Recent Publications

Papers Submitted

1. A. Hamoudi, J.B. Dowd, Housing Prices and Stress-related Health Outcomes (June, 2011)
   (last updated on 2011/12/16)

   Abstract:
   The period from the early 1990s until the middle of this decade saw dramatic increases in property values across the United States, but the rate of appreciation varied widely across geographic locations--in part because communities followed different development trajectories than even their nearby neighbors. We exploit this geographic heterogeneity in order to identify the health effects of these trajectories on homeowners in late middle age and beyond. We draw on two types of datasets--first, detailed longitudinal information on individuals and households available from the Health and Retirement Survey, and second, real estate valuations computed at the ZIP code level by private sector investment advisor firms. Using geocoded information on the residential locations of survey respondents, we are able to combine these data in order to identify those homeowners who lived in communities that experienced more rapid run-ups in housing values from those who lived in communities that fared less well. The identifying assumption in our analysis is that within a geographic cluster of communities, once we control flexibly for initial housing value any subsequent appreciation should be "random" in the sense that it should be independent of any observed or unobserved baseline characteristics. The motivation underlying this assumption is that if there were any baseline characteristics that could predict subsequent appreciation, real estate investors would have incentive to use this information to buy property where prospects were better, thereby bidding up prices in these communities. In this way, baseline valuations should fully summarize all the available information about the future prospects of any community. Our analysis begins with an exploration of the evidence on the plausibility of this assumption. We find that, after controlling flexibly for initial housing values, none of over two dozen individual characteristics could predict subsequent appreciation. The characteristics examined include race, sex, educational attainment, health related behaviors like smoking and exercise, emotional health, previous chronic illness, and performance on cognitive tests. We interpret these results to indicate that our identifying assumption is plausible on empirical grounds, and therefore proceed to the main analyses. In these, we regress individuals' health status in 2006 against the run-up in housing values over the previous decade or more. (Most outcomes were only measured in later years, so we generally are precluded from using longitudinal change in health status in our analyses, but instead are forced to use health status at the end of the period). The outcomes we analyze include stress-related indicators like blood pressure and incident cardiovascular disease, as well as weight-for-height and waist circumference, measures of emotional and psychological well-being, expectations for the future, self-reported generalized health status, performance on cognitive tests, health-seeking behavior, and specific disease conditions. Subject to our identifying assumption, the observed relationships can be interpreted as "causal" in the sense that nearby communities experiencing slower run-ups in housing prices represent a valid counterfactual comparison group for communities that experienced more rapid run-ups. In this sense, we interpret the regression results to indicate a reduced-form effect of both the increased wealth brought about by increases in the asset value of individuals' homes, as well as any potential effects of changes in community characteristics that may have determined the trajectory of appreciation in a community's housing values. Preliminary results indicate that these factors had measurable effects on an array of health outcomes, including anthropometric outcomes, indicators of psychological distress, and activities of daily living. However, they do not appear to be related to mortality risk or the incidence of major health events like heart attack or stroke.