Is Taking a Pill a Day Good for Health Expenditures? Evidence from a Cross Section Time Series Analysis of 19 OECD Countries from 1970-2000

This paper differs in two ways from previous comparative health system research. First, it focuses on the impact of pharmaceutical expenditures on total health expenditures as trends in pharmaceutical expenditures have been blamed of being a major driver of national health expenditures. In addition to pharmaceutical expenditures, other variables of interest are income, public financing, public delivery, ageing and urbanization. Second, the analysis includes a thorough sensitivity analysis on the proposed model using four samples (with and without the US, and imputed and not imputed data) to address the issue of robustness. Based on a typology of health care systems, trends of relevant explanatory variables are described using OECD Health Data 2003 data. Unlike any other of the variables, pharmaceutical expenditures show contradicting trends when measured as per capita pharmaceutical expenditures and pharmaceutical share of total health expenditures. Next, a regression analysis is performed on data from 1970 ? 2000 for 19 OECD countries. Regression diagnostics indicated the absence of multicollinearity but the presence of heteroscedasticity and autocorrelation. Based on the Hausman test, a fixed effect model was chosen. As in all previous empirical research, per capita GDP turned out to be the most influential explanatory variable. While public financing of health care was always three out of four samples significantly inversely related to health expenditures, public delivery as a NHS dummy was always significantly positively related to the dependent variable. Unlike previous research, ageing is consistently and significantly related to higher total health expenditures and, so is urbanization. Finally, all samples show a highly negative relationship between share of pharmaceutical expenditures and health expenditures, suggesting support for the substitution theory.