Modelling issues in the relationship between audit and non‐audit fees

Purpose - The purpose of this paper is to examine modelling issues in the research of the relationship between audit and non‐audit fees by comparing the outcome of a single‐equation model of fees to the results of simultaneous equation model (SEM) of these interactions for a sample of UK listed companies and also by exploring the benefits of using a panel data approach. Design/methodology/approach - The paper is an empirical analysis of audit and non‐audit fees for a sample of 2,072 UK companies. Three types of models, single linear equation, SEM and panel data, are compared and the strengths and weaknesses of each discussed. Findings - The results indicate possible mis‐specification in the single linear equation model and the potential for simultaneous equation basis in the SEM. The panel data analysis confirms the findings of prior literature that there is a relationship between audit and non‐audit fees but statistically insignificant. Research limitations/implications - The implication of these findings is that previously published findings on audit fees, which used single equation, may have overstated the case for a relationship between audit and non‐audit fees due to problems in the econometric models. Practical implications - The issues addressed in this paper are very pertinent to a better understanding of the role of the auditor in the corporate environment. In particular, the findings are relevant to the debate on auditor independence and corporate governance. Originality/value - The study contributes to knowledge of the behaviour of audit fees and non‐audit fees, cross‐sectionally and through time and the most appropriate models for describing that behaviour.