Bounded Rationality in Macroeconomics

To Bobby, Nick, and Jon who can learn at different rates

The Arne Ryde Foundation¶

Arne Ryde was an exceptionally promising young student on the doctorate programme at the Department of Economics at the University of Lund. He died after an automobile accident in 1968 when only twenty-three years old. In his memory his parents Valborg Ryde and pharmacist Sven Ryde established the Arne Ryde Foundation for the advancement of research at our department. We are most grateful to them. The Foundation has made possible important activities which our ordinary resources could not have afforded.

In agreement with Valborg and Sven Ryde, we have decided to use the funds made available by the Foundation to finance major initiatives. Since 1973 we have arranged a series of symposia in various fields of theoretical and applied economics. In 1990 we published “Seven Schools of Macroeconomic Thought” by Edmund S. Phelps, the first issue in a series of Arne Ryde Memorial Lectures. The present book by Professor Thomas J. Sargent, based on the lectures he held at Snogeholm Castle in Skåne in October 1992, is the second issue in this series. We are very glad and grateful that Professor Sargent agreed to come to Lund to give his Arne Ryde Memorial Lecture.

Björn Thalberg

1 Introduction¶

Equilibria and transitions¶

In 1989 the problem of ‘transition dynamics’ forced itself on economists and statesmen. Two generations of work on economic dynamics within parallel traditions in game theory, macroeconomics, and general equilibrium theory have given us theories of dynamics that have their best chance of applying when people are in recurrent situations that they have experienced often before. In Eastern Europe the transition is not like that: people there are confronted with unprecedented opportunities, new and ill-defined rules, and a daily struggle to determine the ‘mechanism’ that will eventually govern trade and production. Economists who dispense advice about governmental strategies to enable transitions to a market economy can do so with ample help from ‘equilibrium theories’ describing how to expect a system to operate after it has fully adjusted to a new and coherent set of rules and expectations, but with virtually no theories about the transition itself. We might have prejudices and anecdotes to guide our preferences among transition strategies, but no empirically confirmed formal theories.^[1]

Against this background, more and more economists have recently ventured into what Christopher Sims (1980) characterized as the ‘wilderness’ of irrational expectations and bounded rationality, aiming partly to create theories of transition dynamics, partly to understand the properties of equilibrium dynamics themselves, and partly to create new dynamics of systems that do not settle down. Beyond confirming Sims’s characterization of this area as a research ‘wilderness,’ no general theory of transitions or of bounded rationality has yet emerged from this area, but much has been learned, and maybe some of it is relevant to thinking about real transitions.

This essay describes and interprets some of this large and diverse body of work. I do not survey very much of the work in the area, but focus on a small number of examples designed to indicate the kinds of questions that are being asked and answered in this research.^[2]

This area is wilderness because the researcher faces so many choices after he decides to forgo the discipline provided by equilibrium theorizing. The commitment to equilibrium theorizing made many choices for him by requiring that people be modelled as optimal decision-makers within a commonly understood environment. When we withdraw the assumption of a commonly understood environment, we have to replace it with something, and there are so many plausible possibilities. Ironically, when we economists make the people in our models more ‘bounded’ in their rationality and more diverse in their understanding of the environment, we must be smarter, because our models become larger and more demanding mathematically and econometrically.

Sketch of the argument¶

The argument in this essay runs as follows. Rational expectations imposes two requirements on economic models: individual rationality, and mutual consistency of perceptions about the environment. When implemented numerically or econometrically, rational expectations models impute much more knowledge to the agents within the model (who use the equilibrium probability distributions in evaluating their Euler equations) than is possessed by an econometrician, who faces estimation and inference problems that the agents in the model have somehow solved. I interpret a proposal to build models with ‘boundedly rational’ agents as a call to retreat from the second piece of rational expectations (mutual consistency of perceptions) by expelling rational agents from our model environments and replacing them with ‘artificially intelligent’ agents who behave like econometricians. These ‘econometricians’ theorize, estimate, and adapt in attempting to learn about probability distributions which, under rational expectations, they already know.

To learn how to build models with boundedly rational agents, I turn to the interrelated literatures on statistics, econometrics, and artificial intelligence. These literatures describe methods for representing and estimating relationships within data. I present a broad brush survey of a few bread and butter statistical methods, and describe how they are related to methods developed in the recent ‘parallel distributed processing’ or neural network literature. I also mention John Holland’s genetic algorithm and classifier system and how they compare with least squares methods familiar to economists. It is from the store of methods built up in these literatures that we shall select the ‘brains’ to give our boundedly rational agents.

Next I shall describe five example economies variously designed to illustrate some of the potential uses and properties of models with boundedly rational agents. Some of these examples illustrate the ability of collections of boundedly rational agents to learn to behave as if they have rational expectations. Other examples illustrate the choices that we the researchers, as the ‘gods’ or creators of these artificial people, have in informing (or ‘hard-wiring’) them about their environments before we turn them loose. Still other examples illustrate how putting boundedly rational agents inside some environments can serve to resolve limitations or puzzles (e.g. equilibrium indeterminacy and odd types of behavior) that rational expectations implies in particular models.

Continuing in the spirit of asking what bounded rationality models can do that rational expectations models cannot, I next describe how models with adaptive agents have been used to interpret experiments for two monetary economies. These two economies have serious problems of multiple equilibria under rational expectations, so it is interesting to see how the experimental results compare with the predictions of particular types of bounded rationality.

The argument concludes with an accounting of the promises and limitations for macroeconomics of models with bounded rationality. On the credit side of the ledger are these models’ success in inspiring useful tests of equilibrium selection; the motivation they have provided for ‘evolutionary programming’ for computing equilibria of rational expectations models; and their suggestion, through the literatures on parallel and genetic algorithms, of new gadgets and statistical methods for econometricians. On the debit side (or at least in the to-be-collected row) of the ledger are bounded rationality’s unfulfilled promise as a device for specifying and understanding out-of-equilibrium dynamics; and its failure thus far to suggest new and fruitful econometric specifications of expectations formation.

The essay ends near where it started, with a difference between the behavior of econometricians and the agents in their models. As I have interpreted it, the bounded rationality program wants to make the agents in our models more like the econometricians who estimate and use them. Given the compliment (‘imitation is the sincerest form of flattery’), we might expect macroeconometricians to rush to implement such models empirically. There has been no rush, maybe for the reason that many macroeconometricians are in the market for methods for reducing the number of parameters to explain data, and a reduction is not what bounded rationality promises.