David vs. Goliath: An Analysis of Asymmetric Mixed-Strategy Games and Experimental Evidence

摘要

Mixed strategies are widely used to model strategic situations in diverse fields such as economics, marketing, political science, and biology. However, some of the implications of asymmetric mixed-strategy solutions are counterintuitive. We develop a stylized model of patent race to examine some of these implications. In our model two firms compete to develop a product and obtain a patent. However, one firm values the patent more because of its market advantages, such as brand reputation and distribution network. Contrary to some intuition, we find that the firm that values the patent less is likely to invest more aggressively in developing the product and will also win the patent more often. We argue that the reason for these counterintuitive results is inherent in the very concept of mixed strategy solution. In a laboratory test, we examine whether subjects' behavior conforms to the equilibrium predictions. We find that the aggregate behavior of our subjects is consistent with the game-theoretic predictions. With the help of the experience-weighted attraction (EWA) learning model proposed by Camerer and Ho (1999), we show that adaptive learning can account for the investment behavior of our subjects. We find that the EWA learning model tracks the investment decisions of our subjects well, whether we hold out trials or an entire group of subjects.