Academic journal article Contemporary Economic Policy

Health Investment and Economic Output in Regional China

Academic journal article Contemporary Economic Policy

Health Investment and Economic Output in Regional China

Article excerpt

Using a 29-year (1978-2006) panel of provincial-level data from China, this article investigates the role of health capital in a human capital model of economic output. Robust evidence is found through panel cointegration analysis that health capital has a significant and positive effect on the Gross Regional Product in China; the effect being stronger in the inland regions compared to the coastal areas based on estimates that account for regional heterogeneity. This article highlights and discusses the potential role of diminishing returns to health investment in this globally important area. (JEL 115, R11, C23)

I. INTRODUCTION

The relationship between health capital and economic output has been the subject of a great deal of theoretical and empirical investigation (Bloom, Canning, and Sevilla 2004; Bloom et al. 2010; Knowles and Owen 1995; to name but a few). While the positive effects of health on income and economic output are well established, the evidence is limited as to whether or how this effect varies with the level of development of a country or region. Conclusively estimating the dynamics of the health--growth relationship via the economic history of just one country would require a long time series to allow for enough variability in levels of health capital and of aggregate economic activity. An alternative is to undertake a cross-country panel study. The diversity of levels of gross domestic product (GDP) and health investment, and of their trajectories across time, creates the possibility of discerning the impact of health capital on growth at various levels of investment and of income. However, there are a number of criticisms of cross-country studies, not least that because of the wide degree of heterogeneity across countries, unified models are potentially implausible (Maddala 1999).

The Chinese experience offers something close to a natural experiment that sheds light on this important question in a way that avoids some of the obvious pitfalls of cross-country studies. Specifically, China's recent history has seen a great deal of geographical diversity in the levels of investment in health, and in rates of economic growth, as a result of largely exogenous policy decisions of a central government. The geographical diversity can be used to model the variations in how health investment affects economic activity. Therefore, regional economic activity in China is the focus of this article. A panel data set of gross regional product (GRP) for all the provinces of China over a 29 year period was constructed. This is used to estimate the effect of health investment on growth in this output measure. By comparing model estimates for the different major regions of China (specifically, the coastal provinces and the inland provinces), insight is provided into how health investment can affect growth differently at different stages of development, as these two regions have developed very differently over the sample period.

Li and Huang's (2009) study provides the most closely related previous work on this topic. They use annual data from 28 provinces and an augmented Mankiw--Romer--Weil (MRW) framework, adopt panel data techniques (fixed-effects [FE] and random-effects [RE] regression analysis), and find that health capital (proxied by the number of doctors per 10,000 people or the number of hospital beds per 10,000 people) has a positive effect on GRP per capita. Their sensitivity analyses found that the health effect is stronger in coastal China compared to inland China. Given that the average number of doctors per 10,000 people was 15.29 in the coastal region and 13.92 in the inland region during the periods from 1978 to 2006,1 this finding from Li and Huang tends to suggest increasing returns to health capital. However, for these long panel datasets, whether variables are stationary or not needs careful examination so as to avoid spurious regression problems. Besides, even if the series are cointegrated, unlike in standard time-series analysis, the FE estimator is inconsistent in the panel data setting if the series are nonstationary (Baltagi 2005). …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.