“The 1970 Phelps volume has been extremely influential in macroeconomics. Three of its contributors went on to win Nobel prizes for work detailed in the book, and it inspired many others who contributed to the small equilibrium models that became the workhorses of macroeconomics. Yet virtually all of these models use the assumption of rational expectations. In this new volume, Phelps and Roman Frydman assemble a new group of scholars to critique the work based on rational expectations. Phelps and Frydman argue that rational expectations destroyed one of the key premises of the original book–that independent expectations are critical for understanding macroeconomic phenomena. The contributors to this follow-up volume make a convincing case for the failure of several models with rational expectations, and present thought-provoking alternatives. Their efforts to build macroeconomic models without the rational expectations hypothesis might have the impact in their areas of research that the original volume had.”–Christopher Pissarides, Nobel Laureate in Economics

“Microeconomic Foundations of Employment and Inflation Theory, edited by Edmund Phelps forty years ago, established the concept of ‘micro foundations’ as an essential macroeconomics idea. Later, ‘rational expectations’ was added as the second pillar of the current standard macro model. Recent events have challenged the validity of that model. This new Phelps volume, coedited with Roman Frydman, challenges and offers alternatives to the second pillar while retaining the first. It is a must-read for anyone interested in modern economic thought and its implications for policy.”–Dale Mortsensen, Nobel Laureate in Economics

“A great volume.”–Peter Howitt, Brown University

“This book brings together an exceptional group of economic theorists who discuss future avenues that the economics profession can take to replace the paradigm of rational expectations. The task is challenging and the outcome of the project still uncertain, but all the chapters are very interesting.”–Fabrizio Coricelli, Paris School of Economics

Roman Frydman is professor of economics at New York University and the coauthor (with Michael D. Goldberg) of Beyond Mechanical Markets and Imperfect Knowledge Economics. Edmund S. Phelps, the winner of the 2006 Nobel Prize in Economics, is director of Columbia University’s Center on Capitalism and Society. His many books include Structural Slumps and Seven Schools of Macroeconomic Thought.

Rethinking Expectations

THE WAY FORWARD FOR MACROECONOMICS

PRINCETON UNIVERSITY PRESS

Copyright © 2013 Princeton University Press
All right reserved.
ISBN: 978-0-691-15523-4

Contents

Which Way Forward for Macroeconomics and Policy Analysis? Roman Frydman and Edmund S. Phelps………………………………………..11 Expectational Coordination Failures and Market Volatility Roger Guesnerie……………………………………………………….492 Learning as a Rational Foundation for Macroeconomics and Finance George W. Evans and Seppo Honkapohja………………………………683 Keynes on Knowledge, Expectations, and Rationality Sheila Dow………………………………………………………………….1124 The Imperfect Knowledge Imperative in Modern Macroeconomics and Finance Theory Roman Frydman and Michael D. Goldberg…………………1305 Heterogeneous Gain Learning and Long Swings in Asset Prices Blake LeBaron……………………………………………………….1696 Opening Models of Asset Prices and Risk to Nonroutine Change Roman Frydman and Michael D. Goldberg…………………………………2077 Animal Spirits, Persistent Unemployment, and the Belief Function Roger E. A. Farmer………………………………………………2518 Indeterminacies in Wage and Asset Price Expectations Edmund S. Phelps…………………………………………………………..2779 The Long Swings of Employment, Investment, and Asset Prices Gylfi Zoega…………………………………………………………30110 Imperfect Knowledge, Asset Price Swings, and Structural Slumps Katarina Juselius………………………………………………..32811 Stabilization Policies and Economic Growth Philippe Aghion and Enisse Kharroubi…………………………………………………35112 Swings and the Rules-Discretion Balance John B. Taylor……………………………………………………………………….37313 Principled Policymaking in an Uncertain World Michael Woodford………………………………………………………………..389Contributors………………………………………………………………………………………………………………..415Index………………………………………………………………………………………………………………………421

Chapter One

Expectational Coordination Failures and Market Volatility

Roger Guesnerie

1.1 Introduction

The stability of market economies has been a recurrent subject of debate since the beginning of the nineteenth century, which is usually viewed as the starting point of economics as a scientific field. “L’offre crée sa propre demande” (“supply creates its own demand”): this formula, a remarkable digest of the argument, is supposed to capture the essence of Jean Baptiste Say’s analysis (1803). It expressed the basis for some of the early protagonists’ strong confidence in the systemic stability of markets. But others strongly disagreed: Jean Baptiste Sismondi and the “catastrophist school”—and its most famous adept, Karl Marx—awaited the next crisis (possibly the final crisis) of the capitalist system. In the middle, Léon Walras thought that Say’s argument was unconvincing but developed an alternative analysis that has long been viewed as supporting the optimists. The skepticism about market economies’ systemic stability reappeared after the 1929 crisis, and Keynes forcefully argued for government intervention to counter markets’ instability.

At the beginning of the current millennium, mainstream economic theory had apparently rallied to Say’s optimism. Even if markets may not be fully self-regulating, our knowledge had improved, so that, according to Robert Lucas (2003), the “central problem of depression prevention has been solved.” Was not the “great moderation” (see Bernanke 2008) proof of the adequacy and efficiency of the prevailing macroeconomic and monetary policies? But the argument reflected an appraisal of the facts that was intellectually and geographically biased (see Reinhardt and Rogoff 2008). Facts, such as crises and bubbles, are stubborn, and their stubbornness challenges the system of explanations that dominate contemporary economics. Neither the volatility of financial markets nor the logic of crises is satisfactorily explained by the best of our existing models.

Many economists, though not all, would agree that the recent crisis has raised questions about the state of economic theory; a subset would probably agree that reading the crisis through the standard lenses of the dominant approaches signals an outright failure of economic theory. Going somewhat further, I am among those who believe that economic theory has a key responsibility in the financial crisis that triggered recent economic events. The bottom line is that economists have provided an overly optimistic view of the workings of financial markets (many specialists’ emphasis on the Efficient Market Hypothesis is a spectacular illustration of this bias). Such an overly optimistic view in the group of financial practitioners—a group already prone to self-satisfaction and reluctant to submit to regulation—triggered the deployment of uncontrolled imagination. Exaggeratedly sanguine conclusions, whether in finance or macroeconomics, reflected uncritical acceptance of modeling principles that, without justification, had become axioms. The question of what went wrong with standard economic theory in general and with its modeling principles in particular is, however, likely to suggest a variety of answers. Let me sketch three.

1. The first answer does not refer so much to modeling principles as to the diversification of modeling. There has been a multiplication of fronts in modern economic research, but progress on each front is increasingly obscure to outsiders, even those from neighboring subfields. I have argued elsewhere that the balkanization of knowledge is more of a problem in the study of economics (or social science more broadly), where social action must rely on all dimensions of understanding, than in a field like physics, where applications depend significantly on specifics. From my teaching experience, I have been struck by the fact that, in the field of finance, the communication between, say, the subfields of standard asset pricing, mathematical finance, “informational” finance, corporate finance, the econometrics of finance, and so forth, is surprisingly limited. Moreover, the difficulty of constructing a synthetic view of the different fronts is exacerbated by the question of systemic stability—not only of the financial system but also of the economy as a whole.

2. A second line of criticism concerns the rationality hypothesis, which has long been a quasi-obligatory ingredient of economic analysis. Real economic agents do not have the supposed rationality of Homo economicus. In fact, the rapid development of behavioral economics in the past 20 years has highlighted the limits of the standard concepts of rationality: behavior under uncertainty, time consistency, and symmetric treatment of losses and gains, to cite a few research themes, have been the subject of critical reconsideration. Hence the problem is not so much a failure to analyze the limits of the standard conception of rationality but rather a failure to focus on the consequences of this reconsideration in, say, finance and macroeconomics.

3. The failure of economic theory stressed here concerns expectational coordination. The changing assessment of expectational coordination in our models reflects major trepidation: either, as in most of the period before the 1950s, expectations are modeled as exogenous (the approach taken until recently in temporary equilibrium modeling of dynamics), or from the mid-1960s they have increasingly become “right on average,” the essence of the Rational Expectations Hypothesis (REH), which emerged in the 1960s. The first option, taking expectations as given, can be at best a preliminary step in a more complex analysis: market expectations have to be explained and not taken as exogenous data. But the ability of decentralized systems, such as markets, to produce good (in the sense of REH) coordination also has to be explained. And it may or may not be possible to produce reasonable explanations. In the latter case, one must conclude that markets are likely to fail to coordinate expectations in a satisfactory way (satisfactory in the sense of REH but also in many other senses of the word).

Section 1.2 returns briefly to the history of economic thought after World War II. It recalls the rise of REH in the field of formalized economic theory. Section 1.3 reviews the critical assessments of REH that have developed, particularly since the beginning of the 1980s. Section 1.4 illustrates how a critical approach, and in particular the one associated with my own research, may trigger a dramatic change in our understanding of economic problems and in how we evaluate economic policies. I stress three examples: the economic role of speculation, the so-called informational efficiency of markets, and the ability of agents with long horizons to predict the future. Section 1.5 recapitulates the arguments and underscores future challenges.

1.2 The Rise of REH in Modern Theory

REH states that economic agents have a view of the future that is “basically right”: this view may vary, depending on individual information, but it is not biased. That the hypothesis is distinct from the hypothesis of individual rationality is obvious but still worth emphasizing. The idea that REH is nothing else than an extension of the rationality hypothesis to expectations has been a misleading but (unfortunately) extremely popular argument in favor of REH. In fact, in a large economy, it is “right” (individually rational) to adopt REH, but only if the other agents adopt it as well. Otherwise it is wrong. To put it in game-theoretical parlance, the rational expectations equilibrium is a Nash equilibrium, not a dominant strategy equilibrium.

The rise of the debate on expectations in economic theory can be traced to the 1950s and 1960s. Although Muth’s (1961) pioneering article addressed a microeconomic problem, the polemics triggered by some so-called Keynesian macroeconomic policies resonated in the background: agents can be fooled once or twice (implicitly, by the government) but not endlessly: they will refer (or one has to model them as if they refer) to the “relevant economic theory.” The Relevant Economic Theory Hypothesis then became the Rational Expectations Hypothesis, a more appealing, though misleading, label. Then REH progressively took over theoretical modeling, or, if one prefers, “formalized economic theory.”

Start with the field of general equilibrium, which emphasized a formally static model (Debreu 1959) that nonetheless supported an atemporal implementation of intertemporal equilibrium. In a truly intertemporal framework, the economy consists of a succession of spot markets for goods and financial markets in each period (replacing the futures markets that existed at the beginning of time in the static theory). The equilibrium becomes an “equilibrium of plans, prices and price expectations” (Radner 1972), an equilibrium that may be interpreted either as involving perfect foresight or, in the broader sense stressed here, rational expectations. In this general framework, if markets are “essentially complete,” the outcome is efficient and replicates the complete markets solution of the static model (see Arrow 1953; Guesnerie and Jaffray 1974). In this setting, both incompleteness and the inadequacy of price expectations are sources of markets failure.

Moreover, since Walras, economists are aware that the market, to deliver the prices that solve the equilibrium equations, has to substitute for a computing machine. According to Walras, the virtues of the market-computing algorithm are not obvious and have to be demonstrated (this is the purpose of tâtonnement theory). In the intertemporal context, the “equilibrium of plans, prices and price expectations” possibly hides two dei ex machina: spot markets clear at a given time (one of the algorithmic virtues of the market, which normally is subject to verification), but this process relies on an understanding of the clearing of the next period’s spot markets, the equations of which are implicitly supposed to be resolved in people’s minds. Hence, part of the market “equilibrations,” in the sense of Perroux (1973), can be viewed as the product of agents’ collective thinking. Such a collective thought process helps to resolve the Walrasian auctioneer’s problem: the mental activity of the agents mimicking the calculation of tomorrow’s market algorithm.

General equilibrium provides a good reference for the development of contemporary economics. Most intertemporal modeling in most subfields—whether with a general equilibrium flavor (as in trade) or with a partial equilibrium focus (labor markets, insurance, etc.)—routinely adopts REH.

Modern finance is partially rooted in general equilibrium, and REH has a clear hegemonic position in the field and underlies its influential theoretical models, for example, those stressing the informational efficiency of markets, even in its most critical versions (Grossman and Stiglitz 1980).

Last but not least, REH has acquired a hegemonic position in modern macroeconomics. Lucas’s (1972) dismissal of voluntarist policies takes place in a rational expectations world. The Real Business Cycles (RBC) models, sometimes presented as “Walrasian,” describe long-lived (identical) agents who anticipate the entire future correctly (in the sense of REH). In this world, there are few spot market adjustments, most adjustments come from expectations, and agents are as good as the best theorist at computing complex equilibria. New Keynesian models echo some of the Walrasian preoccupation with the implementation of the equilibrium. Prices are quoted, not tentatively by the auctioneer, but irrevocably in each round of quotation by firms with market power. The procedure introduces frictions, which indeed have a Keynesian flavor, but the announcement of prices, which provides a non-Walrasian response to the Walras problem, relies on REH: firms have a correct understanding of the future, particularly of the future flows of bids by rival firms. “Good” expectational coordination is also taken for granted.

1.3 Directions for a Critical Assessment

If REH has gained a hegemonic position in formal economic theory, paralleling the rise of the Nash equilibrium in game theory, it has not been unchallenged. I review three avenues of criticism.

1.3.1 Internal Challenges

In economic theory as well as in game theory, an internal challenge (internal in the sense that it arises even if you find the theory’s assumptions impeccable) arises from multiplicity. In a game setting, the challenges take the form of a question: what happens if there are several Nash equilibria? Rephrased in Muth’s defense of REH, the question becomes: what is the relevant economic theory when there are several candidates?

In economic models, the multiplicity challenge is particularly acute in infinite-horizon models. For example, consider the simplest model of this sort, a one-dimensional, one-step-forward-looking model, where the state at time t obtains as a function of the expectation of the state at time t + 1. Typically, such a model has at least a steady state (x* such that x* = f (x*)) but also a continuum of perfect-foresight equilibria, which may (or may not) remain close to the steady state (in the former case, the steady state is referred to as “indeterminate”; in the latter, it is said to be “determinate”). Also, it has been understood that, besides the “focal solution” (the perfect-foresight steady state equilibrium), there could also be well-behaved (stationary) stochastic rational expectations equilibria: indeed, the stochastic stationary beliefs governing sunspot equilibria look arbitrary but are self-fulfilling (see Azariadis 1981; Azariadis and Guesnerie 1982, 1986; Benhabib and Farmer 1994; Farmer and Woodford 1997; for a broader approach and an attempt at synthesis, see Chiappori and Guesnerie 1991; Guesnerie and Woodford 1992; Guesnerie 2001). The lessons drawn from the sunspot literature are fairly general. For example, in a slightly more complicated framework than the one just introduced (one-dimensional, memory one, one-step-forward-looking models), the focal solution switches to a saddle-path trajectory, although the logic of multiplicity (whether sunspot or nonsunspot) remains the same, so that the investigation triggers analogous results. Adding complexity (see Evans and Guesnerie 2005) and/or intrinsic noise does not change the flavor of the analysis.

Another line of research that falls under the heading of a critical assessment of REH is the line associated with the global games literature initiated by Carlsson and Van Damme (1993) and Morris and Shin (1998, 2003). This approach emphasizes the incompleteness of information held by the agents. The initial message (Morris and Shin 1998) is somewhat different from, and to some extent opposed to, the message of the sunspot literature. The simple theory of speculative currency attacks suggests the existence of multiple equilibria, whereas a better modeling of the noisy information used by the agents often leads to the restoration of uniqueness. Taking into account the incomplete information faced by the agents is a key ingredient of global games analysis.

Finally, another line of critique is again internal (it accepts the hypothesis but emphasizes intrinsic difficulties). In the herd behavior literature, the rational expectations equilibrium exists, but the outcome depends on details of the starting conditions and is fragile in the sense that the information transmitted may be unreliable (see Banerjee 1992; Chamley 2002).

1.3.2 External Criticisms

There are also lines of criticism that do not accept REH’s assumptions and that imply a more basic reconsideration of REH. I start from more fundamental approaches, as opposed to those that are more informal or eclectic.

It is natural to interpret Muth’s relevant economic theory as what would be called, in game-theoretic terms, a common knowledge (CK) “theory.” The rational expectations equilibrium is then known to everybody, and everybody knows that everybody knows it. The natural, somewhat fundamental, question to be raised is the following: Is it the case that CK of the world (the logic of interactions) and CK of rationality (the logic of decision) imply CK of the equilibrium? If the answer is yes, then in a sense there is a CK-relevant economic theory. In my terminology (Guesnerie 1992, 2005), the equilibrium is “strongly rational” and globally “eductively stable” (“eductively” because the assumptions trigger a collective learning process that takes place in people’s minds). In a simple variant of Muth’s original model (Guesnerie 1992), the answer to the question may be positive or negative, depending on the characteristics of supply and demand. Interestingly, the answer is closely connected to the convergence of the old-fashioned cobweb tâtonnement (a real-time process that the virtual collective learning process mimics). As global eductive stability along the lines sketched here is very demanding, it makes sense generally to define local eductive stability. This line of investigation has numerous applications, a subset of which is reported in Guesnerie (2005).

(Continues…)

Excerpted from Rethinking Expectations Copyright © 2013 by Princeton University Press. Excerpted by permission of PRINCETON UNIVERSITY PRESS. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.