An Experimental Investigation of Retention and Rotation Requirements

This paper provides results of an experiment designed to investigate how mandatory rotation or retention of auditors may affect auditor reporting. Repeat audit engagements are believed to be a source of economic rents to auditors, and as such they are believed to pose a threat to auditor independence. Our study examines under controlled experimental conditions whether mandatory rotation would materially increase auditors' independence compared to similar settings in which the requirement is not imposed on auditor-client relationships. Our experimental design includes four settings. The first setting has neither mandatory retention nor mandatory rotation and is a benchmark for comparing the other three. The second setting has mandatory retention, without a rotation requirement. The third has mandatory rotation but not retention, and the fourth has both retention and rotation. We find that in the setting without mandatory retention or rotation, managers made investments at levels higher than predicted by our model, and auditors issued a high frequency of reports that favored management. In the retention setting, auditors reported more conservatively than in the baseline setting. Once the retention period ended, however, auditors and managers converged to the same type of cooperative behavior observed in the base setting. In the setting with a rotation requirement, auditors reported more conservatively than they did in the previous two settings. The lower frequency of favorable reports was also observed in the non-rotation periods of the rotation setting. This outcome was not predicted by our model, and it suggests a spillover effect of the rotation requirement to non rotation periods. Our fourth setting included both retention and rotation requirements. We found that this setting had the highest level of conservative reporting but only nominally so relative to the setting with only mandatory rotation. Investment levels were also relatively high, but only in the retention periods and in the period of mandatory rotation; dropping materially in the one period in which the manager had the choice of dismissing the auditor. This result is consistent with the model's predictions since the manager could lower the investment in that one period and still expect the auditor to provide a favorable report.