Exclusive or Not? An Experimental Analysis of Parallel Innovation Contests

We study parallel innovation contests where contest organizers elicit innovative solutions to a set of problems from a group of solvers with limited (financial, time, cognitive) resources. The quality of a solver's solution improves with their effort, yet it is also subject to an output uncertainty, which affects a solver’s control over the link between their effort and winning the contest. Prior theoretical work shows that organizers should discourage solvers from participating in parallel contests in environments with low output uncertainty, where contest outcomes are primarily driven by solver efforts. In this case, organizers should benefit from solvers focusing all of their efforts on a single "exclusive" contest rather than splitting them across multiple "non-exclusive" contests. We test this prediction with controlled laboratory experiments. Our main result (and key managerial insight) is that non-exclusive contests are attractive to organizers even in environments with low output uncertainty where contest theory favors exclusive contests. We link this result to behavioral drivers that induce solvers to exert less effort than predicted on average, while exhibiting a larger-than-predicted variability