Instrumental Variable Estimation of the Causal Effect of Hunger Early in Life on Health Later in Life

Numerous studies have evaluated the effect of nutrition early in life on health much later in life by comparing individuals born during a famine to others. Nutritional intake is typically unobserved and endogenous, whereas famines arguably provide exogenous variation in the provision of nutrition. However, living through a famine early in life does not necessarily imply a lack of nutrition during that age interval, and vice versa, and in this sense the observed difference in outcomes between famine and non-famine exposed individuals at most provides a qualitative assessment of the average causal effect of a nutritional shortage, which is the parameter of interest. Most likely, the causal effect of hunger is under-estimated in absolute value by such a comparison of contextual famine effects. This paper contributes to the existing literature by being the first to combine the occurrence of famines with information on individual hunger periods, to estimate of the average causal effect of nutritional shortages during specific childhood ages on health later in life. Specifically, we estimate the average causal effect of hunger during childhood on health outcomes later in life, using instrumental variable estimation with famines as instruments. We can thus address the problems of endogeneity and underestimation of causal effects. Our individual data are from the Survey of Health, Aging, and Retirement in Europe (SHARE), a European longitudinal survey based on a random sample of individuals aged 50+. The most recent wave asks respondents for retrospective information about hunger periods during their lives. Our sample is composed of cohorts that were exposed to the famines in the Netherlands, Germany, or Greece, in various time intervals in the 1940s. We use parametric and nonparametric two-sample IV estimation to deal with imperfect recollection of conditions at very early stages of life. Our results indicate that among individuals who were 6-16 year old when suffering from hunger undernutrition leads to an increase in obesity risk among females. Besides, males have a strongly elevated risk to suffer from hypertension. If restricted in nutritional supply in utero or until the age of 4, females are almost 42 percentage points more likely to suffer from hypertension as adults. For males, we find a strong negative height effect of more than 3 cm. As a by-product, our study provides estimates of the strength of the association between a famine and an actual hunger episode. The estimated average causal effects often exceed famine effects by a factor three.

van den Berg, Gerard J., Pia Pinger and Johannes Schoch (2012), Instrumental Variable Estimation of the Causal Effect of Hunger Early in Life on Health Later in Life, ZEW Discussion Paper No. 12-019, Mannheim.