6

Inflation and the Rational Expectations Revolution

The task which faces contemporary students of the business cycle is that of sorting through the wreckage.

—ROBERT LUCAS AND THOMAS SARGENT (1978)

As noted in the introduction, this book concentrates on the history of macroeconomic policy, not the history of economic doctrine. Nonetheless, as also mentioned there, the two get intertwined as theoretical developments influence (or fail to influence) policy and as real-world events influence (or fail to influence) theory. The rational expectations revolution offers prime examples of each, and it has had profound intellectual impacts on economic thinking about monetary policy. So a brief digression on rational expectations is appropriate at this point in the narrative.

In stark contrast to the Keynesian revolution, which had tremendous—and, I would argue, mostly salutary—influences on practical policy making, the rational expectations revolution created a deep intellectual chasm between the majority of academic macroeconomists and almost all real-world policy makers. And I include in the latter category the many professional economists serving in government. It is true that rational expectations left major and lasting impacts on the ways central bank economists think about, talk about, and model monetary policy making. But what was truly a revolution in the thinking of PhD economists left only small traces in the world of practical policy making.

Furthermore, the essence of that deep chasm between theory and practice, as we shall see in this chapter, inhered not in the assumption of rationality, but rather in a number of auxiliary ideas (like rapid market clearing) that got attached to rational expectations like barnacles.

In my perhaps biased view, the so-called rational expectations revolution would have set back monetary policy making by decades were it not for a simple fact: Actual policy makers in central banks had the good sense to (mostly) ignore it.

Acorns Quickly Grow into Oak Trees

Intellectual revolutions, like political revolutions, sometimes start quietly, almost unnoticed. The seeds are sown long before open rebellion breaks out. So it was with the rational expectations revolution. In 1961 John Muth, then a young economist at Carnegie Mellon University, published a profoundly important scholarly paper titled “Rational Expectations and the Theory of Price Movements” (Muth 1961). Although the paper was published in one of the leading economic journals, Econometrica, its initial impact was minimal. As Robert E. Lucas Jr., the acknowledged leader of the revolution that came later, recalled, “Of course we knew about [rational expectations]. Muth was a colleague of ours at that time. We just didn’t think it was important. The hypothesis was more or less buried during the ’60s” (Klamer 1984, 38). But it rose like Lazarus during the 1970s.

What became a tremendously successful intellectual revolution started slowly. Lucas, who was another young economist at Carnegie Mellon at the time, presented a short paper at a 1970 Federal Reserve conference in Washington titled “Econometric Testing of the Natural Rate Hypothesis” (Lucas 1972a). Being published in an obscure conference volume, the paper was not widely cited. Indeed, it is hardly ever cited even today. Nor was it well understood at the time. Lucas’s paper argued that simple models of inflationary expectations such as adaptive expectations, which assumed that people corrected their expected inflation rate by a fraction of their most recent forecasting error, were inconsistent with the Friedman-Phelps natural rate theory. The rational expectations model, on the other hand, was consistent with that theory. But testing the theory with rational expectations required more complicated econometric techniques than estimating conventional Phillips curves, which was the practice at the time.

If Lucas’s 1970 conference paper was not widely read and underappreciated at the time, his follow-up paper in the Journal of Economic Theory (Lucas 1972b) was simply baffling. (Try reading it.) The opening paragraph is worth quoting in full, however:

This paper provides a simple example of an economy in which equilibrium prices and quantities exhibit what may be the central feature of the modern business cycle: a systematic relation between the rate of change in nominal prices and the level of real output. The relationship, essentially a variant of the well-known Phillips curve, is derived within a framework from which all forms of “money illusion” are rigorously excluded: all prices are market clearing, all agents behave optimally in light of their objectives and expectations, and expectations are formed optimally (in a sense to be made precise below). (Lucas, 1972b, 103)

Notice the amazing irony here. Lucas offered his Journal of Economic Theory paper as a rigorous foundation for the Phillips curve—micro foundations that he later deemed lacking if not impossible—in a model economy in which the existence of such a link between nominal and real variables seems most unlikely. The key assumptions that made the link unlikely were instantaneous market clearing and continuous optimization, not the rationality of expectations. These are the features that really separated what would come to be called new classical models from traditional Keynesian models. But it took quite a while for the economics profession to realize this.

The central contribution of Lucas (1972b) is elucidated in his footnote 7, where he dwells on the assumed efficiency of markets in the model but then adds, almost as an afterthought, the following pregnant remark: “It will also be true that price expectations are rational in the sense of Muth” (Lucas 1972b, 110). I think it is fair to say that Lucas’s densely mathematical treatment of an age-old idea—the neutrality of money—had relatively little impact at the time. But it was destined to become a classic. In the 1980s and 1990s while teaching macroeconomic theory to PhD students at Princeton, I used to joke that graduate students in top economics departments were being brought up to believe that Bob Lucas invented the neutrality of money in 1972! David Hume? Who was he?

What Lucas actually accomplished in his Journal of Economic Theory paper was to bring Muth’s idea of rational expectations, meaning that subjective expectations matched the mathematical expectations implied by the model, from obscurity into the mainstream and soon into a dominant intellectual position.

But the revolution was still incipient at that point. The Lucas paper that really called attention to rational expectations and to its potentially stunning implications for monetary policy came a year later in the much more widely read American Economic Review. It was an instant sensation. In a sense, Lucas (1973) “publicized” the idea of rational expectations and demonstrated its power in monetary models.

That 1973 paper also reemphasized what would soon come to be called “the Lucas supply function,”¹ the notion that the deviation of actual from potential output depends on the price level “surprise,” that is, on the amount by which p_t exceeds its (rationally) expected value:

(1)

where y is the log of real output, y* is its potential (full information) value, p is the log of the price level, and _t−1p_t is the mathematical expectation of p_t given all the information available up to time t − 1.

In Lucas’s (1973) specific model, such surprises were based on agents’ confusion between changes in relative prices and changes in the absolute price level. Rational expectations, of course, implied that the “surprise” term had to have mean zero, implying that also had to have mean zero. No one—certainly not Lucas—paid much attention to the fact that price level surprises could not be large with monthly data on the Consumer Price Index unless the annual inflation rate was gigantic. And these surprises could not last long. Each month’s Consumer Price Index was publicly announced with only a short time lag.

In stark contrast to his Journal of Economic Theory model, Lucas’s American Economic Review model was straightforward, easy to understand, and easy to teach to students. The paper also purported to offer supporting empirical evidence, though the “good fit” of Lucas’s cross-country regression turned out to depend almost entirely on two outlier countries with very high inflation rates: Argentina and Paraguay. If Lucas’s two papers published in 1972 were like subversive whispers in Fraunces Tavern, his 1973 paper was like Lexington and Concord: the intellectual shot heard ’round the world. Soon the Lucas (1973) model was a staple of teaching in graduate macro courses all over the world. The rational expectations revolution was truly launched.

And also like Lexington and Concord, Lucas (1973) was just the beginning of something much larger. Notice that the Lucas supply function (1) implies that only price level surprises, and therefore only monetary policy surprises, can move output. That point, which was not heavily emphasized in Lucas (1973), was brought to the fore in a celebrated 1975 paper by Thomas Sargent and Neil Wallace.

Sargent and Wallace (1975) originally set out to extend Poole’s (1970) simple analysis of interest rate versus money supply rules in a variety of ways. But it turned out that only one change really mattered: modeling price expectations as rational rather than autoregressive. Soon Poole’s question—should the central bank base its monetary policy on M or r?—was forgotten because, as Sargent and Wallace (1975, 242) put it, if expectations are rational, “one deterministic money supply rule is as good as any other,” which is to say not good at all. This revolutionary idea was soon branded “the policy ineffectiveness result” because, under rational expectations, it is pretty much impossible for a central bank to engineer unanticipated movements in the money supply regularly.

Leader of the Rational Expectations Revolution²

Robert (“Bob”) Lucas, who followed Milton Friedman as the acknowledged leader of the Chicago School of macroeconomics, was born in Yakima, Washington, to parents who admired Franklin Roosevelt and the New Deal. The Lucas family ran a small restaurant that fell victim to the 1937–1938 recession. But Robert E. Lucas Jr., a baby at the time, was far too young to take notice. His interest in business cycles came later and had other sources.

Lucas was an excellent math-science student in high school. He recalls getting his “first taste of real applied mathematics, and an exciting one,” by helping his father, who had not attended college, with a calculus problem. When it came time for college, MIT did not offer Lucas a scholarship, but the University of Chicago did. (The invisible hand at work, perhaps?) So, he matriculated there. Lucas intended to study mathematics but was quickly caught up in the college’s unique liberal arts curriculum, which drew him to both history and an academic career. After a brief stop at Berkeley, where he was exposed to economics, he returned to Chicago for graduate study.

Ironically, it was the “confident and engaging style” of Paul Samuelson’s Foundations of Economic Analysis (1947) that captured Lucas intellectually in graduate school. He recalls “working through the first four chapters, line by line, going back to my calculus books when I needed to,” and winding up “as good an economic technician as anyone on the Chicago faculty. Even more important, I had internalized Samuelson’s standards for when an economic question had been properly posed and when it had been answered.”

The rest, as they say, is history, including being taught price theory by the persuasive Milton Friedman. Lucas recalls that he “tried to hold on to the New Deal politics I had grown up with,” but Friedman’s influence made that difficult. Few of Lucas’s fellow economists think of him as a New Deal liberal today.

The essence of the policy ineffectiveness result is straightforward: only “unanticipated” monetary policy shocks can cause price level surprises, and therefore under the Lucas supply function, only they can move real output. That left counterrevolutionaries only two ways out.

You could dispute the rationality of expectations. But what was the alternative, irrational expectations?³ That route did not seem promising in a discipline built around the hyperrational concept of homo economicus.⁴ Or you could find fault with the Lucas supply function. Defenders of monetary policy effectiveness naturally concentrated their energies there. Stanley Fischer (1977) embraced rational expectations but noted that with long-term contracts, expectations that were rational at the time contracts were written might no longer be rational given current information. John Taylor (1980) took this idea, added staggering of wage contracts, and produced an ingenious model that generated realistic business cycles even with rational expectations. And there was much more.⁵

Sargent and Wallace’s (1975) argument for policy ineffectiveness was based entirely on a priori reasoning—on theory, not on facts. But Robert Barro (1977) soon offered what he claimed to be strong empirical evidence in support of their hypothesis. To do so, he had to devise a way to divide data on money growth into anticipated and unanticipated components and then to show empirically that only the latter moved unemployment. This is precisely what he did. Five years would pass before Frederic Mishkin (1982) and Robert Gordon (1982) effectively debunked Barro’s findings. But it took very long distributed lags to do so, leaving many believers in the Sargent-Wallace proposition unconvinced.

One more academic preliminary is important to our story—very important—though I have barely mentioned it up to now.

At the inaugural Carnegie-Rochester conference in April 1973 (but not published until 1976), Lucas presented one of the most celebrated and influential scholarly papers of the second half of the twentieth century, “Econometric Policy Evaluation: A Critique.” What would soon simply be called the “Lucas critique” was that standard econometric models and methods ignored the fact that changes in policy could cause changes in econometrically estimated parameters through expectational channels. So, adopting a different policy rule might change the quantitative dimensions of the policy’s effects, thereby making the econometric evidence on which it had been based misleading.

We have met one such example already, in chapter 3. Sargent (1971) noted that a change in the autoregressive model generating actual inflation would change how rational agents use past data to forecast future inflation and therefore change the coefficients in an empirical Phillips curve of the form π_t = α(L)π_{t − j} + f (U_t) + ε_t, where α(L)π_{t − j} is a distributed lag of past inflation rates, used as a proxy for expected inflation.

Lucas (1976) showed that the same idea applied to the analysis of the investment tax credit (ITC) and to temporary versus permanent changes in income taxes. All three examples—the Phillips curve, the ITC, and temporary income tax changes—were known prior to Lucas’s 1976 paper. But his demonstration of the underlying unity of the three examples, plus the potential application of the same ideas elsewhere, made the paper an intellectual tour de force. More important, the Lucas critique is absolutely correct. Changes in expectations can render conventional econometric estimates of policy effects misleading. As Lucas (1976, 279) noted in that famous paper, “The argument is, in part, destructive.”

Indeed it was. Lucas’s observation quickly became an understatement, as the verb can got ignored in the academic frenzy—or rather turned into the verb do. Rather than viewing the Lucas critique as one among many pitfalls in doing empirical work with nonexperimental time series data, rational expectations revolutionaries elevated it to primus inter pares without question—and without any evidence of its practical importance. No one stopped to ask, for example, whether Lucas critique problems were more important in practice than, say, bias from omitted variables or endogeneity of right-hand variables.

Rather than serving as a well-founded warning that could have (and should have) improved econometric practice, the Lucas critique led to what might be called econometric nihilism on a grand scale.⁶ Many rational expectations revolutionaries simply asserted that all “old-fashioned” models then being used for policy analysis were systematically misleading and should therefore be ignored. In their place, they offered what? Well, mainly a priori reasoning—and reasoning that struck many mainstream economists as wrongheaded.

In chapter 3, I mentioned a virtually unknown paper of mine, published in 1988, that cast doubt on the empirical importance (not on the intellectual coherence) of the Lucas critique in the Phillips curve context. In a 1984 paper that is perhaps even more unknown, I used three of those “old-fashioned” giant macroeconometric models to appraise the early effects of Reaganomics. After presenting a bevy of estimates, I observed that “there is no doubt that the … models are vulnerable to the Lucas critique” (Blinder 1984, 223). But then I went on to investigate something the rational expectations revolutionaries never did: whether the models displayed large errors in the places and times where we would expect expectational effects to be most important.

For example, the 1981 Reagan tax cuts accelerated depreciation allowances but promised even more accelerated depreciation rules a few years down the line. Standard Lucas-style (1976) reasoning therefore predicts that businesses should have postponed investments until the more generous depreciation schedules went into effect. If that expectational effect was quantitatively important, it should have led models that ignored it to overpredict investment spending in 1981 and 1982. But they didn’t. There were several other such examples, leading to the overall conclusion that “my investigation does not suggest that the Lucas critique is of great empirical importance” (Blinder 1984, 226).

So here was the state of play in academia circa 1977:

·                    Academic economists were madly in love with rational expectations, barely recognizing that the term really meant “model consistent” expectations, no matter how silly the model was.

·                    Sargent and Wallace’s policy ineffectiveness result had convinced many (though not all) academic macroeconomists that central banks that tried to mitigate business cycles were wasting their time because systematic monetary policy reactions to, say, unemployment could never cause unanticipated money shocks.

·                    The rational expectations revolutionaries completely ignored the supply shocks discussed in chapter 5. Rather, they sought to convince everyone that the high inflation the United States had experienced in the early 1970s was evidence that Keynesian economics had failed. To a disconcerting extent, they succeeded in undermining Keynesianism.

·                    And on top of all this, they insisted, no one should pay attention to results generated by models that fiscal and monetary policy makers had used for years to plan and appraise stabilization policies, for they all ran afoul of the Lucas critique. All you had to do to stop the show in those days was shout “Lucas critique” in a crowded seminar room.

The aphorism of the day was that “there are no Keynesian economists under the age of 40” (Blinder 1988, 278). I know that was untrue because I turned thirty-two in 1977. But in any case, citizens of the real world were fortunate that decisions on monetary and fiscal policy were firmly in the hands of people over forty.

Against this anti-Keynesian backdrop, the Federal Reserve of Boston convened a high-level conference on Martha’s Vineyard in June 1978, giving it the provocative title “After the Phillips Curve.” After? Robert Gordon, who was not among the attendees, would certainly have called the year 1978 during the Phillips curve era, not after it. He was busy patching up the traditional Phillips curve to account for supply shocks.

The most famous or infamous paper delivered at that memorable conference was coauthored by Lucas and Sargent and titled “After Keynesian Macroeconomics” (Lucas and Sargent 1978). There’s that provocative word again, after, but now applied more broadly to the entirety of Keynesian economics. Their attack was broad and highly polemical—and not by accident. Lucas and Sargent (1978, 81) wrote after the conference that “since both of us are on record as rather severe critics of Keynesian macroeconometric models, we assumed that we were included in the program to express this dissenting view as forcefully and as accurately as possible.” And they sure did.

Lucas and Sargent (1978, 49) declared that the “predictions” of Keynesian economics “were wildly incorrect, and … the doctrine on which they were based is fundamentally flawed.” Furthermore, they asserted, these criticisms of Keynesianism “are now simple matters of fact, involving no novelties in economic theory. The task which faces contemporary students of the business cycle is that of sorting through the wreckage.” And that was all on the first page. There was more to come.

According to what you might call the Lucas-Sargent critique, the prevailing Keynesian models, particularly their Phillips curve implications, were guilty of “econometric failure on a grand scale” (Lucas and Sargent 1978, 51) for missing the stagflation of 1973–1974 (something that is explained fully by supply shocks). Furthermore, patching up the models—as Keynesians had already done before the Boston Fed conference—was useless because “the difficulties are fatal: … modern macroeconomic models are of no value in guiding policy” (50). Fightin’ words, you might say. Naturally, mainstream Keynesians fired back. Benjamin Friedman and Robert Solow were the official discussants of the Lucas-Sargent paper at the conference, and their remarks displayed a level of disdain that matched that of Lucas and Sargent. More important and less polemically, Gordon (1977) subsequently showed that his vintage 1972 Phillips curve, once modified for supply shocks in straightforward ways, fit the data quite well. But no matter. The Lucas critique was interpreted as telling economists to ignore all such evidence.

The battle was truly joined, lasted for years, and was often less than friendly. (I know; I was among the foot soldiers.)⁷ It was a tumultuous time in academic macroeconomics. Many of the protagonists initially acted as if the rationality of expectations was the key dividing issue. It was not. Fischer (1977), Taylor (1980), and others showed that the true dividing line was rapid (instantaneous in the models) market clearing. Naturally, of course, if all markets clear instantly, you don’t need government actions to fix things up. But in retrospect, it seems a waste of intellectual resources that academic economists expended so much pen and ink arguing over whether all markets—including the aggregate labor market—cleared virtually instantly. But that’s what they did.

Meanwhile, Back in the Real World …

The 1970s and 1980s were also tumultuous times in the real world of monetary policy but for very different reasons. Broadly speaking, around 1980 or so it seemed as though the major industrial nations of the world looked around at the high inflation rates, decided they had had enough, and took actions to get rid of them. With variations in timing, Paul Volcker in the United States, Margaret Thatcher in the United Kingdom,⁸ the Bundesbank in (West) Germany, and inflation fighters in other nations declared all-out war on inflation. If the rational expectations revolution, in particular the policy ineffectiveness result, had any influence on these policy makers’ thinking, it sure didn’t show. Central banks brought inflation down the old-fashioned way: by using tight money to cause deep recessions.

The justly celebrated Volcker disinflation in the United States worked well, but that’s because it drove the unemployment rate to a high-water mark of 10.8 percent in November and December 1982, at the time the highest reading since the 1930s. Given the unending debates in academia over the channels of monetary policy, I once asked Volcker after he had left the Fed how he thought monetary policy worked to crush inflation. His answer surprised me: by causing bankruptcies.⁹

In the United Kingdom, real GDP tumbled 5.3 percent between 1979:2 and 1981:1, and the Thatcher slump raised the unemployment rate from 5.4 percent in 1979 to 11.5 percent by 1983. Just as in the United States, that was the highest reading since the 1930s. But there too, the bitter medicine worked: consumer price inflation crumbled from over 15 percent to under 5 percent. Like Volcker in the United States, Thatcher pledged allegiance to monetarism. And like the experience of the Volcker Fed, money growth in the United Kingdom was far from stable.

Though the details and timing varied, it is hard to make a case that the Volcker and Thatcher recessions, as well as others in other countries, were consequences of unanticipated monetary contractions. Indeed, the iron-willed central banker in the United States and the Iron Lady in the United Kingdom practically shouted their attentions from the rooftops—and then followed through on their promises. True believers in Sargent-Wallace could, I imagine, argue that no one believed Volcker, Thatcher, and others. But that doesn’t seem likely, given Volcker’s reputation as a determined inflation hawk. Rather, the disinflations of the early 1980s look like clear cases of anticipated monetary tightenings causing deep recessions.

Notice the great historical irony here. One main reason why rational expectations macroeconomics, in particular its implication that the Phillips curve for anticipated changes in money is vertical even in the short run, caught on was the allegation that the incumbent Keynesian tradition had failed to either control or explain high inflation. “Failed to control,” I suppose, was true, though the cost of stabilizing inflation in the face of huge supply shocks would have been devastating. But “failed to explain” was clearly a false charge once you gave Keynesian economists a few months to extend their framework to include supply shocks. That’s all it took.¹⁰

Did the rational expectations approach, especially what would soon be called “new classical economics” because of its resemblance to pre-Keynesian thinking, offer a superior alternative? I don’t think either Volcker or Thatcher thought their tough monetary contractions were “unanticipated.” They certainly didn’t intend them to be. But tight money in the United States and the United Kingdom seemed to have powerful effects on real output and employment—Keynesian effects, you might call them. I think both Volcker and Thatcher expected that too.

The fact that the rational expectations revolution swept the academy while all this was going on in the real world stands as a testament to academics’ ability to closet themselves away in ivory towers at times. Had they peeked outside, they would have seen the real effects of anticipated monetary tightenings right before their eyes. Even the Phillips curve worked well, at least in the United States, as we shall see in chapter 8.

Chapter Summary

The storming of the Bastille in 1789 marked neither the beginning nor the end of the French Revolution, but it sure shook up the status quo. The same could be said about the 1970s salvos of the rational expectations revolution lobbed by Lucas, Sargent, and others. In the academy their effects were profound, questioning the very basis—never mind the details—of stabilization policy, especially of monetary policy.

But in the world of actual policy, it is hard to find more than a trace of influence of the rational expectations revolution on policy making. That trace—the idea that policy makers should pay closer attention to expectations formation, including expectations about future policy—was no doubt salutary. And it has (deservedly) lasted. But if rational expectations reasoning made policy makers anywhere believe they could move real variables only through monetary surprises, they sure kept that belief to themselves.

During the notable disinflations of the late 1970s and early 1980s, real-world central bankers in the United States and elsewhere certainly didn’t act as if they were looking to engineer monetary surprises. They looked and acted as if they believed that monetary stringency would beat inflation, and they were glad to announce their intentions. In short, rational expectations and new classical economics rode to academic victory on the backs of inflation just as actual policy makers were conquering inflation the old-fashioned way—with tight money and high unemployment. Ironic.

______________

1. I say “reemphasized” because the Lucas supply function had been introduced in the Lucas (1972a) paper.

2. Much of this material and all the quotations come from Lucas’s Nobel Prize autobiography (Lucas 1995).

3. Years later, behavioral economists would argue that many aspects of human behavior are not rational in the strictly economic sense.

4. The assumption of rationality in expectations is not easy to test. Nonetheless, as the years went by more and more evidence mounted against it. Two of the many such papers are Lovell (1986) and Fair (1993). None of this was known in 1975, of course.

5. The simple Lucas supply precluded serially correlated deviations of output from potential. That was noticed immediately and spawned a cottage industry of more complicated models that provided a wide variety of reasons for serial correlation. See, for example, Kydland and Prescott (1982) and Blinder and Fischer (1981).

6. This statement is not fair to Sargent, Lars Hansen, and others who attempted to devise methods of estimating “deep structural” parameters that would not change when policy changed. These methods, however, relied on so many dubious and restrictive assumptions that many economists found the work technically demanding but ultimately unconvincing.

7. See, for example, Blinder (1987b). Lucas was present at that American Economic Association session and took great umbrage at my remarks.

8. The Bank of England was not independent in those days. Monetary policy was governed from Whitehall.

9. Not that Volcker relished driving businesses into bankruptcy. He just thought that was how the medicine worked.

10. See, for example, Gordon (1975).