Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)

First Advisor

Vincent Brenner


The hospital industry has undergone radical changes in the past fifteen years with respect to the production and distribution of health care services. The introduction of Medicare's prospective payment system, the struggle to retain physicians and competitive bidding for managed care contracts have created increasing risks for hospitals. Coupled with the increased amount of debt sold by health care issuers, these changes have made determining the information utilized in predicting hospital revenue bond ratings a topic of significant interest to investors, creditors and regulators. The primary purpose of this study was to develop an initial model which might be used in predicting hospital bond ratings. In pursuing this goal this study identified a parsimonious set of variables that are significant in predicting hospital bond ratings. These variables might be of interest to those concerned with hospital reporting disclosure and its regulation. A sample of 127 hospitals was selected from a private data base compiled by Van Kampen Merritt. To be included in the final sample a hospital bond issue must have a Standard and Poor's rating of "B$-$" or better, must be free of credit enhancements such as insurance and letters of credit, and must have information on all variables tested. Sixty-four independent variables are initially included in the analysis. Many of these variables share identical values in their numerators or denominators and are, therefore, highly correlated. Factor analysis was applied to the initial group of variables in order to produce a more parsimonious set of independent variables with less correlation. The number of independent variables was reduced from sixty-four to fourteen. Using the reduced set of independent variables, logistic regression was then employed to construct a hospital bond rating prediction model. Five variables were found to be significant in predicting hospital bond ratings: CMA admissions, net take down, fixed asset financing, total outpatient surgeries and percentage population below poverty. The classification accuracy of the model was tested using the jackknife technique. The overall classification accuracy of the model is 37.8% which is greater accuracy than that due to chance.