How Difficult It Is to Leave the Known Path

Pierpaolo Battigalli, Simone Cerreia-Vioglio, Fabio Maccheroni, and Massimo Marinacci, all in the Bocconi Department of Decision Sciences, have recently published an article Self-Confirming Equilibrium and Model Uncertainty in the American Economic Review. The paper builds on Cerreia-Vioglio, Maccheroni and Marinacci's decision theory work on ambiguity aversion and on Battigalli's game theory work on self-confirming equilibrium.

Battigalli sums up the ambiguity aversion in the following way: "People tend to choose actions that entail known rather than unknown risks. For instance, they are more likely to invest in a known technology than in a new and untested one, even after knowing that the former succeeds only 50% of the time. Such attitude is called ambiguity aversion". Cerreia-Vioglio, Maccheroni and Marinacci are among the best-known and most prolific experts on this topic.

Standard economic models leave no room for ambiguity. "These models", Battigalli says, "assume that agents know the objective probability distribution of outcomes implied by each choice, and thus make the best choice accordingly. In game theory, we call this a Nash equilibrium". Battigalli, Fudenberg (Harvard University), and Levine (European University Institute) in their game theory work independently challenge this assumption by arguing that agents interacting in a game can only learn the risks entailed by actions that are repeatedly chosen in a "self-confirming" equilibrium, not the risks of the untested actions.

Building on this, in the article just published in the American Economic Review, Battigalli, Cerreia-Vioglio, Maccheroni and Marinacci show that in a self-confirming equilibrium, unlike in a Nash equilibrium, ambiguity aversion can play a role. In particular, the more ambiguity averse the agents are, the more entrenched they are in their equilibrium choices, that is, the harder it is to disrupt the equilibrium with some exogenous shock to incentives, such as a small change in the tax system. A similar intuition is obtained by looking at an agent who makes repeated choices under uncertainty: higher ambiguity aversion makes him more reluctant to experiment with untested actions.

Ambiguity aversion can make a difference when analyzing economic policy. "Policy makers can only learn the risks entailed by the policies they use", Battigalli says, "whereas alternatives remain untested. If policy makers are ambiguity averse, or want to be re-elected by ambiguity averse voters, they will tend to adopt a conservative stance".