Second Mover Advantage and Bertrand Dynamic Competition: An Experiment

In this paper we provide an experimental test of a dynamic Bertrand duopolistic model, where firms move sequentially and their informational setting varies across different designs. Our experiment is composed of three treatments. In the first treatment, subjects receive information only on the costs and demand parameters and on the price’ choices of their opponent in the market in which they are positioned (matching is fixed); in the second and third treatments, subjects are also informed on the behaviour of players who are not directly operating in their market. Our aim is to study whether the individual behaviour and the process of equilibrium convergence are affected by the specific informational setting adopted. In all treatments we selected students who had previously studied market games and industrial organization, conjecturing that the specific participants’ expertise decreased the chances of imitation in treatment II and III. However, our results prove the opposite: the extra information provided in treatment II and III strongly affects the long run convergence to the market equilibrium. In fact, whilst in the first session, a high proportion of markets converge to the Nash-Bertrand symmetric solution, we observe that a high proportion of markets converge to more collusive outcomes in treatment II and more competitive outcomes in treatment III. By the same token, players’ profits significantly differ in three settings. An interesting point of our analysis relates to the assessment of the individual behavioural rules in the second and third treatments. When information on the behaviour of participants on uncorrelated markets is provided, players begin to adopt mixed behavioural rules, in the sense that they follow myopic best reply rules as long as their profits are in line with the average profits on all markets, and , when their gains fall below that threshold, they start imitating successful strategies adopted on other markets.