Estimation of Subnational Fertility, Mortality, and Migration Across Social and Geographic Di-mensions: A Bayesian Approach Using Survey and Administrative Data

Abstract

Accurate measurement of subnational levels, trends and disparities in fertility, mortality, and migra-tion is essential for informing equitable and targeted policy actions and improving accuracy of sta-tistical demographic estimates. However, demographic estimates in large, diverse countries such as India and the United States are frequently produced only at national or first administrative levels, obscuring important underlying heterogeneity. This dissertation develops high-resolution, intersec-tional estimates of these three demographic components using Bayesian hierarchical statistical frameworks that borrow strength across time, geography, and population subgroups. The first chapter estimates age-specific and total fertility rates in India between 1990 and 2020 by state or Union Territory, education, and urban–rural residence using National Family Health Survey data, showing that fertility decline slowed after the mid-2000s among less-educated women even as it continued among the most educated, with roughly one-third of the aggregate decline attributable to compositional shifts in education. The second study estimates life expectancy at age 25 in the Unit-ed States from 2000 to 2022 jointly by state, race-ethnicity, and educational attainment, adjusting for misclassification on death certificates, and finds that educational gradients widened across near-ly every racial-ethnic group and state while racial-ethnic convergence was confined largely to nar-rowing Black–White gaps, with disparities involving Hispanic, American Indian or Alaska Native, and Asian or Pacific Islander populations remaining stable or highly variable geographically. The third study produces annual interstate in- and out-migration rate estimates by age, sex, and race-ethnicity from 2000 to 2022, revealing that migration systems are racially and ethnically stratified, with Non–Hispanic Asian or Pacific Islander populations emerging as the most mobile group by 2020, Non–Hispanic Black migration increasingly concentrated along Southern and Southwestern corridors, and Hispanic populations exhibiting the lowest overall mobility de-spite growing concentration in Sun Belt destinations. Together, these three studies demonstrate that national and single-dimension summaries mask substantial and policy-relevant heterogeneity in demographic processes, and they establish generalizable statistical methods for producing reliable, intersectional demographic estimates in data-sparse settings.