Sampling Information in Strategic Audit Settings.

Abstract The audit setting is modeled as an interactive game between an auditor and an auditee. In this setting, the auditor's strategies include accepting or rejecting the auditee's reported account balance based on prior information or extending audit procedures to provide additional information regarding the auditee's strategy. The auditor's and auditee's incentives are exogenously restricted to correspond to plausible audit scenarios. In particular, the auditor is assumed to prefer accepting the auditee's reported amounts when "error rates" in the accounts are low and rejecting when "error rates" are high. Similarly, given an accept decision by the auditor, the auditee prefers high error rates (due to the auditee's private costs of reducing errors) and, given a reject decision, the auditee prefers low error rates. We demonstrate that these intuitive restrictions on incentives result in "ran- domized" strategies, thereby endogenously introducing noise in the audit environment. Using the benchmark case of costless information, the value of additional information to the auditor is demonstrated. Next, two cases of costly information, perfect information with fixed cost and imperfect (e.g., sample) information with variable cost, are analyzed. Using these cases, factors associated with increased expenditures on information (i.e., increased audit procedures) are identified. In addition to the cost of information, these factors include the relative payoffs to the auditee for various strategy combinations. As these relative payoffs vary across auditees and over time, different levels of audit activities are expected as equilibrium behavior. Thus, we initiate a descriptive theory of auditor strategies which emphasizes the interactive relationship between the auditor and auditee.