The research program that has dominated macroeconomic theory since the 1980s achieved its hegemony mainly because of its methodological claims of having ‘solid microeconomic foundations’. The core models of new Keynesian macroeconomics are based on intertemporal optimization by representative households, a ‘forward looking’ Phillips curve with nominal price stickiness and staggered priced setting, and rational expectations. Unfortunately, the Euler equations derived from household optimization provide a poor description of the movements of aggregate consumption, the new-Keynesian Phillips curve does not fit the evidence, and claims about rational expectations are undermined by evidence of predictable and systematic expectational biases.
Multiple ad hoc modifications and extensions of the basic model have been introduced to salvage the theory. These attempts to patch it up are unconvincing, and a more radical reorientation is needed. The question is, what should this reorientation look like? This book presents a critique of the prevailing orthodoxy and outlines a structuralist and behavioral alternative.
There are good reasons for the failure of the new Keynesian model. The Lucas critique was valid, but the resolution of these problems – the Lucas solution — that has been developed over the last 40 years is deeply flawed. The presence of well-known aggregation problems implies that a representative agent cannot be constructed except under highly restrictive assumptions. These problems have been swept under the carpet, and there appears to be no recognition that, even in cases where aggregation is possible, the use of this representative agent’s utility function is itself subject to a Lucas critique and that, as a welfare criterion, it leads to a systematic bias in favor of rich.
Even at the microeconomic level the predictions of standard models of rational economic man show systematic and macroeconomically relevant deviations from the behavioral evidence. Some deviations of actual behavior from that of an idealized homo oeconomicus are quite trivial and irrelevant for macroeconomics, but others are systematic and occur in areas that affect aggregate economic outcomes. A large proportion of households have saved very little by the time they reach retirement. Well-documented behavioral findings suggest that present bias and peer effects on consumption contribute to keeping saving rates low. But pervasive uncertainty also poses questions for the general notion that long-term decisions like household retirement saving can be based on meaningful notions of optimization. These complications and systematic deviations of household behavior from the postulates of the model do not merely add random errors that cancel out on aggregation. And they affect the core of the new Keynesian model: it is precisely the use of intertemporal optimization that supposedly establishes the superiority of the model.
The treatment of wage formation and the labor market is another example. The existence of a well-defined ‘natural rate of unemployment’ informs much of economic policy, but the evidence is weak: strong prior beliefs are required to justify interpretations of the evidence as supportive of a natural rate of unemployment. In terms of theory, the standard approach to wage and price setting excludes forces that have a systematic influence on the labor market. Social norms have an impact on wage setting, earnings inequality, and inflation, giving rise to downward nominal wage stickiness and money illusion. The norms also change endogenously in response to persistent deviations of outcomes from the prevailing norms, and induced changes in pay norms can be a source of hysteresis in both employment and relative pay. Increasing earnings inequality, moreover, cannot be explained solely by skill-biased technological change or globalization. Technological and institutional changes have also been power-biased, and the fairness norms with respect to relative earnings have shifted, both endogenously and as a result of broader ideological shifts.
The behavioral evidence must be aggregated and combined in macroeconomic models of interactions between different markets and groups of decision makers. Even if all prices and wages are flexible, there can be no general presumption that these interactions will ensure the stability of a short-run equilibrium with full employment – this, indeed, was the central message of Keynes’s General Theory (a message that should be uncontroversial today: we now have rigorous proofs that the stability of a general equilibrium requires highly restrictive assumptions, even in Walrasian economies with a well-functioning tatonnement process). Keynes’s short-run macroeconomic analysis can be formalized, and the logic is sound. But the analysis has to be extended to address medium and long run issues.
Capitalist economies fluctuate and periodically experience large disruptions in economic activity. New Keynesian models view cycles and periodic crises as driven by a series of stochastic shocks with carefully calibrated statistical properties, the presumption being that the steady growth path would be stable in the absence of these shocks. This presumption of stability is questionable. Capitalist economies contain both stabilizing and destabilizing forces. Depending on the balance of these forces, the dynamic interactions across markets can lead to local instability of the steady growth path and endogenous fluctuations. And the assumptions that produce these endogenous cycles are supported by behavioral and empirical evidence.
Aggregate demand policy can dampen instability and smooth the business cycle. Interventions may also be needed, however, to address persistent aggregate demand problems that jeopardize the existence of a steady growth path (and not just its local stability). If saving rates are high, economies may experience secular stagnation: it may be impossible, without fiscal stimulus, for an economy to grow steadily at the rate that is required to maintain full employment. This potential long-run problem only applies to ‘mature economies’ whose growth rates are constrained by the growth of labor force. Developing economies with large amounts of unemployment are not constrained in this way. Their policy problems therefore become quite different.
Dynamic interactions across markets are hard to keep track of, and formal techniques, primarily in the form of small-scale analytical models, are used throughout this book to develop and analyze these and other issues. The ambition is not to provide a single ‘theory of everything’ but to set out a series of coherent models that can highlight elements of an alternative theoretical framework.
Do we need macroeconomic theory? After the financial crisis of 2008, there appears to have been a growing perception of theory as irrelevant, at best, to real-world issues, spurring a move in the direction of purely data-driven approaches. Big data and sophisticated econometric techniques do not, however, obviate the need for a conceptual and theoretical framework to select and structure empirical studies and to interpret the statistical results that have been collected. The current literature shows many examples of sophisticated and careful econometric work that has then been interpreted, implicitly or explicitly, using theories with little or no behavioral and empirical support.
The theoretical approach in this book draws on recent developments in behavioral economics as well as older literatures from heterodox traditions, including Keynesian, Marxian, institutionalist and development economics. There is no attempt to provide a survey of these theories, and I have deliberately stayed clear of a discussion of debates and disagreements within the diverse heterodox traditions. Instead, the aim has been to develop and combine insights from these traditions, while relating the analysis to weaknesses in the current mainstream. It is my hope that this approach will make the book interesting to open-minded economists with a mainstream background as well as to more heterodox economists. The book could also, I believe, be useful in both heterodox and (as supplementary reading) mainstream graduate courses; early drafts of most of the chapters have been used in graduate classes at both the University of Massachusetts and the New School for Social Research.