Seniors' Health and Residential Sorting

Deciding where to live is a highly consequential decision. Localized amenities - like health care access and quality, environmental amenities such as climate and air pollution, cultural life and potential peers - affect the health, wellbeing, and longevity of the elderly. This study aims to determine the impact of health on seniors’ residential location decisions by combining the ideas of Tiebout (JPE 1956) and Grossman (JPE 1972) in a unified model of health and residential sorting. While considerable knowledge exists about how changes in amenities affect neighborhood demographics through Tiebout sorting (Banzhaf and Walsh, AER 2008) and how demographics and income explain heterogeneity in marginal valuation of amenities (Bayer et al. JPE 2007, JEEM 2009), the role of health in residential location decisions is unknown.

Seniors are an ideal subpopulation to study the effects of health on residential choices. Aged 65 and older, most are retired so not constrained to stay close to job locations. Further, “Health Reasons” is among the top responses that seniors provide in the Health and Retirement Survey as a reason for their most recent move. Panel data from the Center for Medicare and Medicaid Services (CMS) allows me to precisely track the locations of 7.3 million seniors and the presence or absence of over 40 chronic health conditions from 1999 through 2013. To evaluate (1) the relative importance of access to high quality medical care versus private consumption and other neighborhood amenities, and (2) the degree to which health affects these tradeoffs, I estimate a structural model of residential sorting similar to Bayer et al. (JEEM 2009) and Galiani et al. (AER 2015). I innovate on the framework by implementing a flexible CES utility function. This enables me to quantify the complementarity between consumption and amenities, and additionally, trace out relative marginal utilities of health care and other amenities as a function of health. Preliminary results suggest that the importance of health care is u-shaped in health, while the importance of other amenities increases monotonically in illness.

The structural estimates allow me to consistently predict how health shocks affect migration probabilities and destinations. These predictions can be aggregated to counterfactual predictions for migration flows in response to changes in the characteristics of locations, such as improving hospital quality.

While I am currently working to complete the structural model, preliminary reduced form evidence suggests that negative health shocks increase out-of-pocket expenditure substantially and also trigger moves, leading to residential resorting. After experiencing negative health shocks and the accompanying income shock due to increased medical expenditures, low income individuals tend to resort into cheaper and relatively more polluted places, leaving them more vulnerable to subsequent health shocks related to increased pollution exposure (Bishop Ketcham Kuminoff, NBER 2018). Statistical analysis suggests that throughout retirement, low income seniors fare progressively worse in terms of health than higher income seniors. I am working on extending this model to span multiple periods, thereby tracing out health-location trajectories over time, and evaluating the possibility of a pollution-poverty trap for senior citizens.