Technical Efficiency of Swiss Nursing Homes - A DEA and Stochastic Frontier Analysis

Rationale: in 2003, 14.5% of total health expenditures in Switzerland were related to elderly people long term care. To date, this causes an important burden for health insurances and young generations that could be reduced by increasing nursing homes technical efficiency. Objectives: evaluation of the Swiss nursing home efficiency and its determinants using non-parametric (DEA) and parametric (SF) methods. Comparison of technical efficiency measurement methods. Methodology: The data set provided by the Swiss Statistical Office contains information about the 1431 nursing homes in Switzerland for 2004. A set of theoretically relevant variables was selected to conduct the analysis. The non-parametric frontier was estimated with the Data envelopment analysis method (DEA). Labour (medical and non-medical hours) and non-labour inputs (other running costs) were connected to the number of patient days provided, which is a frequently used output to analyse nursing homes' and hospitals' efficiency. Then differences in case-mix and quality of care were taken into account and net efficiency scores were computed. An OLS model was estimated regressing net efficiency on size, status and occupancy rate in order to explain performance variation. A stochastic frontier (SF) was then estimated including control variables in order to better caption the error part due essentially to inefficiency. Results of both methods are finally compared to test the validity and convergence of our results. Results: the mean efficiency score is much higher with SFA. Efficiency scores of both methods are correlated. A clear trade-off between efficiency and quality of care was highlighted. Conclusion: the mean level of efficiency among Swiss nursing homes is surprisingly low. As expected, scores obtained with SFA are higher because this method considers a part of the distance to the frontier as random error whereas DEA method considers the whole distance as inefficiency. To get a better view of nursing home efficiency we should find more precise indicators of case-mix and especially of quality of care