2022 UNSW - UQ
Economic Theory Festival

Causal Reasoning and Economic Behavior   Ran Spiegler 

Economic decisions involve perceptions of causal relations and attempts to quantify them on the basis of observational data. Examples include judging whether attending college will increase one's future earnings, whether choosing a certain lifestyle will improve long-run health, or whether and how interest rates cause inflation. Yet until recently, the only economists who were systematically preoccupied with causal reasoning were theoretical and applied econometricians, whose approach to this subject is normative and prescriptive. Instead, this lecture series introduces a descriptive, behavioral approach to modeling everyday causal reasoning. It will present an equilibrium model of decision making under subjective (and possibly wrong) causal models, borrowing the formalism of Directed Acyclic Graphs (DAGs) from the AI and Statistics literatures. It will show how this formalism enables us to represent decision makers who "confuse correlation with causation" (because they misperceive the direction of causality or ignore confounding variables), attribute outcomes to wrong causes, or neglect causal channels. The applications will shed light on phenomena such as demand for healthcare or education, monetary policy and the role of false narratives in political mobilization. It will link this modeling approach to the broader literature on decision making under misspecified models. Finally, the lectures will briefly introduce basic techniques from the DAG literature and use them to tackle questions such as: When do causal misperceptions lead to equilibrium effects in single-agent decision problems? Can agents with causal misperceptions be systematically fooled? How badly can a wrong causal model distort pairwise correlations?