Recent research on U.S. income mobility and health using community and individual data shows that higher mobility is associated with lower mortality risks and better. That relationship seems to be stronger and more consistent than the relationship between income inequality and health (a topic widely studied), although considerably smaller than the impact of income on mortality. This preliminary evidence suggests that income mobility might be a relevant determinant of health and mortality. Surprisingly, this potential pathway has received little attention in the literature.
This dissertation builds on this small literature and examines the robustness of the relationship between income mobility and health using empirical data and formal modeling. Using different data sources and modeling approaches, I look at the magnitude and variability of this association in the US and explore the plausibility and consistency of explanations offered in the literature. The central argument is that the effect of income mobility on health is stronger and larger than the impact of income inequality and that the mechanisms behind it, although related to income inequality, are theoretically distinct and independent of those of income and inequality, and can have powerful and lasting consequences.
To ground this argument, I use three strategies. First, I analyze aggregate data to assess the magnitude, robustness, and variability of the association of income mobility with mortality. Second, I extend those analyses using individual and longitudinal data to define clearly exposure to an income mobility regime and examine whether some of the potential pathways and mechanisms proposed in the literature are supported by the data. Finally, building on this evidence, I create an agent-based model to assess the conditions and plausibility of the potential mechanisms involved in the association between income mobility and health. This forces me to represent precisely a set of mechanisms likely to bring about the observed patterns. These virtual representations, thus, help me explore the implications of the theory and ask what-if questions, in addition to providing a general framework to assess previous research and help design new studies. The goal is to go beyond statistical models (i.e., the detection of average differences that may reflect the aggregate statistical signature of unspecified underlying mechanisms) and define with plausible (precise) mechanisms, and testable hypotheses.