A comparison of races and tournaments in the field

A wide range of economic and social situations are decided by either a race – a competition to be first – or a tournament – a competition to be best. Examples include international armed conflicts, political elections,... [ view full abstract ]

A wide range of economic and social situations are decided by either a race – a competition to be first – or a tournament – a competition to be best. Examples include international armed conflicts, political elections, patent races, lawsuits, career dynamics inside organizations, and so on. In such contests, agents choose to exert some costly effort to increase the probability of being awarded a prize. Often they make such decision under uncertainty, e.g., uncertainty about the other agents types or actions. A large body of works in economics have modeled this uncertainty, developing multiple predictions on what agents would do in equilibrium, suggesting possible solutions to important questions of contest design. Overall, it is suggested that, whenever the sponsor of the competition prefers competitors’ performance over the time to complete a particular task, the expected outcomes of a tournament setup should be either equal or greater than those of a race. Yet, a race might be more efficient from an economic point of view as it may prevent unnecessary costs due to an excess of participation. In order to examine this trade-off empirically, we report the results of a field experiment conducted on a leading crowdsourcing platform where we compare the outcomes (efforts, quality, and diversity of outputs) of three alternative competitive situations motivated by theory: the race, the tournament, and the tournament with a quality requirement.