Momentum in contests and its underlying behavioral mechanisms

We investigate the existence and nature of momentum in performance in contests and whether momentum arises for reasons in part unrelated to rational strategies in contests. To address this question, we look at a setting where strategic considerations should not generate momentum: a sequence of two rounds of independent contests. We show that if we relax the assumption of payoff maximizing agents, positive momentum (success tends to be followed by more success) or negative momentum (success tends to be followed by less success) can arise through several behavioral mechanisms that have, until now, not been widely considered in the literature. We examine these predictions in an experiment. Using random variations in the participants’ winning chances in a first contest to identify the causal effect of success on later performance, we find that a positive momentum exists. Using several experimental conditions which modulate the effect of the different possible mechanisms, we find that the pattern of momentum is most compatible with players having adaptive preferences, whereby they may gain or lose interest in the second contest after respectively winning or losing the first one. These results suggest that standard models of contests do not fully capture the behavioral dynamics existing in competitive settings.