Governance, Firm Characteristics, and Audit Fees: Panel Evidence from Indonesian Financial Firms

Abstract

This study reassesses governance and firm-level determinants of audit fees in Indonesia's financial sector while explicitly accounting for the repeated firm-year structure of the data. The sample comprises a balanced panel of 59 listed financial firms observed from 2020 to 2024 (295 firm-year observations). Audit fees and total assets are expressed in natural logarithms; the highly skewed counts of internal audit personnel and subsidiaries are transformed using ln(1+x). The analysis replicates conventional pooled ordinary least squares (OLS), then introduces year effects, firm-clustered standard errors, random-intercept estimation, firm fixed effects, and a correlated random-effects (Mundlak) specification. Firm effects are jointly significant, and the Mundlak test rejects the conventional random-effects independence assumption. Accordingly, the correlated random-effects model is treated as the preferred specification. Conventional OLS suggests positive associations for internal audit staffing, audit committee activity, and firm size. Once within-firm dependence and persistent firm heterogeneity are addressed, however, the governance coefficients are no longer statistically distinguishable from zero. Firm size retains a strong positive association in pooled and random-effects models, while its within-firm coefficient remains positive but marginal in the preferred model. Year effects are jointly significant, indicating that common temporal shocks also shaped audit pricing during 2020-2024. The findings show that inference about audit-fee determinants is sensitive to panel specification and standard-error correction. Persistent differences across firms and changes over time explain more of the observed variation than annual changes in the simple governance, leverage, and subsidiary-count proxies used here.