The Impact of Suburbanization on Health Outcomes

We examine the impact of the built environment, with particular emphasis on the degree of suburbanization, on health outcomes.  The population in central cities in the United States has declined dramatically since the 1950s.  Despite the decline in the city center, there has been growth in metropolitan areas as a whole, suggesting that individuals are still living in metropolitan areas but are moving into the suburbs. Baum-Snow (2007) examined this phenomenon and found that approximately one third of the decrease in aggregate central city population relative to the metropolitan area population overall can be explained by the expansion of the interstate highway system.  At the same time as the increase in suburbanization, health outcomes in the U.S. have changed dramatically.  Between 1960 and 2006, the obesity rate increased from 13% to 34% (NCHS, 2008).  Rates of physical activity have decreased substantially at the same time. These trends in physical activity and obesity can cause health-related illnesses, such as heart disease and stroke.

First, we expand upon the Grossman (1972) model of health investment by adding an exogenously determined surrounding environment variable which affects the health investment decisions of individuals.  Individuals cannot directly affect the type of environment they live in, such as how many highways are present and the amount of walking that is necessary, once they have chosen to live in a given neighborhood. Our empirical work focuses on the impact of suburbanization on health outcomes, where we proxy for the degree of suburbanization using the population density of a county. Counties that are more suburbanized will have more urban sprawl, which will cause the county to have a lower population density.   We instrument for population density using a measure of the number of planned highway miles from the 1947 Eisenhower Highway Plan.   We believe this is a valid instrument because it is unlikely that the plan proposed in 1947 is correlated with health outcomes decades later but is correlated with the population density in a given county, we have a valid instrument for our analysis. Data at the individual level were obtained from the Behavioral Risk Factor Surveillance System (BRFSS).  The health outcomes and behaviors we study include obesity, physical activity, heart disease and stroke. We obtained data on planned and completed highways from Baum-Snow (2007). This data set contains the mileage of highways that were completed in each country in each year.  In addition, this data set has information on the original planned highways from the 1947 Highway Plan. Our preliminary instrumental variables results indicate that there is a negative relationship between obesity and population density.