Nobel lectures in economics can be something of an anticlimax. It takes thirty or forty years to win the prize, often in collision with some prevailing orthodoxy. Eventually comes a day of exhilaration, followed by two months of polishing a paper whose kernel may have been tucked away in a drawer, just in case — and then thirty-five minutes of talk in the Great Hall of the University of Stockholm, enforced by a card-wielding disciplinarian.
An hour is time enough to do a distinguished job. Thomas Schelling, of the University of Maryland, gave a brilliant lecture in 45 minutes a few years ago. But when two laureates are required to condense their message into half an hour apiece, the results can be disappointing – the talks, not the messages.
That’s what happened last week to Thomas Sargent, of New York University, and Christopher Sims, of Princeton University. Both are researchers who introduced new kinds of mathematical methods to formulate old questions about cause and effect in economics. But both are committed to proposition that, since their formal methods evolved from natural language, the conclusions they generate can be translated back into relatively straightforward terms. Both are preparing lucid papers. But the talks they gave about them were cut short.
Sargent compared the creation of the fiscal system of the United States in the years between 1780 and 1840 – “time zero” in the language of economic modeling, he joked – with the beginnings of the European Monetary Union after 1990. He then contrasted the US government’s assumption of the original thirteen states’ debts after the War for Independence with its refusal to assume – to “bail out” — the states’ obligations after the collapse of canal and railroad spending sprees in the 1830s, and very lightly drew some parallels to the crisis of 2008.
Sims talked about the evolution over the years of Jan Tinbergen’s original 1938 attempt to build “a mathematical testing ground for theories of the business cycle” with Trygvie Haavelmo’s 1942 criticism of that attempt as insufficiently probablistic (Tinbergen was the co-recipient of the first Nobel prize, in 1969, Haavelmo the twenty-first) with a view to showing that, especially with respect to modeling the behavior of policy makers, scientific progress in economics “is not as pretty as natural science; it’s a little messy.”
Neither lecture worked quite the way the laureates intended. Not wanting to be bad guests, both speakers hurried the pace of their presentation rather than disregard their prompter. (As noted. their presentations are online, you can judge for yourself. Both papers will be published in the American Economic Review next June. Sargent’s, in particular, would make a good little book, along the lines of his two earlier essays, The Big Problem of Small Change and The Conquest of American Inflation.
Given the spirit in which the prize in economic sciences was established – the need not just to fathom the mysteries of political economy, but to convey that understanding to the lay public – the authorities might allocate a little more time to the lectures and a little less to the intense socializing that determines the calendar of the busy Stockholm week.
A good thing, then, that we have Economic Principals’ Translate Thomas Sargent Contest. When he accepted the CME-MSRI Prize in Innovative Quantitative Applications last fall, Sargent summed up the vision of the world on which he worked this way:
The assumption that people share common beliefs about underlying sources of uncertainty [that is, that expectations are rational] underpins influential doctrines in modern economics, including (1) how the distinction between expected and unexpected government actions affect inflation-unemployment dynamics; (2) how to cast optimal fiscal and monetary policy as a dynamic mechanism designed to cope with enforcement and information limits; (3) how timing protocols that capture a government’s ability to commit give rise to optimal fiscal and monetary policies that are time inconsistent; and (4) how reputation can substitute for commitment.
Because central banks and treasuries want to implement solutions of optimal policy problems like (2) in contexts like (1), in which the distinction between foreseen and unforeseen policy actions is important, a time consistency problem like( 3) arises, prompting them to focus on ways like (4) to sustain good expectations. By showing how to modify the common beliefs assumption [the presumption that everyone’s expectations are the same] in ways that take account of both adaptive learning and of economic agents’ response to statistical model uncertainty, economic research has also set down foundations for better models in the future.
EP challenged readers to come up with translations and/or illustrations of Sargent’s description of the human predicament. And several tried. Ken Voytek wrote: “Fool me once, shame on you; fool me twice, shame on me.” David du Plessis wrote: “Economics, because it is wrapped around people, is not an exact science. Our hopes (and consequent actions) are more often not about interest rates but about how we want to see ourselves. Governments should nurture our hopes.” And John Albin wrote: “1. Pay no attention to the man behind the curtain. 2. Pay attention to the man behind the curtain.”
But the realm of strategic interaction routinely overpowers aphorism. The winner therefore is this entry by David Kreps, of Stanford University. His subject is tax amnesties. But notice that his reasoning could apply equally to forgiveness, assumptions, and bailouts of all sorts — including the kinds of the kinds of decisions taken in the formative years of the United States that were the subject of Sargent’s Nobel lecture.
Tax Amnesties and Modern Macroeconomics
Governments, in an attempt to raise funds, impose taxes. Some individuals and firms, in an attempt to avoid being the source of those funds, cheat. They cheat by failing to report income, conducting transactions in cash, and so forth. Governments, to deal with tax cheats, engage in costly endeavors such as auditing. And, sometimes, governments try to recoup taxes they never collected by proclaiming a tax-evasion amnesty.
The idea is simple: The government proclaims to all and sundry: “If you have evaded taxes, come forward now. We won’t punish you, at least not too much.” Certainly, the confessed evader must pay the taxes he or she evaded. And, in addition, interest charges may be imposed, and perhaps a small fine. But criminal prosecution is taken off the table; coming forward and confessing is made relatively acceptable option, attracting a lot of otherwise lost revenue.
Given their frequency and popularity, governments believe that these amnesties are a good thing, flushing out tax cheats and swelling government coffers with tax revenues that would otherwise never be collected. But, in a Wall Street Journal column dated August 7, 1986, Robert Barro and Alan Stockman raised an objection: When individuals realize that tax amnesties will be offered periodically, they can reason that cheating on their taxes might not be such a bad thing: Cheat today, and confess at the next amnesty, or confess at the next amnesty if you feel the hot breath of the tax authorities on the back of your neck. So, Barro and Stockman reason, tax amnesties may actually lower tax compliance, by convincing folks—perhaps in particular folks who are momentarily short of funds but anticipate better days ahead—that another amnesty is always just around the corner.
Moreover, tax cheats, seeing an amnesty in progress, may reason that the government will flush out some tax cheats with this amnesty, but others will remain unflushed. So, once this amnesty is finished, it will be in the government’s interests to offer another amnesty, on even more favorable terms, to flush out those who didn’t come forward in amnesty mark 1. This sort of inference—the government will do tomorrow what is in its best interests tomorrow—may mean less coming-forward in this amnesty. (In the microeconomic context of a monopolist selling a durable good, Ronald Coase’s famous conjecture is that the monopolist will be unable to get anyone to buy at the monopoly price, since, because everyone anticipates that, having made all possible sales at the highest price, the monopolists will then lower his price, to capture those who didn’t come forward – so no one buys at the higher price.)
Of course the government doesn’t want individuals to draw these conclusions. So tax amnesties are frequently accompanied with loud proclamations. “This is it. Last chance. We’re never doing this again.” But while tax cheats may be vile, that doesn’t mean they are stupid. They can do the math. They realize that after the amnesty is over the government will have every incentive to offer another amnesty. The very success of an amnesty, if it is successful, leads to the assessment that it will be offered again. In other words, talk about “last chances” is cheap; incentives after the fact speak louder, and lead individuals to conclude that the government proclamations have no credibility.
Barro and Stockman suggest an ingenious twist on this. Have the government proclaim a tax amnesty. See who comes forward. And then tell those who came forward “Hah!, we didn’t mean it. You are going to prison (or suffer whatever criminal consequences the courts will impose).” (A possible flaw in this scheme is that it depends on a judge to go along with the scam.) This has a wonderful side effect: no individual would ever trust a government-run amnesty again, and everyone knows this. So everyone will recognize that there is no point for the government to try an amnesty in the future. So everyone concludes that there will be no amnesty, and if they evade the tax authorities today, they will never be given a “get out of jail free” card. On the margin, at least, this should increase compliance.
This story (and, by the way, if you can’t find the original and quite wonderful WSJ column, it is reprinted in Eric Rasmussen’s Readings in Game Theory, Wiley, 2001) captures, in a very specific context, what Sargent was describing. To the extent that certain government (and central bank) actions can be anticipated, smart people in the economy will anticipate them and act accordingly. And, it is a matter of doctrine that the proportion of people in the economy who are smart in this sense is larger than you might think, especially when you weight more heavily those individuals whose economic activities have larger impact.
In particular, people anticipate that the government will do next year what is in the best interests of the government, next year. Disclaimers issued today that “We promise not to do X” are not credible, when everyone knows that X will be really beneficial next year. The government, then, has to operate under the constraints this imposes; it must craft its policies with this sort of thing in mind. In some (limited set of) cases, reputation can serve to improve credibility; “we won’t do X next year” is credible if (a) doing X means the government won’t be trusted in other matters and (b) the trust of its citizens in these other matters is more important, going forward, then the short-run benefits (next year) of doing X. But (a) and (b) are real stipulations, not always easy to meet; reputation is not a cure-all by any means.
This scheme is at work in inflation/unemployment dynamics, just as it is in the more specific context of tax amnesties. And this is what Sargent and Sims and their colleagues (prominent among whom, of course, are Barro and Stockman) have brought to the study of the macroeconomy, and for which they are highly deserving laureates.
Ruizimus among the Austrians, EP’s weekly on the misleading parallelism between the careers of Friedrich von Hayek and John Maynard Keynes, occasioned a certain amount of controversy. Readers who took an interest may wish to sample the 43 comments that it elicited, many of them of very high quality.