Real output in most advanced capitalist economies fluctuates around a rising trend. One can argue about whether it is best to think about that trend as passing through successive cyclical averages, defined in one way or another, or best to think of it as passing through cyclical peaks, or some other measure of output. While the outcome of that argument has consequences for macroeconomic theory, I will bypass it for now. The important observation is that, on the whole, the observed fluctuations around trend are contained within a moderately narrow corridor. Unemployment rates tend to run between, say, 5 percent and 10 percent in the United States. (Other countries have different typical ranges, and in each of them, the range can shift from time to time. It is important, theoretically and practically, to understand why; but that remains an open question.) There are notable exceptions to this generalization, of course, the most famous being the depression of the 1930's; but they are exceptions. Again it is important to know why fluctuations are so contained. This could reflect some natural equilibrating process, or it could reflect the intervention of automatic or discretionary government policy, or it could be a mixture of both. That is another issue on which opinions differ. I think it is part of the usable common core of macroeconomics that the trend movement is predominantly driven by the supply side of the economy (the supply of factors of production and total factor productivity) and that the appropriate vehicle for analyzing the trend motion is some sort of growth model, preferably mine. Now, what about those fluctuations around the trend of potential output? A moment ago I put the normal range of unemployment rates at 5-10 percent. By Okun's law I am talking about fluctuations of real GDP with an amplitude of 8-10 percent or so from peak to trough-contained, but not trivial. In my picture of the usable common core of macroeconomics, those fluctuations are predominantly driven by aggregate demand impulses, and the appropriate vehicle for analyzing them is some model of the various sources of expenditure. I am not so obtuse as not to have observed that the whole point of theory is the assertion that these short-run motions of the economy are in fact supplydriven. But my view is that this explanation has been an empirical failure, or at best a nonsuccess. There are now two possibilities. As for the first, I entertain the hope that flexible, observant members of the real-business-cycle school, like Martin Eichenbaum and his coworkers, have come more or less to the same conclusion, and they have found ways to open up the fabric of their underlying model so that it will allow-or insist-that demand-side impulses play the dominant role in short-run macroeconomic fluctuations. Then this proposition is indeed part of the usable core of macroeconomics, and economists can go on to argue back and forth about the best way of modeling those demand-side forces. The other case is that the situation is as before, and the real-business-cycle school holds monolithically to the view that short-run fluctuations are just optimal supply-side adjustments to unforeseeable shocks to tastes and * Department of Economics, Massachusetts Institute of Technology, Cambridge, MA 02139.

I have in the back of my mind a picture of the sort of discipline economics ought to be -or at least the sort of discipline I wish it were. If economics were practiced in that way there would be nothing problematical about its reciprocal relationship with economic history. It would be pretty clear what it is that economic theory offers to economic history and what economic history offers to economic theory. I will try to describe what I mean below. For better or worse, however, economics has gone down a different path, not the one I have in mind. One consequence, not the most important one, but the one that matters for this discussion, is that economic theory learns nothing from economic history, and economic history is as much corrupted as enriched by economic theory. I will come to that, too, later on. You will notice that I am using strong language. I am prepared to admit right away that I may be dead wrong in my judgements. But there is no point in pussyfooting. Bluntness may lead to an interesting discussion. After all, no one would remember the old German Historical School if it were not for the famous Methodenstreit. Actually, no one remembers them anyway. (There must be a lesson in that.) To get right down to it, I suspect that the attempt to construct economics as an axiomatically based hard science is doomed to fail. There are many partially overlapping reasons for believing this; but since that is not the topic under discussion today, I do not have to lay them out in an orderly way. I hope the following hodgepodge will convey what I mean. A modern economy is a very complicated system. Since we cannot conduct controlled experiments on its smaller parts, or even observe them in isolation, the classical hardscience devices for discriminating between competing hypotheses are closed to us. The main alternative device is the statistical analysis of historical time-series. But then another difficulty arises. The competing hypotheses are themselves complex and subtle. We know before we start that all of them, or at least many of them, are capable of fitting the data in a gross sort of way. Then, in order to make more refined distinctions, we need long time-series observed under stationary conditions. Unfortunately, however, economics is a social science. It is subject to Damon Runyon's Law that nothing between human beings is more than three to one. To express the point more formally, much of what we observe cannot be treated as the realization of a stationary stochastic process without straining credulity. Moreover, all narrowly economic activity is embedded in a web of social institutions, customs, beliefs, and attitudes. Concrete outcomes are indubitably affected by these background factors, some of which change slowly and gradually, others erratically. As soon as time-series get long enough to offer hope of discriminating among complex hypotheses, the likelihood that they remain stationary dwindles away, and the noise level gets correspondingly high. Under these circumstances, a little cleverness and persistence can get you almost any result you want. I think that is why so few econometricians have ever been forced by the facts to abandon a firmly held belief. Indeed, some of Fortune's favorites have been known to write scores of empirical articles without once feeling obliged to report a result that contradicts their prior prejudices. If I am anywhere near right about this, the interests of scientific economics would be better served by a more modest approach. There is enough for us to do without pretending to a degree of completeness and *Department of Economics, Massachusetts Institute of Technology, Cambridge, MA 02139.