This paper contains evidence of accounting measurement rules that are negotiated in private corporate lending agreements. The negotiated sets of rules differ from the regulated set of accounting rules (generally accepted accounting principles). Moreover, the differences between regulated and negotiated rules are systematic and consistent with the economic incentives of borrowers and lenders. Private parties in the market for accounting information are able to produce for themselves at least some of the information required for monitoring lending agreements. The evidence and analysis have implications for: 1. The voluntary choice of accounting rules, 2. The superiority of alternative accounting rules, and 3. The demand for a diverse set of accounting rules.

[This paper investigates the extent to which models based on financial and market variables predict auditors' decisions to issue qualified audit reports in situations involving contingencies or uncertainties. A probit model is developed with the dependent variable indicating whether the firm received a qualified opinion, and the independent variables representing publicly available financial and market variables. The estimated model distinguishes between unqualified (clean) opinions and first-time qualifications and between types of qualifications (e.g., going concern, litigation, asset realizing, and multiple qualifications) in the year of the qualification, for both an estimation sample and a holdout sample. The predictive accuracy of the estimated model is evaluated in terms of misclassification costs for alternative costs of type I and type II errors and for specific prior probabilities of qualified and clean opinions.]

This paper investigates the extent to which models based on financial and market variables predict auditors' decisions to issue qualified audit reports in situations involving contingencies or uncertainties. A probit model is developed with the dependent variable indicating whether the firm received a qualified opinion, and the independent variables representing publicly available financial and market variables. The estimated model distinguishes between unqualified (clean) opinions and first-time qualifications and between types of qualifications (e.g., going concern, litigation, asset realizing, and multiple qualifications) in the year of the qualification, for both an estimation sample and a holdout sample. The predictive accuracy of the estimated model is evaluated in terms of misclassification costs for alternative costs of type I and type II errors and for specific prior probabilities of qualified and clean opinions.