The determinants of Canadian provincial health expenditures: evidence from a dynamic panel

Abstract

This article seeks to reveal the magnitude of the income elasticity of health expenditure and the impact of non income determinants of health expenditure across Canada. For this purpose, panel data on gross domestic product, the relative price of health care, the share of publicly funded health expenditure, the share of senior population and the life expectancy at birth have been used to investigate the determinants of Canadian provincial health expenditures over a 28 year period. Dynamic models of health expenditure are analysed via Generalized Instrumental Variables (GIV) and Generalized Method of Moments (GMM). Results indicate that the long run income elasticity of health expenditure is substantially lower than one. Thus, health care is far from being a luxury in Canada.

Keywords:

• health expenditure
• income elasticity
• dynamic panel
• GIV
• GMM

JEL Classification:

• C23
• I18

Notes

¹ The income coefficient due to the exclusion of the health price variable may be biased in either direction. See Bac and Le Pen (2002) and Okunade and Karakus (2001) for example.

² Di Matteo (2000) provides an excellent discussion on the public–private mix of Canadian health expenditures.

³ Macro- and micro-outcomes may differ due to insurance or pooling of resources in which case micro-constraints may exist but macro-constraints may not. These micro- and macro-disparities may render different or inconsistent outcomes.

⁴ According to Kyriopoulos and Souliotis (2002), if the income elasticity of health expenditure is less than one, then the public health sector does not have a high priority among the goals for social and economic development.

⁵ Consumers never face prices for the health services and therefore this variable may be completely irrelevant for the analysis. Second, the price of health care is heavily subsidized in Canada so that even if its effect is not zero, it should be almost zero or negligible.

⁶ If T, G and P denote the real total, public and private health expenditures, respectively, and since , the share of publicly funded health expenditures is . Then, , . Thus, where and . The effect of a marginal increase in on private health expenditures is . The sign of this effect is negative as long as G is positive. See López-Casasnovas and Sáez (2007), Hitiris (1999) and Roberts (1999) for empirical evidence on the magnitude of this effect.

⁷ Kyriopoulos and Souliotis (2002) states that there is no correlation between HE and health status in the Organization for Economic Co-operation and Development (OECD) countries. On the other hand, Maxwell (1981) shows evidence of correlation between the total health spending as a percentage of GDP and infant mortality rate.

⁸ At the data collection level, Disability Adjusted Life Expectancy as a closer proxy for health status than life expectancy at birth is considered. However, it is excluded due to its short time span.

⁹ The endogeneity of income is not considered here and it is argued that causality runs from income to health spending and not the other way. See Ariste and Carr (2003).

¹⁰ See Arellano (2003) and Arellano and Bover (1995) for a technical discussion on this transformation.

¹¹ The F-test has been performed to test the joint significance of the individual fixed effects under the null hypothesis, H_o : μ ₁ = μ ₂ = ··· = μ ₁₀ = 0. The F-test results are 41.81, 22.85 and 131.70 for total, government and private HE, respectively, resulting in favour of rejecting the null hypothesis.

¹² According to Di Matteo and Di Matteo (1998), the impact of the log of the share of the population over the age of 65 on the log of government health expenditures is 0.81 whereas Ariste and Carr (2003) found no evidence on its statistical significance.

¹³ It proved impossible to properly instrument the life expectancy at birth by its lags in the Total HE model.

¹⁴ CANSIM table numbers are 384-0003, 102-0025, 326-0002, 051-0001, 384-0001. http://www.cansim.ca.