A Data Analytic Approach to Predicting Firms that Corrected Prior Period Misstatements

Abstract

When Staff Accounting Bulletin 108 (SAB 108) was issued by the Financial Accounting Standards Board (FASB) in late 2006, US companies were obligated to use the dual approach – the “rollover” and the “iron curtain” approaches resulting in the correction of the prior period errors in both the income statement and the balance sheet. We call these SAB 108 adopters PPA (prior period adjustment) firms. We present a comparative examination of three data analytical models (decision tree, discriminant analysis, and logistic regression) in predicting PPA firms. While analyzing the holdout sample, the decision tree model is more accurate in predicting PPA firms than the other two. Our recommendation is to apply all three models to predict PPA firms and then use the assessment provided by two or more models.