Nursing home spending patterns in the 1990s: the role of nursing home competition and excess demand

Health Services Research, August, 2005 by Dana B. Mukamel, William D. Spector, Alina Bajorska

Interactions between competition, ownership, and NYC location were also explored, as suggested by previous studies (Mukamel and Spector 2000). Only the interaction between NYC and for-profit ownership was significant and are therefore the only interactions included in the models presented here.

One of the difficulties in estimating cost functions is that outputs, days and case mix, may be endogenous with costs. Such endogeneity, if it exists, would lead to bias in the estimated equation. In the case of nursing homes in NYS, however, outputs are likely to be exogenous or only weakly endogenous, because nursing home beds were capped by Certificate of Need regulation, and occupancy rates averaged over 96 percent in each of the three years studied. Thus, total days are exogenously determined, as argued by Gertler and Waldman (1992) and Mukamel and Spector (2000). As excess demand becomes less prevalent toward the end of the decade, endogeneity bias may be of more concern. However, the very high occupancies, even at the end of the decade, mitigate this concern. Case mix may be only weakly endogenous because the payment system links reimbursement to costs for the Medicaid and Medicare patients (Mukamel and Spector 2000).

RESULTS

Figure 1 shows the distribution of empty beds in 1991, 1996, and 1999. In 1991, the majority of markets have a very small number of empty beds, with the mode at 2 and 3. In 1996, the distribution begins to shift to the right with more markets at 5 and 6 empty beds. By 1999, the distribution changes even more. It becomes bimodal with one mode at 5 empty beds and one at 10 and 11, suggesting that there are two underlying distributions, one characterizing excess demand markets with only a few empty beds and one characterizing markets without excess demand and a larger number of empty beds.

Table 2 presents descriptive statistics for all variables included in the analyses. Expenditures data and percent of facilities facing excess demand are shown by year. Data for all other variables are for 1999 only, because there were no large differences over the three years.

Multivariate Results

Table 3 presents the three estimated models, where the dependent variable is log of expenditures for each category. The associations with all variables (e.g., wages and outputs) were as expected. (2) Expenditures changed over time. Clinical expenditures were the same in 1991 and 1996 but increased significantly by 1999. Administrative expenditures also increased, exhibiting a much larger increase than clinical expenditures, increases that were significant in both years relative to 1991. Hotel expenditures on the other hand declined relative to 1991, in both years.

Changes in the Associations between Competition, Excess Demand, and Costs

Nursing homes in more competitive areas had significantly higher clinical (p = 0.0002) and administrative costs (p = .0006). An increase in competition of 0.08 (one standard deviation from the state average at 0.86) was associated with an increase of 2.16 percent in clinical costs and 2.51 percent for administrative costs. Hotel expenditures also increased with competition (at 0.91 percent), although this increase was not significantly different from zero (p = .14).