Dealing with Heteroskedasticity, Autocorrelation and Endogeneity in German Audit Fee Panel Data - Comparing Approaches

In this paper we provide an extensive comparison between commonly used linear econometric methods in the audit fee literature and explicitly address their underlying assumptions. As opposed to common practice in similar papers we explicitly consider violations of the strict exogeneity assumption in terms of unobserved firm-specific effects and argue that endogeneity is likely to be present in audit fee data sets. This leads to significantly different results for magnitude and significance level of most explanatory variables. Additionally, we encourage researchers to use the in the audit fee literature not so widely-spread Hausman-Taylor estimator to benefit from its efficiency