Abstract We use experimental markets to test the Datar, Feltham, and Hughes (DFH) 1991 model of entrepreneur choice of auditor and retained ownership in initial public offerings (IPOs). DFH predict that entrepreneurs use retained ownership to signal IPO value and substitute high‐quality auditors for retained ownership to signal value as the risk of the IPO increases. Given the mixed support for DFH from archival research, we conduct experimental markets that directly operationalize the model's decision variables, which permits a direct test of whether the model is descriptively valid. In addition, our market setting provides a strong test of this theory by including an alternative Nash equilibrium also present in field settings, one in which only auditor quality is used by entrepreneurs to signal IPO value. Our results suggest that DFH predict entrepreneur behavior in baseline markets where both computerized investors and auditors are programmed to price consistently with the DFH equilibrium. However, the DFH model does not describe behavior when “robot” investors are replaced with human investors in the market. The results suggest that entrepreneurs and investors strategically interact in a manner that leads them away from the DFH equilibrium and toward the alternative Nash equilibrium behavior of entrepreneurs with high‐value assets hiring high‐quality auditors irrespective of IPO risk. Our results imply that the DFH model has limited descriptive validity, document the importance of strategic behavior on market equilibrium formation, and suggest that the mixed results found in prior DFH‐based field studies may reflect the model's low descriptive validity.

Abstract This paper explores the behavior of workers in an environment where it is efficient to engage in the mutual exchange of help. Experimental data show that output and workers' payoffs are greater under team‐based incentives than under individual incentives in an environment where coordination is difficult. However, when the environment is more conducive to coordination (that is, a setting where agents interact repeatedly), output and payoffs are greater under individual incentives. Manipulation of the amount of mutually observable information provides evidence that team‐based incentives, relative to individual incentives, create a more difficult coordination problem for workers and that cooperation requires a richer informational environment.