Corporate taxpayers have clear incentives to legally minimize their tax liabilities to the extent that doing so maximizes after-tax returns. Conversely, tax authorities have a clear mandate to ensure that taxpayers meet their tax obligations and enforce tax rules in place to achieve that mandate. These conflicting incentives create an inherent tension between taxpayers and tax authorities wherein both parties use various tactics to achieve their aims. One particular tactic used by tax authorities to increase revenue is to offer a temporary grace period during which taxpayers can remit unpaid or overdue taxes for a reduced penalty or no penalty at all, which we generically label “tax forgiveness”. Tax forgiveness programs are often billed as a one-time chance for a fresh start—the last opportunity for the taxpayer to clear the slate. However, many jurisdictions have offered tax forgiveness multiple times, undercutting their claim that it is truly a one-time chance.
The objective of this study is to learn how firms respond to tax forgiveness programs offered by state tax authorities, especially when those programs are repeated. Specifically, we examine changes in corporate tax avoidance by firms that are headquartered in states that grant the amnesty programs. Because the purpose of many state tax amnesty programs is to enroll and collect overdue taxes from new, unknown, noncompliant taxpayers, prior research has focused on the effect of these “nexus” amnesties on new enrollees or aggregate tax. In contrast, by focusing on firms headquartered in the amnesty-granting state, we examine the effect of tax forgiveness on existing, known taxpayers.
We hypothesize that offering to forgive noncompliant taxpayers can change existing taxpayers’ future tax avoidance for two related reasons. First, such forgiveness programs could act as a signal of a weak enforcement program and reduce expectations of the tax authority detecting aggressive tax positions, because the very need for such a program may reveal a shortfall in the existing tax detection processes. If the expected probability of detection decreases, the rational taxpayer will likely become less compliant. Second, the first instance of tax forgiveness plausibly increases the taxpayer’s expectation of another future tax forgiveness program. That is, an expectation of a future amnesty reduces the expected interest and penalty on any admitted underpayment, again reducing the expected cost of aggressive tax planning. Based on these arguments, we predict that tax forgiveness programs, especially repeated programs, will increase corporate tax avoidance.
To test this prediction empirically, we examine tax amnesty programs offered by U.S. states over the past four decades. Although each state amnesty program has unique attributes, there are three common features associated with state-level corporate income tax amnesties that make them suitable for testing our research question. First, the amnesties occur in different states at different times, which aids in the empirical identification of a tax response by corporations. Second, amnesties are often coupled with various threats of increased enforcement and increased penalties post-amnesty, which could decrease future tax avoidance, providing tension to our main predictions. Third, many states have offered multiple instances of tax amnesty, which allows us to assess the effects of repeated tax forgiveness. During our sample period, 39 states offered a corporate income tax amnesty program at least once; of these 39 states, 30 states have repeated a similar program at least once and several states have offered up to five different amnesties.
We use a difference-in-differences research design, in which amnesty firms are matched to non-amnesty firms in a control sample to mitigate the influence of unobserved, contemporaneous effects. Thus, we test whether firms headquartered in amnesty-granting states exhibit greater increases in tax avoidance following the initial and subsequent amnesties than firms headquartered in non-amnesty-granting states during the same time period. The empirical model also includes firm-level controls, such as the firm’s federal effective tax rate, sales revenues, R&D expense, leverage, and capital intensity to account for firm-level drivers of tax avoidance. Finally, the empirical model includes time-varying state statutory tax rates to account for subsequent changes in the state tax regime, a measure of the headquarter state’s business friendliness index, and headquarter-state fixed effects to account for state heterogeneity (e.g., red state versus blue state).
Overall, the evidence is consistent with firms headquartered in amnesty states increasing their tax avoidance following an amnesty. The average drop in state effective tax rates following an initial amnesty is in the range of 0.44 to 0.99 percentage points, which is an 8 to 18 percent decrease from the average state effective tax rates for amnesty firms. These results are robust to the inclusion of important additional control variables and fixed effects, as well as to several alternate research design choices (such as alternative control groups and matching criteria).
In addition, we find evidence consistent with higher levels of tax amnesty repetition (i.e., when a state offers its second, third, fourth or fifth amnesty) being positively associated with levels of state tax avoidance. Whereas the first amnesty is associated with an average drop in state effective tax rates of 0.64 percentage points in this test, the second amnesty is associated with a drop of 0.69 percentage points and the third (and subsequent) amnesties are associated with a drop of 1.25 percentage points. Overall, our evidence leads us to conclude that tax forgiveness events are associated with increased tax avoidance, and that tax avoidance is increasing in the number of forgiveness repetitions.
In terms of economic magnitude, the coefficient estimates suggest that the first amnesty is associated with a decline in $23.0 billion dollars (aggregating across all states) in state tax expense reported by corporations; the approximate decline for the second amnesty is $7.4 billion dollars and the approximate decline for the third and additional amnesties is $8.0 billion dollars. In sum, the combined effect of granting amnesty has been a $38.4 billion dollar reduction in state tax expense reported from 1980-2012. Overall, these numbers suggest an economically meaningful corporate response to tax forgiveness programs.
This study demonstrates potential firm-level consequences faced by offering tax forgiveness programs, which would be useful for state tax policy makers interested in understanding the consequences of policy choices. Although the economics literature has examined the effects of amnesties on aggregate tax revenues and individual taxpayer compliance (i.e., from the perspective of the government or the individual taxpayer), to our knowledge, this is the first study that examines the effects of tax forgiveness programs on tax avoidance from the perspective of the firm.
To a degree, our results also speak to the potential effects of an additional tax repatriation holiday, a topic of recent debate in the U.S. Although tax amnesties and tax holidays differ along several important dimensions, at a simplified level, they are similar in that both offer a one-time reduction to a firms’ expected tax burden. To the extent that firms anticipate that another forgiveness event may occur, they could become more tax aggressive and recognize more income overseas to reduce U.S. taxes. Indeed, articles in the business press report that U.S. firms are heavily lobbying for a tax holiday and make the conjecture that firms are holding large amounts of cash awaiting a repatriation tax holiday. Our evidence, albeit in a very different setting, is consistent with such a potential response by U.S. firms to another tax holiday.
The preceding post comes to us from Terry J. Shevlin, Professor of Accounting at the University of California-Irvine, Jacob R. Thornock, Associate Professor of Accounting and Gregory Fellow at the University of Washington – Michael G. Foster School of Business, and Braden Williams, Assistant Professor as the University of Texas at Austin – Department of Accounting. The post is based on their article, which is entitled “An Examination of Firms’ Responses to Tax Forgiveness” and available here.