Economics and Finance Symposium - The Evolution of Economic Science: Individual and Firm Behavior

POTERBA: Good morning. As many of you may guess, I am not, as your program suggests, Dean Deborah Fitzgerald. She will be joining us and welcoming all of you this afternoon as we start into the second half of our program today. But as Jim Poterba and one of the organizers of the events that we will be going through in the next day and a half, I want to thank you all for braving the snow this morning. I want to thank David and the members of the MIT 150 committee for selecting the Economics Department in the Sloan School to sponsor this symposium. And I would like to thank both the dean of the Sloan School and the dean of Humanities, Arts, and Social Sciences for agreeing to support this exciting adventure.

When MIT was founded in 1861, the field of political economy was already taking shape as a well-defined academic discipline. The leading text was John Stuart Mill's Principles of Political Economy, which had been published in 1848. And although in the very early years, economics was not yet on the radar screen of MIT, it arrived soon in the history of the young institution.

In 1880, when Francis Amasa Walker was recruited from Yale to come and serve as the third president of MIT, that was a rather unusual thing for an institution of science and technology because Walker was, actually, one of, along with Irving Fisher at Yale, the most distinguished economists in the United States at that point. And Walker would go on to serve as the first president of the American Economic Association for a period of six years in the 1880s. He'd be the president of the American Statistical Association for a span of 17 years in the 1880s and 1890s. And he introduced the first required undergraduate course in economics, or political economy, here at MIT in the 1880s. So the history of economics and the role of economics in MIT'S fundamental fabric goes way back before the origins of the PhD program that MIT created in the early 1940s.

I thought as one way of getting a perspective to introduce and set the stage for this 150-year perspective on economics at MIT, it might be useful to hear a little bit about some of the questions that emerged in the early Francis Amasa Walker course, so that showed up on the exams that we have from that era, and to give you a sense of how economics has changed or not changed as the case may be. In 1895, the exam included the question how are the debts between nations settled? What determines the value of gold? A question that certainly seems to have relevance today. In 1896, the year when the bimetalism controversy was raging, how and when may mortgage indebtedness press upon the farmer, and compare the effects of such indebtedness in different parts of the United States? The role of mortgages in the US seems to have been an issue again in the more recent past.

And to link to tomorrow's finance symposium, I found this fascinating. The 1896 exam asks illustrate by an example the use of an option by a manufacturer to protect himself against the speculative changes in the price of his raw materials. It actually uses the phrase option in what shows up on the exam.

Walker passed away in 1897. Davis R. Dewey, who served as editor of the American Economic Review for over 30 years and was another key figure in the history of the MIT Economics Department, continued to teach the political economy course. In 1898, he asked, for example, what arguments have been advanced in favor of a tax on inheritances? Again, a question that nearly 110 years later, we don't quite seem to have settled.

Although some issues have changed. In 1899, the question, what are the principal economic advantages of employing children in factories does seem to have [LAUGHTER] moved off the stage. Demonstrating that economists then as now have two hands and have difficulty reaching conclusions, in May of 1901, the exam asked, mention six arguments in favor of the imposition of a protective tariff. In the fall 1901, the next semester, the course asked, what arguments have been advanced against protective tariff duties? We don't have an answer sheet to these exams, unfortunately. We just have the questions.

So the history of MIT, as I said, is intertwined with the role of economics. The course then was called Political Economy and Industrial History. And it was basically trying to offer the students an opportunity to learn about the behavior of firms, the role of firms in the economy, and, of course, the role of consumers who are the demanders of those products. And that will provide a focal point for us as we organize our first session this morning.

The quick thumbnail sketch going forward from the turn of the 20th century, of course, is probably well known. In 1940, MIT introduced a Master's program in economics moving from a service department offering the undergraduates an exposure to economics to a department that was then focused more on research and graduate training. A year later, introduced a PhD program. Lawrence Klein, who taught for many years at the University of Pennsylvania, was the first graduate student went on to win the Nobel Prize-- not a bad start for a fledgling PhD program.

And the rest, in some sense, is history. In the post-war period, MIT, with a band of loyal and dedicated faculty, developed a PhD program which truly transformed the teaching of economics and made MIT a center of research activity in economics that attracted students and faculty from around the world, and has really been a central part of the post-war economics policy theory and educational mission in the broader economics profession. And that in, many ways, is what we are here today to celebrate-- the contributions that MIT as an institution, MIT students, and MIT's faculty have made to the advance of economic science, economic policy, and the design of economic institutions.

Today's program is broken into four components. We will begin this morning with a focus on the evolution of economic theory, both at the micro household level and firm level as well as on the understanding of macroeconomics. This afternoon, we'll switch gears to a somewhat more applied focus on macro policy and on regulatory policy. And we have an all-star cast of speakers who will be sharing their thoughts on a variety of these issues with us as we work our way through the day.

What I would like to do is to basically introduce our speakers for this morning's panel and then simply to allow them to take charge and to handle their presentations. We have, by the way, set up some microphones in the audience. And we will have time for some questions and answers at the end of the presentation today, so that I hope we'll be able to draw some audience participation. Let me just introduce our panelists as my final task this morning

George Akerlof, who will lead off, is the Koshland Professor of Economics at the University of California at Berkeley. He's a 1966 PhD from the Economics Department here at MIT. His work in information economics and more broadly in macroeconomics and behavioral economics has cast an incredibly important influence on the evolution of economic thinking in the last few decades. His paper on the market for lemons has been a perennial favorite both with undergraduates and has had a central role in teaching all of us how to think about the research frontier in information economics. He received the Nobel Prize in 2001, shared the Nobel Prize for his work on the theory of information economics. And he will actually be talking about the role of norms in economics with some illustrations from the field of [INAUDIBLE]

Avinash Dixit from Princeton University, the John J. F. Sherrerd University Professor, Emeritus, will be our second speaker. He received his PhD here at MIT in 1968. He is a past president of both the Econometric Society and the American Economic Association. He is a modeler par none who has an unbelievable capacity to distill a range of economic problems to their essence, to construct models which illustrate them, and to enlighten the broader profession and students about the workings of international trade, of growth theory, of a number of different disciplines. He will actually speak about the role of MIT's small models approach to doing economic theory and the influence that it has had on a number of different fields and on the evolution of our disciplines.

Jerry Housman, the John and Jennie MacDonald Professor of Economics here at MIT, will be our third speaker. Jerry is an applied microeconomist and an econometrician who received the John Bates Clark medal in the early 1980s for his work in bringing microeconometric tools to bear on a roll of different questions, taxes and labor supply, telecommunications, the role of transfer programs in effecting household behavior. Jerry has blazed a trail which has taught a generation or two generations of economists how to use the theory that we learn in our theory courses and the econometrics and statistical methods that we learn in our econometrics courses, to bring them together to enlighten our understanding of household and firm behavior. And he will talk about the question, what is econometrics good for?

Finally, the last speaker this morning is a slightly complicated combination. Oliver Williamson, the Edgar Kaiser Professor of Business Economics and Law at Berkeley is our scheduled speaker, but unfortunately could not join us as a consequence of the snow and travel difficulties. Now he's at University of California at Berkeley. He has sent us his prepared remarks, which focus on the role of economics in organizations. Professor Williamson is a 1955 graduate of the Sloan School as an undergraduate. And he received and shared the Nobel Prize in 2009 for his work on the economics of organizations in understanding what happens within firms and other sorts of organizations.

We have prepared remarks from him, but I am delighted to report that my colleague Paul Joskow, who is the Elizabeth and James Killian Professor of Economics, Emeritus, who was a member of the MIT faculty from 1972 until his moving to emeritus status late last year, and is now the president of the Alfred P. Sloan Foundation in New York City-- Sloan, of course, another MIT graduate from the class of 1895. But Paul, who is an expert on institutional and organizational economics, as well as an industrial organization, and who served as the president of the International Society for New Institutional Economics, has agreed to join us this morning. He will present the prepared remarks that Professor Williamson has sent us. And he will also elaborate on those with his own comments and then participate in the discussion that moves forward.

So without further ado, let me welcome all of you to today's proceedings, our panelists. And I'll turn things over to Georgia Akerlof. George?

[APPLAUSE]

AKERLOF: Thank you, Jim. It's tremendously nice to be back at MIT. It's always like coming home to be here. And thank you very much everybody here in the audience for coming, and for everybody else.

So I'm going to speak about the evolution of macroeconomics. And the analysis is going to be based on a book with Rachel Kranton whose title is Identity Economics. So that book is about the role of norms in economics. Now, actually, economists don't talk about norms very much. What is a norm? And norm is things that people think they should or should not do.

And, actually, if you just think about norms, they're tremendously powerful. I think everybody in this room, what we're doing at this very moment, I think a lot of what we're doing is we're here because we think this is something we should do for some reason or other and depending upon what our role is and who we are.

So the book describes a general procedure for bringing norms into economic thinking by including them in the utility function. And then it gives four examples where norms make a very big difference, a very big difference to economics. So those examples are the economics of education, the economics of organizations, the economics of minority poverty, and of gender discrimination.

But then there's another chapter to the book. This is one that we sort of sneaked in where we talk about the role of norms in the economics profession itself. And there we see the norms of how an economist should behave as playing a key role in determining the economics we have. In turn of course, then, this is important for the whole world, for everybody, because our economics plays a key role in the economic system itself.

So let me go back to an old article, which is important. With this preface, let me characterize the norms for how economics should be done. This is how economists think we should do economics. The source for this is Milton Friedman's article on economic methodology. It was written some 50 years ago. And now 50 years later, this article-- which I think was a little bit radical. I think people didn't believe it at this time. This article remarkably well describes how most economists think that economics should be done.

So Friedman says that we should start with a model, a mathematical model, with a null hypothesis, he said, of perfect competition. Now 50 years later, that norm, I think, for the null hypothesis has been broadened. And I think most economists would now say that one should begin with a model where people have only economic motivations.

And then there's been a little bit of an addition to that because we have behavioral economics. And I think behavioral economists say that they'll also allow people to make some kind of cognitive error. Beginning with such an all hypothesis, then, is the first norm for how an economist should proceed in the analysis of any problem. So that's the first norm. And that's in Friedman.

The next norm concerns the use of statistics. This is in Friedman, definitely. Starting with such a rational economic model, the analysis should continue to be the null unless rejected by statistical methods. That means that non-economic motivation is inappropriate. It's inappropriate unless one has statistically rejected all variants of explanation with purely economic motivation.

So what happens, it affects the economics we have. And then it's going to affect, actually, the world we have in so far as economics plays a role. And I think it does play a role. Because such norms privilege economic motivation and therefore they play a major role in the economics we have.

But then further, the institutions of the economics profession itself, especially for acceptance and rejection of economic articles, play a further role. So I'm going to talk a little bit about those institutions of economics.

So those institutions and economics are centered around our journals. Now articles at the journals are refereed by those who have previously published in the same area. So this means that those who accept or those who reject articles have a bias. And this then creates a feedback loop, a feedback loop between those who get their articles published in one generation and the output of the next generation.

So this then poses two questions. First, the norms of economics may be a bit broader than the norms proposed by Milton Friedman 50 years ago, but they're not that much broader. So the question is, do these norms cause economics to evolve toward descriptions of reality? Or do they prevent it from such an evolution? Second, does the system for acceptance or rejection through journal articles lead to evolution toward the truth? Or does it prevent the system from correcting its errors?

So the feedback loop suggests that norms of good economics could potentially get into a stable equilibrium even if much better descriptions of economic reality were possible. It's not only the journal system that determines whether this will happen. It also depends on the interaction. It depends on the interaction with the subject matter itself.

So if, as in physics, models make very specific predictions and empirical tests are powerful in rejecting wrong hypotheses, then these biases will make no difference. But in economics, I don't see that as being the case. In economics, the models make only very general predictions and therefore the power of our tests is extremely low.

So let me just give you one example. So Larry Summers gave us an example of an auto regressive system for stock prices where prices would deviate from the efficient market hypothesis by 30% more than 2/3 of the time. So he gave some kind of hypothesis as a mathematical formula, and the system behaves that way.

Then he asks, how much data would it take for you to refute that system? Stocks are quite a bit away from what the efficient markets hypothesis would suggest. And what he found was it took more than 5,000 years worth of monthly data to reject it with probability one-half. Unfortunately, we don't have 5,000 years worth of monthly data for stock prices. As far as I know, we have no more than 150 years or so.

Now, furthermore, the remarkable thing is this is a case where the model's very precise. The efficient markets hypothesis gives us one of the most strongest possible predictions as to what the model should look like. Usually, it's harder still to reject statistically the rational economics because there are many different models with only economic motivation. Because there are so many models out there, it's a little bit like the trials of Jason and the Argonauts. As soon as you've killed one of those warriors which has come up with their sword who's about to kill you, then that guy dies and you have to kill somebody else. It's really terrible. So as soon as you've rejected one rational model, and new one arises that you have to fight as well.

So then the question is, should we care? We should because without the right economics, we're going to get the wrong economic policy. So I'm going to give you some examples.

Three times in the United States in the last 125 years, we've had major, major downturns. The first was in the vast recession of the 1890s. The second was the worldwide depression of the 1930s. And now, as I speak, we're facing a very deep and a very robust downturn.

And economics gave us the wrong model here because it failed to predict it. So I see this as a tell-tale, a tell-tale that the system's not generating the right economics. And there's a reason to believe this. We can go look at what happened in detail. And we can see why we should believe this.

In 1936, after recession and depression had hit England for more than a decade, John Maynard Keynes wrote the General Theory. It's a long, well-written book, even if here and there it's slightly hard to read. But then John Hicks boiled it down to three easy equations in a journal article.

To remind you what those equations are, they're first the IS and LM equation, which together are the equations for aggregate demand. And then there's a separate equation for aggregate supply.

So you have these three equations. And they are the basis for all macroeconomic models today. But the reduction of the general theory to a journal article spawned a major tragedy, a major tragedy for which today we're paying a great price.

The General Theory did describe Hicks's three equations, but it did much more. The General Theory also described how asset markets worked. And there's a whole chapter, there's a whole chapter there, a wonderful chapter, on stock market irrational exuberance. And it describes in words the origins of this irrational exuberance and also its consequences.

That irrational exuberance explains why we get booms and busts and especially the major sorts of booms and busts that lead to economic calamities such as we had in the 1930s and such as we are experiencing today. This part of the General Theory was not based on standard economic motivation, and so it got lost in translation.

Thus the current crisis and the failure of economists to predict it serves as a potential tell-tale. It's a tell-tale that economics is not systematically getting right answers. If that tell-tale is informative, then we should see answers to other questions that are also wrong.

So now I'm going to give you two other examples. All you need to do to see that is to go to the macroeconomic textbooks, the standard macroeconomics textbooks. So here's an example from the very best of them. It's a wonderful book and very, very well-written and also very convincing.

So it introduces the consumption function with a line from Keynes that says, as a psychological rule-- law-- as a rule, and on the average, as people's income goes up their consumption goes up but by not as much. So if you get an extra dollar's worth of income, you tend to spend some fraction of it.

So we can imagine the sophomores reading this, nodding their heads silently as they read it. And they're saying something like this to themselves. Yeah, when I was 13, my mother got $1,000 bonus. And we went off to Martha's Vineyard. And we stayed at the Fisherman's Lodge. And I stepped on a jellyfish. at the beach. And it's wonderful Keynes understood that even though he didn't know my mother. And he didn't know about that $1,000. He's a really cool guy, that Keynes.

But then that chapter explains that people have other motivations. Yeah, it's not just that. They tend to smooth out consumption as they save for their retirement. And so we can imagine the sophomores still nodding their heads. Yeah, that's right, too.

And then that chapter explains that formal economic reasoning would have predicted that in the beginning. And if we'd use that formal economic reasoning, we would have had a more sophisticated view. And that's how we should proceed.

What we would have done is we would have applied the Friedman-Fischer graph and seen the consumption depends only on lifetime wealth given the rate of interest. And then that's the end of the chapter. So the sophomore says, yeah, I should've used that form of reasoning. And that's what I would have gotten. And I would have been much wiser for it.

But the sophomore is never told-- he's never told-- that this presumed right way of doing economics produces a truly surprising result. This is the result. Given his level of wealth, a consumer's purchases will not change by a single dime as the time path of that income changes. Will not change by a single dime as income goes up and his income goes down given that the total wealth is constant. Says that the rational consumer does not consume more at the time she receives a large bonus check.

That, curiously, is also a result that's contradicted by the data. It also is contrary to what got the freshman to nod his or her head at the very beginning of the chapter. So the freshman got sucked into this, was pulled through this line of reasoning, and now believe something that's truly radical-- not knowing it, either.

I can tell a very similar tale regarding the treatment of the acceleration of the Phillips curve that gets the student to accept the view not just that people try to adjust for inflation in so far as they can, but much more radically that nominal variables corrected for expectations have no effect-- literally no effect-- on how the system works at all. And so that's, I think, seen in almost all macro textbooks. And it is, I think, something that almost every economist believes. It has very important consequences. But yet, when you actually put it the other way, do people have any effect-- do nominal variables have any effect? No. Most people would say no, I think, if most people don't have formal economics training. So that takes us to our conclusion.

So let's go back to the beginning and the question about the evolution of economics. We have institutions of economics and also norms for what's good economics and what is bad economics. But curiously, no one has ever done the analysis for our own institutions and our own norms that we say we should follow as good economists everywhere else. We've never done, for economics, what we've done, what we do for everybody else. No one has ever done an empirical check to see that these are the right norms and the right institutions for an economics that really serves the public. But I've given here at least three examples where we've gotten wrong answers.

In one case, regarding the causes for economic fluctuations, we could have prevented the current world recession. We also saw that the institutions of economics systematically weed out the parts of the analysis that rely on what's called non-economic motivation. For example, it weeded out totally Keynes' explanation for how irrational markets produce boom and bust. And as it did so, we got an economics that gave no appropriate place for macro prudential regulation of asset markets, which we should have had.

So let me conclude. In conclusion, there is an evolution of economics. And there are biases. And it's not clear that it evolves into an economics that systematically gives us the right answers. We should be questioning our journal system for acceptance and rejection of articles. In physics and chemistry, that journal system is really very different. Assistant professors do not have to wait in those fields years to get their article rejected simply because the referees have decided that they do not like it. It's much more easier in these fields to get your article published. And furthermore, it's not going to take you years.

And we should also be questioning-- furthermore, we should be questioning our methodology. We should be questioning our methodology of that dis-privileges non-economic motivations. And it also just dis-privileges careful detailed observation in the place of statistical work, in the place of statistical work that usually has very low power and therefore has a very hard time rejecting a null, even if that null might happen not to be true. Okay, that's all. Thank you.

[APPLAUSE]

DIXIT: It's a real pleasure and privilege to help celebrate MIT's 150th anniversary. Thank you so much, Jim and all the other organizers, for involving me in it. The Economics Department is where I learned my craft of economics. I learned it not just from a galaxy of stellar teachers, but from many fellow students, many of whom went on to join that galaxy. And you will see George is, of course, a prime example. You'll see many others today and tomorrow.

These people were not just brilliant. They're also very collegial. I remember the hospitality of professors at their homes and Friday lunchtime bridge games in the student lounge just as much as I remember classes and workshops. And I'm sure we learned just as much from these social settings as we did in the more formal ones.

I want to do something unusual. Rather than talk about frontiers of research, I want to look back and talk about one specific way in which the MIT department has been instrumental in many of these frontier research developments. I want to talk about what I learned, basically. And it was not any kind of doctrine or ideology. The MIT economics department was never a school of thought in the sense that, let's say, Chicago or the London School of Economics was. It was more a school of research style. Many of its members learned, developed, refined an art of economic modeling that's come to influence a very large proportion of economists the world over in the way they think, write, and do research.

Economic models come in many varieties. Just to contrast two, the very, very general, overarching models of economic systems, general equilibrium where in principle you allow zillions or maybe even an infinite number of goods and services, and try and characterize properties of equilibrium involving demands and supplies for all of these. The problems of devising good social mechanisms for collective decision making, possibility or impossibility of them. These kind of models are very general, very abstract. And the art of doing them is to find the right mathematical framework in which to pause and analyze the question.

Then there are models that are tailored to analyzing very specific issues. For example, what's the relative importance of capital accumulation and technological progress in economic growth, which was a key theme of Bob Solow's researches. The art of these kind of models is to distill the central question that's being asked into small number of special assumptions, setting aside everything else to focus in on that one. And MIT largely, I believe, specialized in the second kind.

That's an art. One can learn science systematically. Art is best acquired by demonstration, by practice. And so let me explain what I mean by giving just a few examples that are my favorites among MIT style of modeling.

Of course, very appropriately, my first choice is a model from Paul Samuelson. It's his exact consumption loan model of interest. The question he asked was, when each person wants to work and save when young, and retire and consume the principle and interest accumulated on the savings when old, how will the interaction between all the people of this kind work out?

So he constructed a model in which there is an overlapping set of generations. Each person lives for two periods, works and saves in the first period of life, retires and consumes the principal and interest on the previous savings in the second period. There is no way of storing goods across time. So you can just store your own output. And there is no durable capital machinery or things of that kind. So you can't, when you're young, build a machine and then consume the output of that when you are old.

Soon as you hear this, all the non-economists, I'm sure, and perhaps many economists as well, say, oh, but that's totally unrealistic. What about childhood? Isn't the working life much longer than retirement? That certainly was so in 1958 when Paul wrote the paper and may become so again as retirement age gets longer and longer.

Of course, there's durable capital. There's uncertainty. There's technological progress. There are business cycles-- all kinds of things.

But, I hold, that it was precisely by ignoring all of these other aspects that Samuelson could zero in on the essential new thing that market for borrowing and lending over the life cycles brings, namely, if a young worker today gives some goods to an old retiree who is not working and not producing any goods, there's nothing the retiree can give to the young person by way of goods and services in exchange. And you can't resolve this problem by using any kind of a longer cycle where a gives to b, b gives to c, et cetera. And z gives back to a, even with a longer, but finite lifetimes.

And this difficulty of organizing exchange creates a new kind of market failure. Samuelson then went on to analyze how that kind of a market failure could be overcome by using social contracts so there is an ongoing understanding where the young person who gives up goods getting nothing directly in exchange is then promised goods when he is retired from the then young, and so on, and so on, and so on. It can also be resolved by creating longer-lived assets, what Samuelson called the social contrivance of money.

So the grossly simplified, stripped-down model gave us conceptual understanding of key aspects of asset markets, of money, and of the possibility of market failures. And all of these deep key issues of the question what has been very hard to spot in a complicated model that tried to bring in all the details of the reality of life cycle saving.

The converse is easier. Once you have the conceptual understanding, you can add all kinds of questions, complexities, bells and whistles; take the model to the data; calculate numerical solutions, simulations; and so on, and so on. And, in fact, that has been done with the result that Samuelson's paper counts well over 2,000 citations on Google Scholar.

The model also proved useful in giving a number of other insights. Building on that, new conceptual questions could be posed, answered, and key insights obtained. The one I want to focus on, again, very appropriately talking at MIT, is Peter Diamond's model of national debt, which combined Samuelson's model of saving across generations with Solow's growth model. So the savings of each period's young could go on to finance the physical capital that would combine then with labor of the future young to produce output, some of which the then-retired could get as returned to their previous saving.

Again, this was an absurdly simple model of a complex reality. But the very simplicity allowed us to understand key economic mechanisms that will work, develop new intuitions. It showed how saving decisions of individuals, each of whom has a finite life, can depart from calculations of socially optimal capital investment, and, again, clarifies how inherent inefficiencies in the operation of markets through time can arise.

Then, when government debt was introduced into the model, it highlighted the effect of saving interest rates and long run capital accumulation, which was how Diamond was able to resolve the conflicting views on national debt, which are still being voiced, one that held that we are mortgaging the future generations, the second that says, oh, we owe the debt to ourselves. So it just cancels out on the balance sheet and doesn't matter at all. Diamond showed that both of these were wrong. And, actually, perhaps particularly interestingly and completely contrary to what raw intuition might have led to suggest, turned out that debt owed to our own nationals has a worse effect on capital accumulation than debt owed to foreigners.

Again, this model has also been the source of a large number of extensions, generalizations, numerical calculations, et cetera, et cetera. Peter built this model soon after he left MIT. And that says how the MIT style kind of spreads to other places. And my third example is a good example of that. That comes from Ronald Jones who got his MIT PhD, went on to be on the faculty at Rochester, and many years later produced a model of international trade that I think is an excellent exemplar of the MIT style of modeling that he must have learned from Samuelson, Solow, and others at MIT.

Again, his economy has two words, but a very natural way to distinguish exports from imports, so very fitting for international trade. He compares it, one, when the economy is close to the rest of the world, and second, when it's freely trading with the rest of the world. And in a very elegant system of just four equations, he captures what before that was a long and complicated series of diagrams and verbal arguments.

I could never understand those. I read the one paper by Ron Jones. And it became crystal clear. He clarified the basics of comparative advantage; effects of labor force, growth, and capital accumulation on the pattern of trade; and the effects of changes in world prices on the returns to a country's factor of production, for example, wages.

Recently, Jones was elected as a distinguished fellow of the American Economic Association. And Gene Grossman, yet another MIT PhD, wrote the citation for that that's worth reading. Said Ronald Jones makes the few equations of small-scale general equilibrium models deliver profound insights. He has applied this talent to the [INAUDIBLE] transfer problem, the effect of tariffs, international factor flaws, and the incidence of technological progress. Jones always had just the right set of equations to capture the sense of a resource allocation problem without unnecessary clutter.

Given enough time, I could produce many, many other examples, a few more of Samuelson's own papers, for example, on the stock market and a couple of models of international trade-- Bob Solo's growth model, of course; Paul Krugman's models of international trade and economic geography; George Akerlof's model of market failures with asymmetric information; and many, many more.

Now, some economists who come from other research styles dismissed these as toy models. But we practitioners of the MIT style wear that label "toy models" proudly. The clarity and intuition that these models convey is absolutely invaluable. And I could cite support from other authorities. For example, Einstein said everything should be made as simple as possible, but no simpler. But I'm going to offer two rather unusual sources of outside support.

One is Hannibal Lecter. [LAUGHTER] Remember what he told Clarice Starling. He said, read Marcus Aurelius. The emperor advises simplicity. Of each particular thing, ask, what is it in itself? What is its nature? And once you understand the innate nature of the problem, the rest is detail.

Detail is important. It needs a lot more painstaking research to bring it to fruition. But it's detail.

And my second outside source, if you don't like Hannibal Lecter being held up as an exemplar, is Mozart. Remember what he told the emperor. I use just the notes I need-- neither too many, nor too few. And to me, that's exactly the sense of the MIT style of economic modeling. Thank you very much.

[APPLAUSE]

HAUSMAN: So since Jim Poterba gave a little history, I'll give a little bit, too. I had a grandmother-in-law who lived to be 108 and probably, apart from vacation, never got more than 10 miles out of downtown Boston, or 15 miles. And I was interested, when I came back from graduate school and started teaching here, she always called MIT the Tech. And so, it turned out that was the old name for MIT.

And she and her husband were always curious because she was actually born just about the time that MIT started. She lived to be 108. And she and her husband always regarded MIT as a school that produced engineers. And those engineers went to work for corporations that her class ran in Boston. And she was always curious that we had economics at MIT.

The other thing which I think she actually got right, which I always found funny, she said that she just didn't believe that all these skyscrapers were ever going to make it because she remembered when MIT was in the Back Bay. And she actually remembered when they filled in the Back Bay with mainly dirt from Needham, which they brought in by train.

And when I first came back, she said, you know, they're building this John Hancock Building. It's just never going to work. I can remember when that was all under water.

And I said, no, no. There are all these MIT engineers. And they dig down to the bedrock. And they float the building. And it's all going to work.

Well, of course, for those of you who live around here, you remember all the windows fell out of the John Hancock. And so I never heard the end of that. That's what grandmothers-in-law are good for, I guess. So that was one thing that, to the extent that engineers were involved, they didn't get quite right.

So I'm going to talk today about econometrics. But I'm going to talk about a particular type of econometrics, which is microeconometrics, which as Jim Poterba said, is sort of what I specialized in. So these are models of how firms behave and how people behave, not about how large economies behave.

And I would say the great advance in econometrics of post World War II-- although it happened during World War II, but the advance happened after World War II-- was the notion of bringing equilibrium into econometrics. So I'll explain that.

So I think the most powerful notion in economics, or certainly one of the most powerful notions, is the idea of equilibrium. So I think most people have an intuitive understanding, which, of course, was discussed by philosophers for years, about cause and effect. Something happens. That's a cause. And because of that cause, there's an effect.

But the idea of equilibrium in economics, which is perhaps not as intuitive to people-- one of the reasons you probably want to study economics-- is you don't have a cause and effect because in equilibrium, something changes, you might think that's a cause. That has an effect. But then that effect has a feedback, which changes the cause.

So prices go up some for a commodity. So the government in its infinite wisdom decides to subsidize ethanol. So the price of corn goes up. When the price of corn goes up, people buy less commodities with corn. And that then has a feedback through getting farmers to plant more corn. And then it subsequently affects the price of corn.

So this notion of equilibrium is an extremely powerful notion. And, of course, Avinash Dixit just talked about that Ron Jones model. I should say that Avinash taught me my first course in economics when I was a graduate student at Oxford. And he said at that time that Ron Jones, that was a great paper. And I'm pleased to see that he hasn't changed his mind in the meantime, since I read the paper very carefully. So that's almost 40 years ago. And it's still a great paper.

So how did econometrics take this into account? Well, econometrics, to a large extent, uses what are called regression models. A regression model is you have a variable you'd like to explain. So the example I'm going to use for a second are house prices. So you want to explain house prices in some suburb of Boston like Lexington, or maybe across suburbs. And you use a regression model to try to figure out how to explain the price.

So what would you take into account? Well, you'd take into account the lot size of the house, the square feet of the house, the number of bathrooms. If you were looking across suburbs, you'd probably look at tax rates and how good the schools are and various things like that.

So that type of regression model was known probably and used from 1910 or 1920 in England. So those models were known. And, actually, regression was known by Gauss, which goes back to probably the 17th century. And that just says, we want to explain some variable like the housing prices. And we're going to use the characteristics of the houses to do it.

That was well known. And, in fact, it was done in England even before computers. So one of the great lines in economics-- this was done at Cambridge University-- and I don't know whether any of you ever seen in the science museum here, but they had these calculators in which you would program in the numbers because you had to invert matrices pretty much by hand. And these old Smith's Marchant calculators would go thunk, thunk, thunk, thunk, thunk as they were doing it.

But anyway, the guy that ran that who did win the Nobel Prize was Sir Richard Stone. And that's always called the Stone age of econometrics in Cambridge because you had all these people who had run these. And then it had to be double and triple checked to make sure it was right because it was quite easy to make a mistake. But, of course, in the '50s, computers came in. And that really gave econometrics a big boost.

So why am I going to talk about microeconomics and econometrics and not macroeconometrics? Well, I'm going to argue that microeconometrics works pretty well. And I'll give you a couple examples where it's led to stuff, which I think has had a big effect on economic policy in people's lives.

Macroeconometrics, a lot was done here at MIT before I came and about the time I came. That's had a much more difficult road. I'm not saying that we haven't learned anything, but the main problem with macroeconomics or econometrics is, I tell my students, is we essentially have only one economy to study. And so things tend to move together. You can look, of course, internationally, other economies, but then that raises other difficulties. But it's very difficult to take quarterly data. George Akerlof mentioned that that's what we often use. And the economy is not all that stable. And you can't use 150 years of data. In fact, you might not want to use more than 10 or 15 years of data. And that makes progress very difficult.

Well, what about microeconometrics? I talked about firms and individuals. Well, starting in the 1970s or 1980s, very large surveys started to be collected. And oftentimes these surveys follow people over time.

So one of the first ones was done at the University of Wisconsin. And when you go into microeconometrics, it's not out of the question-- I mean, the minimum number of observations you'll typically have is about 1,000. But it's not out of the question that you'll have 10,000 or 15,000 observations. And these individuals are often acting in pretty much independent ways. So that's really a lot of data that you can piece things out. And you compare that to 40 observations, if you had 10 years of quarterly data on the macro economy.

And in terms of firms what happened are companies like AC Nielsen and IR-- IRI, in a sense, was partly started at MIT-- but they go and collect scanner data from stores so we know what they're selling, or even better, they now give people wands and so we know everything that a given household will buy. They pass a wand over the UPC codes, which I think were invented at MIT. And anyway, by cellular telephone, it then goes to a database. So we'll have an observation of how 5,000 American households-- they're followed for two years, not only watching TV, but this is everything they essentially buy. So we can do a lot of research on that. And that's led to a lot of advances.

So I'd like to give two examples where this idea of equilibrium comes into play, or the cause and effect is not straightforward, the contrast that to the regression example that I gave before. So I think-- economists, of course, disagree about a lot of things. But I think that one thing that most economists agree on-- and it was also in the President's State of the Union address-- is that the returns to education are quite high.