When the Biden Administration announced its plans to increase both capital gains tax rates and ordinary income tax rates on high-income earners, opponents predictably responded that doing so would reduce economic growth. This objection is supported, in part, by the familiar theory that taxpayers respond to increased taxes on labor by simply working less. Conversely, tax cuts are expected to increase labor effort since the after-tax returns are greater. But the reality is more complicated. As Professors Michael Simkovic and Eric J. Allen explain, “[t]he effect of an income tax cut on work hours can be difficult to predict because tax cuts often produce two opposite effects: a substitution effect, which encourages work, and an income or wealth effect which discourages work.”
As a result of this theoretical uncertainty, countless studies have attempted to test how tax rate changes relate to work effort. In their new article, Simkovic and Allen contribute to this literature by introducing a novel method to test the impact of federal tax rate changes, using the 2003 Jobs and Growth Tax Relief and Reconciliation Act (JGTRRA) as a case study. Unlike Biden’s plan, the JGTRRA lowered the tax rates applicable to both capital gains and ordinary income. Based on their analysis of labor data before and after the JGTRRA, the authors argue that “tax cuts can reduce work hours, especially among high-income and wealthy groups, when tax rates are initially low and when tax cuts are tilted toward investment income.”
The authors base their conclusions on a carefully conducted statistical analysis that exploits differences in the interactions between state and federal tax law. Specifically, a handful of state tax laws—in Alabama, Iowa, Louisiana, and Utah—permit residents to claim an uncapped deduction for federal taxes paid. As a result of these deductions, residents in these states face automatic and offsetting state tax increases when federal tax rates are cut. The net effect of this interaction is to reduce the size of tax cuts received by residents. In other words, residents of these states “received a smaller tax cut in 2003 than residents of ‘no-deduction’ or control states.”
Drawing on American Community Survey data about the self-reported work hours of over 9 million individuals, the authors compared changes in work hours in the pre-JGTRRA and post-JGTRRA period. If the substitution effect of the 2003 tax cuts was stronger than the income effect, then we would expect to see higher increases in work hours in the no-deduction states (which received the highest tax cuts) than in the four uncapped-deduction states.
On the other hand, if the income effect was stronger, then we would expect to see a higher decrease in work hours in the no-deduction states than in the uncapped-deduction states. That is what the authors found. The authors’ analysis showed that people in no-deduction states (where the tax cuts were highest) decreased their work hours more than those in uncapped-deduction states, particularly at the highest income levels: “Overall, the results suggest that a smaller tax cut caused workers to reduce work hours less (or increase hours by more) than a control group that received a larger tax cut.” This suggests that “for the 2003 tax reforms, the income effect was stronger than the substitution effect.”
A question in my mind, as I read this study, was how to account for the lack of salience in the interaction between state and federal law. In other words, if taxpayers in the uncapped-deduction states didn’t realize that their federal tax cuts were offset by automatic changes in the effective—but not marginal—state tax rates, can we really say that their labor choices were related to this interaction? Did taxpayers in Alabama, Iowa, Louisiana, and Utah see the federal tax cuts, consider working less (like residents of the control states), but then reconsider when they realized that their own state tax law was working against them?
The authors are clearly aware of this problem, and they do address it. They cite literature that “taxpayers may become generally aware of even complex income tax changes through a combination of news and advisories, changes to withholding during the year, end-of-year reports from tax preparers showing individualized tax rates, and finally changes to tax refunds.” I think this is probably right in the case of high-income taxpayers. And since high-income taxpayers were the primary group of interest in this particular study, I’m inclined to accept the authors’ findings as both interesting and convincing.
I also think the method itself is pretty cool. I agree with the authors that other researchers can use similar methods to study other problems related to the impact of federal tax changes. But I do question whether the salience problem may be more challenging to overcome when studying a less-sophisticated, lower income population. I also think that this method may be less reliable when studying tax cuts accompanied by reforms that significantly change the tax base, because differences in states’ approaches to conformity may further complicate the analysis.
I enjoyed reading this article for both its substantive and methodological contributions. I recommend this article to anyone interested in tax and economics or empirical tax research.