- The corporate income tax started in 1909 at 1 percent. Today, it's 35 percent.
- A tax on interest can have a very large effect on the rate at which you can trade apples today for oranges tomorrow
- A CRS report that is supposed to inform about the consensus of studies but veers way off is a disservice to Congress and taxpayers
The following are highlights of Kevin Hassett's testimony to the Joint Economic Committee at a hearing entitled
"How the Taxation of Capital Affects Growth and Employment" on April 17, 2012. The submitted written testimony is also available here.
"The corporate income tax has been levied on a permanent basis in the United States since 1909, when it was introduced at the rate of 1 percent. About one hundred years later, the U.S. federal tax rate for most corporations is 35 percent, and state taxes on average add another 4.2 percent tax. With a 39.2 percent combined corporate tax rate, we earned the honor of highest tax rate in the developed world on April 1st when Japan lowered its rate from 39.5 to 38 percent.
Download PDF "In addition to the corporate income tax, the United States also taxes dividends paid out to shareholders and capital gains at the individual level. This extra layer of capital taxation increases the overall effective tax rate that burdens new investment. On the other hand, depreciation and expensing provisions lower the effective tax rates on business income, and numerous loopholes and other tax expenditures lower the rate for industries that happen to be favored in Washington."
Income vs. consumption tax: Like apples to oranges
"A capital tax introduces a distortion into the return on saving and investment, a distortion that 'explodes' over time. Even a small capital tax will not be optimal because the damage it causes will eventually grow without bound. The intuition of this result is quite straightforward.
"Recall that an efficient tax system will cause individuals to change their behavior as little as possible. A huge tax on apples and a small tax on oranges would cause an enormous shift away from apples and toward oranges. A small uniform tax on both would not.
"Think of consumption today as being represented by apples and consumption ten years from now as oranges. If you give up an apple today, you get a number of oranges ten years from now that depends on the interest you got on the money you saved after not eating the apple. At 10 percent interest, a dollar saved today becomes $2.60 ten years from now. If we tax that interest at 50 percent, a dollar saved today only yields $1.63 ten years from now.
"Clearly, a tax on interest can have a very large effect on how much money you have ten years from now, a very big effect on the rate at which you can trade apples today for oranges tomorrow. Indeed, this distortion grows bigger and bigger over time because of compounding. One dollar saved today produces $17.45 thirty years from now at 10 percent interest. If the interest is taxed at 50 percent, then a dollar saved yields only $4.32 over the same time period.
"Since it is not efficient for the tax system to create dramatic changes in the relative prices, it cannot be efficient to rely on a device that produces a distortion that worsens steadily over time. This is why a consumption tax has been found to be optimal."
Double whammy of corporate taxes
"First, the double taxation of corporate income discourages investment in equipment and structures. The dividend tax raises the cost of funds to firms, increasing the hurdle rate for new projects. The accompanying reduction in capital spending reduces economic growth and interferes with the creation of new jobs.
"Second, the asymmetric treatment of debt and equity encourages heavy debt loads and increases the overall level of risk in the corporate sector. Firms that borrow to finance investments are allowed under current law to deduct interest payments associated with that debt. Dividend payments are not deductible. This encourages firms to use debt finance whenever possible. When firms have large debt loads, they are much more likely to enter bankruptcy during difficult times."
What could have been
"An OECD study by Arnold (2008) provides an empirical analysis of the effect of the tax structure on long-run GDP. The main findings include 'Property taxes, and particularly recurrent taxes on immovable property, seem to be the most growth-friendly, followed by consumption taxes and then by personal income taxes. Corporate income taxes appear to have the most negative effect on GDP per capita.'
"This intuition is supported by the review of the literature that I conducted with University of Berkeley economist Alan Auerbach in 2005, which suggested that a transition to an ideal system might increase economic output between 5 and 10 percent.
"This allows us to estimate what our fiscal situation might be today if the United States had implemented a fundamental tax reform ten years ago, and we had achieved the high end growth estimate of a 10 percent long run improvement. GDP would be $17.1 trillion in fiscal year 2012 rather than the expected $15.5 trillion under CBO projections. Moreover, if we assume that revenues stay fixed as a percent of GDP and outlays stay fixed in dollar terms, then the 2012 deficit would be -$830.4 billion rather than the expected -$1.1 trillion under the CBO alternative fiscal scenario. The long run budget deficit would also be substantially improved, with accumulated deficits of $7 trillion from 2013 to 2022 rather than the expected $11 trillion. This illustration suggests that the stakes are very large indeed."
Taking the consumption tax 1X further
"David Bradford ... passed the responsibility for paying taxes on wages on to the workers, and then taxed their wages using a graduated rate system. In principle, such an approach could allow for any possible level of redistribution, substantially weakening the logical basis of opposition to a consumption tax on social-justice grounds. ...
"Because the X-tax remains relatively unfamiliar, my AEI colleague Alan Viard and Robert Carroll of Ernst & Young have set out to introduce the Bradford X-tax to the broader public in their forthcoming book which we can arrange to send to each member of this committee upon publication. Their book sets forth solutions to commonly perceived problems concerning the taxation of pensions and fringe benefits, business firms, financial intermediaries, international transactions, owner-occupied housing, state and local governments, and nonprofit institutions, and the transition. By adopting these proposed approaches, the United States can move to a progressive tax system that no longer penalizes saving and investment."
The other side of the ledger
"Much political courage is needed to propose and achieve fundamental tax reform, but there are other smaller compromise actions that can be taken to improve the current tax system. One of the main steps towards consumption taxation, without full-blown tax reform, is the implementation of permanent business expensing. In other words, allowing firms that purchase new machines and other capital goods to be able to write them off immediately, instead of over many years.
Much political courage is needed to propose and achieve fundamental tax reform, but there are other smaller compromise actions that can be taken to improve the current tax system.
"A well-developed body of research by economists confirms what businessmen will tell you if you ask: When the cost of capital is low, firms are much more likely to expand their capital stock. And full expensing can reduce the cost of capital significantly. Future deductions are not as valuable as current deductions because of the time value of money, and because these deductions are not indexed for inflation. Expensing gives firms the entire deduction up front, and with full expensing, the value of the deduction will exactly offset the present value return on the investment over its lifetime, so the effective marginal tax rate on investment will be zero."
Dividends: To tax or not to tax
"The literature on dividend tax policy and investment has had a rather contentious history. Theoretically speaking, it is possible to derive cases where dividend taxes have a large effect on investment, but other cases exist that are equally plausible that suggest that dividend taxes have a smaller effect. ...
"The data seem to favor the new view of the user cost effect, which suggests that the impact of the dividend tax change would be small for most mature firms. However, there is evidence that immature firms responded quite a bit. On balance, then, one should conclude that as was the case with earlier studies of the new and old views, firm heterogeneity seems to be quite important in evaluating the investment response to the dividend tax reduction. ...
"Overall, it is safe to conclude that a near tripling of the dividend tax rate proposed by President Obama’s latest budget would have negative consequences on investment and growth."
Pushing businesses away
"[T]axation at the corporate level has the undesirable tendency to drive capital overseas. Much of the early research on the corporate income tax examined its effect in a closed economy, in other words an economy where capital is contained. In this situation the corporate income tax can be viewed as, essentially, a direct tax on the owners of capital. More recent research, however, has begun to reflect the fact that the US economy is certainly best characterized as an open one.
"In an open economy, if corporate tax rates are high, then investors and firms are free to move capital to other countries with more favorable tax treatments. If an American firm locates a plant in the U.S., for example, it will after state and local taxes keep only 61 cents of every dollar the facility earns. If it locates the new plant in Ireland, it keeps 87 cents of unrepatriated earnings. There is a large literature that finds that firms are incredibly skilled at moving money around to minimize their taxes."
The wrong side of the Laffer curve
"[M]y work with AEI colleague Alex Brill also finds strong evidence that a Laffer curve exists in the corporate sphere and that the revenue maximizing rate has fallen from about 34 percent in the 1980s to 26 percent in the early 2000s. If you take the Brill Hassett estimates seriously, then the U.S. could increase tax revenue by 767 billion dollars over the next ten years if it reduces its rate to 26.4 percent, and it would have to cut the rate all the way to 17.8 percent if it wanted to enact a revenue neutral reform.
"A final argument in favor of cutting the corporate tax rate is that it would benefit workers. This channel was recently discussed in a Senate Budget Committee testimony by the former director of the Brookings-Urban Tax Policy Center Roseanne Altshuler, who wrote, “Moreover, any increase in the corporate income tax rate will reduce domestic income and lower wages (through an outflow of capital) and adversely affect economic efficiency.
"The idea that workers may bear a portion of the corporate income tax is neither surprising nor new. Basic incidence analysis suggests that the burden of the tax will always be larger on the side of the market that is more inelastic. In the short run, the incidence will necessarily be borne out of the earnings of fixed capital since the supply of capital is fixed. However, it is the long run effects which are of greatest theoretical and practical interest. Since capital is relatively more mobile in the long-run than labor (which is relatively inelastically supplied), labor could bear a larger portion of the tax burden.
"There are two important implications of this capital mobility. The first is that the United States would likely draw more capital by lowering its corporate tax rates. It may also be on the wrong side of the 'Laffer curve' and be able to raise more revenue from a lower rate. The second implication is that the gains from a corporate tax cut would likely flow through to labor. As capital returns to the American economy, each worker will have a relatively larger stock of capital to work with, and the marginal product of labor will rise."
Response to criticism
"In a recent research report prepared by Gravelle and Hungerford (2011) for the Congressional Research Service, the authors take issue with a couple of my studies on corporate taxation, along with almost every other paper in the literature. They argue that the Laffer curve results from the Brill paper are the result of an 'econometric error.' The authors also go to great lengths to criticize my work with Mathur.
"The 'error' they accuse Brill and I of making is that we do not use fixed effects in our panel regressions. While this is a technical issue that is far beyond the scope of this hearing, it is important to note that this choice (which has been made by other authors in the literature for the same reason we make it) is a specification choice, not an 'error.'
"As I taught my students when I taught graduate level econometrics at Columbia, if one is running a cross section regression, one cannot control for fixed effects. If one has a large panel data set with many countries and years, and ample variation for the relevant variables, one can. The sample we use has ample cross section variation, but not an enormous amount of time series variation for each country, as corporate tax changes are infrequent. The specification preferred by the CRS throws out the variation across countries, and focuses only on those countries that change their rates. In their specification, it is irrelevant that the U.S. is now the high tax country, since differences across countries are thrown out. It should be no surprise that, since changes are relatively infrequent, throwing out all other variation makes it difficult to find statistically significant results.
"The logically correct statement one might make given their results is that if one controls for fixed effects, the Laffer curve is not statistically significant, and if one does not then it is. It might be that this is because the CRS estimator eliminates much of the variation, or it might be for the reasons highlighted in the CRS report. The fact that the report immediately jumps to the conclusion it does reveals a tendency in the report that is repeated often as it turns to other papers. When the authors finally find some specification that agrees with their biases, then they conclude that only that specification is correct.
"The problem with such an explicit data mining approach is that it has very little potential to reveal the truth. The first sign that they have not done so is that there are a number of other papers ... with similar findings. A second sign that the authors have not shed light on the truth on this issue is the logical problem presented by their results. If there is no Laffer curve in the data in the range of current tax rates, then they would also have to reject the large literature mentioned above that finds that corporate income is highly mobile, seeking out the lowest tax countries. The authors also would be unable to explain why countries around the world have been cutting their corporate tax rates. If Brill and I (and the other authors mentioned) are right, then all the pieces tie together sensibly. Countries reduce their rates because they see the harm that is done when their own rate disadvantages them as a location.
"The same approach is taken later in the study when the authors turn to my paper with Aparna Mathur on the effect of corporate taxes on worker wages, where the authors again discard much of the variation for poorly motivated reasons, making statistical inference more difficult, and then declare victory when statistical significance diminishes.
"In one of the first empirical studies on the topic, (Hassett and Mathur, 2006, revised 2010) we use a unique, self-compiled dataset on international tax rates and explore the link between taxes and manufacturing wages for a panel of 65 countries over 25 years. We find, controlling for other macroeconomic variables, that wages are significantly responsive to corporate taxation, in that higher corporate tax rates depress wages. We also find that tax characteristics of neighboring countries, whether geographic or economic, have a significant effect on domestic wages. These results are consistent with the frequently employed assumptions in the public finance literature that capital is highly mobile, but labor is not. Under these conditions labor will bear the burden of capital taxes, after some lag while firms observe productivity gains and workers renegotiate fixed wage contracts. The study uses a standard specification drawn from the existing literature on wage variation across countries.
"My colleagues Aparna Mathur and Matt Jensen summarize these results concisely, 'the results suggest that a 1 percent increase in the corporate tax rates leads to a 0.5 percent decrease in wage rates. For example, if the corporate tax rate increases from 35 percent to 35.35 percent, a 1 percent increase, a 10 dollar per hour wage rate will decrease 0.5 percent to $9.95. Using information from the United States wage bill and tax revenues, this implies that every additional dollar of tax revenue leads to a $4 decrease in aggregate real wages. Examining the effects of tax rate changes one year later, rather than five, we find that a $1 increase in tax revenues leads to $2 decrease in wages.'
"The CRS first tested our data and results to see if the results were in fact replicable. They reported that the results in fact did match the results presented in our paper, and the variable of interest, the corporate tax rate was indeed statistically significant. They then tested to see if alternative specifications of the regression equation would still produce significant results. For example, they suggested that using annual exchange rates to convert the national wage data may be inappropriate and that purchasing power parity conversions were needed. When they did the conversions using PPP, the results were very similar to those with the exchange rate data and were still statistically significant. They then went on to suggest that inflation-adjusted PPPs were an even better method for obtaining comparable real wages across countries. Using this measure as the dependent variable did produce a change in the magnitude of the coefficient—it decreased marginally. However, the coefficient was still statistically significant at conventional significance levels. The authors then concluded that their methodology had yielded less robust estimates of the effect of corporate taxes on workers, even though the negative and statistically significant effect on wages was robust.
"As a final check, they attempted to replicate the results using only a balanced panel (i.e. using only those countries which had the full five years of wage data). In this case, the results were insignificant. However, this is not surprising since imposing the condition of full availability of data implied that the sample size dropped significantly (by 30 percent) and most importantly, would have eliminated a lot of small, developing countries whose wage response to corporate tax changes would likely be more pronounced. In fact, in results not reported in the paper, we show that small, open economies have a higher elasticity of wages to corporate taxes than larger economies. In addition, econometric analysis can be conducted using unbalanced panels, particularly when the nature of the data is such that it is difficult to obtain consistent, good data for all countries.
"A final specification that the authors ran was to use annual wage data and then use several lags of the corporate tax variable to see if the coefficients were significant. They report that the coefficients are insignificant. However, this is clearly incorrect since our paper does present one specification using annual wages as the dependent variable and we find statistically significant coefficients on the corporate tax variable. One reason the authors may be obtaining their results is due to an incorrect specification of the regression equation. In a typical annual wage regression, one would have to include lagged wages as an additional regressor on the right hand side to account for persistence in the wage data. Current wages are likely to be highly dependent on lagged wages, at least for one period before. Further, the lagged dependent variable and the error term are likely to be correlated due to serial correlation in the error term. We attempt to control for all these misspecification problems in the regression reported in our paper, and use GMM to instrument for the lagged dependent variable. As mentioned earlier, our results suggest that the elasticity of wages to corporate taxes is lower in the annual data, but still statistically significant. ...
A CRS report that is supposed to inform about the consensus of the literature that veers so far from that activity, is a disservice to Congress, and the taxpayers.
"I add this digression to my testimony because the CRS report is radically at odds with the literature. I relish academic debate, and think that authors serve a valuable service when they challenge research. But a CRS report that is supposed to inform about the consensus of the literature that veers so far from that activity, is a disservice to Congress, and the taxpayers."
Oh, and don't forget the Buffett Rule
"I would be remiss if I did not offer a few words on the so-called Buffett Rule, which appears to be the focal point of President Obama’s decidedly domestic policy agenda. The Buffet rule proposes to apply a minimum tax to ensure that all taxpayers with income exceeding $1 million pay at least 30 percent of their income in taxes. It raises relatively little revenue ($160 billion over 10 years compared to a current policy baseline, $47 billion compared to a current law baseline, and even less than that if all of President Obama’s other proposals were to become a reality). Additionally, the Buffett rule does not apply equally to all income classes. It focuses on capital gains and oddly omits interest income from municipal bonds. From an economic standpoint, the Buffett Rule is merely a stealth tax on capital gains and other forms of capital income. Once again, there is no support in the literature for such a tax."