Content area
Abstract
The purpose of this study is to determine what variables are significant in predicting the type of professional liability insurance, if any, an accounting firm would have. The three dependent variables are firms that: purchased commercial liability insurance, are members of a captive company or risk retention group, and have no commercial liability insurance coverage. Independent variables include characteristics of the firm (form of the business, size of the firm) and of the client (recently merged clients, clients that are financial institutions). The study uses the multinomial logit model with maximum likelihood. The Hausman-McFadden test is used to test for independence of irrelevant alternatives. Various methods are used to detect multicollinearity, the most sophisticated of which is the Belsley, Kuh, and Welsch test. Correlation tables and stepwise backward elimination are also used. The seven significant variables identified by the model as useful for predictors are: (1) Is the firm a sole practitioner? (2) What is the size of the firm? (3) What is the firm's review and compilation revenue as a percent of total revenue? (4) Does the firm audit, review, or compile for companies recently merged? (5) Does the firm perform a "cold" review of audits? (6) Does the firm evaluate new clients, and (7) Does the firm use engagement letters?
