Testing the appropriateness of nonprofit hospitals' tax exemptions with logistic regression analyses of hospital financial data

This study's main objective is to test whether tax-exempt and investor-owned taxable hospitals differ on measures of charity and indigent care provided and profitability. Relieving government burden through the provision of charity and indigent care is frequently cited as justification for nonprofit hospitals' tax exemptions. Also, economic theory suggests that tax-exempt hospitals should require a lower rate of return on equity than taxable hospitals. A secondary objective of this study is to establish a statistical technique that can help governments identify nonprofit hospitals that may not be earning their tax-exemptions. Analyses are performed on hospital data from Tennessee, Florida, West Virginia, and Arizona. Logistic regression with tax-status as the dependent variable is used to analyze the data and test hypotheses. Several control variables are tested for significance before charity and indigent care and profitability variables are added to logistic models. Results indicate that tax-exempt hospitals provide significantly more charity and indigent care than investor-owned taxable hospitals. Contrary to expectations, results indicate that tax-exempt hospitals are more profitable than taxable hospitals. Models developed exhibit better classification ability than is expected by chance, but classification analyses indicate that some tax-exempt hospitals provide less charity and indigent care than do some investor-owned taxable hospitals. Findings related to charity and indigent care provided indicate that the current approach taken by governments of selectively re-evaluating nonprofit hospitals' tax-exemptions rather than revoking exemptions of all nonprofit hospitals may be proper. Results from classification analyses demonstrate that the statistical technique used in this dissertation could help governments identify nonprofit hospitals that warrant having their tax exemptions re-evaluated.