September 20, 2011

The Obama-Buffett Siren Call

The president’s new tax plan defies the basic laws of economics.

The topic of tax increases is again very much in the news. President Obama has just announced his intention to put forward yet another tax plan to close tax loopholes and increase the marginal rates on the rich, which he now restricts to persons whose income is $1 million or more a year. In order to lend legitimacy to this campaign, he has chosen to invoke the confessions of Warren Buffett to justify his proposal. After all, if one of the richest of men alive is willing to accept tax increases on the rich, how could any mere millionaire oppose that position?

I have already written at length about why Mr. Buffett should follow the advice oft given to the humble shoemaker by sticking to his last. However successful he is in the world of finance, his skills lie in working within the system, not in figuring out the rules of the game.

Epstein
Illustration by Barbara Kelley

Mr. Buffett and the president are not alone in their determined inability to understand how economic systems work. In the past several weeks, I have been peppered with pointed criticism insisting that any concern with the effect of marginal rates of taxation on high-income earners is overwrought.

As one critic put it to me bluntly over coffee, "do you think that an increase in the marginal tax rate from 35 to 37 percent will change anyone’s behavior?" His next sentence was, in its entirety, "Get real!" The point of this barb was to pinion me on the horns of a dilemma: either concede the truth of the objection or admit a congenital overreaction to small differences that really make no difference at all.

In yet another confrontation, I faced a more personal version of this challenge. "Professor Epstein," I was asked with cool disdain, "have you ever refused to do anything because of an increase in marginal tax rates?" Again the dilemma was real. I could save face by denying that I had succumbed to some base motive. But I then had to concede the liberal point that higher marginal rates had no effect on conduct. Yet if I admitted that taxes help shape my behavior, then I would have showed a smallness of mind that would have disqualified me from having a credible voice in one of the major social debates of our time.

Mr. Buffett and the president do not understand how economic systems work.

Needless to say, I refused to "get real" on the impact of marginal rates and gamely insisted that there were occasions in which changes in marginal tax rates influenced my behavior. But on reflection, the interesting point is that it is exceedingly difficult for me to point to examples in day-to-day life or to locate a specific decision that showed this response to tax changes. A question therefore arises: if the personal accounts are so elusive, why take the position that tax rates do matter, and indeed, have to matter?

The answer starts with basic economic theory: the laws of supply and demand. Supply curves slope upward, so that the higher the price (on the y-axis), the greater the quantity supplied on the x-axis. The relationship is the opposite with demand curves: the lower the price, the higher the demand. The law of supply and demand predicts that the quantity and price of a given good will be set by the place at which these two curves meet or intersect.

The question then is how to measure changes along either curve. The key point for today’s vigorous tax debates is that both curves are continuous, meaning that there are no gaps anywhere along either the supply or demand curve. In most places, both curves are also smooth, which means that there are no kinks along the way. Supply and demand curves have this essential property: whenever the price increases, the supply increases. Whenever the price decreases, the demand increases. The same principles work in reverse for price decreases.

These functions are termed "strictly monotonic," meaning that no matter how small the interval, the change in price of either the good supplied or demanded will elicit some change in response. No interval is too small not to have that effect.

It is at just this point that the critics I mentioned necessarily get the basic economics wrong. They assume that we are always at a point where there will be no reduction in output for any small increase in tax rates. But this point has to be wrong as a matter of first principle, in spite of the anecdotal reports of sharp interlocutors.

To see why, answer this question: what is the likelihood that someone will not buy a Rolls Royce if the price is increased by a penny? It is easy to think of fat cats and to laugh at the suggestion that a single penny could ever make a difference in their decision-making. But "not ever" is not the same as "highly unlikely," because zero is not the same as a very small positive number.

Will increasing marginal tax rates change anyone's behavior? Certainly.

So suppose we are confident that there is no chance that the penny makes a difference. Can we arrive at the same conclusion about the second penny? Yes, but with a little less confidence than with the first penny. If you try this trick 100,000 times, it is quite possible that one or more potential buyers will be driven away by a $1,000 increase in price. Assume that this number is 1 percent of purchasers. At a first approximation assume that the demand for Rolls Royces is linear. It follows that if the increment of any given price increase is a penny, then we divide that overall 1 percent figure by 100,000, so that the odds of losing a given purchaser for the change of a penny is about one in ten million. As there are far fewer than ten million buyers of Rolls Royce cars, in most years we will observe no change at all. But at some point we will come to that point where one person flips over and will decide not to buy the car.

The same point becomes clear if we introduce another key principle of economic theory into the debate: the reservation price, which is defined as the maximum that a given person will pay relative to the market price. The reservation price for a consumer less the market price equals consumer surplus, or the gain to the consumer from the deal. In most instances, the fluctuations in prices are smaller than the difference between the reservation price and the market price, so we observe no change in behavior as a function of a change in price.

It is, however, a sheer delusion to assume that consumer surplus will always exceed any variation in price. By definition, if the market price is one penny more than the reservation price, then the buyer will walk away. If the market price is one cent less, then the buyer stays in the market. Hence, no consumer has to determine the size of the consumer surplus so long as he knows that it is positive after a price change. So most of the time we go about our business in relative bliss: why spend resources to find out one’s reservation price if it is not going to influence any immediate decision?

Yet in a small fraction of cases the consumer will surely make this calculation. For consumer goods that are sold to millions of people, it is clear that the penny change in price could easily influence behavior, especially on people buying products lower in price and higher in volume than Rolls Royces. That is what is observed whenever a person takes a package off the shelf, looks at the price, pauses a second, and then returns it to the shelf.

In a market economy, failed enterprises lose support. With Medicare, it's the opposite.

Indeed, even when a customer buys the product at the higher price, she may consume less of it. About a generation ago, the late sociologist and statistician Hans Zeisel told me a story about some consumer research he did on the sale of cases of bottled soft drinks. He was trying to find out whether price increases influenced behavior, and was met with the objection that no customer was observed taking one bottle out of the package before purchasing the other 23. What Zeisel reported was that their behavior did change at the margin—but subtly. The reduction in consumption with higher prices was measured by the time between purchases, which increased as prices went up. Thus, it is quite possible for changes in response to fly below the radar.

All of this has a profound effect on the tax debate. Does an increase in tax rates from 34 percent to 36 percent make a difference on behavior? To answer this question, we have to know something about the slope of the response curve. If the tax increase reduces the return from a dollar of labor from 66 to 64 percent, we should expect the reduction in revenue from labor to reduce income by about the same percent. The exact number cannot be pinned down in advance. But the one number that is certain to be wrong is zero.

Yet could the effect be as large as three percent? Why not. After all, there are so many ways in which productivity could decline. People may turn down overtime or extra jobs. They could take a bit more time off between jobs. Or they could switch from a higher powered job to a lower powered job and take out the gains in leisure. Or they could retire earlier or take part-time status. The reason why these changes are so hard to track in individual cases, is that they could take so many different forms, and could easily be bundled with changes in other factors that obscure the long-term economic effects of the overall tax increases.

Any decision to prop up Medicare with added tax dollars, as the president is trying to do, will postpone needed reforms to the Medicare program. In a sensible market economy, failed enterprises lose support. With Medicare, the opposite is true. The use of tax increases supports systems that will continue to bleed cash and provide inferior care for years to come. Sooner or later this game has to end. It would be far better if we ended it now, by rejecting the false Obama-Buffett siren.


Richard A. Epstein, Peter and Kirsten Bedford Senior Fellow at the Hoover Institution, Laurence A. Tisch Professor of Law at New York University, and senior lecturer at the University of Chicago, researches and writes on a broad range of constitutional, economic, historical, and philosophical subjects. He has taught administrative law, antitrust law, communications law, constitutional law, corporate law, criminal law, employment discrimination law, environmental law, food and drug law, health law, labor law, Roman law, real estate development and finance, and individual and corporate taxation. His publications cover an equally broad range of topics. His most recent book, published in 2013, is The Classical Liberal Constitution: The Uncertain Quest for Limited Government (2013). He is a past editor of the Journal of Legal Studies (1981–91) and the Journal of Law and Economics (1991–2001).


Letters to the editor

It's Obvious, Isn't It?

I think people look at the tree so closely that they miss the forest. Of course increasing marginal tax rates changes behavior. If the marginal tax rates increase then marginal income decreases. The total money available for consumption, investment, and giving becomes smaller. I will still consume, invest, and give but there will have to be adjustments made in the total amounts for each. Why is it so difficult to see this?

---Richard Brimage

A Fool's Errand

In my opinion, Professor Epstein is quite correct to make the obvious point that, ". . . zero is not the same as a very small positive number."

As you point out, the effect of minute changes in a large economy are exponentially multiplied. Further proof is offered by the fact that, by the government's own statistics, the rate of federal tax revenue to GDP has remained remarkably stable since the end of the Second World War. Thus, during the exorbitantly high income tax periods of the late 1950s (maximum rate over 90 percent) all the way through the lower rates of today (maximum rate of 35 percent), federal revenue remains stable at about 18.8 percent of GDP.

If tax rates did not effect behavior, then we would see this ratio rise and fall as tax rates changed, but it doesn't with any significance. Why? Because changing tax rates, as you point out, changes behavior, which in turn causes the GDP to expand or contract.

Clearly those who pursue tax increases as a means of increasing federal revenue are on a fool's errand.

---Anthony Benvin

Taking Irrational Behavior Into Account

I would argue that much economic modeling relies too much on quantitative data without taking into account psychologically motivated real behaviors. We have recently seen (and continue to see) that financial markets are built on the abstraction called confidence. When confidence erodes, asset prices fall. 'Irrational exuberance' is an emotional construct to which the human species seems inherently prone to exhibit from time to time.

Perhaps it's time that the laws of supply and demand get a rethinking, with allowances made for irrational behaviors.

---Gregory F. Zerovnik

Tax Rate Increases Will Hit Employers the Hardest

I enjoyed Professor Epstein’s explanation of how an increase in marginal tax rates leads to reduced economic activity. It’s a good discussion of a frequently misunderstood topic; however, it understates the extent of the problem. In today’s tough economy, few wage earners would turn down additional work. Many are working fewer hours than they would like or are working at low paying jobs while hoping for something better to come along. Few have satiated themselves to the point that they would “take out the gains in leisure” in response to a slight increase in their marginal tax rates.

While workers might not respond to a slight increase in marginal rates, their employers undoubtedly will. The problem is that when an employer turns down the next opportunity, the employer’s workers are idled. Each employer that responds to an increase in tax rates by foregoing opportunities may idle dozens if not thousands of employees. And, because Obama’s tax hikes would hit employers the hardest, the ill-effects on the economy would be greatly magnified.

---David K. Walser


Letters to the editor may be sent to definingideas@stanford.edu. Editors reserve the right to reject or publish (and edit) letters.