Rethinking the Role of the Representativeness Heuristic in Macroeconomics and Finance Theory

Abstract

We propose a novel interpretation and formalization of Kahneman and Tversky's findings in the Linda experiment which implies that subjects are rational in the sense of Muth's hypothesis and provides an approach to specifying rational assessment of uncertainty in macroeconomic models. Behavioral-finance theorists have appealed to Kahneman and Tversky's findings as an empirical foundation for a general approach replacing rational expectations. We show that behavioral models' specifications of participants' irrational forecasts and predictable errors are incompatible with Kahneman and Tversky's findings. Our interpretation of Kahneman and Tversky's findings is supportive of Lucas's compelling critique of inconsistent macroeconomic models.