Journal of Economic Literature Vol. XXVIII (December 1990), pp. 1645-1 660 A Quick Refresher Course Macroeconomics BY N. GREGORYMANKIW Harvard University and NBER This paper, though new, draws heavily on my previous paper, "Recent Developments in Macroeconomics: A Very Quick Refresher Course," Journal of Money, Credit, and Banking, August 1988, Part 2. I am grateful to Moses Abramovitz, David Laidler, and Thomas Mayer for comments, and to the National Science Foundation for jinancial support. Introduction Council (MPS) model. The job of refining these models generated many disserta- WENTY YEARS AGO, it was easier being tions. Private and public decision makers a student of macroeconomics. Mac- confidently used the models to forecast roeconomists felt more sure of the an- important economic time series and to swers they gave to questions such as, evaluate the effects of alternative macro- "What causes output and employment to economic policies. fluctuate?" and "How should policy re- Today, macroeconomists are much less spond to these fluctuations?" sure of their answers. The IS-LM model At the textbook level, the accepted rarely finds its way into scholarly jour- model of the economy was the IS-LM nals; some economists view the model model. It was little changed from John as a relic of a bygone age and no longer Hicks' (1937) interpretation of John May- bother to teach it. The large-scale mac- nard Keynes' (1936) once revolutionary roeconometric models are mentioned vision of the economy. Because the IS- only occasionally at academic confer- LM model took the price level as given, ences, often with derision. A graduate a Phillips curve of some sort was ap- student today is unlikely to devote his pended to explain the adjustment of dissertation to improving some small sec- prices. Some thought the Phillips curve tor of the MPS model. had the natural rate property, implying In contrast to this radical change in that the economy was self-correcting in the way academic macroeconomists view the long run. their field of study, applied macroecd- At the more applied level, this consen- nomists have not substantially changed sus was embodied in the large-scale the way they analyze the economy. The macroeconometric models, such as the IS-LM model, augmented by the Phillips MIT-Penn-Social Science Research curve, continues to provide the best way 1645 1646 Journal of Economic Literature, Vol. XXVZZZ (December 1990) to interpret discussions of economic pol- Approximately five centuries ago, icy in the press and among policy makers. Nicholas Copernicus suggested that the Economists in business and government sun, rather than the earth, is the center continue to use the large-scale macro- of the planetary system. At the time, he econometric models for forecasting and mistakenly thought that the planets fol- policy analysis. The theoretical develop- lowed circular orbits; we now know that ments of the past twenty years have these orbits are actually elliptical. Com- had relatively little impact on applied pared to the then prevailing geocentric macroeconomics. system of Ptolemy, the original Coperni- Why is there such a great disparity be- can system was more elegant and, ulti- tween academic and applied macroeco- mately, it proved more useful. But at the nomics? The view of some academics is time it was proposed and for many years that practitioners have simply fallen be- thereafter, the Copernican system did hind the state of the art, that they con- not work as well as the Ptolemaic system. tinue to use obsolete models because For predicting the positions of the plan- they have not kept up with the quickly ets, the Ptolemaic system was superior. advancing field. Yet this self-serving view Now imagine yburself, alternatively, as is suspect, for it violates a fundamental an academic astronomer and as an ap- property of economic equilibrium: It as- plied astronomer when Copernicus first sumes that a profit opportunity remains published. If you had been an academic unexploited. If recent developments in astronomer, you would have devoted macroeconomics were useful for applied your research to improving the Coperni- work, they should have been adopted. can system. The Copernican system held The observation that recent develop- out the greater promise for understand- ments have had little impact on applied ing the movements of the planets in a macroeconomics creates at least the pre- simple and intellectually satisfying way. sumption that these developments are of Yet if you had been an applied astrono- little use to applied macroeconomists. mer, you would have continued to use One might be tempted to conclude the Ptolemaic system. It would have that, because the macroeconomic re- been foolhardy to navigate your ship by search of the past 20 years has had little the more promising yet less accurate Co- impact on applied economists, the re- pernican system. Given the state of search has no value. Yet-this conclusion knowledge immediately after Coperni- also is unwarranted. The past 20 years cus, a functional separation between aca- have been a fertile time for macroeco- demic and applied astronomers was rea- nomics. Recent developments have just sonable and, indeed, optimal. not been of the sort that can be quickly In this paper I survey some of the re- adopted by applied economists. cent developments in macroeconomics. My intended audience includes those ap- A. A Parable for Macroeconomics plied economists in business and govern- A tale from the history of science is ment who often view recent research helpful for understanding the current with a combination of amusement, puzz- state of macroeconomics. Because I am lement, and disdain. My goal is not to not an historian of science, I cannot proselytize. Rather, it is to show how sev- vouch for its accuracy. But regardless of eral recent developments point the way whether it is true in detail, the story toward a better understanding of the serves nicely as a parable for macroeco- economy, just as Copernicus' suggestion nomics today. of the heliocentric system pointed the Mankiw: A Quick Refresher Course in Macroeconomics 1647 way toward a better understanding of tion in unemployment. planetary motion. Yet just as Copernicus The breakdown of the Phillips curve did not see his vision fully realized in and the prescience of Friedman and his lifetime, we should not expect these Phelps made macroeconomists ready for recent developments, no matter how Robert Lucas' (1976) more comprehen- promising, to be of great practical use sive attack on the consensus view. Lucas in the near future. In the long run, how- contended that many of the empirical re- ever, many of these developments will lations that make up the large-scale mac- profoundly change the way all econo- roeconometric models were no better mists think about the economy and eco- founded on microeconomic principles nomic policy. than was the Phillips curve. In particular, the decisions that determine most macro- B. The Breakdown of the Consensus economic variables, such as consumption The consensus in macroeconomics and investment, depend crucially on that prevailed until the early 1970s fal- expectations of the future course of tered because of two flaws, one empirical the economy h4acroeconometric models and one theoretical. The empirical flaw treated expectations in a cavalier way, was that the consensus view could not most often by resorting to plausible but adequately cope with the rising rates of arbitrary proxies. Lucas pointed out that inflation and unemployment experienced most policy interventions change the way during the 1970s. The theoretical flaw individuals form expectations about the was that the consensus view left a chasm future. Yet the proxies for expectations between microeconomic principles and used in the macroeconometric models macroeconomic practice that was too failed to take account of this change in great to be intellectually satisfying. expectation formation. Lucas concluded, These two flaws came together most therefore, that these models should not dramatically and most profoundly in the be used to evaluate the impact of alterna- famous prediction of Milton Friedman tive policies. (1968) and Edmund Phelps (1968). Ac- The "Lucas critique" became the rally- cording to the unadorned Phillips curve, ing cry for those young turks intent on one could achieve and maintain a perma- destroying the consensus. Defenders of nently low level of unemployment the consensus argued that users of merely by tolerating a permanently high macroeconometric models were already level of inflation. In the late 1960s, when aware of the problem Lucas defined so the consensus view was still in its heyday, forcefully, that the models were nonethe- Friedman and Phelps argued from mi- less informative if used with care and croeconomic principles that this empiri- judgment, and that the Lucas critique cal relationship between inflation and un- was right in principle but not important employment would break down if policy in practice. These defenses were not makers tried to exploit it. They reasoned heeded. that the equilibrium, or natural, rate of As I have mentioned, the consensus unemployment should depend on labor in macroeconomics broke down because supply, labor demand, optimal search of two flaws. Both were crucial. Neither times, and other microeconomic consid- the empirical flaw nor the theoretical flaw erations, not on the average rate of was, by itself, sufficient to cause the money growth. Subsequent events breakdown. As an exercise in intellectual proved Friedman and Phelps correct: In- history, it is instructive to consider two flation rose without a permanent reduc- counterfactuals. 1648 Journal of Economic Literature, Vol. XXVIZI (December 1990) Suppose the macroeconometric mod- teric and useless. Indeed, for practical els had failed to explain the events of purposes, it is. the 1970s, but macroeconomists had felt Let me divide recent developments in confident in the theoretical underpinning macroeconomics into three catagories.