussed in the crowding-out literature. The theoretical papers by Warr (1982, 1983), Roberts (1984, 1987), and Bergstrom, Blume, and Varian (1986) show that government contributions to charity completely crowd out private contributions. The reason for this result is that they assume that individuals are purely altruistic, caring only about the total contributions to the third party, but not the relative magnitudes of the private and government contributions that the total contributions are composed of. This setting implies that private and government contributions are perfect substitutes and, therefore, individuals will simply substitute government contributions for private contributions dollar for dollar as their taxes increase. As a result, taxes are neutral, having no effect on total contributions.
This neutrality or complete crowding-out result has been challenged by some theoretical examinations based on impure altruism. For instance, besides caring about the total contributions Andreoni (1989, 1990) assumes that individuals also obtain some private benefit from the act of giving, which he refers to as "warm-glow." The "warm-glow" component makes government contributions imperfect substitutes for private contributions and, therefore, individuals will reduce their own contributions less than dollar for dollar as government contributions to charity increase. This partial crowding-out result has been supported by many empirical studies (Abrams and Schmitz, 1978; Reece, 1979; Schiff, 1985; Kingma, 1989; Khanna, Posnett, and Sandier, 1995; Payne, 1998, 2001; Khanna and Sandler, 2000; Okten and Weisbrod, 2000; and Hungerman, 2005) and some experimental examinations (Andreoni, 1993; Chan, Godby, Mestelman, and Muller, 2002), in which the evidence often exhibits slight or modest levels of crowding-out, and sometimes even a crowding-in.
Despite these findings, a common feature of the above-cited studies is that they examine the crowding-out hypothesis in a climate without tax evasion. In particular, in the experimental studies cited above, the public goods are funded by lump-sum taxes. Furthermore, since in the theoretical settings of Warr (1982, 1983), Roberts (1984,1987), and Bergstrom et al. (1986) the public good consumption comes from the sum of government and private contributions, subsequent experimental studies generally follow this setup, implicitly assuming that the two sources are used to fund the "same" public good. This implies that private and government contributions yield the same marginal benefits to taxpayers, excluding any possibility that taxpayers may benefit more from either one than the other, and the subsequent influences that taxes may have on private contributions.
In sum, although Alm and his coauthors apply the voluntary contribution mechanism to tax compliance experiments, in their experiments the public good is funded by taxes alone, not private contributions. Therefore, in their experiments the interaction between taxes and private contributions cannot be investigated. In the experimental literature on crowding-out, the public good is funded by both taxes and private contributions, but the taxes are lump-sum taxes. Therefore, only the effect of taxes on private contributions can be examined, while how the incentive to pay taxes is affected by private contributions remains unanswered. Since in the real world some people do have the incentive to evade tax, and besides paying taxes to fund the publicly provided public goods, people often make contributions to the privately provided public goods, the literature on tax compliance, crowding-out, and the voluntary provision of public goods is complementary.
This paper attempts to connect the three avenues of the experimental literature by incorporating voluntary contributions into a tax compliance framework. Unlike the crowding-out literature in which taxes and private contributions are used to fund the same public good, this paper assumes that taxes are used to fund a publicly provided public good, and private contributions are used to provide a privately provided public good. The reason for this setting is that people may give more and evade more tax if the privately provided public good is relatively more valuable as compared with the publicly provided public good. On the contrary, people may evade less tax and give less if the publicly provided public good is operated more efficiently than the privately provided public good. Assuming that there is only one public good fails to explore the interactions between tax compliance and private giving when the publicly or the privately provided public good is valued relatively more highly than the other.
Three baseline treatments are employed here. The first baseline treatment includes only the public good funded by taxes and the second one includes only the public good established through voluntary contributions. The third treatment involves both public goods. The three baseline treatments have the same marginal per capita returns (MPCRs) of the publicly and privately provided public goods, and the same tax rate, audit probability, and fine rate whenever these fiscal variables appear. The MPCRs of the publicly and privately provided public goods, the tax rate, and the audit probability...
|
