Biography

Tony Berrada is Professor of Finance at the University of Geneva (Geneva Finance Research Institute and Swiss Finance Institute). His research interests are in asset pricing, and particularly the role of learning in models with incomplete information. His research in finance was published in top ranked journals in the field, such as the Journal of Financial Economics, the Review of Finance, Management Science and the Journal of Banking and Finance. His research at the intersection of neuroscience and finance appeared in top ranked biology journal Current Biology. Professor Berrada is a regular speaker at leading finance conferences and workshops worldwide. He teaches executive education courses on portfolio management. He is the head of research at the GIWM (Geneva Institute for Wealth Management), a non–profit foundation created to promote international partnerships with the University of Geneva in post-graduate education, executive education, research and knowledge transfer in wealth management.

Abstract

Financial theories rely on a central assumption: agents optimize behavior to maximize utility. This fails to accurately capture behavior in many instances and has thus been challenged. Herbert Simon proposes that agents do not optimize but rather satisfice, i.e. agents optimize up to some point of satisfaction. To date, however, few formal models have been proposed to capture how satisficing unfolds. Here, we use Model Reference Based Adaptive Control (MRAC), a well-established robust control technique in engineering, to develop a formal theory of a satisficing investor and of the resulting financial market equilibrium. To ensure robustness in a nonstationary world, an MRAC agent chooses portfolios that generate return distributions that minimize surprise with respect to a desired reference distribution. Our results reveal that under most circumstances MRAC mirrors predictions of CAPM or multifactor asset pricing models, i.e. the agent acts “as if” optimizing. At the same time, the model delivers a sharp condition under which these predictions fail. The investor’s problem is characterized by a discriminant that governs whether optimal choices lie on the mean–variance frontier or within an interior zero-surprise set. Through market clearing, this mechanism determines whether the market portfolio is mean–variance efficient. When equilibrium choices align with the frontier, CAPM pricing obtains; when they do not, CAPM pricing fails. More generally, the model implies that the validity of CAPM pricing is state-dependent and governed by this underlying mechanism. We take this prediction to the data using a proxy for the distance between the market portfolio and the efficient frontier. Using a panel of U.S. equities, we show that cross-sectional CAPM pricing errors increase strongly and significantly with this distance. This result is robust across multiple measures of pricing error, including the mean absolute alpha, the root mean squared alpha, and the cross-sectional dispersion of alphas. Our findings provide a new perspective on asset pricing by showing that deviations from CAPM arise endogenously as a state-dependent feature of equilibrium behavior under target-based decision rules.

Join the NIPE seminars on Google calendar: https://bit.ly/2LKkPyV