We examine auditor switching conditional on the compatibility of clients and their auditors using a unique text‐based measure of similarity of financial disclosures. We find clustering of clients within an audit firm based on this measure. We find that clients with the lowest similarity scores are significantly more likely (9.4%–10.6%) to switch auditors, and will change to an audit firm to which they are more similar. Regarding the effect on audit quality, we find that discretionary accruals are lower when similarity is higher. However, accounting restatements are more likely when text disclosures that are unaudited—business description, and management discussion and analysis (MD&A)—are more similar. We find no such similarity effect for the audited footnotes. Finally, we find that firms that are more similar are less likely to receive a going concern opinion (GCO), but the GCO reporting decision is more accurate. It is unclear if this reflects higher or lower audit quality since firms that are candidates for a GCO are intrinsically different from the average firm in an auditor's portfolio due to their financial distress. One implication of these results is that auditors might have greater involvement in the quality of the text disclosures that are currently not audited.

We propose financial statement similarity as a measure of financial reporting comparability. The firm‐pair version of our measure reflects the degree to which two firms report similar relations within their financial statement items; this version can help managers and market participants identify peer firms. The firm‐year version of our measure reflects the degree to which a firm reports financial statement relations that are similar to other members of its industry; this version can help market participants, regulators, and auditors screen firms for further attention. Our measure uses the presence and amounts of almost all financial items reported by a firm. We validate our measure in four sets of analyses to establish concurrent validity and in three sets of analyses to establish predictive validity. In all these tests, we contrast our measure with the comparability measure in De Franco et al. (2011) and a multivariate measure that considers the presence, but not amounts, of financial statement items. Our measure outperforms the alternatives and can be a useful tool for users.

Natural language is a key form of business communication. Textual analysis is the application of natural language processing (NLP) to textual data for automated information extraction or measurement. We survey publications in top accounting journals and describe the trend and current state of textual analysis in accounting. We organize available NLP methods in a unified framework. Accounting researchers have often used textual analysis to measure disclosure sentiment, readability, and disclosure quantity; to compare disclosures to determine similarities or differences; to identify forward‐looking information; and to detect themes. For each of these tasks, we explain the conventional approach and newer approaches, which are based on machine learning, especially deep learning. We discuss how to establish the construct validity of text‐based measures and the typical decisions researchers face in implementing NLP models. Finally, we discuss opportunities for future research. We conclude that (i) textual analysis has grown as an important research method and (ii) accounting researchers should increase their knowledge and use of machine learning, especially deep learning, for textual analysis.