Allocative Inefficiency in the Production of Hospital Services

Researchers have offered numerous theoretical models of hospital behavior that suggest inefficiency in the provision of hospital services.' Explanations of inefficiency are tied to the organization and regulation of the industry, principal-agent problems, and the dominance of third-party reimbursement. Eakin and Kniesner [3] use a non-minimum cost function to obtain estimates of allocative inefficiency for 331 U.S. short-term hospitals. In this paper, I fill a void in the hospital cost literature by investigating variations in allocative inefficiency in the short-term hospital industry. This is done by regressing the allocative inefficiency estimates from Eakin and Kniesner against hospital characteristics predicted by theory as determinants of inefficiency. I define allocative inefficiency at the firm level, but from a social point of view. Specifically, allocative inefficiency results from the technically efficient employment of inputs in a combination which is not cost-minimizing. A hospital may choose a non-cost-minimizing set of inputs if it evaluates the shadow prices of inputs as being different from the actual prices. A shadow price is the hospital's internal perception of the unit cost of an input. This perception may differ from the actual input price if, for example, the input enters directly into the hospital's objective function. The hospital is assumed to minimize shadow cost by equating the marginal rate of technical substitution and the ratio of input shadow prices. Thus, it is the shadow cost function that is dual to the underlying technology. The hospital is modelled as a multiproduct firm. A modified translog cost function is used to develop a system of observed cost and share equations which is estimated via seemingly unrelated non-linear regressions. Empirical results indicate that overcapitalization and underemployment of physicians increase short-term hospital costs by about 5 percent. Hospital care expenditures are about 4.5 percent of the gross national product, which is almost three times the percentage in 1960. Thus, 5 percent represents a large amount in actual dollars. Further investigation identifies hospital size, market share, regulations and geographic region as the hospital characteristics related to the measure of allocative inefficiency. This paper is organized as follows. In section II, I develop the empirical model based on a non-minimum cost function. The traditional minimum cost function is a special case and thus