Abstract
This paper estimates a multiple-output hospital cost function using a panel data technique that allows for correlation between unobservable individual effects and observable determinants of behavior. Analysis of 1733 facilities for the period of 1987-1991 yields estimates that differ widely from those obtained from a more standard cross-sectional procedure. While the latter method results in negative and stable measures of ray economies of scale, the panel model indicates positive economies of scale that fall slightly over time. © 1997 by the President and Fellows of Harvard College and the Massachusetts Institute of Technology