UC Berkeley

In the three chapters of this dissertation, I examine the effects of changes in income on health measures for three different populations. Each of the three populations live in countries with developing economies: Whites in the post-bellum U.S.; black ex- slaves and freedmen upon Emancipation in the U.S., and individuals living in Indian households at the end of the 20th century. In the first and second chapters, I look at the effect of unearned income on morbidity and mortality, while in the third, I investigate how spatial and temporal differences in energy requirements affected food consumption at the end of the 20th century in India. Taken together, the three chapters focus on the effects of income on individual health measures.

In the first essay, I investigate how increases in individual income contributed to improvements in adult health during the late 19th and early 20th century. To disentangle the effect of income as opposed to medical advancements or public health interventions, I use exogenous variation in income from the first wide-scale entitlement program in the United States: the Union Army pensions. Documenting that Republican Congressional candidates boosted veterans pensions in order to secure votes, I exploit exogenous increases in income stemming from Republican corruption to estimate income effects on morbidity and mortality. The effects of income on disease onset are large - an extra $1 of monthly pension income, a 9% average real income increase, lowered the probability of infectious disease onset by 38%. In addition, I find that an extra $1 of monthly income lowers the crude death rate by .008. I find the largest income effects for infectious illnesses, smaller effects for respiratory and digestive illnesses, and no effect for the onset of most endocrine diseases. Results from this chapter help shape our understanding of the U.S. mortality transition and inform todays debates on the health benefits of cash transfers to adults in regions with wide SES gradients in health, as was the case in the U.S. a century ago.

In the second chapter co-authored with Trevon Logan, we investigate the effect of increases in Union Army pension income on the mortality rates of black veterans who served for the side of the North in the Civil War. It is not possible to use the same source of exogenous variation in pension income as in the first chapter since the 2

majority of black veterans live in the South and thus are unable to vote. Therefore, they are not a constituency of interest to Republican Congressional candidates. To circumvent endogeneity bias, we use propose a new instrument for pension income and an econometric framework to determine black-white differences in the effect of pension income on health. In addition, we investigate the determinants of differential pension rates among black veterans and present new results.

In the third chapter co-authored with Nicholas Li, we explore the effect of declining energy requirements on the demand for calories in India. Despite the large growth in real expenditure and positive calorie-expenditure elasticities, calories consumed per person in India declined between 1983 and 2005. Similarly, rural households are found to be poorer than urban households but consume more calories on average. We test the energy requirements hypothesis of Deaton and Dreze (2009) as an explanation for these missing calories by using time-use data to impute household energy requirements. To show the differential effects of energy requirements and other factors on food quality and expenditures, we use a simple model to provide intuition and motivate our empirical approach. We find that variations in household characteristics such as size, age, education and occupation predict highly correlated fluctuations in caloric intake and requirements. Labor-saving durables also play an important role in caloric intake. Quantitatively, energy requirements can explain most of the missing calories between urban and rural areas. Over time, differing energy requirements explain approximately one-half of the changes in food quality but only a modest share of the total missing calories, which implies that other factors are important. Our results shed light on the importance of considering variations in energy requirements when formulating welfare and poverty measures.