We study a Tullock contest game where players’ effort strategies generate different random evolving lotteries depending on how and when the contest outcome is announced. We characterize its symmetric Nash equilibrium under the assumption that players have peak-trough preferences (see Gul et al. 2021). Analyzing an optimal design problem in which the contest designer can choose how (at once or gradual) and when (early or late) to announce the outcome, we show that (i) the total equilibrium effort exerted is influenced by the designer’s information disclosure strategy and (ii) any specification under consideration can be optimal for some values of preference parameters. Interestingly, a player may end up exerting a lower equilibrium effort in his most preferred specification (i.e., where he collects the highest utility from any given strategy profile compared to the other specifications). Our analysis highlights that complex strategic interactions, even in simple static games, can yield outcomes that are seemingly at odds with the insights gained from individual decision-making contexts.
