Interview with Edward C. Prescott

When Prescott was named a Regents' professor at the University of Minnesota this summer—an honor reserved for just 20 active professors at any time—it was not only a recognition of his contributions to economics research, but also his dedication to the classroom. Prescott enjoys working with his students, and this commitment has inspired devotion in many of them: On the occasion of his 50th birthday, nearly all of his former doctoral students attended the celebration, many traveling from Europe. Also, foreign exchange students—with considerably more knowledge about soccer than Prescott—have volunteered in the past to help their economics professor coach youth soccer teams.

Prescott has been at the university since 1981, the same year he joined the Federal Reserve Bank of Minneapolis as a consultant in the Research Department. According to Arthur Rolnick, the Minneapolis Fed's director of research, Prescott has great influence within the profession and has made important contributions on many issues. "Economists pay attention to what Prescott has to say," says Rolnick, who conducted the following interview.

For example, in a 1986 paper in the Minneapolis Fed's Quarterly Review, Prescott described the state of knowledge about real business cycles, a line of research to which he and economist Finn E. Kydland had previously contributed pioneering work. Kydland and Prescott argued that technology shocks drive the economic system and that costly policies aimed at stabilizing these shocks are counterproductive. In that same issue, Lawrence Summers, then professor at Harvard University and currently deputy secretary of the Treasury, responded with no uncertain skepticism: "Let me follow Prescott in being blunt. My view is that real business cycle models of the type urged on us by Prescott have nothing to do with the business cycle phenomena observed in the United States or other capitalist economies." In the same issue, Prescott then followed Summers with a response of his own, concluding: "My guess is that the reason for skepticism is not the methods used, but rather the unexpected nature of the findings."

For more on this issue—and for a discussion of economic growth, the wealth of nations and other topics—read on.

The following interview was conducted by Arthur J. Rolnick, senior vice president and director of research at the Minneapolis Fed.

ROLNICK: Ed, over the years you have worked with some of the best economists in the world. Which ones have influenced your research the most?

PRESCOTT: My thinking has been heavily influenced by Bob Lucas. I met him in 1964, which is pretty far back. He was a new professor and I a new graduate student at Carnegie-Mellon. As a graduate student, however, he did not influence my thinking much. His impact on my economic agenda began a few years after graduate school—beginning about 1969—when we collaborated on the paper, "Investment Under Uncertainty." That collaboration along with his seminal paper, "Expectations and the Neutrality of Money," forced me to rethink completely how to do macroeconomics. So, when I returned to Carnegie-Mellon in 1971, this time as a colleague of Bob Lucas, I decided not to teach macro until I understood what it was and how to teach it. As a result, I never did teach macroeconomics at Carnegie-Mellon. I only resumed teaching this subject when I moved to Minnesota in the spring of 1981.

At that time, the foundations of dynamic economic theory had been laid, and the task was to figure out how to use this theory to quantitatively study economic fluctuations. At that time there was a theory of secular growth and this theory was being used to address public finance issues. Virtually everyone, including myself, was certain that this theory, without money introduced, would not be useful for studying economic fluctuations and assumed that some other theory was needed to account for the fluctuations about the secularly increasing trend. Finn Kydland and I developed practical methods for deriving the quantitative implications of growth theory for fluctuations. We thought that with some extensions, in particular longer time requirements to build new factories and office buildings, that growth theory might provide the mechanism for the propagation of monetary shocks. When we derived the implications of growth theory without any monetary shocks, however, we found to our surprise that business cycle fluctuations are just what that theory predicted.

These were exciting times, particularly here at the university and the Minneapolis Federal Reserve bank, where much of the progress in this research program was made. What we were doing was controversial. There was a taboo then against doing quantitative theoretical exercise in macroeconomics. This is no longer the case. Even my undergraduates at the university run their business cycle experiments using model economies with people deciding how much to work and consume, and firms deciding how much to produce and invest in dynamic environments characterized by uncertainty.

ROLNICK: Is this what the profession calls "micro foundation of macroeconomics?"

PRESCOTT: That's what some have labeled it, but I think the labeling is antiquated. Let me explain why. In the '60s the profession, including myself, thought Keynesian macroeconomics was a great theory that could be used to eliminate costly economic fluctuations. Naturally the profession tried to develop the theoretical foundations for the equations that constituted the Keynesian macroeconometric models. But, Bob Lucas established that this is impossible. The assumptions underlying the macroeconometric models were logically inconsistent with dynamic economic theory. This conceptual problem, along with the failure of these models in the '70s, resulted in the demise of Keynesian macroeconomics. With this demise, again there is just economic theory.

ROLNICK: Over 30 years ago, Milton Friedman made the argument that we must consider the economy as a whole. He warned the profession to beware of conclusions based on partial equilibrium analysis. Why has it taken this long for macroeconomists to pay heed to his warning?

PRESCOTT: There have been breakthroughs in the way we model the economy. Dynamic economic theory was not very far advanced 30 years ago. In addition there were neither the computers to compute the equilibria of dynamic economies with uncertainty, nor the rich data sets to restrict the models economies constructed.

ROLNICK: So are you saying that economists now have the tools that allow us to solve dynamic general equilibrium models —tools that we did not have 30 years ago?

PRESCOTT: Yes, but even today, Friedman and many others of his generation do not think in terms of dynamic equilibrium.

ROLNICK: But in fairness to Friedman, he didn't have the tools.

PRESCOTT: That's true.

ROLNICK: Who else has influenced your work?

PRESCOTT: Another person that influenced my thinking a lot is Robert Townsend. He was a graduate of the University of Minnesota. He came to Carnegie-Mellon, around 1976, as an assistant professor. He introduced me to some of the ideas at Minnesota—in particular to mechanism-design theory, which can be used to study economic situations where there are contracting problems. This is proving to be increasingly useful in constructing models with financial intermediation. Such models are needed to understand issues such as bank regulation. The Minneapolis Fed has been, and is, a leader in this important development.

Let me elaborate further. In much of economics, we abstract and study idealized worlds where people can write any type of contract and where people honor their contracts. In fact, for some issues, this does not provide a reasonable abstraction—in particular issues concerned with financial intermediation. There are moral hazard and adverse selection problems. Borrowers from intermediaries, for example, do not always take the promised action and may be less than honest as to the riskiness of the projects they want to fund. There are events which result in borrowers not being able to honor their contracts. Introducing features such as these into applied general equilibrium is proving difficult but progress is being made. Only with their introduction can we assess whether there is a market failure and if there is one, what intervention will mitigate it.

ROLNICK: Is there anybody else you want to add to your list of people that have influenced your views on economics?

PRESCOTT: There are two other people for whom I have incredible respect for making economics better. These two people have been associated with the Minneapolis Federal Reserve bank—Tom Sargent and Neil Wallace. They along with Lucas were the ones responsible for the use of dynamic economic theory to study macroeconomic phenomena. I, of course, was influenced by the writings of Arrow, Debreu and McKenzie, the great general equilibrium theorists, but they being pure theorists were not concerned with macroeconomics.

ROLNICK: With regards to your views on the research programs of Tom Sargent and Neil Wallace, I'd like you to go back in your career to when you started at the University of Minnesota in 1980-81, and your affiliation with the Minneapolis Fed's Research Department. How did these programs, along with your own, evolve and influence each other? What were the biggest changes?

PRESCOTT: Under Tom Sargent's leadership, in the mid- and late-'70s, the Minneapolis Fed and the university were at the lead in the rational expectations revolution. I am sure that rational expectations would have eventually dominated, as it has, but Tom made it happen much sooner than it would otherwise have. Tom, however, did not change my views. I was already convinced when I came here.

When I came here, the University of Minnesota had the only graduate program that trained students in the use of dynamic economic theory in the study of macroeconomic phenomena. Neil Wallace and Tom Sargent were responsible for this. I like working with students, and having students trained in dynamic theory made Minnesota a very attractive place for me. The Minnesota graduates in macroeconomics have had a major influence upon economics and many are now in positions of leadership in the other top economics departments.

When I came here, Neil Wallace was leading a program to lay theoretical foundations for monetary economics. In the late '70s this was a controversial research program, with which I was sympathetic. Neil clearly had a major but indirect influence upon my thinking. His research program of laying theoretical foundations to monetary economics was controversial, as was Sargent's research program. It implicitly said that there was no hope for the old, largely empirical approach to money. The old research program was to use balance sheets to construct some monetary aggregates and then to examine the statistical relation between these monetary aggregates and real output. The weakness of that research program was the lack of theoretical mechanisms. You always worry or hesitate to rely on just empiricism when you can't run experiments.

ROLNICK: When you say you worry about just empiricism, are you worried about drawing implications from empiricism about causality?

PRESCOTT: Yes. With the monetarist research program, the appropriate monetary aggregate was the one that was highly correlated with output and led the cycle. That one was assumed to cause output fluctuations with a "long and variable" lag. I want a mechanism by which these monetary aggregates have consequences for output. At a deeper level, all that we can hope for is a theory of the equilibrium process of the economy given the rule by which the money supply or interest rates are selected now and will be selected in the future. We can say this monetary policy shock did this or that. But, we can't say what the effect will be of a particular policy action. All that dynamic economic theory can do is to predict how the economy will behave given the rules by which policy will be selected in the future.

As I said, the problem in monetary theory is the mechanism by which money has real consequences. In the 1890s and 1900s, which is prior to the creation of the Fed, there were financial crises followed by steep declines and rapid recoveries in real output and employment. Subsequent to the creation of the Fed, the economy has not been characterized by that type of crisis, or at least not as frequently.

ROLNICK: The Great Depression aside?

PRESCOTT: The Great Depression is another matter. I also was hedging my statements a little bit because of the 1937 experience. In that year there was a large, sharp decline in output and a very rapid recovery that was associated with a temporary large increase in the reserves that banks were required to hold. And then there was a similar but much less dramatic event in the first half of 1980.

I think the evidence strongly indicates that Fed actions can cause sharp declines by taking actions that result in a credit crisis or by failing to take an action that would avoid a credit crisis. Before the creation of the Fed there were a number of financial crises with each being followed by a large decline in output and a rapid recovery.

ROLNICK:We have talked about how people have influenced your thinking and research agenda, but your work in turn has been very influential. I'd like to turn to some questions about your own work, partly to get you to expound on the nature of that work and partly to get you to respond to some of your critics. It was the fall of '86 that we devoted an issue of our Quarterly Review to your research on real business cycles—research that has attracted considerable attention from both the media and academia. While much of that attention has been laudatory, some has been critical.

You are associated with the view that technology shocks, those affecting production opportunities, are the major driving force behind business cycles, accounting, according to your estimates, for about 70 percent of these fluctuations. How would you respond to the following observations: There seems to be overwhelming evidence that the Great Depression and many of the major recessions in U.S. history had monetary origins and that monetary factors were very important in determining the depth of the downturns as well as their length. How does that observation square with your 70 percent estimate? Second, it is very hard to identify specific technology shocks that have coincided with economic fluctuations.

PRESCOTT: First, the 70 percent number is for the post-World War II period. This is a period for which we have pretty good measures of the inputs—in particular the labor input—and for that matter, quarterly measures of output. Such data are needed to carry out such an analysis. I suspect that the estimate would be different for the pre-World War I period if the needed data were available to carry out the accounting exercise.

With regard to the Great Depression, I have not studied it in detail. I don't know of anybody well-versed in modern dynamic economic theory who has looked at it carefully. You have to be a pretty good historian as well as a good theorist to do this task. My own view is that printing a little bit more money would not have been the solution. Third World countries have tried without success to avoid depressions by printing money. With regard to why there was a Great Depression in the United States in the '30s, I just haven't seen the evidence that points the finger to the monetary side. Canada had the persistent and big decline in output—every bit as big as the United States—but not the bank failures.

ROLNICK: And Canada experienced a much smaller decline in its money supply. Of course, some would argue that Canada's economy was so closely tied to the U.S. economy that the decline of ours spilled over to theirs. In other words, Canada's experience over this period does not represent an independent experiment. The emphasis you put on intermediation may some day provide an answer. There was a significant upheaval in financial intermediation during the Great Depression. Something real was lost—some information was lost and that possibly could have turned a typical recession into a depression.

PRESCOTT: My own speculation would have something to do with the unstable political times. I have looked at the time series of factor income shares. Labor's share of total income or, equivalently, of total output, was more or less constant up to the beginning of the Great Depression, at which time it jumped about 5 percent. Subsequently it has remained at that higher level. This strongly suggests there was a big change in the rules of the economic game that occurred just after the beginning of the Great Depression. Incidentally, there was a large drop in total factor productivity, which matches well with the technology shock story. I emphasize that technology shocks are changes in the total productivity of the business sector. The nature of the rules of the game that govern business activities have consequences for total factor productivity.

The early '30s were not politically stable times. Even in peaceful, lovable Minneapolis there were some serious conflicts. Teamsters struck and shut down businesses. The police not being able to handle the situation deputized a citizen army. In subsequent conflicts between the truck drivers and the police, many people were injured and a few killed. In neighboring states farmers were setting up private armies to block the export of agricultural products in the hope of raising the price of agricultural products. These were politically unstable times.

ROLNICK: So, that's what you mean by some political instability?

PRESCOTT: There also were movements throughout the world that must have worried people, namely the rise of fascism and of communism. In economic terms fascist Italy and communist Russia were perceived as doing well. The fear that such systems would be adopted in the U.S. must have deterred investment. The fear of changes in the rules of the game can have large real consequences.

ROLNICK: Part of your response to your critics, therefore, is that your 70 percent estimate is for the post-World War II period. What about the recession of 1980-81?

PRESCOTT: Here I've looked at the data carefully. Central to the business cycle is how many hours are worked per adult in the market sector. This time-series started a large decline in the fourth quarter of 1979, which is significantly before the 1980-81 recession, and bottomed out in the third quarter of '82. The 1980-81 recession was a small blip down, associated with the temporary institution of certain reserve requirements. But, the recession had started before the tightening of monetary policy and the decline is consistent with technology shocks being the factor.

There were probably some disruptions due to uncertainty. Some people bet wrong about interest rates. There were a lot of bankruptcies during that period. Mortgage markets did not function as well as they normally do. Some of these factors probably contributed. I only claim about 70 percent of business cycle fluctuations are induced by technology shocks. That leaves 30 percent for other factors.

ROLNICK: So you concede that monetary factors can explain some real fluctuations in the economy?

PRESCOTT: Yes. The question is how much.

ROLNICK: Second observation: It seems difficult to point to specific technological factors that led to each of the several recessions and recoveries during the post-World War II period. Yet, economists who emphasize monetary shocks can easily point to many recessions and many recoveries that seem to be caused by specific monetary shocks. How would you respond?

PRESCOTT: First, the finding is that recessions or economic fluctuations are the result of the sum of many random causes. We don't have a theory of what causes economywide productivity to change. We can measure how big the changes are, and we can use dynamic theory to predict the consequences of these random changes. Now, the question is: Can we identify specific shocks? My answer is no. We can't even identify why today total productivity of labor and capital is four or five times higher here than in India. Given this, how can we hope to identify why this productivity grew by 2 percent less than expected over some two-year period? Such an occurrence is all that is needed to induce a recession.

Let me try to explain better what I mean by business cycles being the sum of random causes. If you take a coin and flip it repeatedly, assign a one to heads and a minus one to tails and sum up the values of the last 15 flips, the resulting time-series will display cyclical fluctuations, that is fluctuations that look a lot like cycles. This is Eugene Slutsky's (Russian economist and econometrician) observation in his famous 1927 paper: that cycles can be the sums of random causes—a translation of which appears in the 1937 Econometrica. With real business cycle models, a given shock's effect declines only 5 percent per quarter. This implies after three and a half years, the shock still has half its effect. The current state of the business cycle depends upon things that have happened over the last four years.

Returning to your question of what are the factors leading to technology change, new knowledge is only one factor. I think from the point of business cycles, changes in our legal and regulatory system are the big factor. Some of these changes are good because they permit an industry to develop, which in turn increases the business sector productivity. Other changes may be good even though these reduce productivity because they lead to a smaller externality—say less pollution or more safety—but, they are still negative shocks. Others may be bad because they result in resources being allocated to unproductive type activities. Having a theory of these shocks is tantamount to having a theory of international income differences.

ROLNICK: The other point your critics have raised is related to measurement errors in estimates of productivity. A major piece of evidence that you cite in favor of your view of technology shocks is that labor productivity is procyclical. But your critics note that measurement error is biased in your favor. Consider the following example: Sales services provided by a sales clerk in a store that is open eight hours a day go up or down depending on how many shoppers show up, if the sales clerk's recorded hours are more or less the same, all of the increased sales activity shows up as movements in productivity, even though there is no change in technological opportunities. Further, empirically possible factors like monopolistic competition, increasing the terms to external economies and overtime labor can explain the observed degree of pro-cyclicality in labor productivity while attributing much less importance to technology shocks. In other words, how do we know this evidence you cite on productivity is in fact evidence in favor of technology shocks?

PRESCOTT: Yes, it is possible that due to mismeasurement of the labor input, productivity may not be as pro-cyclical as measured, or it may be more pro-cyclical than measured. One way to challenge our finding is to introduce some feature of reality from which we abstracted and establish that the answer changes. Some researchers, including Finn Kydland (Carnegie-Mellon) and I, have introduced features such as a variable workweek length for the production units, monopolistic competition and factors that result in idle capacity in equilibrium. The estimate has stood up to these challenges. Researchers are restricted as to how they introduce these features by theory and by micro observations as well as by macro observations. Probably the key macro observation is the small correlation of the cyclical components of the labor input and labor productivity. There is one study that introduced a feature that gave rise to labor hoarding in equilibrium. This study found that firms must commit their employment levels for ridiculously long periods for this feature to matter. At the time these challenges were articulated by Larry Summers in a 1986 Quarterly Review paper, the studies I refer to had not been carried out. In responding to his challenge a lot was learned.

ROLNICK: So I take it that you are not swayed by the labor hoarding explanation?

PRESCOTT: When times get hard, there are a lot of people out there who would take a job. Workers worry about keeping their jobs. I suspect, if anything, they work harder and are absent less in bad economic times.

Let me review some history. Economists, since the late '30s or maybe earlier, have been trying to make the case that there is labor hoarding in bad times. This is because they were bothered by the fact that when output was low, labor productivity was low rather than high, as standard production theories predict. They rationalized this observation by hypothesizing an error in measuring the labor input. In the ensuing years no measures of the amount of labor hoarding were provided that show how it varies with the cycle.

There is a problem with the labor hoarding story on the theoretical side as well. Theoretically, labor hoarding results in response to temporary and not to permanent shocks. Technology shocks that induce business cycles are highly persistent. Cyclical variation in the intensity with which workers work hasn't panned out.

ROLNICK:Other research of yours that has received a lot of attention and has intrigued me personally is your joint work with Finn Kydland that appeared in the Journal of Political Economy in 1977, titled "Rules Rather Than Discretion: The Inconsistency of Optimum Plans." It has been cited by the Economist as one of the 10 modern classics in economics. What is the thesis of that work and what implications do you think it has for the Federal Reserve in our search for an optimal monetary policy?

PRESCOTT: Finn Kydland and I began working on the optimal policy problem within the framework of dynamic equilibrium theory. Once we worked out the logic of it, we found that there was a fundamental problem. The problem in a nutshell is as follows: The best policy plan has the property that after following the plan for a while, everybody agrees that there is a better alternative then to continuing the original plan. But, if a group of people cannot commit to its original plan, the ex ante best plan is not feasible.

The problem comes up in the design of the legal system. Suppose the best policy is to punish people who take certain actions and that punishment is costly to society. The best plan, which is time inconsistent, is to punish people who take the action in the future. But the time consistent solution is not to punish in the second period of the plan those who took the action in the first, as costly punishment will not deter actions already taken. The solution to this problem is to have rules or laws and to follow them. The same logic holds for fiscal policy. It is always best to tax the returns on past investments but not on new investments. But with the passage of time, the present becomes the past. This is why constitutional provisions that make it possible for the government to commit to not taking property without full compensation are desirable. A big problem facing many Third World countries is their limited ability to so commit.

One problem is that sometimes the environment changes and the old rule no longer works well, which necessitates a change in the rule. But such changes should be made only after a lot of discussion with past commitments being honored as much as practical. Credibility of policy permits much better outcomes to be achieved.

A secondary implication of this is that the Fed should not try to trick people. It should stick to rules, and I think implicitly it has over a considerable period of time now, since the Volcker/Greenspan era. There has been some important subsequent work in this area, some of which is done here at the Minneapolis Federal Reserve bank by V. V. Chari and Patrick Kehoe on sustainable plans. Using the language of modern mechanism of design, they examined the entire set of policy plans that are sustainable, that is plans with the property that people will not want to change the plan subsequent to instituting it. And then they looked at which plan is the best within that set.

An independent Fed, I think, is something that is a valuable commitment technology for Congress, for the same reason an independent judiciary is a good arrangement.

ROLNICK: Let me go to your third area of influence on the profession. That's your more recent work—you touched on this earlier but I'd like you to elaborate a little bit more. In the spring of '93 we published your joint work with Steve Parente (Northeastern University) in our Quarterly Review. This is your work on understanding changes in wealth of nations. This work has also received quite a bit of attention and there's been a considerable amount of additional work since the '93 piece. In your mind, what have we learned from this work, and again, what policy implications, if any, have you uncovered?

PRESCOTT: I think the question, "Why isn't the whole world rich?" is the most important question facing economists. I think we've learned that just accumulating more capital—that is more machines, factories, and roads—is not sufficient to become rich. Accumulating more human capital, as well, is not sufficient either. Both are important and essential but, given the economywide productivity parameter, these factors of production will be accumulated. As I see it, what we need is a theory of this parameter, and I expect that the rules of the game a country sets up will account for the big difference in this number across countries.

ROLNICK: That parameter being ...

PRESCOTT: The parameter is total factor productivity. In rich countries the same quantities of the inputs produce more output. This parameter summarizes the technology level of a country.

ROLNICK: Is this what is left over after you take account of infrastructure and human capital?

PRESCOTT: Yes, differences in this parameter is what is left over after accounting for differences in all the inputs.

ROLNICK: And you're saying that "Why isn't the whole world rich?" is the paramount question for economists?

PRESCOTT: Yes, I think that is the question. We may not be any more successful this time around than the development economists were in the '50s. I hope we are.

ROLNICK: Are you saying that answering this question is more important than, say, finding better policies for smoothing the business cycle?

PRESCOTT: Yes, I think we found out that these fluctuations are not costly to society, and that the response of the economy to these shocks is approximately optimal. What we should be worrying about is increasing the average rate of increase in economywide productivity and not smoothing business cycle fluctuations.

ROLNICK: I understand that your son Ned has recently been hired by the Research Department at the Richmond Fed. What type of research is he doing and does he ask you for advice on his research?

PRESCOTT: One thing he's made a point of throughout his life was never to ask his father for advice. I offered it for free. We never discussed economics and he was annoyed when in college he found that he liked doing economics. He is doing some research at the Richmond Fed that appears quite interesting in the area of theory of financial intermediation and theory of the firm.

ROLNICK: We should let the readers know that Ned was a graduate of the University of Chicago, studied under Robert Townsend and graduated in 1995. Let me finish with one last question that a lot of people wanted me to ask. I have been told, mostly by you, that you were a pretty good football player in college and that your bowling average was close to 170. Some say you are no longer the athlete that you once were. Do you accept that view or do you think you still have a few good years left?

PRESCOTT: Well, I just made a big decision: I'm going to start playing golf.

ROLNICK: In other words, no. Thank you, Ed.

More About Edward Prescott In June 1996 Prescott was named a Regents? Professor at the University of Minnesota, which is the highest honor bestowed on a member of the university faculty; he joined the University?s Department of Economics in 1980. He is also a senior consultant in the Research Department at the Federal Reserve Bank of Minneapolis, which he also joined in 1980. Walras-Pareto Lecturer: Barriers to Riches, University of Lausanne, 1994. Fellow, American Academy of Arts and Science, 1992. Prescott has held visiting professorships at the Kellogg Graduate School of Management, Northwestern University, the university of Chicago and the Norwegian School of Business and Economics. He has also held positions at Carnegie-Mellon University and the University of Pennsylvania. Recent papers include: "Real Returns on Government Debt: A General Equilibrium Quantitative Exploration," with J. Diaz-Gimenez, European Economic Review, forthcoming; "The Discipline of Applied General Equilibrium," with T.J. Kehoe, Economic Theory, June 1995; "Economic Growth and Business Cycles," with T.F. Cooley, Chapter 1 in T.F. Cooley, ed., Frontiers of Business Cycle Research, Princeton University Press, 1995.

Top