Optimal dynamic auditing based on game theory

A dynamic model based on game-theory is proposed to address the problem of fraud detection in auditing under non-linear payoff functions. Non-linearity is introduced by incorporating learning and sympathy effects in the audit process. It is proven that the audit/fraud detection game between two new engagement parties has a unique mixed strategy equilibrium, between an experienced auditor and a client has a unique pure strategy equilibrium, whereas in the long-run the game converges to a pure strategy equilibrium. In addition to this, to ensure an acceptable level of quality in the audit process, a closed form formula used to estimate the optimal auditor’s replacement time is extracted. The validity of the proposed scheme is tested on empirical data and modeling results comply with the International Standard of Auditing that requires the key audit partner to be rotated after a predefined period.