Article ID: | iaor20122194 |
Volume: | 136 |
Issue: | 2 |
Start Page Number: | 297 |
End Page Number: | 305 |
Publication Date: | Apr 2012 |
Journal: | International Journal of Production Economics |
Authors: | Valdmanis Vivian, Leleu Herv, Moises James |
Keywords: | design, combinatorial optimization |
Authors of past studies focusing on returns to scale in hospitals proffered mixed results. These seemingly contradictory findings have probably arisen due to different methodological approaches (parametric or non parametric), different aggregation levels of analysis (hospital/department/units), nature of data (quantity data or economic values) and also due to technological improvements operating in hospitals and case mix adjustment to account for the severity of patients' conditions. In this paper, we apply a new approach to determining returns to scale for single and multi‐output homogenous technologies, which is different from traditional DEA models. Our approach is characterized by (1) a non parametric approach based on quantity data that allows us to avoid assumptions on cost minimization or profit maximization behavior of hospitals, on relevancy of economic values for hospitals (costs, revenues and prices) and on a priori specification of the health care production function, and (2) an analysis of optimal productivity size at both the disaggregated level of intensive care units and at the aggregated hospital level. The methodological advantage is that we can unambiguously define increasing returns to scale, which is lacking in more traditional non‐parametric approaches because of the convexity assumption imposed earlier. We apply the methodology to intensive care units (cardiac care (CICU), medical/surgical care (MSICU), pediatric care (PCIU) and neonatal care (NICU), which are operating in 235 general short term hospitals of Florida state in 2005. We also consider the hospital level by analyzing the general activity of the hospitals in our population. To summarize our findings, we find that 60% of intensive care units are operating at increasing returns to scale, 10% are operating at optimal productive size and 30% are characterized by decreasing returns to scale. In average intensive care units operate 40% under the optimal size. The policy implication of this result should be an increase of the size of all types of intensive care units to meet productivity gains. The picture is completely reversed at the aggregate hospital level. Here decreasing returns to scale prevail for 65% of hospitals while only one fourth are operating at increasing returns to scale. In average hospitals' number of beds should decrease by 40% to reach the optimal productivity size. One policy solution may include reallocating resources from general beds to the more specialized beds.