Keynes, Lucas, and Scientific Progress OR。 Alan s. blinder The American Economic Review, Vol. 77, No 2, Papers and Proceedings of the Ninety-Nint Annual Meeting of the American Economic Association (May, 1987), pp. 130-136 Stable url: ttp: /inks. jstor org/sici?sici=0002-8282%28198705%2977%3A2%3C130%3AKLASP%3E2.0.CO%3B2-Q The American Economic Review is currently published by American EconomIc Association Your use of the jStoR archive indicates your acceptance of jSTOR's Terms and Conditions of Use, available at http:/lwww.istor.org/about/terms.htmlJstOr'sTermsandConditionsofUseprovidesinpartthatunlessyouhaveobtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the jStOR archive only for your personal, non-commercial use Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support(@jstor.org http:/www.jstor.org Fri mar1610:59:532007
Keynes, Lucas, and Scientific Progress Alan S. Blinder The American Economic Review, Vol. 77, No. 2, Papers and Proceedings of the Ninety-Ninth Annual Meeting of the American Economic Association. (May, 1987), pp. 130-136. Stable URL: http://links.jstor.org/sici?sici=0002-8282%28198705%2977%3A2%3C130%3AKLASP%3E2.0.CO%3B2-Q The American Economic Review is currently published by American Economic Association. Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at http://www.jstor.org/about/terms.html. JSTOR's Terms and Conditions of Use provides, in part, that unless you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you may use content in the JSTOR archive only for your personal, non-commercial use. Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at http://www.jstor.org/journals/aea.html. Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed page of such transmission. JSTOR is an independent not-for-profit organization dedicated to and preserving a digital archive of scholarly journals. For more information regarding JSTOR, please contact support@jstor.org. http://www.jstor.org Fri Mar 16 10:59:53 2007
Keynes, Lucas, and Scientific Progress By ALan S BLInDer In one of those marvelous coincidences of I. Are Expectations Rational intellectual history, Robert Lucas was born the year after the publication of Keynes Keynes, though no stranger to probability General Theory. For the first thirty-five years ocal in his of their mutual lives, the two apparently denial that expectations are what we now coexisted in harmony. But their relationship call rational since. Lucas has frequently criticized Keynesian economics as large prop our poor science, and it is precisely in that spirit that i want to address the debate today mism rather than on a mathematical We all know the old joke about the profes- expectation..Only a little than sor who uses the same exam questions year an expedition to the South Pole. is it after year, but changes the answers. That based on an exact calculation of ben- efits to come. Thus if anima joke encapsulates all too well what has hap- pened to macroeconomics these last fifteen dimmed and the spontaneous opti mism falters, leaving us to depend on years and seems to reflect poorly on econom nothing but a mathematical expect ics as a science. Or does it? On second tion, enterprise will fade and d thought, the best answers to scientific ques [1936,pp.161-62 tions do change as new observat ns are made, as new experiments are run, and as That attitude left a big loose end in The better theories are developed. The issue is General Theory. Business investment is sup whether the answers to important questions posedly driven by"the state of long-term in macroeconomics have changed for good expectations, but expectations are not scientific reasons or for other reasons pinned down by the theory, leaving substan The joke provides the framework for my tial room for gyrations in macroeconomic Ik. I will pose eight exam questions; and activity driven by autonomous changes in for each one I will summarize the answers animal spirits. That hardly constitutes a tight iven by Keynes, by Lucas and his followers, scientific theory; but Keynes was probably and by modern Keynesians. I pick only happy to leave the loose end loose. Modern questions that are answered differently by y“ sunspot theorists” have tightened the Keynesians and Lucasians and that are argument considerably, in ways that Keynes central to contemporary macroeconomic de- might have found congenial bates. The focus is on whether the Keynes- Lucas, of course, changes the answer to ian or new classical answers have greater yes. Was this change motivated by empirical claim to being"scientific "Each student evidence that subjective expectations match must answer every question the conditional expectations generated models-or even that actual expectations are unbiased and efficient? No. Indeed, Edward Prescott has boldly asserted that"surveys Princeton University, Princeton, NJ 08544. I cannot be used to test the rational expecta or correspondence tions hypothesis"(1977, p. 3). Rather, econo- th Ben Bernanke, Andrew Caplin, Mark Gertler, mists are supposed to convert to rational Reperi solow and Lawrence summers. a version of expectations(re) because of the unlove- avain paper with footnotes and complete references request to nisms that preceded it and because RE is
VOL. 77 NO. 2 THE CONTRIBUTION OF KEYNES A FTER 5O YEARS 13/ more consistent with their (unverified) the unemployed are engaged in intel- worldview that people always optimize at all ligent search or purposeful intertemporal mas Sargent put it:“Re ubstitution. He scoffs at the Keynes search in rational expectations . has a mo- mentum of its own . that, stems from the that unemployment be classified as in- logical structure of rational expectations as a voluntary.simply cut itself off from seri modeling strategy"(1982, P. 382). The ous thinking about the actual options unem- momentum, you will note, does not stem ployed people are faced with"(1986, p. 47) from empirics. I leave it to you to decide This is a tough question to adjudicate on hether these criteria are more like those scientific grounds since the issue is largely that led physicists to dump Newton in favor definitional and, as Lewis Carroll pointed of Einstein, or those that led artists to out, everyone is entitled to his own defini a bandon manet to follow Picasso tions. In Lucass view, a person laid off from Modern Keynesians are split on this ques- a job can, presumably, shine shoes in a tion. To some, the theoretical appeal of re railroad station or sell apples on a streer and the general idea that expectations should corner. If he is not doing any of these things respond to policy changes are sufficient rea- he must be choosing not to do so. Both son to conclude that"rational expectations statements like this and reactions them is the right initial hypothesis. Others harbor tend to be polemical. I guess dogmatism is in doubts. I think the weight of the evidence- the ear of the beholder both from directly observed expectations and However, a few pertinent facts should from indirect statistical tests of rationality leaven the ideological debate. First, when the (usually in conjunction with some other hy- unemployment rate rises, it is layoffs, not pothesis)is overwhelmingly against the re quits, that are rising while consumption falls hypothesis. Furthermore, RE is not without rather than rises -all of which are bad news theoretical difficulties. We all know that re for search theory. Second, real wage move models often have multiple equilibria. More ments are close to a random walk-which is fundamentally, RE is theoretically coherent bad news for the intertemporal substitution only in the context of a single agreed-upon approach. Third, unemployment is heavily model. In an economy in which different concentrated among the long-term unem- people hold different views of the world, the ployed; in 1985, for example, people whe very notion lacks clarity. For example, if ere jobless for 27 weeks or more con Paul Volcker announces today that on New stituted 54 percent of all unemployment and Year's Day he will raise Ml by 20 percent, I the expected duration of a complete spell of magine Lucas and I will make different unemployment was 31 weeks. Can that be revisions in our expectations for, say, real intertemporal substitution? Fourth, unem GNP in 1987. Whose expectations are"ra- ployed workers normally accept their first tional? "Heterogeneous beliefs pose serious job offer, and those who are looking for heoretical problems for RE. As scientists, work spend an average of only 4 hours per then, I think we should be hesitant to em- week on search activity That hardly suggests ace r predominant role for search in explaining unemployment Il. Is there Involuntary Unemployment? Ill. Do Wage Movements Quickly Clear Keynes said, nay screamed, yes. Lucas not he labor market only says no, but questions whether the phrase has meaning. In his words, "To ex- Keynes certainly thought not, for such plain why people allocate time to . unem- reasons as trade union siVeness, custom ployment we need to known why refer and inertia, and outright stubbornness. lucas it to all other activities"(1986, p 38). Notice answers yes-though perhaps only in a broad the words allocate and prefer. In his view, sense. He has, for example, cited approv-
32 AEA PA PERS AND PROCEEDINGS MAY 1987 ingly the competitive contract equilibrium to clear the labor market in a world of approach in which workers have 100 percent imperfect information--not even in the long unemployment insurance and, because of in run. Of course, that efficiency wage models divisibilities, are chosen randomly to work can be built does not imply that they de ing, of course, that unemployed workers have clearing models have no particular claim to higher utility than employed ones. In Lucas's the theoretical high ground opinion, there is“ no reason to believe”that In sum, the scientific basis for modeling competitive models of labor markets that labor markets-or goods markets for that treat unemployment like leisure commit matter-as continuously clearing escapes me. serious strategic error. To reason? I think the preponderance of IV. Is the Natural Rate of Unemploymer the evidence says otherwise. Unemployment a Strong Attractor for the insurance replaces only about 40 percent of Actual Rate of Unemployment? lost earnings. Lately, only about one-third of the unemployed collect it. Where is the Keynes thought not. Indeed, in his revolu evidence that the unem ployed are happier tionary zeal, Keynes spoke loosely (loose than the employed? Most economists think talk was a problem for Keynes) of an Lucass distinguished predecessor at the unemployment equilibrium"which would University of Chicago had it right when he seem to deny the natural rate any attractive wrote,"Under any conceivable institutional force at all. Lucas answers in the affirmativ arrangements, and certainly those that now Modern Keynesians have long had trouble prevail in the United States, there is only a with the master's notion that the econom limited amount of flexibility in prices and ould equilibrate below full employment; wages. And it is hard, for me at least, to they prefer to think of unemployment as a look at what has gone on in this country long-lasting disequilibrium. In the United not to mention in Europe-since 1974 and States at least, the validity of the natural rate see clearing labor markets. That the market- hypothesis has not been at issue for a long clearing approach caught on in this environ- time. The argument, instead, is over whether ment is testimony to Lucas's keen intellect the speed of convergence to the natural rate and profound influence, not to economists' is rapid or glacial espect for facts. On this. the American evidence is un- More than just casual empiricism supports equivocal and the European evidence is this view: numerous formal econometric overwhelming. The U.S. civilian unemploy studies reject the market-clearing hypothesis ment rate peaked at 8.9 percent in May I against some sort of disequilibrium alterna- and then took almost three years to get back tive. Unfortunately, it is usually spot-market down to 6 percent. It then peaked again at clearing that is rejected. Equilibrium con- 10.7 percent in November-December 1982 tracting models in which the wage plays little now, four years later, it has yet to fall below or no short-run allocative role are difficult to 6.7 percent for even a single month. Some formulate econometrically, much less to re- will argue that 7 percent is now the natural ject. Indeed, it is hard to know what ob- rate, without worrying much about how it servations could contradict such models; grew so high. My view is that a theory that theory just leaves too many open possibil- allows the natural rate to trundle along after the actual rate is not a natural rate theory Nonetheless, certain observations are at all worth making. For one, several authors have In Europe, the evidence is far more com- pointed to interindustry wage differentials pelling. Unemployment rates rose more or that are persistent across both time and less steadily from 1974 to 1985-from 3 to space-differentials which are not easily over 13 percent in Britain, from 2.8 to 10.5 quared with market clearing. Theoretically, percent in France, and from 1.6 to 8 percent we know that the wage rate may not be able in Germany. Some young men in these coun
VOL. 77 NO. 2 THE CONTRIBUTION OF KEYNES AFTER 50 yEARS 13 tries have never held a job and may never be bater's tactic, and a poor one at that, to productive workers. Facts like these have claim that supply shocks are outside the prompted several authors to seek models purview of Keynesian economics which explicitly reject the natural rate hy pothesis in favor of hysteresis. And recent VI. Does a Change in the Money Supply econometric work suggests hysteresis in have Real effects? postwar U.S. real GNP as well. It may well be that Keynesians caved in too readily to Keynes and the Keynesians answered yes, the natural rate hypothesis. without bothering to distinguish between an ticipated and unanticipated changes. Lu v. Is there a Reliable Short-Run Philips Curve? and the Lucasians answer that money has real effects only if it is misperceived. In their Keynes, of course, did not answer this view, a properly perceived injection of mon- question; the Phillips curve came later. I ey is like a currency reform include it on the exam because Lucas and Here, again, the weight of the econometric Sargent made it central to their attack on evidence(though certainly not all of it)sug- Keynesian economics. The alleged failure of gests that Keynes had the right answer after the Phillips curve was their main piece of all. Robert Barro's alleged empirical demon- evidenc at empirical Keynesian models stration that only unanticipated money has were wildly incorrect, and that the doctrine real effects did not hold up.Perceived on which they were based is fundamentally changes in money are not neutral flawed. "(Please notice the adverbs. This charge was repeated so often VIl. Does Social Welfare Rise when with such certitude that it became part of Business Cycles are Limited? conventional wisdom. Unfortunately, it is, to coin a phrase, wildly incorrect. The fact is Keynes tacitly, but that, the Lucas critique notwithstanding, the swered yes. If asked for proof, he probably Phillips curve, once modified to allow for would have chuckled with the condescension supply shocks(any one of several variables of the British upper crust-which is hardly a will do), has been one of the best-behaved scientific attitude empirical regularities in macroeconomics- fully agnostic, but clearly much better behaved, in fact, than we had leans toward the answer no. He has long any right to expect. A long list of studies been sympathetic to the idea that successful pports this conclusion. Nonetheless, Lucas stabilization policies that smooth business ontinues to speak of the Phillips curve as an cycles may actually decrease welfare. Pres- econometric basket case cott is less circumspect. Without bothering Let me anticipate the obvious objection to draw any distinction between modeling a that saving the Phillips curve after the fact conclusion and proving it, he asserts that by adding a supply variable does not absolve "costly efforts at stabilization are likely to be it of its ex ante forecasting errors. It is true counterproductive"because"economic fluc- that, while Robert Gordon,s latest Phillips tuations are optimal responses to uncertaint curves fit the data well, his pre-1972 equa- in the rate of technological change. "Clearly ions do less well. And they did not predict Harberger triangles look bigger and Okun e rise and fall of OPEC. But there is no gaps smaller near lakes than near oceans. Is sense in which new classical models either Prescotts attitude more scientific than tion: pated the error or pointed to the solu- Keynes ke Keynesian models, they were de- I think it is worth taking a moment to signed to analyze demand shocks. Events in explain why Lucas believes that the potential the 1970,s and 1980s demonstrated to gains from stabilization policy are so small Keynesian and new Classical economists The postwar standard deviation of log alike that Marshall's celebrated scissors also quarterly consumption around trend is about comes in a giant economy size. It is a de-.013. Lucas asks an infinitely lived consumer
A EA PAPERS AND PROCEEDINGS MAY198 living under perfect capital markets how since every down is matched by a subse much he would be willing to give up to quent up. This is Lucas's world reduce this small standard deviation to zero Now change utility to the Stone-Geary Unsurprisingly, the answer comes back: not form: U= log(C -$1500). Here a 4 percent much. So Lucas concludes that"the post-war drop in consumption reduces utility by 8.3 business cycle is just not a very important percent. That seems a bigger deal. Finally, problem in terms of individual welfare. That let the cycle instead reduce the consumption a stunning assertion, especially when of 10 percent of the population by 40 per juxtaposed against the conventional wisdom cent while the other 90 percent loses noth- hat governments rise and fall on the vicissi- ing. (Note that I am allowing very generous tudes of the business cycle unemployment insurance here. )With the Lucas's conclusion, it seems to me, ignores Stone-Geary utility function, mean utility few pertinent facts. First, the cycle is not declines 16.1 percent. Now were talking real mainly in consumer expenditures, much less in consumption. Indeed, there is virtually no Lucas will, of course, counter that any cycle at all in spending on nondurables and such problem is best dealt with by better services. Are large swings in consumer dura- unemployment insurance, not by stabiliza bles, in inventories, and in fixed investment tion policies that interfere with free-market all socially costless? Don't these ups and allocations. The same logic says that fire and downs impose serious adjustment costs and theft insurance-where moral hazard prob- cations on lems are certainly less severe-obviate the Second, Lucas's calculation assumes that need for fire and police departments. Isn't cyclical fluctuations take place around an prevention better than insurance? unchanged trend, with booms as likely a However, Lucas's challenge to the Keynes recessions. But what if recessions leave per- ian presumption that smaller cycles are bet manent scars on either labor or capital or ter cycles needs to be addressed scientifi productivity? What if there is hysteresis, so cally. And, since we can,t observe cyclical the natural rate hypothesis fails? What if fluctuations in utility, that requires the use there is a systematic tendency le Keynesian beginning to be developed in the burgeoning or output to of theory. The relevant theory is, I think, just be too low on average? Then th goal of filling in troughs without shaving off literature on monopolistic competition and peaks starts to make sense aggregate demand externalities. It would be Third, Lucas ignores a variety of psycho- foolish to say that a definitive answer is in logical, sociological, and physiological costs hand; but some good answers may be on the hich many people feel are important. horizon Against Lucas's benign view of the cycle compare the opinion of Martin Luther King, VIll. Must Macroeconomics be Built Up it is murder from Neoclassical First Principles? psychologically, to deprive a man of a job or n income. You are in substance saying to Keynes answered no. a practical man liv that man that he has no right to exist. "The ing in a complex world, he would not close truth, I think, lies somewhere between Lucas his eyes to apparent deviations from nar and King row-minded concepts of optimizing behavior Finally, it is important to remember that -nor even to gross deviations from rational little while a minority suffers much. Let me in labor supply. e modeling behavior as it yclical losses are not distributed uniformly, ity. He believed in as Lucas assumes; instead, most people lose was. Witness hi ense of money illusion illustrate with some simple calculations. Sup pose everyone has log utility and consumes Now ordinary experience tells us. that $3000 per quarter. Let a severe recession a situation where labor stipulates.. for educe consumption 4 percent. Utility falls 1 percent, which is no big deal, especially so far ey-wage rather than a real wage, ng a mere possibility
VOL. 77 NO. 2 THE CONTRIBUTION OF KEYNES AFTER 50 YEARS 35 It is sometimes from solid microfoundations. Thermody namics and chemistry, for example, have reduction of money-wage done pretty well without much micro theory but not to resist a reduction of real wages,. But, whether logical or il- Boyle's Law applies directly to aggregates, logical, experience shows that this much like the marginal propensity to con how labour in fact behaves p sume. And the microfoundations of medi cine are often very poor; yet much of it works. Empirical regularities that are for- Lucas and other new classicists take a mulated and tested directly at the macro different view. They emphasize the impor- level do have a place in science ance of building up macroeconomic rela Second, it is far from clear that the par tionships from sound microfoundations, by ticular first principles selected by new classi- which they mean the solutions to dynamic I economists deserve to come first. Why stochastic games. Lapses from what Lucas don' t people know the money supply or the alled"the only 'engine for the discovery of price level within very small margins of er truth'"are one of the chief grounds or? Who imposed a cash-in-advance con- which Keynesianism is branded unscientific. straint? Why should price move to equate Now, neither side is hostile either to first supply and demand in markets with asym principles or to factual accuracy. We all metric information? Why, Keynes might ask agree that the ideal macro theory would be are these postulates more acceptable as first built up logically from first principles and principles than nominal wage contracting would explain the data well. But we also Third, the model of man as a strongly agree that such a theory is a long way off. rational maximizer is not the only option The issue is how religiously we must adhe open to theorists. There are theories of to frictionless neoclassical optimizing princi bounded rationality” and of“ near ration ples until that glorious day arrives. Here the ality. " Even within the strict optimizing devoutness of American economists dis- framework, neoclassical tangencies are not tinguishes us from our colleagues in other the only, nor even the most likely, alterna- lands. But which attitude leads to better tive. Pervasive lumpy transactions costs lead science? Is it better to start deductively from to"the optimality of usually doing nothing, axioms or inductively from facts? When the meaning that it rarely pays to change your time comes to choose between internal con- decision variable, even if it is not set at the sistency and consistency with observations, frictionless"optimal "value. In a word, near which side should we take? Must we be rationality is full rationality. It is continuous restricted to microfoundations that preclude optimization that would be irrational he colossal market failures that created Direct empirical evidence on individual macroeconomics as a subdiscipline? behavior is difficult-some would say im Here followers of Keynes and followers of possible-to come by. But what little we Lucas often part company. Like Keynes, know from experiments by psychologists like modern Keynesians are inclined to begin by Daniel Kahneman and Amos Tversky and taking things as they are; rigorous opti- others does not suggest that homo sapiens mizing explanations for what they observe behaves like homo economicus( Perhaps that such as nominal wage contracts)can come is why they have different names. )Incon later. The important thing is to make sure sistent choices are common. People put too our models are congruent with the facts. much weight on what has happened to them Lucasian. it everse the se- and to their friends and too little quence. They want to begin with fully cal evidence. Framing of the question mat articulated, tractable models and worry later ters. The von Neumann-Morgenstern axioms about realism and descriptive accuracy are routinely violated. It is remarkable how This is a judgment call; but I judge the little impact this evidence has had on mod Keynesian approach more scientific. First, ern economics. Is that scientific detachment good science need not always be built up or religious zealotry
136 A EA PAPERS AND PROCEEDINGS MAY/98 So I have come to the end of my exam Lucas, Robert E,Jr, ""Models of Business with the conclusions you might have guessed Cycles, "paper prepared for the Yrjo at the outset: that when Lucas changed the Jahnsson Lectures, Helsinki, Finland answers given by Keynes, he was most- mimeo. March 1986 ly turning better answers into worse ones Prescott, Edward,"Should Control Theory be that modern Keynesian economics, though Used for Economic Stabilization? "in Karl far from flawless. has a better claim to Brunner and Alan Meltzer, eds, Optimal being“ scientific” than does new classical Policies, Control Theory, and Technological Exports, VoL. 7, Carnegie-Rochester Con ferences on Public Policy, Journal of Monetary Economics, Suppl. 1977, 13-38 REFERENCES Theory Ahead of Business Cycle Measurement, "Federal Reserve Bank of Friedman, Milton, "The Role of Monetary Pol- Minneapolis Research Department StafT icy, "American Economic Review, March Report 102, February 1986 1968,78,1-17 Sargent, Thomas J,"Beyond Demand an Keynes, J. M, The General Theory of Employ Supply Curves in Macroeconomics, ment, Interest, and Money, London: Har- American Economic re court, Brace and World, 1936 May1982,72,382-89
