Three Essays on Measurement and Evaluation of Mortality

This dissertation is composed of three essays studying the measurement and evaluation of mortality.;In the first chapter, written in collaboration with Lowell Taylor, we investigate an important methodological issue that arises in studies that use panel data to examine the forces that shape racial disparity in mortality. An important example is the National Longitudinal Survey of Older Men (NLS-OM), which collected data for men aged 45--59 in 1966 and several subsequent years, and then also reported deaths as indicated by death certificate data collected in 1990. An important problem with longitudinal data such as the NLS-OM is a particular form of measurement error---missing data on death. In the NLS-OM, for example, the matching procedure appears to have missed a substantial number of deaths, most likely in systematic ways. We then demonstrate that improper handling of the measurement error in death records would cause serious inference problems in racial mortality estimation.;In the second chapter, written in collaboration with Dan Black, Seth Sanders, and Lowell Taylor, we establish a generalized method of moments (GMM) estimator for studies that have problems in combining data from separate sources for the purpose of estimation. Using the GMM approach, we conduct estimation for mortality using data from the U.S. Census and from Vital Statistics. Specifically, the GMM approach allows us to credibly estimate mortality rates in mid-life (ages 40 to 60) by race, gender, birth cohort, and State of birth for cohorts born 1930--1939. Our estimates document very large variation in State-of-birth effects for men, especially black men, though not for women. Moreover, since we are studying cohorts born in the 1930s, who have experienced the Great Depression during childhood, we are able to see if mid-life mortality measured by State of birth is correlated with State-level average household conditions for these individuals when they were children. We use the data from the 1940 U.S. Census to examine State-level characteristics of households that had children born 1930-1939. Our results show that for both black and white men, mortality in mid-life is negatively correlated with household income in childhood at the level of State of birth, suggesting that mortality is generally higher for individuals born in relatively low-income States.;In the third chapter, written in collaboration with Dan Black and Lowell Taylor, we examine the effect of schooling on older-age mortality. We study the extent to which the relationship between early-life education and later-life mortality is causal. There is a growing literature which highlights the effect of education on mortality. These studies typically link cross-State policy variation in the U.S. to the early-life human capital accumulation of individuals born in the early twentieth century. Recent research seeks to identify the causal impact of schooling on subsequent mortality by exploiting variation in resource allocation to public schools and in compulsory school attendance laws as a means of finding exogenous variation in schooling. For our application, we investigate the extent to which these forces thereby affect mortality at older ages. We first use the GMM approach developed in the second chapter to form mortality estimates by birth cohort, birth State, sex, and race using both U.S. Census and Vital Statistics data. The GMM estimates of mortality form the basis for the analysis that follows. We then take two empirical exercises. First, using measures of compulsory schooling laws as instruments, we estimate the impact of completed education on mortality at older ages for white individuals. We find that increased schooling leads to modest declines in mortality for both white men and women. Overall, the estimated causal effect is larger for women than for men. Our second exercise examines if variation in the allocation of resources used to educate black and white students plausibly contributes to the racial mortality gap at older ages. Using the variation in the quality of educational opportunities provided to black and white children across the Southern United States in the early twentieth century, we estimate the effect of relative school quality on the gap in mortality between blacks and whites. Our results show that the black-white gap in mortality at older ages is relatively larger for individuals born in States that had larger black-white quality gaps in school quality. Also, the estimated effect seems to be larger for women than for men.;(Abstract shortened by UMI.).