The Effect of Education on Obesity: Evidence from College Attendance during the Vietnam War

A large number of studies have found that educational attainment is strongly and positively correlated with health.  This paper estimates the causal effect of education on one particular measure of health – obesity - which is of considerable interest because it more than doubled in the U.S. since 1980 and it raises the risk of heart disease and diabetes, which substantially increases medical care costs.  We examine data on over 400,000 individuals in the National Health Interview Surveys of 1982-1996 who were of college age during the Vietnam War.  We estimate the effect of education on obesity using the method of instrumental variables.  The instrument for education is the risk of being inducted into the U.S. military for each birth-year cohort and gender.  (As a robustness check, we instead use the risk of being killed in action in Vietnam, and find similar results.)  Men’s school enrollment was elastic to these risks; the instrument far exceeds the minimum standards of power in the first stage (F>10).

Before any correction for the endogeneity of education, the basic pattern in the data is that body mass index and the probability of obesity decrease with years of college education.  However, when we instrument for education, each additional year of college is associated with a 3 percentage point higher probability of obesity. The same IV approach and data indicate that additional years of college education decrease the probability of smoking, so our unexpected finding for obesity is not an artifact of an unusual dataset or instrument for education.  The unexpected result for obesity may instead be due to college allowing men to enter sedentary occupations; this represents a rare example of additional education decreasing health.  An alternate explanation is that we measure a very specific Local Average Treatment Effect; men who are especially averse to exercise may be particularly likely to enroll in college to avoid military service.