Modeling a World of Imperfect Knowledge

Does it matter if the Rational Expectations Hypothesis is unrealistic?

Does it matter if the Rational Expectations Hypothesis is unrealistic?

Not according to New York University Prof. Roman Frydman, head of the Institute for New Economic Thinking’s research group on Imperfect Knowledge Economics (IKE).

Speaking at a seminar organized by the Institute and Columbia University and chaired by Columbia Prof. Joseph Stiglitz, who’s a member of the Institute’s advisory board, Frydman stated that the Rational Expectations Hypothesis, or REH, is simply an abstraction of reality and therefore it is misleading to criticize it for abstracting from many significant aspects of the real world economy. To Frydman, instead, the more pressing questions raised by REH are, what kind of world is it an abstraction of? And can this explain why the models fail so dramatically?

Frydman’s point is that the kind of the world that REH models are abstracting actually does not exist. A REH model assumes that people need to make decisions on how the economy works and the problem that they face is how to forecast when there are so many ways to do so. The relevant economic models, according to John Muth (1961), could serve this purpose. But the current version of REH modeling is empirically non-testable, which creates significant controversy.

Having said that, there also is no proof that other versions of REH models can work. Although we do not know if the model empirically works, we can determine in which “domain” the model works. This basically means that we have to investigate the following question: How does change proceed in REH models over time? As Popper (1957) argues, “If there is such a thing as “growing human knowledge,” we cannot anticipate today what we shall know tomorrow.”

Frydman is focusing his work on understanding “imperfect knowledge” to find a solution for the REH’s irrelevance in representing rational decision-making in the real world economy. Frydman and his co-author, Michael D. Goldberg, also argue that they can empirically test this model by formulizing market understanding. Rational forecasting, therefore, needs to be related to the markets’ understanding of the price process. Market understanding is indeed the function of some causal factors. Therefore, one could formulize the market understanding of the price changes in order to empirically test the model.

The domain that one could choose for the models is macroeconomics or the financial market. By choosing a specific domain we are actually making assumptions about the different processes of change. Mathematically speaking, choosing the domain keeps the model open to an unobserved change.

One of the characteristics of the models that are widely used in the current economic literature is that they are closed, so it is assumed that the structure of the model does not change over time and the model is predetermined. In the REH model, therefore, it is assumed that the human knowledge does not change. This is very contradictory because the REH models are indeed invented to represent the growth of human knowledge. Put differently, economic agents in the REH model assume that there would be no unexpected change. If the market does not assume the same thing, the agent is considered incoherent.

Behavioral economists also tend to forget about this fundamental contradiction in the REH model since REH plays a crucial role in their models as well. In their view, the REH model adequately represents rationality, but the market participants are irrational. They therefore ignore the fact that in these models knowledge does not grow by claiming that the market participants actually behave according to determinate models.

In his critique of this approach, Frydman creates new economic thinking by arguing that the key to incorporating the insights of REH models and behavioral economics is to jettison the determinate models. By eliminating determinacy Frydman puts forward a model in which, even though there is some regularity, it is not completely open and is not completely determinate. The Contingent Expectations Hypothesis, or CEH, that is introduced in the latest paper by Frydman and Goldberg is somewhere “in between”.

Ultimately, Frydman argues that in order to push the boundaries of economic literature there is no need to throw out all the things that we have learned. Instead, we just need to discard the current popular assumptions that we know how markets work and we have the same information as the market does.

Of course, behind this methodological shift from the REH model to the CEH model are overarching sharp contradictions of all future outcomes and their probabilities. It is therefore important to understand that the ultimate ambition of the REH model is beyond the reach of the economic analysis.

The balanced approach, as identified by Frydman, is that the markets are essential to the modern economy, but they have bad tendencies to create extreme results, such as extreme booms or busts. It takes new economic thinking like Frydman’s CEH model to break through and change current standards.

Share your perspective

Featured in this article