Does Korea’s Seminal Audit Reform Improve Audit Quality?

[Purpose] We examine whether the Korea Audit Reform Amendment (KARA) of 2017 improves audit quality for all listed firms. Instead of investigating the effect of individual KARA elements that take effect sequentially, we treat all KARA elements as a package and investigate their effect from the KARA’s first year of 2019. [Methodology] We use all KOSPI listed firms as our sample subject to data availability since KARA affects all listed firms. We use discretionary accruals as a measure of audit quality. For the estimation of discretionary accruals, we use two accrual models (Modified Jones model and a new model) using two alternative estimation approaches to enhance the robustness of audit quality measures. We employ both univariate between－period difference tests and multivariate regression analyses. We test both mean and variance differences between periods for the univariate tests. Also, we use a new metric called ‘excess discretionary accruals (EDA)’ as a modified version of the difference－in－differences test. We use a dummy variable, KARA, as the main variable of interest for the multivariate regression methods. [Findings] We document that KARA significantly improves audit quality. Discretionary accrual levels and variances in the post－KARA period (2019－2020) have been significantly lower than in the pre－KARA period (2011－2018). These findings are robust to differences in financial performance, leverage ratios, auditor reputation, and discretionary accrual models and estimation approaches. [implications] Our research has strong theoretical and policy implications. We contribute to the literature by using a new accrual model and a modified version of the difference－in－ differences method when a new regulation affects all firms. Finally, our finding that KARA improves audit quality provides positive policy implications to the government’s regulatory policy amidst companies’ opposition to the regulation because the regulation comes with high compliance costs.