Can Agents with Causal Misperceptions be Systematically Fooled?∗

Can a policy maker lead an agent to form systematically biased economic forecasts? This possibility is ruled out by the conventional rational-expectations postulate. I revisit this question and assume that the agent misperceives causal relations among economic variables, which may lead to non-rational expectations. The agent forms forecasts of economic variables after observing the policy maker’s action. The agent’s forecasts are based on fitting a subjective causal model formalized as a direct acyclic graph, following the “Bayesian networks”literature to objective long-run data. I show that the agent’s forecasts can be systematically biased if and only if the agent’s graph is not perfect i.e., if the direction of at least some of the causal links he postulates is empirically meaningful. I characterize the policy maker’s optimal strategy for several examples, mainly a stylized “monetary policy”application. ∗Financial support by the Sapir Center is gratefully acknowledged. I thank Kfir Eliaz, Rafaella Giacomini, Heidi Thysen, Michael Woodford and Vasiliki Skreta for helpful discussions. †Tel Aviv University, University College London and CFM. URL: http://www.tau.ac.il/~rani. E-mail: [email protected].