ABSTRACT Collusion between managers who share private information represents a significant control concern for firms. Prior research suggests that mutual monitoring contracts that incentivize honest reporting do not prevent all collusion, making it important to understand how elements of the control environment may facilitate collusion, as well as how control choices—and collusion itself—affect subsequent behavior within the firm. Results of two experiments show that the frequency of collusion between managers in a repeated‐interaction setting is greatest when one can view the other's reports before making their own (“open internal reporting”) and slack obtained from misreporting is shared with a non‐reporting employee (“shared interest”). Moreover, although open (versus closed) internal reporting increases collusion in a single‐shot setting with or without a residual claimant to managers' budget reports being present, neither openness alone nor shared interest alone increases collusion in a repeated‐interaction setting. Results further suggest that collusion improves managers' perceptions of autonomy and group identification, resulting in greater cooperation on a subsequent task and potentially reducing the costs of some collusion to the extent such cooperation benefits the firm. Finally, open internal reporting worsens managers' perceptions of autonomy and group identification which, for managers who did not previously engage in collusion, leads to less cooperation on a subsequent task. In total, these results highlight to firms that the costs of collusion in repeated‐interaction settings frequently found in practice may be greatest when the common control choices of open internal reporting and shared interest are made in tandem, and that open internal reporting may carry another unintended cost in the form of lower cooperation on other tasks.

Firms allocate significant amounts of common costs, and these allocations have implications for performance evaluation and remuneration. Non‐insulating cost allocations distribute costs based on same‐period relative performance, creating a contemporaneous interdependence between managers that in turn adds uncertainty to the link between effort and performance. In contrast, insulating cost allocations are independent of relative performance during the period and can thus be determined with greater certainty ex ante. In an experiment, we predict and find that managers' effortful performance in a stable production environment—where the pre‐allocation return for effort is constant—is higher when costs are allocated via an insulating allocation compared to when costs are allocated via a non‐insulating allocation. We further find that in an unstable production environment—where the pre‐allocation return for effort can vary from period to period—there are no differences in performance between the allocation methods when managers face a lower return for effort. Conversely, when managers face a higher return for effort in this environment, performance is greater when costs are allocated via an insulating allocation. Taken together, overall performance in the unstable production environment is greater when managers work under insulating cost allocations, suggesting the net effects of cost allocation methods are similar in each type of production environment. As such, our study identifies an important cost—lower effortful performance—of using non‐insulating methods to allocate common costs.