Econ Focus

The recent recession and relatively sluggish recovery have prompted much discussion about what policymakers can — and cannot — do to stimulate economic activity and foster a healthy financial system. Much of that discussion has focused on monetary and regulatory policy, of course. But fiscal policy can play an important role as well in such periods.

Joel Slemrod, an economist at the University of Michigan, has spent his career working in the field of public finance. His research has spanned a number of areas, including how sensitive businesses and individuals are to tax rates within and across countries, and how that sensitivity affects their location decisions; the degree to which households, especially high-income households, alter their behavior due to tax policies; conditions that may affect people's savings behavior; and the level of noncompliance with tax laws in the United States and abroad.

Slemrod, who directs the Office of Tax Policy Research at Michigan, also has held numerous appointments in Washington, D.C., with such institutions as the U.S. Department of Treasury and the President's Council of Economic Advisers. Aaron Steelman interviewed Slemrod in the fall of 2010.

RF: In a 2005 paper, you described your "Beautiful Tax Reform." Could you briefly discuss the system that you laid out in that paper? Relatedly, do you think it is possible to divorce normative concerns from positive concerns when thinking about tax policy, or will equity issues always arise?

Slemrod: Let me take the second question first. Although this paper does lay out a framework for my preferred tax system, it also argues that one's preferred tax policy is inevitably a mixture of what one thinks about how the economy works — for example, behavioral responses to tax rates — and also value judgments, which aren't subject to economic analysis and probably are very hard to persuade others to adopt. So I thought that I could make a bigger contribution to this edited volume (which included papers from many people saying what tax system they would prefer), if I stated explicitly how my own preferred tax reform depends on both my views on how the economy works and on what my values are. So my answer to the second question is no — what tax policy is best will always depend on both positive and normative judgments.

Let's now come back to the first question. In the paper, I made the point that the simplest tax system isn't necessarily the best, in part because of the trade-off between what one might call efficiency and equity. Consider that the simplest tax system is probably what one might call a lump-sum tax, where everyone pays basically a fixed amount. It wouldn't be trivial to enforce, but it certainly would be a lot simpler than what we've got now. However, I think most people, maybe not everybody but most people, would find that objectionable because they think the tax burden ought to be related to a household's well-being, whether that be assessed by income or wealth or consumption or some other measure. So that's why I wouldn't favor replacing our graduated income tax system with a lump-sum tax or a value-added tax — because the distribution of the tax burden is not progressive enough for me, and that, I emphasize, is "for me." If your values were such that you were happy with an approximately proportional tax burden, where the tax burden is approximately proportional to lifetime income, there's a lot to be said for just relying on a value-added tax. But I'm not willing to do that, so I'm sticking with a tax system that relies heavily, maybe not entirely, but heavily on a graduated income tax.

The rest of the paper talks about fairly standard ways to clean up the income tax, because I think a lot of the exceptions to a straightforward tax levied on income — a lot of the credits and deductions, for instance — are examples of the government favoring particular people or particular activities that result in inefficiency. So, for example, I would get rid of the itemized deduction for state and local property taxes; I would get rid of the preferential tax treatment of employer-provided health insurance; and I would clean up the implicit subsidy to owner- occupied housing, and although I don't have a clean solution for how to do that, we know how to move things in the right direction. Then in the paper I talk a little bit about how the process of filing taxes could be simplified, so that a large fraction of Americans wouldn't need to file a tax return at all. Other countries do this, including 15 OECD countries, and we could do it, too, regardless of how hard the Internal Revenue Service (IRS) now says it would be. And, finally, I talk a little bit about cleaning up how corporate income is taxed in the United States. Almost no economist thinks we have a very rational system. The most obvious problem is that income is taxed first at the corporate level and then at the personal level, but the underlying issues are that the rate of tax on corporate income is different than on other income and the tax base is capricious, depending on things like the financial policy of the corporation; I discuss some ideas about how to clean that up.

RF: You mentioned the way we currently tax employer-provided health insurance. What do you think are the benefits as well as deficiencies with that policy?

Slemrod: Well, it certainly reduces the after-tax price of health insurance for people. The problem is that it reduces the price below the true social cost, so that people acting in their own family's interest, are, at the margin, buying insurance where the value to them is actually less than the true cost. In a word, we are subsidizing high-deductible, low copay insurance policies and, given the upward trend we are seeing in the fraction of our gross national product that goes to health care, I think we ought to be moving toward reducing or eliminating such subsidies. Not only that, it's a very unattractive sort of subsidy, because the subsidy rate is dependent on the household's marginal tax rate, so the subsidy rate is highest for the highest-income people. And I just don't think that even people who would argue for a subsidy would favor such regressivity if they were designing a subsidy scheme from scratch. The reason to be wary about abandoning the subsidy is that it supports the system of employer-provided health insurance, which spreads risks across employees and offsets the problem of adverse selection that can plague health insurance markets; before we eliminate the subsidy entirely, we need to have other policies in place to prevent a collapse of efficient markets for health insurance.

RF: What does our recent experience tell us about the effectiveness of tax rebates in stimulating economic activity in a recessionary period?

Slemrod: Actually, we have had three tax rebate policies enacted in the last decade — 2001, 2008, and then again in 2009. I have done a fair amount of research on this topic with Matthew Shapiro, my colleague here at Michigan. The research methodology is based on posing the following questions to a sample of people: What did the tax rebates lead you to do? Did they lead you mostly to increase spending, mostly to increase saving, or mostly to pay down debt?

When we focused on 2008 and 2009, in both cases we found that only a small fraction of people said it led them to mostly increase their spending. In 2008, less than a quarter of people said that and in 2009, only about 13 percent said that. So we concluded that the stimulus to spending that works through the marginal propensity to spend would actually be quite modest as a fraction of the total tax cut. Because these tax cuts were pretty large, the dollar stimulus was not trivial, but certainly relative to the tax cut, the stimulus probably was fairly modest. Our surveys tell us two other interesting things: One, contrary to conventional wisdom, we found no evidence that low-income people would be more likely to spend the money they received from the tax cuts. Second, we found that, with the 2009 tax cuts, which were delivered in the form of reduced employer withholding, people actually had a lower marginal propensity to spend, which was contrary to what a lot of economists had opined when the delivery mechanism for the tax cuts — rebate payments versus reduced withholding — was being discussed in early 2009.

WEB-EXCLUSIVE

RF: Why was the conventional wisdom wrong?

Slemrod: I'm not sure we know the answer to that. My sense is that one of the things that affect a household's marginal propensity to spend after a windfall increase in after-tax income is how they're feeling about their financial situation generally. And in both 2008 and 2009, many people were feeling like their situation was dire. If they owned stocks, their wealth was down. If they owned a house, their wealth was down. They might have been out of a job or worried about being out of a job. So I think a lot of people were just worried that their asset cushion wasn't big enough, or their debt was too high, and they looked at these tax cuts as a way to get their assets or debt back into balance. It's reasonable to believe that the people who were most concerned about that were low-income people, so if these concerns were indeed significant factors in getting people to save or pay down debt, it might have offset what people had found in earlier situations, that is, that the marginal propensity to consume was higher for low-income folks because of liquidity constraints.

RF: We often hear the claim that taxpayers vote with their feet, leaving relatively high-tax states for relatively low-tax states. What does your work on the estate tax tell us about that claim? And what do you think of it more generally?

Slemrod: I think there's substantial evidence that people and businesses, when they consider where to locate, think about the financial implications of where they're going, including the kind of taxes and the tax rates they would face. There's also a lot of evidence that they think about the other side of the government budget too, that is, what government provides. For example, there is evidence that, other things equal, some people will migrate to where welfare benefits are higher. It's also very clear that people don't migrate simply to where the taxes are the lowest, because if you look at the United States, the states with the lowest taxes ( and, consequently, relatively low levels of public services) are not attracting masses of people. I have one article, now several years old, which looks at one aspect of this issue: whether differences in the estate tax rates of American states affect migration. It uses a fairly simple research design that analyzes data about the state of residence at death of people whose wealth is high enough that they would be subject to estate taxes. We looked to see whether, over time, when state estate tax rates change, if deaths of high wealth people go up when estate tax rates go down, and vice versa. And we do, in fact, find that to be true. That's consistent with the fact that, other things equal, at the margin, some rich people will move when they're getting on in years to places where the estate tax is lower. Another possibility is that they just manage their affairs so that their legal residence is in a state with a relatively low tax rate. One or the other, or some combination of the two, appears to happen, although the magnitude is not very large.

RF: This is a broad question: But does Atlas, in fact, shrug?

Slemrod: You're referring to the book I edited titled Does Atlas Shrug? The Economic Consequences of Taxing the Rich. It seems like that this issue never goes away in policy debates. The book is a collection of articles by different economists and lawyers who don't all come to the same conclusion. My own view, based in part on the research discussed in this book, is that certainly high-income people notice taxes, and they react to taxes in ways that lower their exposure to taxes. The evidence for taxes substantially affecting what one might call "real" behavior, such as labor supply or savings, is not as strong as the evidence regarding another class of behaviors we might label "avoidance."

There are a lot of examples of high-income people taking avoidance steps to reduce their exposure to taxes. For instance, when tax rates are known in advance to change from one year to the next, we see high-income people shifting their taxable income into the lower tax rate year. If the relative tax on corporate income versus individual income diverges a lot, there's evidence that high-income people will change the form of their business, from a corporation subject to the corporate income tax to a business not subject. Thus, my overall conclusion is that Atlas does shrug, but not in the way that some might think.

Now, that being said, the public economics field has moved toward the view that if you're trying to measure the efficiency cost of state income taxation, the best summary measure of that is not the elasticity of labor supply — it's the elasticity of taxable income. Taxable income is certainly affected by labor supply, but also by all the other things people might do to lower their taxable incomes, such as avoidance, evasion, increasing tax deductible activities, and so on. So I don't mean to say that these other sorts of "avoidance" responses are not relevant for policy. They absolutely are.

RF: How large of a problem is tax evasion in the United States? That is, what is the magnitude of tax evasion and what could be done to decrease that number in a way that is not socially harmful?

Slemrod: The most comprehensive attempts to assess the magnitude and nature of tax evasion have been done in the United States by the IRS. For obvious reasons this is not an easy question to answer, even with a careful, comprehensive study. So, with some margin of error, the IRS thinks that for the income tax and other taxes that the IRS oversees, the rate of noncompliance is about 13 or 14 percent — about 13 or 14 percent of what should be paid is not paid.

What should be done about it? First, note that just because there's a 13 or 14 percent noncompliance rate does not mean that we have vastly too little enforcement. For sure the optimal noncompliance rate is certainly not zero, just the way the optimal burglary rate is not zero — it would just require too many resources to completely eradicate either of these things. What would I do? The most effective way to reduce noncompliance is to have third-party reporting. In the United States for most wages and salaries, your employer sends a report to the IRS stating how much you have been paid, and now their computers are good enough that if you don't report those wages and salaries, there's a very high likelihood that you are going to get a computer notice from the IRS asking you why. Thus, the chance of getting away with understating your wage and salary taxable income is very low and, consequently, the IRS has estimated that the rate of noncompliance for wages and salaries is 1 percent, while the rate of noncompliance for self-employment income is 57 percent. The former is subject to withholding and information reporting, and the latter is subject to neither. So one thing we should consider doing is extending information reporting further. Most other countries have it for interest and dividends; many countries have withholding for those kinds of payments, as well. We should pursue, as the IRS has been doing recently, informationexchange agreements with other countries, because it's become quite clear that a lot of the noncompliance of highincome people involves offshore accounts or transactions, and transparent information exchange among countries reduces the attractiveness of noncompliance.

I and a co-author just recently completed a study using these data from the IRS about the distribution of noncompliance by income class, which suggests that the rate of noncompliance goes up with income class, except at the very highest levels of income. No one had a good sense of the distributional pattern of noncompliance until this analysis. One commonly hears that "the poor evade but the rich avoid," the idea being that high-income people don't need to do illegal things because they have plenty of legal ways to reduce their taxes. But our analysis suggests that this is not true — the rate of noncompliance does generally go up with income class, except, again, at the very highest levels. Now, all of these studies are fraught with problems. It might be, for example, that the kind of evasion that really high-income people engage in is very difficult for the IRS, even with a very intensive audit, to discover. The IRS is certainly aware that even an intensive audit isn't going to uncover all noncompliance, and it's going to uncover some kinds more than others. They try to make up for this by estimating multiplicative factors that adjust for the fraction of noncompliance they think they have missed.

WEB-EXCLUSIVE

RF: Is noncompliance particularly a problem for self-employed people, who, one would think, would have greater opportunity to not report income? And, if that is true, does noncompliance differ depending on income class or occupation? For instance, do relatively low-income people who take part-time, odd jobs for which they are paid in cash often not report that income?

Slemrod: In the study I referred to earlier we also examined sources of reported income. Part of the variation in noncompliance rates by income class is, as you suggested, due to the different composition of people's income sources. But that's not the whole story by any means. As to your point about the cash income of relatively low-income people, the classic example being a house painter, it could be that the IRS's methods for measuring noncompliance in such cases just completely miss this. It could be that if people don't report any self-employment income on their tax returns, or don't file tax returns at all, it's really tough for the IRS to know about this. And as I said, the IRS has ways to try to put that into its overall estimates, but they're unlikely to be perfect.

RF: Many developing countries have much higher rates of tax evasion. Is this simply because their collection systems are less inefficient? Or might there be cultural reasons such as the populace may have less trust in their government and feel less obliged to support it?

Slemrod: Well, I think the first aspect — variation in tax enforcement effectiveness — is certainly a big part of it. In countries where the tax administration is severely constrained for resources and the enforcement is very weak, the return to evasion is high. People don't want to be perceived as suckers in countries like that, where they see everybody else getting away with it.

Whether part of the story is that people evade more when they don't trust their government, including regarding spending their money wisely, is an interesting question, and a lot of social scientists argue that it is important. My own view is that we don't yet have a lot of hard evidence on the question. There is a positive cross-country correlation between the fraction of people who say they don't trust their government and measures of tax evasion, but that doesn't compellingly tell us that the lack of trust causes the higher rates of tax evasion. What causes what is tough to nail down. For example, you can't really do a field experiment, where you go into one part of a country and change how people feel about the government, and don't do that it in another part, and then compare changes in tax evasion rates. Trust in government could be very important, but it is just very hard for social scientists to pin down its behavioral implications.

WEB-EXCLUSIVE

RF: How sensitive are second earners in a household to changes in marginal tax rates? And has that sensitivity changed over time?

Slemrod: That is not directly my field, but I can say as an outsider looking in that the consensus among labor economists seems to be that the response of prime-age workers, of prime-age males especially, to changes in tax rates is relatively small. The labor supply response that is not small is the labor force participation decision of the second earners, often females. So their decision whether to work or not seems to be significantly affected by the rate at which that extra income would be taxed. In the U.S. system, unlike many other countries, income tax liability is based on family income, so the spouses' incomes are added. That means if there's one spouse working and the second spouse is considering working, the marginal tax rate on the second spouse's first dollar of income depends on how much income is earned by the first spouse. The marginal tax rate could be as high as 35 percent or more if you add state income taxes. So my sense is that the participation decision of the second worker is quite sensitive to the tax treatment of his or her prospective wages. Note, though, that second earners comprise a fairly small fraction of total labor supply, so the highly responsive portion of labor supply makes up only a fairly small fraction of total labor earnings.

RF: Are there certain goods for which consumption seems relatively unaffected by higher taxes? For instance, certain "sin goods" such as cigarettes? If so, what is a policymaker who, otherwise would prefer to use that instrument to reduce consumption in an effort to improve public health, to do?

Slemrod: Goods vary quite a bit in their price elasticity, that is, their responsiveness to tax-inclusive prices. The evidence suggests that the consumption of cigarettes is relatively price inelastic. So while you could potentially see an alliance between people who care about public health issues and people in the government who care about raising revenue, those two constituencies differ in their "preferred" elasticity. If cigarette purchases were inelastic to a taxinduced price increase, this would disappoint people who want to reduce smoking, but it is going to raise more revenue than if demand were highly price elastic.

My own work has addressed how the possibility of tax avoidance affects the impact of raising state cigarette taxes. Consider what happens when there are ways to avoid a state's cigarette taxes without actually smoking less — for example, traveling across the border to buy cigarettes in a state which has much lower taxes, or in the modern version, going on the Internet and buying apparently taxfree cigarettes. Then the state faces a tricky dilemma. If it tries to raise rates, it's not going to get as much revenue as it otherwise would. And for a lot of people the effective price has not gone up, because it just drives them to the Internet. My research on cigarettes suggests that the elasticity of taxed sales in a given state has gone up over recent years, as these tax-free alternatives, for example through the Internet, have become more widely available. And I say "apparently tax-free" because, although it is quite easy to buy untaxed cigarettes over the Internet, technically, if I did that, I am supposed to remit tax liability to the state where I live. The tax applies depending on where you smoke them, not where you buy them. But everyone knows that almost nobody actually remits these taxes.

On this topic, there is a wonderful piece of work by an economist from the University of Illinois at Chicago named David Merriman who had his students collect discarded cigarette packs all over the Chicago area. You can tell from the stamp on the pack where it was purchased. And sure enough, there are a lot of packs apparently consumed in Chicago that were purchased in Indiana, where the taxes are lower, and the fraction increases as you move closer to the Indiana border.

RF: What does your research tell us about the effects of tax policy on foreign direct investment (FDI)?

Slemrod: My own research and research done by others suggest that a host country's tax policy does have a significant effect on the amount and the type of FDI it attracts. My own research has tried to differentiate two aspects of why a low-tax rate may make a country more attractive for FDI. One is that it just lowers the effective tax on income. The other aspect is that, with a low tax rate, once there is activity in the country, most multinationals have the incentive to shift their taxable income into your country. They have many ways of doing this — such as establishing subsidiaries in low-tax countries. From a policy point of view, I feel quite differently about these two aspects. I have no problem with a country levying a low effective tax rate on income to try to attract real investment. I have a bigger problem with a country inviting, even encouraging, multinationals to shift income from higher-tax countries into their country, because to me this is parasitic on the treasuries of these other countries and isn't productive at all from a global point of view. In fact, I think this is welfare reducing because the higher-tax countries expend resources to try to keep the revenue from leaving and the companies expend resources to camouflage the income shifting. In the news recently is a country that has been quite successful at both of these aspects. For a long time Ireland has had a 12.5 percent corporate tax rate. This means there is a relatively low-tax rate on income from investment in Ireland. But I think the bigger issue is that this provides a tremendous incentive for, say, U.S. car companies to build maybe only a single plant in Ireland, and then shift the taxable income earned in its hightax locations into Ireland and thus lower their worldwide tax burden. That's a "beggar-thy-neighbor" policy of Ireland.

RF: The savings rate is affected by many things but one possible factor that many people may not have considered is the threat of a catastrophic war. What does your research tell us about this question?

Slemrod: Congratulations for waiting about an hour to ask me about one of my quirky papers! I have studied a few issues people seem fascinated by, and this is one of them. I have three articles that try to estimate whether, when people seriously think there's a chance of a nuclear conflagration, this belief affects their saving behavior. In short, do people believe we ought "to eat, drink, and be merry, for tomorrow we die?" To test this hypothesis I looked at aggregate saving over time in the United States, across countries, and micro data within the United States, and in all three cases found that when people think, or profess to think, there's a chance of a nuclear war, their saving rate goes down, just as economic theory would predict.

RF: You mentioned you have published other supposedly quirky papers that have garnered a lot of attention.

Slemrod: Yes, my co-author, Wojciech Kopczuk of Columbia University, and I actually won an "Ig Nobel" prize from a publication called the Annals of Improbable Research. We won it for a serious economics paper that was eventually published in the Review of Economics and Statistics entitled "Dying to Save Taxes: Evidence from Estate Tax Returns on the Death Elasticity."

We looked at estate tax return data from the history of the U.S. estate tax and found that when the estate tax was going to change — go up or down — in an anticipated way, then the distribution of deaths around that date was not symmetric. When the tax rate was going to increase, more people died before the rate rose, and when the tax rate was going to be lowered, people held on and more people died after the decrease. Since we wrote the paper, the general "death elasticity" finding has been replicated using data from episodes in Australia and Sweden when they ended their estate taxes. Those studies found evidence that people delayed their death to save their heirs' money, in some cases, millions and millions of dollars. We wrote this paper before the 2001 U.S. tax changes, which phased down the estate tax over the subsequent decade and which eliminated it completely for 2010, only to reinstate it in 2011. So there now are two recent episodes to further investigate our hypothesis. Between 2009 and 2010, some people should have been hanging on to "get" the zero estate tax rate. And now, right now (November 2010), the morbid part of the hypothesis applies, because someone who is going to leave a huge estate — well, they've got four weeks to get on with it estate-tax-free.

WEB-EXCLUSIVE

RF: There is currently a lot of uncertainty about both economic conditions and the way public policy issues will be determined. How do you think that uncertainty affects the decisions of market participants?

Slemrod: I think it depends on what kind of decision you're considering. My sense is that there's a lot of heterogeneity, so some people are probably making decisions based on a set of assumptions regarding what they think will happen, and other people are doing the same but based on completely opposite assumptions. For example, I have colleagues here at Michigan whose opinions I respect a lot who were quite convinced that income tax rates on the wealthy were going to increase in 2011 and other colleagues whose opinions I also respect very much who were quite convinced that the Bush tax cuts would not be repealed. One member of the first camp even told me that if he could find a way to bet on his forecast, he would, and so I tried to broker a bet. I don't know if they actually did bet, but it shows that even very well-informed people can work from very different assumptions about the future, and this may affect their decisionmaking. So I think the effect of uncertainty depends on the situation. In some cases uncertainty means you should be cautious and put off important decisions until the uncertainty clears up.

RF: Are there papers you have been working on recently that you would like to discuss?

Slemrod: I am working on a paper about the effect of public disclosure of income tax returns. The issue is, what would be the impact if there was public disclosure of income tax liability and taxable income, as there is in several countries today and there was in the United States in the 1920s and again in the 1930s? In Norway, for instance, you can go online and see anybody's taxable income, income tax liability, taxable wealth, and wealth tax liability. The people who think this is a good idea argue that it dampens noncompliance, because if your neighbor sees that you have reported $10,000 of income and has reason to think it should be $100,000, he might provide that information to the tax authority. The ongoing research is, as far as I know, the first empirical study on the impact of disclosure.

It uses data from Japan, which had disclosure from 1949 until 2004 for both individuals and corporations. We examine what happened when disclosure ended, and take advantage of the fact that disclosure was required only for people and corporations with taxable income and tax liability over some threshold. So we look at the distribution of taxable income reports and observe that it tightly fits a Pareto distribution until you get very near the threshold, where there are noticeably fewer reports than would be expected under a Pareto distribution. This is completely consistent with the notion that both individuals and corporations near the threshold are understating their income in order to avoid disclosure. Also, in Japan — this is wellknown among accountants, apparently — there were so-called "39 companies," referring to the fact that the disclosure threshold level for disclosure was 40 million yen. These "39 companies" arranged their affairs so they wouldn't have to publicly disclose their income. That's the first half of the paper: that a nontrivial amount of individuals and corporations apparently take actions to avoid disclosure.

In the second part of the paper, we look at whether we can see a disruption in corporations' reporting of taxable income in their financial statements when disclosure ended. Remember, in 2004, everybody could see what your taxable income was, but in 2005 nobody could see, other than Japan's version of the IRS. Did we suddenly see taxable income and tax payments go down, because they didn't feel this pressure of public disclosure? And the answer seems to be that no, we didn't see that. This might be so because in Japan there is a very high degree of conformity between the tax return measure of income and the financial statement measure of income, which means there's already quite a lot of information in the public domain about taxable income for big public companies, so disclosure was not that big a deal — when it ended, there wasn't a big response. But for smaller companies whose income was near the disclosure threshold, which were mostly private companies, the public tax disclosure was the only information out there, so it mattered more for them.

Something else that I have been thinking about lately is what I call "policy notches" — where a very small change in behavior can lead to a large change in tax liability. A good example is fuel economy policy. The "Gas Guzzler" tax is notched. So when the fuel economy of a car (but not a truck or SUV) changes from 16.4 to 16.5 miles per gallon, there's a several-hundred dollar reduction in the tax, throughout the whole range of the tax. The same is true in the Canadian system. Do automobile manufacturers respond to those notches? Do they make sure that the fuel economy measure on which the tax is based is just over the notch to get the lower tax?

A former graduate student of mine, Jim Sallee, who's now at the Harris School of Public Policy Studies at Chicago, and I have written a paper called "Car Notches" that reports pretty convincing evidence that car manufacturers are well aware of these notches and are re-engineering their cars to take advantage of it. Unfortunately, a notched policy like this is inefficient because it induces auto manufacturers to spend a lot of effort and resources re-engineering some of their cars that are near the notch just barely over it and provides no incentive at all to do the same on cars that aren't near the notch. That just isn't an efficient way to encourage fuel economy.

RF: Which economists have been most influential in shaping your research agenda and your thinking about economic policy issues?

Slemrod: I was incredibly lucky to go to graduate school at Harvard at a time when some of the really great contributors to my field were on the faculty. My advisor was Marty Feldstein, who was teaching one half of the two-semester public finance sequence. Richard Musgrave, who had written probably the most influential book on public finance while he was teaching at Michigan in the late 1950s, about 15 years before I got to graduate school, taught the other half. Marty was a leader in the new public finance, taking seriously rigorous normative models of optimal taxation and applying frontier empirical methods to estimating the impact of taxation on behavior, and he had a tremendous influence on how I think about research. But I was tremendously influenced by Dick Musgrave and his views about the importance of the normative issues that inform economic models. The most stimulating economic intellectual experience I have ever had was the weekly public finance seminar at Harvard, when — metaphorically speaking — Marty would sit in one corner of the room, Dick Musgrave would sit in the other corner, and we graduate students, who at the time included tremendously smart people like Larry Summers and Alan Auerbach, would be in the middle. Marty and Dick would — generally very respectfully, but always forcefully — give their often contrary perspectives on the issues of the day. That experience was very important and formative for me.

My first job as an assistant professor was at the University of Minnesota. I never became a rational expectations guy like many others there, but one thing I respected tremendously about the faculty there was that they took economics very seriously. It wasn't a game. It was important, and it needed to be rigorously based, whether they were talking about theory or empirical work; I hope I picked up some of that seriousness. I also have had long-time colleagues and collaborators who have been incredibly important to me. One is Shlomo Yitzhaki, an Israeli economist who I worked with continually for 20 years on one project or another, and who convinced me that aspects of taxation that were then at the periphery of the standard models, such as avoidance, evasion, and enforcement, were actually central to the economics of taxation. Another big influence on my research career was Roger Gordon, who was my colleague at Michigan and was the reason I came here. The serious and eclectic intellectual environment in the economics department at the University of Michigan, and the great graduate students here, keep me stimulated and motivated.