Economics and Finance Symposium - The Evolution of Economic Science: Macroeconomics, Growth and Development

Search transcript...

[MUSIC PLAYING]

ACEMOGLU: Welcome back, everybody. It's a great honor to be here for MIT's 150th to introduce you to the panel on Macroeconomics, Growth, and Development. I think we've had a great start. It's kind of reminded me how lucky I am to have chosen this profession. I think that economics is not only an exciting discipline with a great history of ideas and questions, but it's also amazingly fortunate to have all of these questions more or less open. So it kind of reminds me, seeing all these debates by the luminaries of our field, that we still have a lot of work to do ahead of us, and lots of things are still up for grabs, not only in terms of substantive questions, but in terms of methodology, and how to approach, and how to think about these big questions. And I think the same applies even more to the topics of this panel, which is macroeconomics, economic growth, and development. And I can't think of a better panel to kind of give us perspectives on this.

We're going to get started with Robert Solow, who I don't think needs any introduction. And he's going to give an broad overview of the development of macroeconomics and economic growth over the last 60 or so years. And then Peter Diamond is going to follow up with microfoundations of macroeconomics, in particular emphasis on search. Robert Hall is going to talk about diagnosing more short-term issues, or the long slump, which is somewhere between short and long. And Esther Duflo is going to talk about the where micro and macro meet, and in particular in respect to development economics, which is one of the central applications of the economics discipline, which has made tremendous progress over the last few decades. So without further ado, I will leave it to Bob Solow.

[APPLAUSE]

SOLOW: No, I'm going to speak from here.

[LAUGHTER]

I think better sitting down. Like any good economist, I asked myself what my comparative advantage was here this morning. And I decided it must be related to my extreme age.

[LAUGHTER]

So I thought I would say a few words about how macroeconomics has changed and worked out during a very long period of time. When I took my first course in economics-- which, believe it or not, was in 1940, that's 70 years ago-- there was no such thing as macroeconomics, and certainly no such word as macroeconomics. There were courses in money and banking. But all you learned that had macro implications or scale in money and banking courses was you learned about the quantity theory of money, and you learned about banking crises.

There were also courses in business cycle theory. And they were certainly macroeconomic in intention. But the whole history and volume of business cycle theory had no general framework within which you could combine and compare various ideas about fluctuations. There was, in short, in 1940, no inkling of a theory of aggregate income determination that you could use to sort of standardize stories about fluctuations in aggregate incomes. It's hard for people now to visualize. But this frame of mind, without any macroeconomic notion at all-- what we would recognize as macroeconomic notion-- persisted for a very long time.

If you look at the Burns and Mitchell Measuring Business Cycles book, which was published just after the war, I suppose, 1946 or so, you will see that. And I will tell you a story which illustrates it, and dates only from 1961. In 1961 and '62 were the only period of time in which I ever worked in Washington, at the Council of Economic Advisors. And we-- mostly Jim Tobin, Art Oaken, an I, on this particular issue-- had some notions which were put about, about the evolution of employment, about the nature of the status of the labor market then, in 1961 or '62.

These notions were strongly attacked by Arthur Burns in a bank newsletter in a couple of articles we found rather puzzling. Henry Wallach also found them puzzling. Henry Wallach was a very good economist, former professor at Yale, and a member of the Board of Governors of the Federal Reserve System. And Henry Wallach went to see Arthur Burns, and said to him, and what trajectory of GNP do you imagine to be consistent with what you have said about the evolution of employment? And Burns's reply was, oh, I don't think in GNP terms. So it was simply not part of the vocabulary.

The fact is that macroeconomics in that sense in which I'm talking about it was really born in Keynes's general theory, in 1936, as AC Pagout later acknowledged very generously. And the Hansen Samuelson multiplier accelerator model and Lloyd Metzler's inventory cycle model are the earliest examples of business cycle models within the framework of aggregate income determination, at least aggregate income determination on the demand side.

The word, macroeconomics, I didn't try to date. It's almost certainly due to Ragnar Frisch, and at some later time.

Now, if you leave aside the particular formulations that people think of as, quote, "Keynesian," end quote, which I'm not interested in here, the important characteristics of Keynes's economics for this kind of history, of historical view of the evolution of macro theory, were two things in my mind. First, he wanted to take seriously the possible occurrence of persistent aggregate excess demand or aggregate excess supply.

And secondly, he wanted a macroeconomics that would keep closely in touch with data and would keep closely in touch with events, not just the second-order moments of time series, but actual events. It's no accident that the evolution of that kind of economics went hand in hand with the development of the national income accounts. I think either of those is inconceivable without the other.

Now of course the 1936 book, The General Theory, is all about persistent excess supply. But when the war broke out, a little booklet on how to pay for the war was all about excess demand. By the way, someone once accused Keynes of inconsistency. And his reply was, when the facts change, I change my mind-- what do you do?

[LAUGHTER]

Now, from 1950 to the middle to late 1970s, macroeconomics was all Keynesian in the sense that I have just mentioned. And most controversy and discussion was about accounting for facts, not about points of theory. And macro economists were interested in such things as the measurement of potential output, the numerical value of multipliers, the numerical value of the elasticities of key behavioral relations, especially the components of aggregate demand, like consumer expenditure, business investment expenditure, exports, imports, the demand for money. I want to be very clear about this. The great divide in macroeconomics in those days was between monetarists and Keynesians. But apart from debating tactics and oneupmanship, that whole argument could be carried on in terms of the shape and variability of Hicks's IS and LM curves, plus some information about the supply side and about inflation dynamics. Those were all empirical questions, but they're hard empirical questions, so that there was lots of room for a priori argument and for ideology to be dragged in. And of course they did get dragged in.

There was a sea change in the late 1970s, and that was primarily event-driven. What happened was a period of inflation without visible excess aggregate demand. The word, stagflation, was coined for that. That is to say, there were episodes which could not be understood without talking about supply shocks. All the earlier attention-- except for growth theory, of course-- had focused on demand-side disturbances. So neither the Keynesian nor the monetarist inheritors had any really ready-made analysis to deal with that kind of inflation, driven initially by the oil shocks, of course-- well, oil, and grain as well.

So there was a temporary gap. The gap was filled in fairly soon, but it provided an opening for opponents of this way of doing macroeconomics. Remember, I'm lumping Keynesians and monetarists together. And the opposition to this way of doing macroeconomics got its energy from two sources. One, I think, you can think of as ideology. Think of it as part of the general turn to conservatism that took place then. And the other was to tap into the latent desire of all economists for theoretical neatness. And this is not something I just think, this is something I feel, since I have it as much as anyone else.

So after the 1970s, paying close attention to events got to be a bad thing. After the 1970s, to describe something as ad hoc became a pejorative term, roughly the equivalent of child molestation.

[LAUGHTER]

And this is a very serious matter. And all of the discussion turned to microfoundations, which is what Peter is going to talk about in a minute. Bob Lucas made a very well-known remark, a little bit later, about how it would be desirable to eliminate the very distinction between microeconomics and macroeconomics. There should just be economics. I have to say I think that was a profoundly foolish thing to suggest, mainly because it's impossible. You can't answer the questions macroeconomics is asked to answer by modeling the whole economy as the interaction of tens of millions of households and millions of firms and products, all of them different, all of them idiosyncratic. Even the so-called agent-based computer models that one sees now are not trying to explain events. It would be like trying to design an airplane molecule by molecule.

For that reason, what we mostly got, apart from things like search models that Peter will talk about, which arose out of microeconomic, everyday research, what we got were primarily representative agent-based models where there's a set of consistent preferences that imposes some kind of order and coherence on observed the macroeconomic outcomes.

So I'm a crank about this. And I won't go on about it. Just to say that all that talk about microfoundations in practice turned out to be a word for less plausible macrofoundations for macroeconomics, and was more or less divorced from the need to account for macro events. This is related to the absence of financial intermediation of any realistic type. By the way, nobody covered themselves with glory on that score.

But the logic of representative agent models almost precludes serious autonomous financial intermediation with serious frictions. Now, I want to take one more minute, since I think I was expected to say something about growth theory in the course of this discussion, and I got wound up in what I've just been telling you. So I'll take one or two more minutes, and just say that, so far as the future is concerned-- and this fits in with some of the micro discussion that we had earlier in the morning-- what old growth theory established that the main force driving trends in potential output is total factor productivity. So the natural direction of research would be to endogenize part or all of the evolution of total factor productivity. And a very good, very productive literature went down two paths, one using the accumulation of human capital as the main way of doing that endogenizing, and the other using technological progress, innovation, as the way of doing that. And obviously that has to be an important interaction between those two.

If you ask how much progress has been made along those lines, I think there's some, but not much. No one seems to have developed a lot in the way of stylized facts or stylized near-facts, near-theories along that line. Maybe the truly exogenous element is too large, maybe not. But the thing that has always nagged me about this part of the literature is how hard it is to-- and this is another micro/macro line, and the last remark I want to make-- how hard it is to tell a coherent story that has total factor productivity on one side and a history of concrete innovation on the other. How do we go in either direction from the story that involves stocks of capital, human and physical, and stocks of labor, or flows of labor of various degrees of skill, and some undefined total factor productivity on one side of the line, and the other side of the line is the transistor, the cell phone, the Kindle that will replace us all, and things like that. I think that's work still to be done. And it's going to be a neat combination of qualitative and quantitative. Thanks.

[APPLAUSE]

ACEMOGLU: Thank you. Thank you, Bob. Peter Diamond is going to speak next. And I think he will use the podium, and he will have a presentation.

DIAMOND: Jim Poterba assigned me the topic of the microfoundations of macro. And I, like everyone else in the department, always does what Jim suggests. When I agreed to speak here, my memory flashed back. I was a graduate student at MIT's 100th anniversary. And I actually remember sitting in Kresge listening to Harold macmillan, although I could not, for the life of me, tell you anything he actually said.

[LAUGHTER]

Even though we didn't coordinate-- and this is, I guess, part of Jim's genius-- my talk feeds directly off Bob's in several ways. Growth theory was a hot topic in lots of places, but obviously particularly at MIT, in the '60s. And growth theory was very odd, probably unique at the time, and still very unusual, in that it was both a micro subject and a macro subject. And by subject, I mean the people working on it, some of my identified themselves as macroeconomists, some of them identified themselves as microeconomists. The modeling assumptions, elements like that were the same across this divide. The uses being made of it were not identical.

But it was something that crossed that micro/macro barrier. And what made that work was a lack of a deep contradiction between the standard general equilibrium model that microeconomists grow up with and start thinking from and the basic growth model, which was going to be used for the kinds of macro purposes-- not the short-run stabilization issues that Bob was talking about, but was actually addressing things where the micro assumptions seemed to work for the macro ideas.

Since the '70s, we have seen analysis based on the idea-- and this is the stuff that Bob was criticizing. I'm going to criticize it a bit differently, but we have the same warm view of how good it is-- based on the idea that the same dual use would work for short-run macro for business cycle analysis. And of course Lucas and Prescott are the key generators of this literature. The idea that a market-clearing approach would work well for short-run macro seemed to me deeply implausible then, and after watching it play out over several decades, every bit as implausible now.

If I could actually put Bob Lucas in Bob Solow's camp for a moment, in 2003, Bob Lucas gave an address at a HOPE conference-- that's an acronym, not a subject of the conference-- and he was commenting on-- it's titled "My Keynesian Education--" and he was commenting on the kind of financial crisis we had recently, Latin America, Asia. And he said, the models we know and work well with have nothing to say about this subject. I take that to be an acceptance that event-triggered analyses call for different kinds of modeling.

In the early '70s, there were some-- and I feel strongly Jim Tobin was in that school, and Bob and his remarks tiptoed up to it-- who were opposed to seeking a micro foundation for macro, I think because they thought it would end up following the same basically unproductive market-clearing line. And Bob also was particularly vocal on representative agents.

I had a different view. And I want to talk about that today. And not surprisingly, it's going to connect to search theory. The late '60s saw a rise of a group of economists working to construct a microfoundation for macro out of search models. Different researchers had different starting places. The desire for this came from a sense that standard macro modeling involved micro concerns, but did not incorporate them in the same systematic way with underlying modeling that was standard in micro. And the presumption was that doing microfoundation modeling would give rise to new insights and a better basis for even short-run macro phenomena.

The famous Phelps volume had its eye on getting foundations for the Phillips curve, which related unemployment and inflation. My own starting place in the '60s was rather different, having cut my teeth initially as a general equilibrium theorist as an undergraduate studying with Gérard Debreu. It was a desire to do micro better, to overcome the limitations in relevance coming from the excessive degree of coordination, that the standard general equilibrium model assumed.

I thought that a proper microfoundation for macro would also lead to a better micro, as well as a better macro. There are two key Keynesian ideas that do not fit comfortably in the basic general equilibrium-- and I'll refer to it as GE for shorthand-- model. One is the large importance of current income in thinking about things, exactly what's behind the stimulus logic. And secondly the stickiness of wages, the fact that wages don't adjust quickly, and that somehow that's connected with having a lot more unemployed at some times than at other times.

The challenge, as I saw it, was to have a framework describing the determination of an equilibrium that had room for both of these phenomena. Alas, I can report that we are still working at trying to do this. And while there's a bit of insight into these pieces, not much. And where we have made a lot of progress is on focusing on the labor market as a partial equilibrium piece of thinking about the whole economy.

Bob Solow, in 1964, gave the Wicksell lectures. And the Wicksell lectures were, as I realize right now, a rebuttal to Arthur Burns on why, at the time, aggregate demand stimulation was the right thing to be pursuing. And the idea that there was just a lot of structural unemployment out there, and aggregate demand wouldn't do anything about it, that that view was wrong. And that lecture goes through a lot of discussion of details of the labor market, and does them-- no surprise-- very intelligently, very insightfully, but in a way that was lacking an underlying uniform theoretical underpinning. And that same kind of debate between structural unemployment and the need for aggregate demand stimulation is going on today. And today, the discussion takes a rather different form because search theory has grown up, over the last 40 years, to have a whole bunch of different models to address different parts of what goes on in the labor market once you assume what's going on in the rest of the economy. So it's partial equilibrium.

And the starting place for this is a focus not on the stock of unemployed and the stock of the employed, but labor market flows. And here a 20-year average calculates, from month to month, what fraction of unemployed workers in one month are employed the following month. And the answer is 37%. And what fraction of unemployed workers are actually out of the labor force-- they're not employed and they're not looking for employment-- the following month. And that's 33%. And if you look at the people out of the labor force, and there's of course a vast number of them. Many of them have no interest in jobs. Many of them are not actively looking for jobs, but actually are available for work. And 4% of the "out of the labor force" people, labeled I, for Inactive, in that figure, are directly employed a month later. And 3 and 1/2% are now officially among the unemployed.

And what happens on the employment side, well, roughly 2%, every month, flow from employment to unemployment, and roughly 3% flow every month from employment to outside the labor force. Over that 20-year period, roughly six million workers moved into employment each month, and roughly the same number moved out of employment each month. And the change in unemployment was based on the small difference between the large gross flows into and out of unemployment. And the thrust of the modeling and the empirical work that has gone with it is to say, let's focus on the flows and assume the lessons from the stocks will be derived by arithmetic from having looked at the flows.

Now, the US is a bit of an outlier in the size of the flows. But the flows are large in lots of other countries as well.

Now, I've given you the picture from the worker side. There's also a great deal of work in a wonderful new data set called JOLTS, the analytical work led by Steve Davis and John Haltwanger, which looks at flows again from the employer side. So you interview employers, and this is here quarterly, not monthly, and you say, do you have more employees now than you did last quarter? And relative to employment, the increase in employment where that happens is 7 and 1/2% of employment. And then you look at other firms, and they have fewer employees than they had. And so that's labeled job destruction. And you see it's almost the same size. The decrease in employment in firms having decreases and the increase in employment in firms having increases is very close to each other. And the difference between them is what's happening with total employment. Again, the flows are huge. The change in stocks is small.

If you're looking at the full pattern, hiring is about twice as large as job creation. A worker leaves, gets replaced, that's additional hires. And if you look at workers having left an employer, job destruction is part of that. But quits is a big part of that. So you see quits and layoffs are about the same size, on average. Although in the course of a business cycle, quits and layoffs play very, very differently.

So these large flows, and modeling the flows, and modeling how they feed into wage determination is critical for two kinds of questions. One question is we set up unemployment insurance programs. They work at what I might call ordinary times. And they play a key role in times like now, with a big recession, high unemployment, where sometimes automatically and sometimes by explicit new legislation, we get extended benefits, we get changes in benefit levels.

And so the question is, how do you want to think about those? First of all, you recognize you have policies that are in place in good times and bad times. Marginal tax rates is another example. Where in good times we are focused on the impact of inefficiency, you go into a recession, they're a big part of the built-in stabilizer that hold down the size of a recession. So how do we want to think about that? And a key part that came out of formally modeling search is the idea that this process is not inherently efficient. Unlike standard modeling of markets in micro, where demand, supply, the market clears, and things are efficient. People searching for each other-- workers looking for jobs, employers looking for workers-- aren't in contact with each other. Their behavior effects each other. And so there are externalities, there are inefficiencies. And you can't start thinking about the role of unemployment without thinking about that. And you can't start thinking about the role of unemployment without thinking about the feedback onto aggregate demand, the piece that isn't in the literature.

So we're still a long way from getting to where some people wanted to start, with the Phillips curve. Search is being built into otherwise conventional macro models, looking at things like sticky prices, and wages, and how they affect the economy. But it's a microfoundation piece, not a full-blown microfoundation model. And I look forward to and hope for a big expansion in the range of microfoundation models that will be consistent with the general equilibrium views of thinking about the whole economy, and the Keynesian view of recognizing the fact that we get, as Bob said, events that really affect things. And how the economy responds to that is critical. Thank you.

[APPLAUSE]

ACEMOGLU: Thank you very much, Peter. And the next speaker is Robert Hall, who will talk about the long slump.

HALL: Okay, well, it's a great pleasure to be here to recognize this important event in MIT's history. MIT was very important to me personally, first as a graduate student in the mid-1960s, where my classmates have all become just a hugely important generation, including Avinash Dixit and George Akerlof. Avinash and I shared an office, and I learned an enormous amount from him.

In addition, MIT is important to my family in that my older sister was an undergraduate here graduating in 1963. How many women were in the class of 1963? I think it's, you know, two hands' worth. And she went on to and continues an interesting career, most famous in the family for the fact that she worked for the guy who put the @ sign in email addresses. And my kids think that this is absolutely legendary. More recently, my daughter received a PhD in economics from the economics department, six years ago, in 2005.

Now, the master of this event, Jim Poterba, assigned me to talk about something which he knew that I was thinking about. It was not a random choice. And that is to present a diagnosis of what's happened in the US economy. I'll take the US economy as the factual base for my discussion. But similar things happened in quite a number of other advanced countries, and are continuing to happen. So you can think of this as more of a worldwide analysis than just focusing on data from the US.

So let me give you just a quick account of the message of this analysis and this diagnosis. There were three factors that built up in the US economy in the past decade. The first was a tremendous splurge of spending for housing and consumer durables, which left, coming into the crisis and slump period, an overhang of durables-- houses and cars in particular-- that meant that we didn't need to build so many afterwards. And that overhang problem is very much still with us.

The second was-- and closely related to the first-- was an overhang of high consumer commitments to pay off debt. It's not just that debt existed, but as we went into the crisis, lenders forced consumers to repay their debt, and that had a huge effect on the decline in consumption that occurred in the contraction, and it's still weighing heavily on and preventing a brisk expansion. And the final thing is the weakness of financial institutions that occurred as a result of the financial crisis itself in the fall of 2008 resulted in a big increase in financial frictions.

Now all of this interacted with another factor, which is that we quickly moved to an economy that didn't have the normal ability of an economy to absorb a shock through its interest rate. We'd had a period of low interest rates. The Fed responded to the crisis almost immediately, pushing interest rates down to zero-- the interest rate that it controls, the Fed funds rate. But that wasn't nearly enough to bring the economy back to life. And this problem of relatively little headroom in monetary policy-- because when you have low interest rates to begin with, the stimulus you can get is strictly limited by the fact that interest rates can't be negative. And that gave us the slump in the slump can be characterized most efficiently in terms of data, just as a big, rapid increase in unemployment that started in 2007, but greatly increased in 2008 especially toward the end of 2008.

And we have this very long period-- the dashed line on the right there shows the normal level of unemployment. Current forecasts don't have us getting back to normal until 2015. So we still have a long way to go according to expert forecasters. I don't forecast. One of the reasons is that Paul Samuelson, who taught me so many things, taught me, if you have to forecast, forecast often.

[LAUGHTER]

And there is a very deep wisdom in that remark, which says that forecasting, if you get into it, is always a full-time business. And if you want to be an idea person, a research person, you shouldn't also be a forecaster.

So this condemnation of my beleaguered profession of being a macroeconomist for having failed to forecast the slump, well of course we didn't. We don't forecast. We don't even do that.

So I think the real test of our performance as a profession is not our forecasting, but rather are we doing a good job of understanding what happened, and had we done research before that helps us understand? And my reading as a biased-- I have to say that I'm biased, of course-- as a biased reader of that is that we actually did a lot of research that turned out to be highly material, especially on the financial side. There is a paper by D. Bernanke, Gertler, and Gilchrist, published in 1999, that is a roadmap. And that's not an obscure paper. It's cited more than 1,000 times. A lot was known that reads directly on what happened on the financial side, and why financial frictions rise when you have a financial crisis in particular. So I want to defend my profession against some of the things that have been said, even by its friends here.

Okay, so the factors leading up to the slump-- I want to talk mostly about the slump itself, but I found that everyone wants to know my view of what caused these problems. First of all, there was a dramatic easing of lending standards during the first decade of the century. That resulted in a lot of new mortgages and bidding-up of housing prices. At the same time, so-called deregulation-- and I condemn the notion that it was ever even properly called deregulation-- of financial institutions occurred.

I think that the one most important failure of regulation is easy to identify. In 2004 the SEC removed its capital requirements on investment banks. So what are the names of some investment banks? Well, Lehman. What about Lehman? So the reason that Lehman was able to do what it did, which proved so destructive, was that it had no limits on the amount of leverage that it could adopt. And we learned that it adopted a huge amount, and then it fell apart. As soon as asset prices declined even a little, Lehman necessarily fell apart because it was so highly levered. So I think that the regulatory side of this is a very important one.

I mentioned before, there was a buildup of consumer debt, and of a form with the expansion of debt, in effect, people were borrowing more to pay the interest on the debt they already had. It was unsustainable. So the result was an overshooting of housing prices. And then-- it's a very simple principle, the housing prices, as in most asset markets, if it's gone up a lot, it'll go down. And that's what happened. And when it happened, it affected these highly-levered financial institutions in a way we learned so much about in on September 15 of 2008.

So there was a sudden realization that capital had been depleted, there was potential insolvency of a great many institutions, and panic, and some very aggressive-- and I think appropriate-- government action, making up for the failure of government action in earlier years. And all kinds of institutions besides banks-- investment banks, AIG's hedge fund, money market funds-- all proved to be vulnerable to this decline in real estate prices.

There was a big increase in financial friction as a result. There was an increase in spreads. Let me pause for a second. I'm going to de-synchronize the slides here. This slide shows this big increase in the relationship between durables, household capital, houses, and cars and other consumer durables. The red line on top, you can see, rose a lot. Nothing similar happened on the business side. The boom was very focused on households and very focused on real estate. There wasn't much suspicious or interesting going on in the business sector. That was pretty much normal. What happened was a big expansion of debt, and a corresponding expansion of building of houses, and a tremendous splurge of buying cars.

On the financial friction side, one of the ways we like to measure that is a spread, which is the difference between, for example, what businesses charge on loans and what their funds cost. And that, this is really key to the diagnosis that I'm talking about here. The spread rose in 2008, and it's still high. We still have an economy with excess financial friction relative to normal. And it hasn't gone back to normal. And that's one of the important factors that's impeding recovery.

If you look at that on the consumer side, you see exactly the same thing. Credit card interest rates have remained high at the same time that the cost of funds to banks to provide credit to credit card borrowers has declined substantially. And that's resulted in increase in the spread. So there's another friction. In addition, there's been a huge restriction in the availability of credit.

If you haven't used it already, there's a service at Google called Google Insights that is just tremendously valuable. And I looked at a variety of searches. This shows that there was an explosion of people looking for information about withdrawal penalties. Now, what does that tell you? It tells you that they'd exhausted their normal ways of dealing with a financial emergency at home, and had to go to things like taking money out of a retirement fund, which involves a very substantial tax withdrawal penalty, or taking money out of saving instruments that have withdrawal penalties. And you can see, at the normal level, it was cruising along, and then suddenly it's shot up. And again, it hasn't come back down very much. And that's again, a very important issue with respect to the continuation of the slump.

Okay, so what happened, turning to the interest rate side, the Federal Reserve responded very quickly. By the end of October, the Fed had done everything that it can do with conventional monetary policy. It had pushed the fed funds rate down to almost zero. And it's still there today. And forecasts shown here on the right side suggest that it's going to remain at zero. And certainly the most recent decision was emphatically to keep it at zero. We're far from the time when we should begin to raise it. This scenario suggests that next year it will begin rising, but only by a little, and won't get back to normal again until 2015, when the economy is thought to be returning to normal.

So we're in an extended period when the Fed's hands are tied with respect to its ability to expand the economy further. So that means that the normal mechanisms that should operate in the economy, of lowering interest rates, stimulating spending on, for example, business capital to replace the fact that people don't want to spend any more on houses. We built, in some sense, way too many houses through the year 2006. We can rest on our ores with respect to house production. We should be producing other stuff. But that would require a decline in interest rates. And that interest rate decline cannot happen because interest rates can't be negative.

And the result is the normal equilibration process of the economy fails in a situation like this. There's been a huge outpouring of research that I've contributed to on this question of what happens in an economy when its interest rate can't go below zero. That's called a Zero Lower Bound, or as we say, ZLB. There's now a whole branch of macro called ZLB studies. And it's extremely active right now. Okay, so what else happened that's important? Fortunately it turned out that nothing much happened to inflation. Inflation cruised at right exactly around 2%, which was the Fed's desired level of inflation, for an extended period. At the beginning of the crisis, it fell just a bit, by about almost exactly a percentage point. And since then, it's cruised at 1%.

Now, you might think all inflation is a bad thing, but that's not at all the way macroeconomists think about it-- quite the reverse. I've said-- on several occasions, but this is the most appropriate one-- Ben Bernanke would surrender his MIT PhD in economics if we could have 3% inflation today. And why is that? It's because what we really care about is the real interest rate, which is the ordinary interest rate, the nominal interest rate, minus the rate of inflation. What we'd really like is for people to perceive that now is a great time to buy stuff instead of later.

Well, there's nothing like inflation to get that mentality going. Normally we don't like to see that. We want to keep inflation under control, because it's a sign of an overheated economy. Well, we'd like a little bit of that overheated mentality now. In the overheated economy, people are buying. now instead of later. They're doing so guided by financial incentives, the interest rate. We'd like a nice low real interest rate, which says buy now instead of later. And that inflation could deliver that. Well, you can't revoke Ben Bernanke PhD. By the way, Franco Modigliani once suggested that my PhD should be revoked when I said something outrageous at a conference. But it didn't happen. I still have it, I believe.

[LAUGHTER]

And so we're lucky that we have this continuation of a moderate amounts of inflation. Okay, so all these factors that I've identified are going to continue as far as anyone can see. Gradually, things are going to get better. The overhang of overbuilding in one sector is going to gradually come to an end. We're gradually going to fill up those houses and build more houses. The households are gradually working down this excess debt that they accumulated in the last decade. Financial frictions are gradually getting back to normal, at least in some areas. And all that's going to give us a gradual glide path of unemployment back down to normal. But another four years, maybe more.

With respect to policy, the policymakers that were going to talk this afternoon, I understand they're all still waiting for their planes in DC. So maybe I can say a bit now. First of all, I think the Fed has basically done all it can. It's made all the right decisions as far as I'm concerned. But there's not much more it can do. And it is saying just steady as you go, and not engaging in any more attempts at expansion. Because it's basically done everything it can.

There was a lot of interest when the Obama administration came in. And MIT economists were central there. Larry Summers, who basically got his PhD from MIT because he took graduate courses, including one for me-- a memorable experience, I have to say, and actually quite a pleasant one-- and Christy Romer, who has an MIT PhD in economics, were very instrumental in designing early policies. One of them was to try to take advantage of the fact that when the government buys more stuff, every model agrees that that stimulates the economy. Any economist who suggests that that doesn't stimulate the economy hasn't looked at a wide range of models, all of which agree. It's a numbered question, but that number is positive.

And furthermore-- and this is really important-- in a zero lower bound situation, which is what existed when the Obama administration came in, everyone also agrees that the multiplier is higher when the zero lower bound is in effect. So the numbers that were put out for what you could get from increasing government purchases, I think were about right. And that was a big effort of the administration. But it failed completely, not because the multiplier was low, but because it's just impossible to get the government to buy more stuff. The only government agency in the federal government that bought any more stuff was the Pentagon. But there was no increase in other types of federal government purchases. So we just didn't do it. And that's a repetition of experiences we've had before.

The only thing that's under discussion now that would be interesting-- and it's an idea that quite a number of people have come up with-- is that if we could use taxes to create the equivalent of inflation, that would give us a huge stimulus. And it could be done in a way that didn't have any budgetary impact. But that's an idea that's borderline crank. So I don't think it's going to happen. But it's an idea that, if economists ran the show, it's probably something we'd be doing.

And the last comment I'll make is that all this happened because of regulatory lapses. The most important thing, which we've made almost no progress on, the Dodd-Frank bill, only scratched the surface of what needs to be done. We need to have robust financial institutions with lots of capital. When I say lots of capital, I mean like 50% capital. There's no reason why we shouldn't have a heavily capitalized financial system. And if that had been in place and enforced, none of this would have happened. It's entirely the result of a very substantial enforcement policy lapse, of allowing extremely highly-leveraged financial institutions to have important roles in the economy without proper regulation. Thank you.

[APPLAUSE]

ACEMOGLU: So I agree with Bob that we should be careful in making forecasts, especially about the future. But it seems that I just made a good forecast. I asked Bob, so will you be talking about policy? And he said, no, no, no, that's the panel in the afternoon. I said, oh, I would have thought you would talk about policy. And he did talk about policy.

[LAUGHTER]

So we're very happy to have Esther Duflo, who is going to talk about, kind of, an economic development. It's going to be a gear change, I guess.

DUFLO: I was trying to figure out what could possibly be my comparative advantage. And I couldn't find one, except that perhaps just before lunch you might want to see some pictures. So here is a picture of Egypt. In 2010, Egypt got $1.3 billion in military aid and $250 million in economic aid. Egypt is the country that gets the most aid from the US, except for Israel. Unemployment rate is about 25%. And as you can see from this picture, there is deep, deep frustration against a regime that is both so corrupt and extremely tired, both physically and institutionally.

If we had given all these aid that went to Egypt to Africa, we could have given $20 per year per child for every child under 5 who lives in Africa today. So far some people, like Jeffrey Sachs, this is a sign that we could have ended poverty if we had had the political will to do so. With this $20 per child per year, you could have bought bed nets to protect her from malaria, deworming medicine to make her stronger and make her go to school more, all of the immunization that she would need, and you would have plenty of cash left over.

So if we don't do that, it's because we don't want it because our priorities are wrong. On the other side, we have Bill Easterley, who's actually a graduate from MIT, who says that, no, in fact this is a symptom, this is a sign-- the Egypt case, as many others, is a sign that aid, in general, fails, and that it's useless-- or maybe worse than useless, because it contributes to prop up regimes which would have gone long ago.

So the issue is how do we say who is right. And it is very difficult to say who is right. If you look at the aid to Africa, it's been increasing steadily. Those are the blue bars here on the picture. Now of course growth has not really been stellar. The GDP per capita is the red line.

So some, like Bill Easterley, would conclude, well, it means that aid has dramatically failed. Of course we have no idea what would have happened without the aid. Maybe it would have gone downwards, like here, or maybe it would have gone upwards for that matter. And this is like-- the problem is what Jerry Hausman said, we only have one world, we only have one Africa, and we just don't have a counterfactual to know what could possibly have happened in this case.

So when people don't use this kind of figure, they use anecdotes. So one that is a very popular one on both sides of the aisle, are Rwanda got tremendous amount of aid after the genocide, and it did very well. And now the president, Paul Kagame, wants to wean the country off aid, and that's tough to do so. So one that is claimed as a success both by Jeffrey Sachs and by Dambisa Moyo-- by Jeffrey Sachs as a success of aid, and by Dambisa Moyo as a poster child for self-reliance-- of course it's very hard to arbitrate between this he said, she said, because we have no idea what would have happened in Rwanda if the policies had been different.

So what has happened in development economics over the last few decades is to try and move away from those big, big questions-- can aid make a difference-- to much, much smaller questions, much more focused questions to which we might possibly have an answer. So it's a little bit, in a sense, the empirical equivalent of what [INAUDIBLE] said is an MIT-style theory. This is, in my view, MIT-style empirical work, too, not only MIT obviously.

So one of these questions, for example, is malaria. Malaria is a focus question. I would agree it's not a small question. It's pretty serious. It kills about 900,000 people every year, give and take a few, 91% of them in Africa, 85% of them under 5. In fact, that's the leading cause of under-5 mortality.

So how do we go about it? Well, there's one thing we do know from various experiments done by doctors, is bed nets are an effective way to prevent the spread of malaria. Long-lasting insecticide bed nets are relatively cheap. For about $10, you can manufacture one, ship it to Africa, and teach a family to use it. The problem is that $10 is still too much for many African families when the GDP per capita for a country like Kenya is about $300, for example. So only about 10% to 20% of children at risk sleep under a net. And of course as conventional public finance would tell us, we should subsidize bed nets, because malaria is highly contagious. So if I sleep under a net, that helps the kid next to me. And in fact, the externalities are not linear. Because the estimates are that when 50% of people sleep under a net, the mosquitoes don't have time to consummate the parasite before they die, and therefore there is no transmission anymore. So you can achieve herd immunity from 50%. That would give you a very strong argument for subsidizing net purchase.

In fact, we might even want-- there's no reason to stop at zero. It's not like inflation, which for some reason is kept at zero-- interest rates, sorry, which for some reason is kept at zero. We could go and pay people to use nets. And maybe that's what we should be doing. So this is a policy that hasn't been argued, for example, by the WHO or by Jeff Sachs.

But then there's always a new economics to come forward. And there is this very old principle in economics called sunk cost fallacy, which is what you paid for something shouldn't influence what you do with it. Because once you've paid for it, you've paid for it. And these arguments in coming from psychology and economics saying that maybe the sunk cost fallacy isn't a fallacy to the extent that people actually are influenced by it. So there could be what people call a psychological sunk cost effect, which is unless you pay for something, you consider it worthless. So the argument is that if you give bed nets away, people are not going to use them as bed nets, but as wedding veils or fishing nets, like on this picture.

So the argument is very heated. It's a very, very ideological argument with very strong position. Of course there are dollars involved, but not very many dollars when we compare to the US economy or anything. But a lot passion and ideology.

And sitting from where I sit, this is a bit strange that this debate is dominated by ideology. Because it seems we could pretty much resolve it. It seems that what we do need to do to know whether bed nets should be distributed for free, or we should pay people to use them, or we should make people pay a little bit, or make people pay full fare, is number one, what's the price elasticity of the demand for bed nets? Do we lose a lot of people if we start charging for them? So that's the standard, sort of economics 101 question, the empirical equivalent of that.

Second is, if people get bed nets for free, will they use them or will they waste them? So that's the more sort of behavioral sunk cost fallacy hypothesis. And third one, which is something that also come forward, is that are people are going to be used to handouts. So you give them the bed net for free, and maybe they use the bed nets, but you can never ever sell them a bed net again, or clothing, or school meals. They will always expect handouts.

And all of these questions, in principle, you could answer by looking at people at what people do, with the caveat that you have the Jewish mother or Asian mother problem, which is whoever decides to pay for a bed net is probably not comparable than someone who will only get a bed net if they get it for free. So a big movement in development economics in the last 10 to 15 years has been to try to create a situation where people are exposed to different conditions or treatment that are in fact comparable. And the way you do this is the equivalent of the randomized controlled trials that are used in medicine to test the effectiveness of drugs.

So to give you an example here, this is an experiment that was carried out by Pascaline Dupas, who is assistant professor at UCLA and a former student of mine. What she did is to randomly select households all over the place, and offer them nets, either for free, or with partial subsidies, or with full subsidies. And she randomly broadcast her vouchers for the various prizes for the nets so that people are strictly comparable. And when we look at that behavior, we know that it's the effect of the price they had to pay, as opposed to any differential selection.

And what do we find? The first finding is that the elasticity of demand for bed net is extremely strong. So for free, almost everybody get it. As soon as you make people pay just a little bit, the demand drops. And when you get to $3, which is only about less than half of the full price, the demand is basically all gone. So the price [INAUDIBLE] is very high. So if you start getting people to pay, you will lose a lot of people.

Second question is, are they going to use it if you make them available for free? And there you find sometimes no evidence for this psychological sunk cost effect. People who paid for the bed net and people who get it for free are about as likely to use it. I should say, it means both there is no psychological sunk cost effect, and the selection of people who decide to purchase the net is not affected by the price they have to pay.

And the third question is whether getting a free bed net discourages future purchases. And you find, in fact, if anything, the opposite, which is people who got their first bed net for free are somewhat more likely to buy a second one when it's offered to them about a month later at an average price. So it seems people do not get used two handouts, they get used to nets. Which is why, in the end, there is this learning and information effect going on.

So what do we get from this? Well number one, I think we answer a policy question of key interest. When that paper of Pascaline's came about, that sparked a lot of debate, and that did change policies. For example, a big social marketing organization PSA changed their policy and started to give bed nets away. The government of Kenya is also now-- this experiment was done in Kenya-- government of Kenya is distributing bed nets for free in the maternal health clinic.

Number two-- and that's perhaps where it becomes more interesting from sitting in an economics department-- is this gives you a window on how and why the poor make the decisions they are making. And here the big puzzle is, given that people are willing to use the net, seem to be valuing them, why is it that they are so price-elastic, at least at the first purchase? And that's not an isolated thing for bed nets. The same thing is true for immunization.

Immunization is the cheapest way to save lives. It's very, very cheap to immunize a kid. The GAVI Alliance and the Gates Foundation are pouring millions and millions of dollars on the problem. And yet, in a country-- there is about 20 million children every year who don't get the vaccination they should get. In a place like Rajasthan, India, where this picture was taken, 5% of children get the full set of immunizations they should get. This is a tremendous waste. It's hard to imagine a worse problem.

And what's going on here, well, we did an experiment try to figure out whether small demand-side incentive would make a difference in the decision to get immunized. And what we find-- this is a baseline, you see almost no one gets immunized-- after a year, there is some increase in immunization in the control group where we did nothing. When we introduced some camps to make immunization easier, some increase, to 17%. But then when you start giving this small demand-side incentive-- we're talking about a kilo of lentils-- the full immunization rate jumps from 17% to 38%.

So it's a little bit the same thing, where you get these very, very high price elasticity for preventive care, which contrast with huge demand for curative care. And in fact, in experiments I'm going to spare you the detail of, very low price elasticity for curative care. So people are willing to pay just about any price to cure their kid for malaria once they have malaria, but they won't get a bed net. So that is kind of-- that's the mystery to what could be going on. Well, part of it could be misinformation and mistrust. That's not only developing countries. In this country as well there are ways where people decide that getting immunized is not so necessary anymore. Part of it could be procrastination, a natural human tendency to wait to do something. With preventive care, the cost you have to pay, like going to the center or paying for your bed net, is today, but the benefit is in the future. So you might as well wait a little bit to do it. And that's why offsetting the small cost with a small benefit, like for example in the case of the immunization program, is so effective.

Or maybe it's a combination of the two things. Regardless of what it is, understanding exactly what's going wrong is key to design effective health care policy in developing countries. For example, what are the lessons that come out of each one? One is absenteeism. This is a closed health center in Udaipur District, 63% of centers are closed when they should be open. Absenteeism is very [INAUDIBLE] and it's amazingly costly, because it introduces both mistrust and increases those little costs that stand in the way of people doing the right thing.

Secondly, information is important, but we have to recognize that information is going to be difficult. Because learning about preventive care is very, very hard for people who had no high school biology, and even for people who had high school biology. So there is a lot of trust. And I think the right message and way of diffusing the information is hard.

Thirdly, we cannot possibly overemphasize the importance of making things easy. And this is something that, from the comfort of our lives, we don't fully realize. For example, we don't realize that water comes chlorinated in our tap free of charge when we moan about the fact that the poor don't boil the water, when we never had to boil our water. We don't realize that we have to immunize our children, because otherwise we can't send them to school. So that is a very powerful incentive.

And then we-- we meaning the policymakers, experts, et cetera-- make big speeches about the importance of the poor really understanding what they need to do, and we shouldn't bribe them to do the right thing. We are constantly bribed and helped to do the right thing in our lives.

So why did I spend so much time on that question of health care? Just because it's, I think, a good illustration of the modus operandi in development economics today. Take a pretty focused question, and bring it all your attention and your focus, and try to answer it very well. The question, of course, is whether when we do that, we sort of give up on the large questions of how our economy's going to grow, and are we going to get treat of poverty one day, whether we have completely exploded the problems into a myriad of small problems, and we are too far from the big answer.

So obviously if I thought the answer was yes, maybe I would not have phrased the question that way. I do think this very patient, step-by-step approach is a very productive way of trying to understand how the poor behave, why they behave the way they do, and therefore how we can possibly help them get out of poverty trap or get going.

And it's not going to be easy. Because this is from the accumulation of these experiments and experiences that a set of general principles can maybe start to appear. But something that-- it needs a lot of replication. Something that works somewhere may not work elsewhere. It also means that in a lot of cases, you have to be like Keynes and change your mind when the facts change. Because the facts easily change on you.

But I don't think there is another way, really. And in a sense I think there is an analogy between this way of proceeding and 20th and 21st-century medicine, where you have a back and forth between trying to understand what a particular molecule do, and then going to a big project like the Human Genome Project, where the two kind of feed off each other to develop medicines. And there are no miracle cure to anything in medicine, there are just thousands and thousands of drugs, and a better and better understanding of the human body. And together, that amounts for millions of lives saved.

So I also wouldn't like you to be fooled by the apparent modesty of the objective. Deworming sounds sort of-- very, very small thing. It's take a pill twice a year. How can that be possibly important. It turns out that deworming-- kids who have been dewormed for one more year than comparable kids, when they were in school, earn 20% more every year when they are adult. 20% more every year is a lot of years of good growth in Kenya. This study was done in Kenya. And there has not been that many good years of good growth in Kenya anyway. So these are actually things that can quantitatively be extremely important here and now, and have policy implications.

A few years ago, I was in Davos, started this Deworm the World organization, which is kind of get all of these rich people to give money to deworming, which is sort of pleasant anyway, and gets this-- making some progress in deworming as many kids as possible.

So where are we? Well, in a sense, I think there is a pretty clear moral case for trying to help here and now if we can. I think there is an efficiency case as well. It seems like a considerable waste not to do the maximum we can to help kids like these kids get educated and healthy. My personal view is that it can't possibly be bad for growth, in that when and where the growth spark starts, people who aren't properly educated and healthy will be probably more likely to take advantage of it. Even if it were wrong, I don't really see a strong case not to do the maximum we can. And in a sense, I don't think we have a choice. I think that's a political imperative as well.

An go back to Egypt-- I think the poor are not going to sit silently while we are trying to figure out how we can make the economic [INAUDIBLE] or the labor market work better.

And I'll finish with Roosevelt. I replaced the world "country," which he was using in 1932. "The world needs and, unless I mistake its temper, the world demands bold persistent experimentation. It is common sense to take a method and try it. If it fails, admit it frankly and try another. But above all, try something. The millions who are in want will not stand by silently forever while the things to satisfy their needs are within easy reach. Thank you.

[APPLAUSE]

ACEMOGLU: Thank you very much to all four panelists for these very interesting and thought-provoking comments. I would like to see whether any one of the panelists would like to make any follow-on comments. And if not, I think we have a little bit of time for questions. And the microphones are up front again. Please come to either of these two sides.

AUDIENCE: Thank you, everyone. My name is [INAUDIBLE] from Northeastern graduate student. Last week, one of my students asked me a simple question in Professor [INAUDIBLE] test book about why the human capital is never decreasing. I found it very difficult to answer such questions, not because I don't know the answers in test book, but because of my experience before two years. I was hired by the Chinese government to do a survey in Jiangsu Province about economical growth. So I spent half a year to do survey. But what I see is all the local governments they do is introduce capital, introduce technique, and spend a lot of money to develop educations. Because China always lack of these things.

But the problem is that all the foreign factories invest in China, they just use China as a factory, not industry. China become a part of productions for those firms. As a result, it has no such demand for human capital and techniques because there's no design department, there is no [INAUDIBLE]. So as a result, all the people have a high education level, the [INAUDIBLE] to outside, not stay at those countries, stay on those less developing areas.

So for those less developing countries, even they spend a lot of money, time to develop a technique, develop human capital, as a result, they never get benefit from that. So my question is that, based on Professor Solow's series, the difference [INAUDIBLE] of economic growth, they don't have stable solutions in the equations. So whether it is a time for us to reconsider the power and function of human capital techniques for developing countries. Because these powerful factories never belong to those developing countries. Thank you.

SOLOW: I'm sorry. I would like to reply if it's directed at me.

HALL: Let me try to translate.

SOLOW: I don't hear very well, and I could not make it out.

AUDIENCE: I'm sorry about that.

[INTERPOSING VOICES]

ACEMOGLU: Bob Hall, I think, will say something.

[LAUGHTER]

HALL: Well, a one-sentence version is, what's the role of human capital in growth theory?

SOLOW: Oh.

[LAUGHTER, APPLAUSE]

I said earlier that, for the long run, it's extraordinarily important-- supremely important-- to understand the evolution of total factor productivity, with an eye, of course, to asking the question what a country or some other policy unit can do to accelerate the growth of total factor productivity. And the literature has produced these two interrelated paths by which total factor productivity improves. One is by improving the skills of people, and the other is by technological innovation. And they can't be independent, because most technological innovations require some set of specific skills which need to be imparted. And as production gets more technologically sophisticated, greater skills are needed.

It's very hard, from the sorts of time series that we have, to isolate the relative contribution of these or any other particular source of total factor productivity that you would like to think about. There is certainly no doubt that human capital is extremely important, not only in the growth of advanced countries, but also in the development of poor countries. And the big thing that China has brought to the world economy is a very large source of low-wage labor, first of all employed in functions that require very little in the way of human capital. And then the question is, can the investment in education facilitate, make possible, and even make necessary the move to more and more sophisticated lines of production? In the case of China, that appears to be so.

I don't know whether I've answered the question in particular. It's certainly the question I would have asked myself.

ACEMOGLU: So we'll have a couple more questions. If you can make the questions relatively brief, because I think we have limited time.

AUDIENCE: Good morning. My name is Karen, and my question is for Professor Diamond. And I know you were nominated by President Obama to the Federal Reserve Board once again. So I want to hear your opinion on the effect of QE2. And I once heard you say you are a believer of the market and capitalism. And I want to share your opinion that do you believe the US economy can work its own way out when the effect of QE2 wears off this time. Thank you so much.

DIAMOND: The first rule I was told after I was nominated for the first time, and the second time, and the third time-- which I think should put me in the Guinness Book of World Records-- is don't answer questions like that, it will just get in the way--

[LAUGHTER]

--of your being confirmed. So let me skate around it. First of all, Bob Hall said the Fed has pushed the interest rate down to zero, and there's very little more it can do. QE2 is something more. And I don't think anyone thinks of it as a huge deal by itself. But I share the view that has been expressed by the existing governors-- I think I can say that in public-- that it is helpful in the right direction. But I share the view implicit in what Bob said, that it isn't as powerful as an ability to lower interest rates 2%, 3%, or 4% more.

ACEMOGLU: Okay, one-- so I think we should-- yeah.

AUDIENCE: Bill Carlson, '68. And I'm going to ask this question of you, Daron, and you can allocate it to the panel. You've published research on dual labor markets, the so-called "good jobs, bad jobs" literature. And I'm curious whether this distinction between good jobs and bad jobs is so fundamental that it should be in the microfoundations of macro models whose intent is to predict the effect of stimulus policies, especially when you take into account Williamson's work on the structure of firms.

ACEMOGLU: I think that's a wonderful question. But I'm not going to answer it myself. And I'll see whether-- Peter.

DIAMOND: Let me just say that a major part of the flows in the slides that I showed are flows of employed people going to other jobs. And so it's a quit, and it's a re-employment, and that that generally represents an increase in the efficiency in the economy. You could call it moving from a bad job to a good job, but it may be a series of small steps. That's a nice vocabulary for something that's a continuum.

And there is a significant part of the literature exploring what drives it, why it disappears. Quits disappear we go into a recession. As Bob has noticed, when the quits are gone, then the replacements disappear, and that that affects hiring rates. So this issue, in a different vocabulary, is very much there. It's not in the simplest basic starter model but I think it's the kind of thing that is recognized as an important part of the cost of a recession, that these things aren't happening.

HALL: So David Autor, in the MIT economics department, has done some very interesting research that suggests that we really need to think about three groups in the labor market, not two. There's one group of people who have very little training and basically work with their hands-- they do gardening, and they take care of people-- and they're doing all right by historical standards. Then there's the people who have graduated from MIT, who are ready to go out and develop new products, and do intellectual problem-solving work. And they're doing extremely well. And then there's a very sad group in the middle who are high school graduates, but are the ones whose jobs have been taken away, so to speak, by computerization. And they have suffered the biggest disadvantage.

So rather than talking about a dual labor market, I think the next step is the three-part, trinary labor market, to get a good picture of what the US labor market looks like.

ACEMOGLU: Yes, please.

AUDIENCE: [INAUDIBLE], Sloan School of Management, '74. I took a course from Professor Hall. I don't really remember what I learned.

[LAUGHTER]

HALL: That's good, because I don't remember what I taught.

[LAUGHTER]

AUDIENCE: But I remember it was very important. And I wish I had learned more of it.

[LAUGHTER]

My question sparks from your chart that showed the spreads on lending, which are very high for both consumers and corporates, and they remain high. And the observation I want to insert into that, and then ask a question, is that the shadow banking system is effectively on its back. Structured financings are not happening, and they're not likely to happen in any way in the size that they're used to, for a variety of reasons, mostly good but many bad, overregulation being one of the bad ones.

Given the lack of lending that is taking place in the high spreads, how do you see the evolution of the decrease? The observation I had about the structured finance market, it basically has slowed the velocity of money. And I wonder to what extent you see reduction in spreads and incentives that might be given to reduce those spreads from lenders.

HALL: So on the business side, the bulk of American business is not dependent on borrowing from financial institutions. There's relatively little leverage. Where I live, corporations don't borrow from banks at all. Quite the country, they accumulate almost mindlessly. Like Apple-- Apple has $50 billion in short-term instruments on its balance sheet, not borrowing, but lending. And the shareholders are getting a little restless about that.

Of course there are certain industries that are credit-dependent. And they are still definitely-- there's no question that there's an impairment of credit flows in certain areas.

I think, though, and I think there's been a big shift of the thinking of macroeconomists in this direction, that the real credit-dependent sector is not business, but it's households. And the real questions are the squeeze that's gone on in the household, where they were permitted to borrow large amounts in the last decade, and then forced to repay that over the last few years. That, in terms of GDP-- the way I think-- is very much larger than what I would say is a relatively marginal issue with respect to businesses. Most GDP arises in businesses that have direct access to credit markets, if they need credit at all-- or finance. They're mostly equity-financed.

The US business sector is not highly levered. Only certain sectors are. Airlines are. We used to say that auto companies were. But generally speaking, leverage is not a factor in business. It is a huge factor in financial institutions. But how important that is for the business sector I think is still up in the air. It's the household side that I think it deserves the most attention today.

DIAMOND: Let me put a different wrinkle on the interpretation of the household sector and the word, business. From the employment point of view, there's a great deal of employment that goes on with small business. And a lot of small business is financed by banks, and is financed by drawing on-- there's the household link-- their own wealth, the wealth of families and friends. And what we have been seeing in this great recession is that employment growth by small business relative to employment growth by large business is way down. And the cyclic behavior of the two generally is different. And what we're seeing now is an unusual amount, which I think is connected to exactly what Bob was talking about, the financial stretch-out of households and of banks. I think that affects the financing not of big business, but of small business and in terms of hiring. That's an important part of the story.

ACEMOGLU: Since we are running late and we don't want to eat too much more into the lunch time, I think we're going to have only time for one more question. Sorry for the remaining people who wanted to ask questions.

AUDIENCE: Hopefully this is worth your while, then. I'm Doug Carr, a Sloan grad. Obviously the issue of the size of government is a huge controversy today. There are studies and economics-- Barrow, I believe, has found that larger size of government relative to the economy has associated with it a lower rate of long-term growth. And I think there may be other studies that refute his study. Within the field of growth theory, is there a framework for analyzing this issue?

SOLOW: Certainly a framework for analyzing it. The difficulty in dealing with associations like that is to translate it into a causal mechanism one way or another. It is of course perfectly possible that large government budgets waste resources and affect the level of output that way. It's also possible that slow-growing economies tend to depend more on governmental intervention than fast-growing ones. My own feeling about this is that what's more important to begin with is to talk less about growth and more about levels. One could easily make a sensible argument one way or another about the size of government relative to the economy and the efficiency of the economy. But that's a matter of a level change, not a change in something that's potentially part of a steady-state rate of growth.

So I tend to be suspicious of all simple generalizations that go from some current fraction to an implication about steady-state or near-steady-state rates of growth.

DIAMOND: Bob has-- if I can characterize that-- says he's skeptical about an answer to that question, Barrow's or the opposite. I'm skeptical about the value of the question. And this also relates with what Esther was saying. Esther was saying, more aid, less aid, wrong debate. What we want is good aid. More government, less government, wrong debate. There's good things the government spends on that helps, if you'll excuse the expression, economic growth, or efficiency in the economy. And you do more of it, then the level goes up, and that's growth.

Regulation-- is it more regulation or less regulation? Now, Bob was saying it's the quality of the regulation. I was a little surprised that dropping the regulation on leverage, which you referred to as a policy lapse, isn't in the category of a bad deregulation. Not that the whole system was deregulated, if that's what the term means, but I think, on regulations, there were good and bad ones, on aid, there's good and bad, on government spending, there's good and bad. And the idea that you look at the whole and then move from that, somehow, into policy and politics seems, to me, misguided.

SOLOW: Let me add one more thing. This is what they call a teachable moment, I think.

[APPLAUSE]

AUDIENCE: Teaching me?

SOLOW: No, no, no. No, no, no. No, teaching the world.

[LAUGHTER]

A government that spends money, which would appear as a larger government, supporting higher education, supporting technological innovation, supporting R&D generally, certainly has a prayer-- more than a prayer-- a likelihood of affecting the steady state rate of growth.

A government that botches food and drug regulation or that spends more or less will affect the current efficiency of the economy, but won't affect the rate-- to that extent, I think that's what Peter had mind by saying he disagreed with the question.

ACEMOGLU: Bob Hall.

HALL: Okay, great question. All right, so as it happens, I've done research on this question, published a paper on it, which agrees extremely strongly with both Bob and Peter. In the first place, it was in a levels framework, which I strongly agree with. What matters is the level. Growth is just a way you get to the level.

Second, it examines this question of how government relates to economic success. And the answer is, very clearly, size is not important. And you can see that in Scandinavia. Scandinavia has governments that historically have been very involved in the economy, but has achieved excellent results. And then you see some economies that have very small governments that have terrible results.

What shows through unambiguously is competence. What we want is a government that knows what it's doing, everywhere. Now, luckily, that's pretty much what we have. The US does have-- it stands at the very top in terms of output per worker, and generally, with a government that does a good job of providing an environment in which private activity can flourish with the government being assigned the right things to do.

There's a few exceptions, like the post office. But generally, this country gets things right. And if you go to another country and see governments floundering, then you can appreciate that.

ACEMOGLU: Okay I think I would like to thank the four panelists again for--

[APPLAUSE]