This new application proposes to analyze the financial performance, operating characteristics and market area conditions of approximately 4,000 U.S. short-term general hospitals from l985-l993, to answer questions about the effects of hospital mergers. The proposed work will build on previous work by this team of investigators, extending their previous work to include additional years of data and to address additional issues. The previous study used a panel of American Hospital Association (AHA) data from 1985 to 1991 to look at hospital mergers. They found that mergers were associated with an average savings of 5%, which was shared between the hospital and consumers. The savings were greater for mergers of similar-sized hospitals and for mergers of hospitals in competitive markets. In near monopoly markets, there were no savings. The proposed study will use a longitudinal data set for the years 1985-1993 to examine the effects of hospital mergers on costs and consumer prices. It will extend the previous work by: 1) Extending the panel through 1993, which will increase the number of mergers observed; 2) Adding data that allow the examination of the impact of HMO penetration; 3) Examining the impact of geographic proximity of the merging hospitals; 4) Looking to see if the savings associated with mergers vary with the degree of overlap in the services offered by the merging hospitals; and 5) Conducting more in-depth analysis of the interaction of market concentration and mergers in their effects on hospital prices and costs. These objectives are formally stated in a long list of specific hypotheses. The model will be a modified Cobb-Douglas model that has been used by other investigators in this area. The investigators acknowledge most of the potential econometric problems and note how they will address them in the study. The panel data will be used to estimate fixed effects models that remove the potential bias of unobserved hospital-level effects. Different specifications of the merger variable and its interaction with the market conditions as measured by the Herfindahl-Hirschman Index (HHI) will be tested. The investigators address the problem that mergers cause by changing the sample size, and describe two possible ways of addressing the issue: combining the data for the merging hospitals for the pre-merger period, and using fitted pre-merger values for the pre-merger data.
