Does Automation Improve Financial Reporting? Evidence from Internal Controls

Musaib Ashraf

DOI: https://doi.org/10.2139/ssrn.4709274

2024-01-01

SSRN Electronic Journal

Abstract:Automation -- such as machine learning, robotic process automation, and artificial intelligence -- is the next major technological leap in accounting and financial reporting, and I empirically study whether (public) firms’ use of automation technology improves their financial reporting, specifically focusing on the internal control environment. I document two critical inferences. First, I find evidence which suggests that automation improves financial reporting quality. Specifically, firms’ use of automation in the financial reporting process is associated with a reduction in internal control material weaknesses. This association is consistent in a levels analysis with firm and year fixed effects, in a changes analysis, and in a propensity score matched difference-in-differences analysis. Second, I find evidence which suggests that monitoring of the financial reporting process is decreased after automation, likely because of a perception that automation reduces the need for monitoring vis-à-vis stronger internal controls. Specifically, automation is associated with higher external audit fees and audit committee meetings in the initial years after a firm implements automation but associated with lower external audit fees and audit committee meetings in subsequent years. I also find evidence which suggests that this decreased monitoring may be costly: when internal control failures do happen for firms with automation, the failures appear to be more material, as proxied by stronger negative market reactions. In aggregate, my evidence provides nuanced insights regarding whether automation technology improves financial reporting.