Life After “Rational Expectations”?

1. Life After “Rational Expectations”? Imperfect Knowledge, Behavioral Insights and the Social Context Roman Frydman and Michael D. Goldberg Institute for New Economic Thinking Inaugural Conference King’s College, Cambridge, United Kingdom 8-11 April 2010

2. Roman Frydman and Michael D. Goldberg , Illusions of Stability: Financial Markets, the State, and the Future of Capitalism, forthcoming from Princeton University Press in 2010. Roman Frydman and Michael D. Goldberg, Imperfect Knowledge Economics: Exchange Rates and Risk, Princeton University Press, August 2007. http://press.princeton.edu/titles/8537.html Roman Frydman and Michael D. Goldberg "Macroeconomic Theory for a World of Imperfect Knowledge,“ Capitalism and Society:, 2008, Vol. 3: Iss. 3, Article 1, http://www.bepress.com/cas/vol3/iss3/art1/

3. The upswings in housing and equity markets that preceded the financial crisis are seen as evidence that market participants are irrational, emotional, and ignore fundamentals. • Blaming market participants forbeing irrational sidestepsthe key issue: • the impossibility of establishing a standard of rationality that would enable an economist to model how a rational individual makes decisions in every situation.

4. Imagining Rationality as Mechanical • The vast majority of economists, including those following the behavioral approach, consider the “Rational Expectations Hypothesis” (REH) to be the cornerstone of the contemporary standard of economic rationality. • REH instructs an economist to model the way a rational individual forecasts market outcomes with his own model of these outcomes.

5. Whenever an economist constructs a model, he has a mechanical way for modeling expectations on an individual and aggregate level. • Because this can be done in every model, REH appears as both straightforward and universally applicable.

6. Behavioral economists and non-academic commentators often criticize economists’ standard of rationality as implying that • rational individuals make decisions as if they had superhuman abilities to understand the future – that they can compute correctly the consequences of alternative options available to them.

7. REH presumes no such thing. • It ignores that rational individuals may interpret outcomes in ways different from a particular economist’s model. • Moreover, for a contemporary economist a model is not just a mathematical account of market outcomes.

8. Over the last four decades, economists have come to believe that, to be worthy of scientific status, their models should generate “sharp probabilistic predictions”: • they should account for the full range of possible market outcomes and their likelihoods.

9. In order to generate such predictions, economists must pre-specify – in full -- how an individual alters the way she makes decisions and how aggregate outcomes unfold over time. • Thus, contemporary models disregard the key feature of markets: non-routine change that cannot be prespecified in advance with mechanical rules. • We refer to such models as fully predetermined.

10. Because every REH model is fully predetermined, it implies that a single, overarchingstrategy can adequately portray the forecasting by market participants both individually and in the aggregate: • it specifies just one interpretation of the process driving market outcomes, which, up to a random error term, relates these outcomes exactly to a set of causal variables at every point in time, past, present, and future.

11. Thus, by design, REH ignores the diversity of “reasons” upon which individuals in real-world contexts might base their forecasts. • Moreover, because each REH model disregards non-routine change, it presumes that participants act as if nothing genuinely new would ever happen that would lead them to change the way they think about the future.

12. But profit-seeking individuals cannot afford to adhere to a fixed overarching strategy in real-world markets, in which non-routine change is an ever-present possibility. • Thus, REH presumes that market participants forgousing whatever cognitive abilities they dohave in thinking about the future and making their decisions.

13. Indeed, we demonstrate that economists’ characterizations of rational forecasting would appear to any reasonable person, let alone a profit-seeking participant in financial markets, to be obviously irrational, in the sense that their trading decisions forgo profit opportunities. • Thus, REH imagines a place devoid of the crucial features that characterize, even in the most rudimentary abstract way, how individuals think about the future in real-world markets.

14. The Incoherence of the “Rational Market” • An alternative interpretation of REH is that it is a hypothesis about forecasting on a “rational market” level rather than on an individual level. • REH is widely seen as an “approximation” that captures in a parsimonious way the complexitiesof revisions and diversity of market participants’ forecasting strategies.

15. In order to ensure that REHholds together in the aggregate, it cannot “approximate” the micro-level diversity that underpins the outcomes in real-world markets. • Instead, we demonstrate that REH imposes “pseudo-diversity:” • market participants’ revisions of the way they think about the future are tied to each in a fully predetermined way.

16. Moreover, diversity of views on the individual level renders the very notion of the “rational market” incoherent.

17. Diversity could persist only if individuals were obviously irrational, ignoring systematic forecast errors and forgoing obvious profit opportunities endlessly. • Thus, REH’s so-called “rational markets” are populated by • either irrational individuals, • or those who presume that there is no diversity on the micro-level --all market participants think exactly alike about the future.

18. Emerging Alternative? • Long before the financial crisis, Maurice Obstfeld and Kenneth Rogoff concluded that, • the undeniable difficulties that international economists encounter in empirically explaining nominal exchange rate movements are an embarrassment, but one shared with virtually any other field that attempts to explain asset price data • (Obstfeld and Rogoff, 1996, p. 625).

19. Behavioral economists also uncovered massive evidence that conventional models – based on REH – are grossly inconsistent with empirical evidence. • This evidence is important because it opened economics to alternative explanations of individual decision-making and market outcomes.

20. However, in building “behavioral finance” models formalizing their insights, behavioral economists retain the core of the contemporary approach: • Like their REH predecessors, behavioral finance models formalize empirical findings about how individuals act with fully predetermined mechanical rules.

21. Moreover, behavioral economists did not interpret the failure of REH models as evidence that fully predetermined models of rationality cannot adequately capture how rational individuals make decisions. • Instead, they concluded that market participants are “irrational.”

22. Given this interpretation, and their shared belief in the “scientific” status of fully predetermined accounts of economic outcomes, behavioral economists ended up embracing mechanical models of “irrational” decision-making.

23. To be sure, not all behavioral economists embraced fully predetermined models. • Notably, George Akerlof and Robert Shiller rely on a largely narrative mode of analysis. • This enables them to develop richer descriptions of fluctuations than fully predetermined models of “rational” and “irrational” behavior deliver.

24. Modeling Without Universal Rationality • There is an alternative interpretation of the failure of REH models to the prevailing diagnosis that participants are irrational: • purposeful decision-making in capitalist economies, particularly individual forecasting, cannot be adequately portrayed with an economist’s fully pre-specified mechanical rules.

25. This alternative interpretation implies that fully pre-determined models of “rational and irrational” behavior – whatever that means -- are outside the reach of economic analysis. • We urge conventional and behavioral economists to consider this premise and abandon REH as model for individual and market forecasting as well as policy analysis.

26. IKE begins with this premise. • Like the contemporary approach, IKE bases its models of aggregate outcomes on mathematical representations of individual behavior. • Unlike contemporary models, IKE attempts to come to terms with the modesty of Keynes, Knight, and Hayek about how exact their representations of individual behavior should attempt to be.

27. IKE presumes that purposeful behavior cannot be fully pre-specified in advance. • It explores the possibility that the ways in which market participants make and alter their decisions may be formalized with qualitative conditions. • IKE uses non-standard probabilistic formalism to formulate its models.

28. In contrast to REH, multi-disciplinary insights play a crucial role in constructing IKE models. • Instead of the contemporary belief that economics explains outcomes in other fields of human inquiry, these disciplines play an integral role in IKE-based accounts of economic phenomena.

29. IKE models formalize the psychological insights and findings of behavioral economists. • These models also formalize observations about the social context within which individuals make decisions, including the historical record, conventions, and norms among market participants.

30. While IKE models, by design, do not imply sharp predictions of change, they do generate qualitative predictions that enable an economist to distinguish empirically among alternative explanations of economic phenomena.

31. IKE as a Boundary of Macroeconomic Theory • In our book we show how IKE models shed new light on the salient features of the empirical record on exchange rates, which have confounded international macroeconomists for decades. • Although these results are promising, it is much too early to claim broader usefulness for IKE in macroeconomic and policy modeling.

32. In contrast to the conventional and the behavioral finance models, which seek to understand economic decisions with fully predetermined rules, IKE’s constraints are qualitative and context-dependent. • If qualitative conditions can be found in contexts other than asset markets, IKE provides a way to model individual behavior and market outcomes based on those conditions.

33. But in contexts in which revisions of forecasting strategies cannot be adequately characterized with qualitative conditions, empirically relevant mathematical models of the observed time-series may be beyond the scope of economic analysis. • In this sense, IKE provides a boundary for what modern macroeconomic theory can deliver.