Close Nav

Tax Increases and Behavioral Responses


Tax Increases and Behavioral Responses

December 2, 2010

Yesterday, Senate Republicans threatened to block legislative action on every issue during the lame duck session until the dispute over the looming tax increase is dealt with and an extension of current government funding is approved. While some may view this as obstructionist, the importance of the resolution of these two legislative items can not be overstated, and with only a few weeks left time is running short. With focus narrowing on the tax increase debate, it’s important to consider all the ramifications of the proposed hike, especially the economic impact that such an increase could have.

The looming expiration of the Bush tax cuts has focused attention on the Obama Administration’s plan to extend most, but not all, of the tax cuts. Individuals earning more than $200,000 (and couples earning more than $250,000) would face higher marginal tax rates. Understanding the behavioral response of this group to proposed higher taxes is especially important given that the top one percent of taxpayers generate more than one-third of all federal individual income tax revenue.

In opposing the extension of tax cuts for the highest tax brackets, the Administration has argued that the nation cannot afford expensive “giveaways” to the “wealthiest” individuals. Republicans who support the extension of all tax cuts have instead claimed that registered income at the highest tax brackets frequently represents small business income, and that hiking taxes during a recession would harm the recovery.

However, both sides have largely dodged the issue of how high-income households would respond to these tax hikes. If there is a strong connection between taxes and behavior, then there may be a high price in failing to extend the Bush tax cuts for high-income taxpayers. A body of academic evidence suggests this is in fact the case.

Elasticity of Taxable Income

Higher tax rates induce a variety of behavioral responses. People may choose to work fewer hours, take jobs that offer less pay, or drop out of the workforce entirely. Over the long run, people may avoid making investments that boost income in the future, such as pursuing more education, because the return on this investment is lower (in the form of reduced after-tax income). These tax responses all lower economic efficiency. Alternatively, individuals may restructure their declared income to lower their tax liability -- for example, by engaging in economic activities that generate more deductions. While these actions may not necessarily lower economic output, they lower the revenue governments receive from higher taxation.

While economists largely agree that higher taxes have some impact on both economic efficiency and declared income, the quantitative impact can be difficult to gauge. In calculating this effect, economists emphasize the importance of one parameter: the elasticity of taxable income (ETI). This parameter measures the extent to which individuals adjust their taxable income in response to a rise in their marginal tax rates. It is an attractive instrument because it allows an analysis of the impacts of taxation without needing to specify the particular pathways by which taxation impacts individual behavior.

For example, suppose taxpayers earned $100 billion of income taxed at the top federal income tax rate of 35%. If individuals did not respond at all to higher taxes, the effects of tax hikes would be straightforward to analyze. A hike in the top marginal tax rates to 40% would mechanically raise tax revenue by $5 billion. If individuals adjust their reported income in response to tax hikes, higher tax rates raise less revenue. With an ETI of .40, the federal government raises only two-thirds of the expected revenue, or $3.3 billion. With a value of .80, the same tax hike raises only $1.5 billion.

The ETI is also valuable for other calculations. Forexample, an ETI of .40 produces a revenue-maximizing tax rate (corresponding to the top of the Laffer Curve) of 61%, while an ETI of 0.8 yields a revenue-maximizing rate of 44%. At tax rates higher than these values, additional tax revenue can only be raised by cutting tax rates.

Another key metric is the marginal excess burden of taxation, which corresponds to the economic loss incurred by higher taxation. It can be thought of as the additional cost of taxation and should be balanced against the benefits of spending. If we hike taxes to pay for a particular program, we should insist that the benefit of that program exceed the cost of the associated taxes. An ETI of .40 implies that a tax increase of $100 on top earners corresponds to a marginal burden of $53, while an ETI of .80 implies a marginal burden of $222.

The impact of taxes matters a great deal in structuring government spending programs. If statutory tax rates exceed the revenue-maximizing rate, the government can only raise additional money by cutting taxes. But even if taxes are below the revenue-maximizing limit, there may be good reasons to avoid additional taxes. For instance, the high marginal cost of taxes when the ETI is .80 is linked to the fact that the revenue-maximizing rate under that estimate (44%) is not much higher than the top federal income marginal tax rate (35%). While setting taxes at the top of the Laffer curve maximizes tax revenue, it is advisable to set marginal tax rates below this level. As tax rates approach the revenue-maximizing rate, the marginal cost of additional taxation grows infinitely large.

Responsiveness of High Income Individuals

The impact of higher marginal tax rates depends on the magnitude of behavioral responses. The more people change their behavior in response to higher taxes, the greater are the economy-wide losses. At high response rates, additional increases in taxation result in negative consequences for economic growth overall, and offer little additional revenue to government coffers.

However, the responsiveness of individuals to taxes is not constant across the population. Richer individuals, in particular, are more likely to respond to higher taxes for a variety of reasons. One reason is their access to a greater variety of tax avoidance options, such as itemized deductions. Richer individuals also pay higher marginal tax rates to begin with, so additional hikes hurt them proportionately more. As Alan Viard explains:

The fact that taxpayers can alter their taxable income implies that income taxation is distortionary, meaning that it imposes an economic burden beyond the actual tax payments... The distortion depends upon the marginal tax rate, the fraction of each additional dollar of income that is absorbed by tax. Moreover, the distortion rises disproportionately with the marginal tax rate, roughly quadrupling when the marginal tax rate doubles. (Emphasis added)

A large economic literature has documented that the reported taxable income of high earners is more responsive to taxation, and provides quantitative evidence for the magnitude of the effect.

Emmanuel Saez and Jonathan Gruber provide one such study. Though they found that the ETI for the population as a whole was .40, the authors found that this elasticity was .57 for Americans making over $100,000 a year in 1992 dollars (comparable to about $150,000 today). It was even higher -- .66 -- for those in that group who itemized deductions.

Several other academic estimates of high-earner ETI are summarized in the following table, along with corresponding estimates of economic costs:

(Click the table to see a larger version)

Though precise estimates vary from study to study, all are consistent with high-income taxation imposing large costs on the economy. Estimated elasticities for high-income earners tend to average between .50-.60, except when looking at extremely high earners, for whom the estimates may be far larger. These rates imply that the behavioral responses to taxation take up to half of the overall expected gains in revenue. They are also consistent with revenue-maximizing rates that are not much higher than the top statutory rates currently prevailing, once Medicare, income share of corporate taxes, and state and local taxes are accounted for.

While these studies remain some of the best available in gauging the quantitative impact of higher taxes, they are subject to potential biases in both directions:

Upward Bias: All referenced studies use for their estimate of taxable income the declared individual taxable income of individuals. However, high-income individuals often have the option to shift their income into different categories. For instance, a doctor may respond to higher income taxes by doing business in corporate form. Though this will result in lower declared individual taxable income, the doctor will declare corporate income and pay corporate income tax. Simply looking at individual taxable income therefore overstates the actual behavioral response of individuals if they can shift into other types of taxed income. To address this issue, a separate column in the above table computes results assuming that half of the change in reported income represents a shift in tax declaration and that this alternate income is taxed at 30%.

Downward Bias: There are other adjustments individuals could make that are not captured in these estimates.

For example, responses to taxes may be larger in the long run, when individuals have more time to adjust their patterns of work, leisure, and consumption. While entrepreneurs may not quit immediately in response to higher taxes, they may choose to retire early. Yet the academic estimates presented here are evaluated over the short- and medium-run and therefore do not capture the long-run effects of higher taxes.

One study that has tried to calculate effects over a slightly longer horizon is that by Auten, Carroll, and Gee, who find an ETI of 0.67 for the population as a whole, based on data for the 2000-2005response to the Bush tax cuts. This estimate is higher than past estimates, and the high-income response would presumably be even higher. This response could also come from general increases in income and cannot be interpreted as simply reflecting the effects of the Bush tax cuts. However, it suggests that the long-term effects of tax responses may be sizable.

Another example of this broader response comes from the dramatic rise in tax revenue reported by the richest earners. As James Poterba and Daniel Feenberg have suggested, the rapid rise in reported income coincided with the dramatic decline in the top marginal tax rates that began in the early ‘80s:


This dramatic rise in earnings for high earners no doubt has multiple causes. Yet the close connection between reported income over a period of time and top marginal tax rates suggests that changes in tax policy may generate substantial long-run results. These results are also consistent with the cited studies in suggesting that high-income individuals are highly responsive in adjusting their behavior in response to changes in taxation.

Institutional Factors

These considerations suggest that taxes have effects not only on the particular individuals involved; they also impact broader social choices on the labor-leisure tradeoff. To evaluate these overall macroeconomic effects, some researchers have pointed to a comparison between America and Europe. America has lower taxes than Europe, and also has a far higher income. An ETI estimated on the basis of this difference is extremely high, and would suggest that the response to taxes is massive, and occurring through channels that are not captured by the current economic literature.

It seems unlikely that the extremely high estimates of labor responsiveness derived from cross-country comparisons represent the true effect. There are many reasons why America and Europe have different levels of income other than their differing tax regimes. However, it does suggest that there are broader responses to taxes may be quite large.

One researcher who has tried to account for more flexible responses to taxation is Raj Chetty, someone whose work we have featured here in the past. In a paper examining evidence from Danish tax records, Chetty and co-authors emphasize that estimates like the ones presented here capture only the individual responses to changes in tax rates once a job has been acquired. However, over the long run, higher tax rates also affect the types of jobs firms offer and employees accept. Government policies affect not only individual choices, but also the social norms and institutions that govern labor-leisure tradeoffs and coordinated work responses.

The existence of search costs may attenuate these effects in the short term. But over time, the coordinated response to taxes will be higher. The results also raise the possibility that taxes that affect only a small group of individuals may ultimately impact other workers, if firms change job offers to many workers.

In a separate paper, Chetty focuses on the frictions that may affect short-term responses to tax hikes. There are costs to adjusting to new tax regimes, while individuals may also be unaware of changes in taxes. Chetty develops a new technique to calibrate the responsiveness to taxes, incorporating small frictions. His results suggest that the ETI for the population as a whole may be close to .50, higher than most conventional estimates. The comparable figure for high-income earners may be even greater.


Analyzing the behavioral response to tax hikes is a tricky subject. Hard estimates do not exist to gauge how high-income individuals will choose to respond to higher taxes. However, reasonable estimates from the most rigorous work in the field suggest that the costs of high-income taxation are quite high. Because individuals with high incomes are more responsive to taxation, strategies to “soak the rich” will exact a large economic burden while delivering little tax revenue.

If we account for the fact that taxes exert social changes and manifest themselves over time, we find even larger effects -- effects sufficiently large to suggest that the scope of additional taxation on the rich is small and the marginal costs on society is high.

By itself, this analysis does not suggest that tax hikes for the rich are a bad idea. If the goal is, as Robert Frank might prefer, to clamp down on the consumption habits of the wealthiest, then high-income tax hikes might be a good idea despite the high costs on the overall economy.

Nor will these estimates remain valid for different tax regimes. The high response of individuals who use deductions suggests that it is the design of the tax system itself — filled with deductions, and based on income — that is responsible for the high costs of high-income taxation. It’s likely that several European countries with far higher tax/GDP ratios, and with far higher marginal tax rates on the rich than the US, are able to manage that trick through a more efficient tax system that relies more heavily on consumption taxes.

Yet maintaining the Bush tax cuts without making broader changes to patterns of government spending will not bring reprieve to taxpayers. As Milton Friedman emphasized, "To spend is to tax." The problem with the fiscal deficit is not the Bush tax cuts, which have maintained federal tax/GDP ratios within historical norms. It's the looming threat of higher spending as a portion of the economy driven by entitlement programs, as well as the potential burden of paying for future banking and housing bailouts. If spending is not restrained, higher taxes are inevitable.

If the goal of the tax system is to generate tax revenue with a minimum of economic loss, policymakers should think hard before agreeing to raise taxes on the rich, given the existing tax code. Available evidence suggests that these taxes will come at a large economic cost. If policymakers choose to further tax the rich, they should be sure they are spending the money on programs valued by the public.

Arpit Gupta is a Research Coordinator at Columbia University, where he focuses on consumer finance, real estate, and banking.

e21 Partnership

Stay on top of the issues that matter to you most





























Main Error Mesage Here
More detailed message would go here to provide context for the user and how to proceed
Main Error Mesage Here
More detailed message would go here to provide context for the user and how to proceed