Housing Costs and Commuting Distance
Author: Kevin A. Park
Publisher:
Published: 2016
Total Pages:
ISBN-13:
DOWNLOAD EBOOKHouseholds face a tradeoff between housing costs and commuting costs. Using a database that connects residence and workplace neighborhoods in eight larger metropolitan areas, we model the difference in housing costs as a function of estimated commuting distance. Based on the linear distance between residence and workplace neighborhoods, average commuting distances are roughly 10 to 15 miles. Dissimilarity indices show that roughly half of workers would need to relocate to a different neighborhood to balance jobs and workers across regions. Higher income jobs and workers are less likely to be balanced than lower income jobs and workers. Consequently, higher income workers typically have longer commutes.In only three metropolitan areas are house values found to decline with distance from the Central Business District as predicted by the monocentric model widely used in urban economics.However, only 37 percent or workers commute in the general direction of the Central Business District, undermining a basic assumption of the monocentric city model. Instead there is substantial reverse and cross-commuting in metropolitan areas. Unlike estimates derived from a monocentric model, models based on actual commuting behavior yield consistently negative estimates for the effect of commuting distance. Specifically, house values decline between 0.05 and 0.65 percent for every additional commuting mile and gross rents decline between 0.07 and 0.26 percent. In dollar terms, these gradients correspond to an average decline in house value of $792 per mile and average decline in annual gross rent of $22 per mile in 2011. Consistent with economic theory that higher income workers have a higher shadow value of time that increases commuting costs, these workers were found to have steeper value and rent gradients. Meanwhile, public transportation involves a fixed cost but lowers the marginal effect of commuting distance.