This study attempts to determine if the language of Form 10-K, the press release, and the annual report were consistent in reporting year-end results. A general semantics model was used to analyze linguistic qualities such as completeness, qualification, facts versus generalizations, appositives and references versus evaluations, signal, and propagandistic words. The Form 10-K and the press release sections were not significantly different when describing the same topics. The language of the annual report was significantly different from both Form 10-K and the press release. The results suggest that when describing the same topics, the language of the press release may be more similar to Form 10-K than may have been assumed by professional and governmental groups. Other reporting elements such as format requirements may play a greater role in influencing the perception of readability of disclosure documents than may have been commonly assumed. The press release may be more useful in providing 10-K topics which not only are more timely but also are consistent with the 10-K in the qualities of language employed.
The risk classification of commercial bank loans is performed by loan officers, bank controllers, auditors, and bank examiners. Despite the importance of this classification decision, little empirical research has been performed to explain this subjective evaluation procedure. In this paper, a simple linear model is developed which reproduces most of the lending officer's classification decisions. Two variables, a debt-to-total-assets ratio and a funds-flow-to-fixed-commitments ratio, provided most of the explanatory power, but a sales trend variable was also significant. For some of the loans for which the model and the actual classification differed, the model's classification was found to be an advance indicator of a subsequent reclassification by the lending officer. The simple three-variable linear model provided much better predictions of loan risk classification than did two popular bankruptcy prediction models.