Per capita measures of state government expenditures are used to investigate how per capita health expenditure affects state government finances for the period 1992-2004. Table 14 provides descriptive statistics for these variables.
The changes over time in the outcomes and also in per capita health care expenditure are now analyzed for each of the 50 states. The average change in each variable over these 13 years is calculated for each state, and the mean changes over time (computed across all 50 states) are presented in Table 15.
Next, the changes over time in various outcomes are compared against the change over time in per capita health care expenditure. Specifically, the average change over time for each outcome is plotted against quartiles of the change in per capita health expenditure (Figure 10).
Figure 10 suggests weak positive correlation between the average increase in per capita state government expenditure on health and change in per capita health expenditure.14 There appears to be somewhat stronger correlation between increase in per capita health expenditure and the average increase in per capita state government expenditure on all other categories. Specifically, per capita state government expenditure on all other categories registers an increase of $130 in the first quartile and a somewhat higher increase of $170 in the fourth quartile of the change in per capita health expenditure. The same holds true for per capita state government debt that increases by $70-80 in the first three quartiles, and by $100 in the fourth quartile of the change in per capita health expenditure.
Figure 10: Average Change in Per Capita State Government Expenditures &
Debt By Quartiles of Change in Per Capita Health Expenditure
Note: All variables are in thousands of dollars
Q1,mean change: health-.03, debt-.08, all other-.13, total-.16; Q2,mean change: health--.03, debt--.07, all other--.14, total-.17; Q3,mean change: health--.03, debt--.08, all other--.14, total-.17; Q4,mean change: health--.03, debt--.10, all other--.17, total-.21.
This descriptive evidence seems to suggest that state government expenditures on health and on other sectors tend to move together over time. This could be because sectors such as health and education compete for funding so an increase in health expenditures is accompanied by a similar increase in several other sectors. Alternatively, this relationship could reflect unobserved state-specific factors. For example, states with rising per capita income would have larger increases in per capita health expenditures, and also have faster increases in various categories of state government expenditures, including health.
To further investigate this issue, a regression analysis is now carried out that incorporates state and year fixed effects. The regressions also control for state-specific factors that vary over time, e.g., per capita income, per capita state government revenue, and unemployment rate. As before, all regressions were estimated with HEW or robust standard errors to account for heteroskedasticity in the data (White, 1980). Results from both specifications – in levels and in logs – are reported in Table 16.
The regression analysis suggests a significant positive relationship between per capita health expenditure and state government expenditures on health and other sectors (columns (1) and (2) of Table 16). State government expenditures on health (as well as on most other sectors) tend to move together with increase in total spending on health; in addition, the growth rates of state government expenditures on health exceed the rate of increase in population so that the per capita measures keep rising over time.
There is no evidence that rising health expenditure crowds out other components of state expenditure. On the contrary, there is some evidence that the government’s total expenditure and debt increase over time (columns (3) and (4) of Table 16) as health care costs and state government expenditures rise in tandem. However, while all other changes are significant even in percentage terms (or, in terms of elasticity with respect to per capita health expenditure), change in government debt is not significant in percentage terms when we look at the results from the log specification in column (5) – (8). Further investigation revealed that the level of debt in 1992 was positively correlated with increase in health care costs. So, even though states in the fourth quartile of the change in per capita health expenditure experienced the largest increase in debt (Figure 11), it was a small increase in percentage terms because these states already had high debt.