On the international stability of health care expenditure functions: are government and private functions similar?

Abstract

This paper studies the stability of health care expenditure functions in a sample of OECD countries. We adopt the cointegration approach and the results show that there is a long-term relationship between total health care expenditure (HCE) and gross domestic product (GDP). However, the existence of cointegration is only shown when we admit the presence of some changes in the elasticities of the model. Our results also provide evidence against the existence of a unique relationship between health and GDP for the sample. Thus, we can conclude that the differences in health systems may cause differences in the aggregate functions. Additionally, we examine aggregate health functions for government (GHCE) and private expenditures (PHCE), again finding evidence of different patterns of behaviour. Finally, we open a discussion on the character of health as a necessary or luxury good. In this context, we find differences between the government and the private function. In order to illustrate these findings, we propose a theoretical model as an example of the influence of political decisions on income elasticity.

Introduction

Total health care expenditure (HCE) represents some 7.8% of the gross domestic product (GDP) for the OECD countries. When some leading members are considered, this percentage rises, for example, to 9.3% in the case of France and 12.9% in the USA, with all these values referring to 1998. Thus, given the clear importance of the health sector in the GDP, the growing volume of research into its evolution and determinants that we have witnessed in the last decade is quite understandable.

Furthermore, there are additional arguments in favour of analysing the evolution of HCE from an aggregate or macroeconomic perspective. From amongst these, in this paper we are particularly interested in three: the recent creation of new supra-economies, such as the EU; the increase of the private health care sector that some OECD countries have experienced; and the debate on the characterisation of health as either a luxury or first necessity good.

The first question that we set out to analyse is the stability of the HCE function given the EU integration process. Hitiris and Nixon (2001) found a special situation related to health convergence in EU countries, where the “general move towards closer economic integration … does not only result in convergence of income, … but it also extends to other sectors and the health care sector in particular”. That is, economic integration and GDP convergence drive health care expenditure convergence. Thus, we consider that EU convergence is an important factor that influences on health sector, but there is no clear evidence in favour of the existence of a single health function for all the EU member states. The situation is even more complex if we consider the existence of different systems for financing health, or if we extend this analysis to the other OECD countries. So far as the second question is concerned, we should note that most of the studies which analyse health expenditure have exclusively focused on total health care expenditure, without considering its components. This approach may mask the existence of a different pattern of behaviour in government and private health expenditures and, therefore, it is more appropriate to carry out a disaggregated analysis.

In spite of the fact that a number of recent papers have investigated these issues, it is also true that some work still seems to be necessary in order to provide an appropriate response to these aspects. Various authors have studied the performance of the health function, with most of these works being based on a simple relationship between HCE and the GDP: here, we can highlight the contributions of Hansen and King (1996), Blomqvist and Carter (1997), McCoskey and Selden (1998), Gerdtham and Löthgren (2000), Karatzas (2000) or Roberts (2000). It is possible to draw some important conclusions from this body of work. First, we should note that the variables commonly employed are not stationary. Thus, the OLS techniques are no longer valid and, as a result, the cointegration approach has to be used. Secondly, there is no unanimous conclusion in favour of the presence of a cointegration relationship between HCE and the GDP: the results obtained by the different authors are somewhat contradictory. The contribution of Gerdtham and Löthgren (2000) has helped to shed some light on this puzzle. However, their results, based on the use of panel unit root tests, are not free of criticism, as is reflected in Taylor and Sarno (1998) or Banerjee et al. (2000). Furthermore, we should note that the panel data methods implicitly involve the assumption of the homogeneity of tastes, preferences or/and production function. Additionally, these factors can change in a non-simultaneous way in different countries, and consequently the heterogeneity in the health care expenditure functions could be wider. Therefore, the homogeneity hypothesis is a quite restrictive and, furthermore, unrealistic assumption for health economics. Thus, we consider that these methods are not able to offer an appropriate performance in this area and that it is necessary to adopt a different approach.

Against this background, a methodological approach that has not yet been considered and that, in our view, could help us to understand the behaviour of health expenditure is the possible non-stability of the health care expenditure/GDP relationship. We should recall that most of the empirical analyses use a sample period which runs from the 1960s until the 1990s. It is obvious that, during this sample period, the OECD economies have experienced some structural changes, crises and booms. Thus, the assumption of stability in the health care expenditure/GDP relationship seems too restrictive. If this hypothesis is correct, we should use those methods that allow us to test and estimate the cointegration relationship in the presence of structural breaks; otherwise, the results on both inference for cointegration and on the estimation of the long-run relationship may be biased and incorrect.

If we can provide evidence in favour of the existence of a long-run relationship between health care expenditure and the GDP, we could then focus on another interesting topic in health economics, namely, whether health is a luxury or a necessity good. Once again, there is no unanimous conclusion in the literature with respect to this question. For example, Getzen (2000) has recently pointed out that “health care is an individual necessity and a national luxury”. However, the empirical evidence does not sustain this claim. A possible explanation for this result is the presence of an aggregation problem, in the sense that most of the studies in this area have focused exclusively on the analysis of HCE. By contrast, in this paper we will carry out the analysis for both government health care expenditure (GHCE) and for private health care expenditure (PHCE), in order to determine whether there are differences in elasticity according to the government or private character of the expenditure.

Finally, this empirical research requires theoretical support. In this regard, it is interesting to develop some theoretical model that may help us to explain the different behaviour of government and private health expenditure. Thus, we present a simple demand model where various political measures can replicate the empirical results. This model is no more than a first approach and, consequently, its function is simply illustrative.

The rest of the paper is organised as follows. In Section 2 we discuss the methods employed to analyse the existence of a cointegration relationship between three health care expenditures (HCE, GHCE and PHCE) and the GDP. Section 3 presents evidence in favour of the presence of non-linearities in the relationship. When this presence is captured by some dummy variables, we can provide a significant body of evidence against the non-cointegration null hypothesis, confirming that the starting hypothesis holds. However, we also show that the inclusion of these dummies does not imply an increase in the evidence against an income elasticity greater than 1. In order to explain why this phenomenon may occur, in Section 4 we develop an economic model where we discuss how a particular combination of parameters may imply that the income elasticity is indeed greater than 1. Finally, Section 5 closes the paper with a review of the most important conclusions.