Spatial Health: How Health Behaviors Shape Our Neighborhoods

Neighborhoods profoundly influence children’s life outcomes (Chetty et al. 2016, 2018). While existing research emphasizes educational spillovers and their role in residential sorting (Benabou 1996; Durlauf 1996; Fernandez and Rogerson 1996, 1998), neighborhoods also exhibit substantial disparities in health outcomes. Spatial variation in mortality rates is argued to be driven primarily by health behaviors such as smoking rather than healthcare access conditional on income (Chetty et al. 2016). Peer effects in smoking and related behaviors are well documented (Argys and Rees,2008; Lundborg, 2006; Card and Giuliano, 2013; Bor et al., 2024). These findings suggest that local health externalities may reinforce educational spillovers in shaping residential choices.
We develop a general equilibrium life-cycle model incorporating both education and health behavior spillovers to evaluate neighborhood-based interventions. Building on recent work evaluating rental vouchers and place-based policies (Eckert and Kleineberg, 2021; Chyn and Daruich, 2022; Zheng and Graham, 2022;Fogli et al., 2023), our framework extends these by incorporating health behavior channels. Parents choose neighborhoods, recognizing that local environments shape their children’s decisions about attending college and smoking.
In particular, the model features two local externalities: living in more affluent neighborhoods increases children’s probability of attending college, while living in areas with higher smoking rates increases their likelihood of smoking. These mechanisms interact through complementarities between education and health. College-educated individuals earn higher incomes, making longevity more valuable and creating stronger incentives to avoid smoking. Consequently, wealthy parents tend to prefer neighborhoods with higher income levels and lower smoking rates, thereby generating reinforcing sorting patterns.
The model follows individuals through three life phases: adolescence (when college and smoking decisions are made), prime age (divided into a consumption-smoothing stage and a stage when children are born and parents choose neighborhoods), and retirement. Parents are altruistic, following Barro and Becker (1988). Neighborhoods are characterized by average income and smoking rates, which determine local spillover effects.
We calibrate the model using the Add Health survey data from the US, tracking 9,034 parent-child pairs, matching conditional probabilities of smoking and college attendance across neighborhoods. The calibrated model successfully replicates observed patterns of health and income segregation. Counterfactual analysis of rental vouchers targeted at low-income households shows that enabling moves to more affluent neighborhoods increases aggregate college attendance and decreases smoking rates. Welfare gains from improved health and longevity are comparable to those from higher earnings resulting from increased college enrollment.