Further Evidence on Auditor Selection Bias and the Big 4 Premium

Abstract

A revised version was published in: Accounting and Business Research, Vol. 39 (2), 2009.

A large number of prior studies of the pricing of audit services find a large auditor premium using OLS regression. However, recent research concludes that this premium is explained by auditor selection bias and that when this is controlled for using a two-step Heckman procedure, the premium vanishes. In this paper we discuss some of the problems involved in specifying the Heckman model and examine the potential sensitivity of the estimates it produces. Using a sample of over 36,000 independent UK private companies, we employ decomposition methods not previously used in the audit literature in a comprehensive analysis of the big 4 premium. We also re-estimate the big 4 premium employing a relatively new development in the applied econometrics literature - propensity score matching. Our results suggest that reports of the large auditor premium vanishing when selection bias is controlled for do not generalise and that the Heckman two-step procedure is highly sensitive to model specification and to changes in samples. Matching results indicate that auditees of similar size, risk and complexity pay significantly higher fees if they are audited by the big 4.

Keywords: Audit fees, big 4 premium, propensity score matching, selection bias

JEL Classification: C14, C21, M41, M47, M49