Is Health Care a Necessity or a Luxury? Evidence from Local Quantile Regressions

In this paper, local relationship between per capita health expenditure and GDP is investigated with local quantile regressions. The advantage of using local quantile regressions is the assumption of homogeneity on per capita health care expenditure could be relaxed so that number of countries in a cross-sectional model can be enlarged. Per capita GDPs and health expenditures of 151 countries in 2000 and 25%, 50% and 75% quantile regressions are considered in this paper. To check significance of the relationship in local quantile regressions, significance tests for nonparametric regressions suggested by Racine(1997) and Ait-Sahalia et al.(2001) are implemented. The adequacy of of these tests are investigated by Monte Carlo simulations. Based on our primary simulation results, these test are suggestive. Our empirical results show that per capita GDP has significant effect on per capita health expenditure. The sizes of effect are quite different among 25%, 50% and 75% quantiles. Besides, we also find the effects are distinct among different ranges of per capita income. In other words, the per capita health care expenditures are heterskedastic. The conditional distribution of health care expenditure on per capita GDP is asymmetric and skewed to the left. For low per capita GDP countries, the effect of income on health expenditure is insignificant, which indicates the health care is a necessity. On the contrast, the effect becomes larger for the countries with high per capita GDPs, which implies the health care is a luxury for these countries