This study assesses the applicability of a game theoretic concept, known as the forward induction, in resolving coordination failures in a very simple two-player situation in lab experiments. If players believe their opponents to be rational, actions taken by the opponents in the earlier stages of the game might convey information about their intended actions in the later stages of the game. Then it might be easier for the players to coordinate towards a mutually beneficial outcome.
The problem of selection in a situation with multiple equilibria has always posed a challenge to game theorists. In the absence of coordination this problem often leads to outcomes that are inefficient in terms of the payoffs of the parties involved. Over the years we have seen the development of different approaches towards the refinement of equilibrium concepts but hardly any of them seems to be satisfactory especially from the practical viewpoint. In this context, the concept of forward induction has the potentiality to be a practical solution for the coordination problem.
Broader Impact: Coordination failure is a source of concern in any discipline, such as management, marketing, information technology and economics, where we analyze strategic interaction and the stability properties. The concept of forward induction seems to be a fairly simple and intuitive solution to this problem, and its use could avoid coordination failures and the loss of efficiency in many real life situations. Also, the concept of forward induction is based on the belief about rivals' rationality. If players are truly rational, then forward induction should be an effective solution to coordination failures. Therefore, the results of this study will also be an indirect test of players' rationality and their belief about opponents' rationality.
Previous experimental studies on forward induction did not find much support for forward induction and suffered from inconclusive results. This inconclusiveness might arise from some flaws in the design of these earlier studies. With improved experimental design, this problem can be avoided and at the same time valid conclusions on the effectiveness and applicability of forward induction can be drawn. Also, the games used in the previous studies are fairness games. The current study controls for the effects of other-regarding from being confounded with the forward induction effects. This is achieved with particular treatments. In sum, there is much to be explored in this area that will enrich our knowledge.
