The long-term health and economic consequences of the 1959–1961 famine in China

Abstract

This paper, using a difference-in-differences method, tries to quantify the long-term effects of China's 1959–1961 famine on the health and economic status of the survivors. We find that the great famine caused serious health and economic consequences for the survivors, especially for those in early childhood during the famine. Our estimates show that on average, in the absence of the famine, individuals of the 1959 birth cohort would have otherwise grown 3.03 cm taller in adulthood. The famine also greatly impacted the labor supply and earnings of the survivors with famine exposure during their early childhood.

Introduction

China's 1959–1961 famine stands out as the worst in human history: there were about 15–30 million excess deaths and about 30 million lost or postponed births.¹ The startling severity of this famine of modern times has provoked persistent efforts in understanding this unprecedented catastrophe and quantifying its devastating effects. Many previous studies have explored in great detail the causes of the famine and its impact on mortality and childbearing behavior (Ashton et al., 1984, Peng, 1987, Lin, 1990, Lin and Yang, 2000, An et al., 2001). Nevertheless, very few studies have touched on the long-term health and economic consequences for the survivors of the great famine, and little is known about the magnitude of these consequences.

The 1959–1961 famine created a large undernourished population during the three-year period. Malnutrition and exposure to famine-related diseases adversely affected the health of the survivors. In particular, it may have had adverse consequences for children who were conceived or born during the famine. There is a large body of evidence in population health literature showing that exposure to malnutrition or other adverse environments in the fetal period and early childhood exerts significant lasting effects on health, well-being and competence (Barker, 1989, Barker, 1992, Hertzman and Wiens, 1996, Hertzman and Power, 2003, Heymann et al., 2005). Economists have also increasingly recognized the vital role of health status as a dimension of human capital in economic development.² In particular, recent research provides strong evidence on the long-term effects of childhood health and economic circumstances on adult health, employment and socioeconomic status (SES) (Case et al., 2005). Combining these findings from different disciplines suggests that negative health consequences in early life may translate into negative economic outcomes in adulthood. From this perspective, it is both interesting and important to understand to what extent the 1959–1961 famine exerted sustained health and economic effects on a large surviving population, mostly in rural areas.

A rigorous and systematic analysis of the long-term impact of the famine, however, is hindered by a number of problems. The first is limitations of data, such as the lack of data tracking individual characteristics of the population during and after the famine. Collection of data during the famine would have been virtually impossible. A more realistic source of data comes from surveys sampling the population exposed to, and surviving the famine. But even if we have such a dataset, identifying the effects directly caused by the famine is still very challenging. An analysis based upon a sample of the surviving population may potentially suffer from a sample selection problem if the childbearing decisions during the famine respond to the famine and systematically correlate with the heterogeneity across individuals, such as socioeconomic status. If fertility was more responsive among those families with higher SES, then more children may have been born into poor or less healthy families. This selection effect will cause a spurious correlation between exposure to famine and its effect on health and economic outcomes. Leaving aside this sample selection bias, we also encounter the problem of how to build a causal relationship between famine and its long-term effects when many other unobserved characteristics may confound our inference.

Existing literature on the sustained effects of famine has not satisfactorily addressed the issues of sample selection and causal inference. One of the main approaches adopted in current literature to estimate the effects of famine, relies on variations in exposure to famine across cohorts.³ A key idea behind this approach is that individuals born after a famine will not be affected by it, while those who are born during the famine may be affected differently from those conceived during the famine. Therefore, this source of variation across birth cohorts can be used to identify the effects of famine. But this approach may suffer from the possibility of confounding the effects of famine with intrinsic cohort effects. In other words, the observed difference in demographic effects (e.g., mortality rate) across birth cohorts may be a reflection of general cohort effects even without exposure to the famine.⁴ Unless we can control the same set of cohorts in some regions not exposed to famine, it is hard to attribute these observed effects directly to famine.

In this paper, we are interested in identifying the long-term causal effects of the 1959–1961 famine on the health and economic outcomes of survivors. In particular, using a cross section sample of Chinese rural people born between 1954 and 1967, we try to estimate precisely to what extent the attained height of rural cohorts in adulthood was reduced by the exposure to famine in their early childhood, and how the resulting adverse health consequences impaired their labor supply behavior and family income. To our knowledge it is the first attempt to present the estimates of long-term effects of the 1959–1961 famine in the literature.

We find strong evidence that the great famine in 1959–1961 caused serious health and economic consequences for the survivors, especially for those who were in early periods of life during the famine. Our estimates show that on average, in the absence of the famine, individuals born in 1959 would have otherwise grown 3.03 cm taller. Quantile regressions present some evidence that the height reduction appears to be larger for the individuals with shorter heights. The famine also greatly impacted the labor supply and earnings of the survivors with famine exposure during their early childhood.

China's 1959–1961 famine provides a unique case for studying the long-term effects of famine in general. The uniqueness of this famine not only lies in its long time span,⁵ unprecedented severity and scope of incidence, but also in its substantial variation across regions. The effects of famine were much more devastating in rural areas than in urban areas. Exposure to famine also varied greatly across provinces due to the variance in population density, exposure to bad weather and provincial response to food shortage, among many other contributing factors. Regional disparity in famine severity, combined with variations in health consequences across cohorts commonly exploited in the literature, provides us with a crucial source of exogenous variation to identify the causal effects of famine. A major innovation of this paper is to combine the variations of famine effects across regions and cohorts to construct a difference-in-difference (DID) estimator.⁶ More specifically, we will introduce a key variable—the interaction terms between dummy variables indicating the birth cohorts and excess death rates of 1960 in the region where the cohorts were born to capture the causal effects of the famine on this particular cohort. Here, excess death rate of a region is a proxy for the severity of the famine.

China's institutional context during the famine period greatly alleviates a host of potential concerns surrounding the legitimacy of our DID estimation. As discussed above, sample selection problems constitute a serious challenge to any attempt to identify the causality between famine and its outcomes. However, China's rural situation in 1959–1961 considerably mitigates the potential selection into childbearing by social-economic status of parents. Following a series of political movements commencing in 1949, such as Land Reform, Agricultural Collectivization and the Great Leap Forward movement, there was no private ownership of land and other production tools and vehicles, and the virtual equalization of SES within and across communes occurred. In addition, China's residence registration system, which aimed at restricting migration and relocation, helps establish a very high correlation, especially in rural areas, between the region of birth and growing-up and the region of residence at the time of survey (1991). The strict prohibition of migration from rural areas during the famine period also limits the potential selection through migration.⁷

This paper contributes to our better understanding of the long-term effects of the worst famine in history. Evaluating the long-term effects of famine in the long run has traditionally been the sphere of medical researchers, demographers and historians and has received scant attention in economics literature. In this study, we try to conduct a serious econometric analysis and obtain more reliable estimates of the causal effects of famine. In many ways, our study also addresses the broader questions of how events in early life affect long-term health and economic outcomes and of how adverse health effects in early life affect labor supply and income in later periods. The research conducted in this paper closely relates to a burgeoning literature on the linkage between health, labor market outcomes and economic development (Strauss and Thomas, 1998). Fogel, 1992, Fogel, 1994 has convincingly established the relationship between aggregate movements in adult height and long-term changes in standard of living. Some other studies have devoted efforts to identifying the effects of health on wages and labor supply at an individual level (Strauss, 1986, Schultz and Tansel, 1997; Thomas and Strauss, 1997). In this paper, we offer indirect and supportive evidence on the linkage between health and labor market outcomes activated through the effects of famine.

The remainder of this paper is organized as follows. In Section 2 we briefly introduce the 1959–1961 Chinese famine. In Section 3 we describe the data. Section 4 discusses the identification strategy used to evaluate the effect of the famine on height and economic outcomes. Section 5 presents regression results on the sustained effect of the famine on the height, labor supply and income of survivors. Section 6 discusses the validity of the DID estimator and the fitness of OLS. Section 7 briefly concludes this research.