Bounded rationality and financial markets: an experimental approach.

Summary This thesis studies price formation in experimental financial markets, with common value and asymmetric information, inspired by Plott and Sunder (1982, 1988). The first chapter presents the interest of the experimental method in finance related to the control of the environment and the quality of the observation, and proposes a review of the literature on experimental finance (CAPM, price efficiency, microstructure, market rationality). In the second chapter, I) a double auction, II) a fixing followed by a double auction, and III) a pre-opening period and a fixing followed by a double auction are experimented. Equilibrium strategies are computed. In our experimental markets, the pre-opening facilitates the discovery of equilibrium strategies. This discovery is improved by the experiment when subjects are given the right incentives. The relationship between the psychological traits of investors and their behavior is analyzed in the third chapter. The traits (impulsivity, social intelligence, overconfidence, availability and representativeness rules, and confirmation bias) are measured by administering a questionnaire to subjects. These subjects then participate in experimental financial markets. Impulsive subjects place more orders but do not suffer more losses than other individuals. Overconfident subjects realize more losses than others, and confirmation bias and the use of the representativeness rule slow down the learning of equilibrium strategies. The fourth chapter compares a Walrasian trial and error (TW) and a fixing market (MF). These two institutions are strategically equivalent. In our experimental markets, price efficiency is almost perfect in both market structures. On the other hand, transaction costs are lower in the TW than in the MF where investors are unable to discover equilibrium strategies.