Abstract

This study utilizes “doubts of going-concern”, “lack of consistency” and “realized investment income based on non-audited financial statements” opinions provided by auditors to construct sixteen predictive models of financial distress. Two different approaches: single-period logit model and discrete-time hazard model are developed to forecast financial distress and the results obtained are compared among the various models. Furthermore, additional tests revise to adopt the auditors' opinions from the most recent auditors' reports in the predictive models. If a firm suffers financial distress prior to April 30, additional tests choose the auditors' opinions in the year before the last. The results show that the accuracy of discrete-time hazard model is superior to that of single-period logit model. In addition, the coefficients of auditors' opinions are statistically significant, which implies that the auditors' opinion provides incremental information content in explaining financial distress. Finally, the results of additional tests show that the auditors do not timely issue “going-concern” opinion before financial distress. Even the Big-4 CPA firms only perform slightly better than the other CPA firms in predicting financial distress. The findings do not strongly support the views that auditors meet the ethics requirements in financial distress prediction and the issuance of modified opinions.