WEDNESDAY, OCTOBER 20th, 2021, 13:00 h. @ ICAE Seminar Room (N101- Prefabricado)
Álvaro Name Correa (Carlos III)
TITLE: Multiple Prizes in Tournaments with Career Concerns (joint with Huseyin Yildirim, Duke)
ABSTRACT
In a first period a designer selects the prize structure of a rank-order tournament in a setting where risk-neutral agents decide whether to participate or not. The score of a participant agent is determined by the sum of his (natural) talent, effort, and the noise term. Players besides caring about winning a prize, also have career concerns, and, as it is common in that literature, they do not know their talents. The market, however, learns about them through their performances. In a second period agents enter the (labor or talent) market regardless of their participation in the tournament. The market then makes each agent a wage offer equal to his inferred talent.
We show that when agents are more patient or the variance of the error term is low, the tournament offered by the designer becomes less competitive (with more prizes) in order to induce participation. In the first case, career incentives are strong, in the second case, both career and tournament incentives are intense. Thus, given the market's expectation to perform, agents would be reluctant to enter the tournament, requiring the designer to announce a high number of (small) prizes. Furthermore, the optimal number of prizes is single peaked in the population variance of talent. This is because the most motivated agents in the tournament would be those who are moderately well-known by the market. Thus, given the market's expectation such agents would also be the most reluctant to enter in a competitive tournament.
We also examine an alternative setting where only rankings are observed, and show that the designer offers a steeper prize schedule with fewer prizes with respect to the benchmark. This is due to career marginal incentives being less powerful in that case.
You can join the seminar through Google Hangout Meets at the following link: