Skip to main content

Psychology and Economic Behavior: Theory, Tests and Applications

Periodic Reporting for period 4 - SalThApp (Psychology and Economic Behavior: Theory, Tests and Applications)

Reporting period: 2020-03-01 to 2021-02-28

This project develops a model of how individual form beliefs and make choices that is grounded in human psychology. Economics views humans as fully rational agents, who use the information at their disposal to form the most accurate beliefs possible and to make the best possible choices. In the last thirty years or so, psychologists and behavioral economists have collected vast experimental and field evidence showing that human beliefs and behavior depart in certain systematic ways from the predicates of full rationality, and that these departures have profound influence on the working of real world markets (e.g. financial markets).
Notwithstanding this evidence, the rational approach is still dominant. This is due to the lack of a generally applicable model of human psychology. On the one hand, different departures from rationality are often explained invoking different psychological principles. The proliferation of factors and “explanations” makes it hard for researchers to obtain general and widely applicable lessons. On the other hand, existing attempts to model psychological influences are tailored to specific markets or decisions, which makes it hard to transport these models into different contexts. Against this complex and intricate web of results, the great advantage of the rational approach is to offer a general, coherent, and widely applicable model of human behavior.
In this project we purport to develop a psychologically founded alternative to rational behavior by following a two pronged strategy. The first is unification. We seek to explain a large number of departures from rational behavior by relying on a simple yet powerful general principle: the idea that our belief or decisions are excessively influenced by information that is unusual or different, while we tend to neglect relevant but non salient information. This principle is grounded on a vast literature in the cognitive sciences stressing the importance of limited selective attention. The first goal of the project, then, is to show that limited and selective human attention offers a parsimonious way to rationalize many disparate anomalies in human behavior. The second pillar of the project, then, is to build a general economic model embodying selective and limited attention and to apply the model to analyze different domains of real world behavior and markets. In particular, we used our approach to analyze the formation of beliefs in financial markets, the determinants of beliefs and stereotypes in social domains involving for instance racial, gender, or political groups, as well as human choice among different consumer goods.
Developing an psychologically founded yet generally applicable alternative to rational behavior is very important for two reasons. In the first place, because it deepens our understanding of social phenomena, be they market or non-market ones. In the second place, because having a more realistic and accepted model of human behavior has critical implications for public policy. If individual decisions are rational and optimal under given conditions, then there is limited room for policy to improve upon social and market outcomes, and relevant policy tools effectively amount to material incentives. If instead phenomena like financial crises, discrimination, political polarization are due to erroneous beliefs and decisions by the market participants or relevant players, the role and tools of social policy have to be rethought. In particularly, policy is likely to play a more important role in this world, and material incentives may be less relevant. Corrective action on beliefs may achieve larger results, as several recent studies are indeed starting to indicate.
So far, we have constructed a formal model in which human judgments exaggerate the probability of events that are particularly representative or salient. This model offers a generally applicable formalization of the famous “Representativeness Heuristic”, documented by leading psychologists Khaneman and Tversky in several experiments, starting around the early 1970s. We have shown that this formalization can unify the explanation for many anomalies in human belief formation that were previously explained invoking different psychological principle. We have then shown that our model is applicable to analyze both the formation of stereotypes about social groups, as well as the formation of expectations about future economic and financial conditions.
We have next developed specific applications of this idea. The first application is gender stereotypes. Using experimental data, we show that our theory of stereotypes can explain why women are underconfident in typically male subjects such as mathematics. In a second set of papers we apply out theory of non-rational beliefs to expectation formation in financial markets. The paper “Diagnostic Expectations and Credit Cycles” shows that this approach helps account for expectations of market participants about credit market conditions, for economic fluctuations and for financial crises as the result to over-reaction to news. In the paper “Diagnostic Expectations and Stock Returns” we show that the very same expectation formation process accounts for anomalous stock price behavior. Finally, the paper “Overreaction in Macroeconomic Expectations” shows that diagnostic expectations can account for systematic errors in the predictions made by professional forecasters about several macroeconomic variables. Overall, this body of work shows that the very same psychological principle is be able to shed light on behavior in different social domains, offering a generally applicable and valid alternative to rational decision-making. Finally Andrei Shleifer and I have written a book using psychology to understand financial crises. The book, entitled “A Crisis of Beliefs” was published by Princeton University Press in the early fall of 2018.
The project is evidently very ambitious and risky. We have some encouraging early findings. But there is still significant uncertainty. A lot more work must be carried out before we have enough confidence to make wide projections.
PI at work