Regulators worldwide are considering expanding current audit reporting model to include key audit matters (KAM). Proponents argue that current audit reports are standardized and uninformative to financial statement users. Auditors in current reporting regime can choose to add explanatory notes in audit reports, however, few current studies have investigated the information content of these explanatory notes. This thesis conducts a textual analysis of explanatory notes in auditor reports and examines the predictability of auditors’ explanatory notes consisting of both unqualified and qualified opinions in determining the incidence of subsequent restatements. I hand collect material accounting restatements disclosed by the public companies in China from 2003 to 2017 and obtain modified audit reports from the CSMAR database during the period between 2003 and 2015. Based on a sample of 22,850 firm-years from 2003 to 2015 in China, I find that modified audit opinions, in general, can communicate financial misstatement risks, and the probability of such risks increases when the type of audit opinion is more severe. I also find that compared with unmodified audit opinion, modified ones containing explanatory notes have a higher possibility of subsequently being restated. Further, the predictive power is not the same across different types of explanatory notes. I have shown that explanatory notes including notes emphasizing contingencies and uncertainties and those relating to audit scope limitations have greater predictive power in explaining subsequent accounting restatements. My findings are robust with a set of additional tests. The findings of this thesis indicate that auditors do communicate financial misstatement risks in modified audit opinions (MAOs). Moreover, the findings are consistent with and provide evidence to support policy changes in developing new enhanced auditor reports introduced by the standard setters in China.

Prior research has provided evidence that the disclosure of internal controls material weaknesses (ICMWs) is a powerful input attribute in misstatement prediction. However, the disclosure of ICMWs is imperfect in capturing internal control quality because many firms with control problems fail to disclose ICMWs on a timely basis. The purpose of this study is to examine whether the existence of ICMWs, including both the disclosed and the undisclosed ICMWs, improves misstatement prediction. I develop a two-stage machine learning model for misstatement prediction with the predicted existence of ICMWs as the intermediate concept; my model that outperforms the model with the ICMW disclosures. I also find that the model incorporating both the predicted existence and the disclosure of ICMWs outperforms those with only the disclosure or the predicted existence of ICMWs. These results hold across different input attributes, machine learning methods, and prediction periods, and training-test samples splitting methods. Finally, this study shows that the two-stage models outperform the one-stage models in predictions related to financial reporting quality.