Audit and Non‐Audit Fees: New Zealand Evidence

This study adds to the evidence concerning the link between the fees that auditors charge for audit and non‐audit services to the same client. We argue that audit fees will be cross‐subsidised in a particular way: they will be reduced dollar for dollar by the expected profi ts from nonaudit work that is ‘tied’ to the engagement in the sense that such work generally goes to the incumbent auditor. This argument is tested using New Zealand listed company data for 2002. A nonlinear model for audit fees is developed that respects the mathematical relationship between audit and non‐audit fees implied by the argument from tied profits, and which includes standard control variables in a mathematically consistent way. When linearized for estimation the model has a form that differs from the linear regressions commonly used. The coefficient which tests for cross‐subsidisation is not significant, and the confi dence interval for that coefficient excludes plausible numerical values. Changing the way in which non‐audit fees are scaled and using a conventional (but unjustified) control for subsidiaries leads to a significant but spurious positive association between audit and nonaudit fees. This indicates the importance of avoiding mathematically infeasible specifications in audit‐fee research.