Housing Market Spillovers in a System of Cities

It is well-documented that housing leads the business cycle at the national level. Dating back as far as the great depression, nearly every recession or boom has been preceded by a respective drop or rise in residential investment (Green, 1997; Leamer, 2007). However, neither the business cycle nor the housing cycle is truly national in their nature. Historically, the severity and timing of both vary greatly across cities. This has been acknowledged in the literature (Ghent and Owyang, 2010), but the possibility that different cities housing markets and economies could be interacting and spilling over amongst one another has not yet been considered. I use a Global Vector Autoregression to model building permits and employment at the metro level for 78 cities in the U.S from 1990-2015, then link these models together in a system of cities. The model reveals several results of interest. First, the status of residential investment as a leading indicator for the business cycle is questionable at the metro level, and is found to be true in only 47 of 78 cities. Second, shocks to housing in cities do spill over into other cities. In fact, in many cases the responses to shocks are larger in other cities than in those where the shocks originated. Thus, the national relationship between housing and the economy appears to be created by a collection of cities housing markets spilling over into other cities. Third, the largest spillovers are often not found in adjacent cities to those where the shock occurred. This indicates that factors other than physical distance, particularly trade, are the channels shocks are transmitted through. Overall, the results suggest a role for place-based policy, where stabilizing new housing development in particular cities could result in some stabilization of the national business cycle