This study compared the economic efficiency of Veterans Administration (VA) Nursing Home Care Units (NHCUs) and State Veterans Homes (SVHs). Technical efficiency was estimated through use of a stochastic Cobb-Douglas frontier production function with a composed error term. Inputs included full-time equivalent RNs, LPNs, nurse assistants, rehabilitative personnel, other indirect care personnel, and number of beds. The output measure was the weighted RUG-II casemix adjusted average daily census. Allocative efficiency was evaluated through estimation of cost-minimizing first order conditions utilizing the price of each input.

Both NHCUs and SVHs were found to be technically inefficient on average, suggesting that the total cost of providing care to veterans may be higher than necessary. Given the assumption of equal quality levels, no mean differences in technical efficiency between NHCUs and SVHs were found. When the technical efficiency estimates were regressed on six patient condition quality variables, three quality variable coefficients were significant and negative. This suggests that quality may account for use of resources beyond that accounted for by the RUG-II casemix. Inclusion of the quality variables resulted in group membership becoming significant, suggesting that, on average, SVHs may be more technically efficient than NHCUs.

Allocative inefficiency appears to be present in both NHCUs and SVHs. However, marginal price measurement complications, coupled with theoretical problems in abstracting allocative inefficiency from statistical noise, generated difficulties in interpretation. Given certain assumptions, the results indicate a possible systematic bias toward hiring more highly skilled personnel on the part of NHCUs. This bias may be aggravating the extent of NHCUs' allocative inefficiency.

These findings provide insight into the possible effects of current VA policy. Shifting more patients from NHCUs to SVHs may promote better management of the finite resources available to the federal government since SVHs may be more technically efficient. Moreover, given equal levels of technical efficiency, SVHs may produce care of a higher quality than NHCUs, if the quality variables are actually capturing quality level differences. Finally, the possible NHCU hiring bias implies that the policy may promote better use of lower skilled, and potentially less costly, personnel.