This study aims to identify predictive patterns of material accounting misstatements using detailed audit adjustments and preaudit financial statement data to improve early detection of reporting risks among listed companies.
A regularization-based prediction algorithm is applied to identify patterns in preaudit data that signal the risk of material misstatements before audit completion. The analysis uses preaudit and audited financial statements from Croatian listed companies.
The key predictors differ systematically across types of misstatements. The most relevant income-related misstatement predictors include absolute total accruals, company size, high return on assets, receivable turnover and sales. However, accounts payable turnover is the strongest predictor of material adjustments across operating, investing and financing cash flows. A distinct predictor set is associated with the likelihood of receiving modified audit opinions. Smaller firms exhibiting financial distress indicators (going concerns and losses), higher soft assets and greater leverage are more likely to receive modified opinions.
The findings offer valuable insights for auditors, audit committees, financial statement users and other stakeholders by enabling earlier identification of misreporting risks, thereby enhancing audit efficiency and financial reporting quality.
Using a unique institutional setting in which firms disclose both preaudit and audited financial statements, this analysis extends prior research by examining predictors of non-income-related misreporting alongside income-related misstatements. The dataset mitigates common challenges in fraud and restatement prediction, including selection bias and severe class imbalance, and provides novel evidence on predictors of accepted versus waived material audit adjustments reflected in modified audit opinions.
