Exploring psychiatric hospital performance using data envelopment analysis and cluster analysis

In this paper, we apply Data Envelopment Analysis (DEA) and cluster analysis (CA) to assess psychiatric hospitals operating in the US. DEA is utilized to examine the relative performance of these hospitals in terms of input use. CA is applied for two purposes. First, we use CA to find similar hospitals prior to the DEA to facilitate peer-groupings; and second, we can also use CA to identify factors that distinguish efficient from inefficient hospitals. To preview our results, we find that even though public hospitals appear to be relatively less efficient than either not-for-profit or for-profit hospitals, no one ownership group dominated any other in terms of efficiency. Using CA, we identify which groupings of hospitals (irrespective of ownership) make up peer groupings based on hospital characteristics. This approach is superior to simply stratifying by ownership which may lead to biased efficiency results.