An experimental study of sorting in group contests

Date: 2015-12
By: Philip Brookins (Department of Economics, Florida State University)
John P. Lightle (Department of Economics, Virginia Commonwealth University)
Dmitry Ryvkin (Department of Economics, Florida State University)
We experimentally explore the effects of sorting and communication in lottery contests between groups of heterogeneous players whose within-group efforts are perfect complements. Subjects are assigned a type — A, B, C or D — that determines their cost of effort, with A having the lowest cost and D the highest cost, and are then assigned to one of the two two-player groups competing in the contest. Theory predicts that aggregate contest output increases in the variation in abilities between groups, i.e., the output is maximized by the most unbalanced sorting of players into groups — (A,B) vs. (C,D) — and minimized by the most balanced sorting — (A,D) vs. (B,C). That is, the equilibrium prediction goes against the “competitive balance" heuristic. In the absence of communication, this prediction is directionally confirmed, although the effect is not statistically significant. In the presence of within-group communication, however, we find that total output is 33% higher under the balanced sorting as compared to the unbalanced sorting — a reversal of the prediction, but in line with the heuristic. This result is driven by an increase in output by (B,C) groups under the balanced sorting and a strong decrease in output by the underdog (C,D) groups under the unbalanced sorting, relative to no communication. These results are at odds with previous studies that find that within-group communication always increases output, and suggest that the effect of communication depends strongly on the configuration of heterogeneity between and within groups. Competitive balance is confirmed as a robust sorting heuristic for sustaining competition and high effort provision in group contests.
Keywords: group contest, sorting, complementarity, heterogeneous players, experiment
JEL: C72 C91 D72 M54