A Search for the Underlying Structure Driving House Prices in a Distressed Environment

This is the third of three papers prepared for the Lincoln Institute of Land Policy about the housing price bubbles and busts in the US. The first two by Follain and Giertz (2011, 2012) use lengthy time-series data for over 300 MSAs to build models with the potential to predict what actually occurred in the last several years. This paper focuses upon the period 2005–2010, which includes the last years of the boom and the several years of its aftermath. County level data are used instead of MSA level, which offers the opportunity to take a more geographically granular look at the crisis. Annual data for 416 counties are available all six years; data for 439 counties are available for 2006 through 2010. Counties from 47 states and the District of Columbia are included. These data draw upon the American Community Survey, information provided by Collateral Analytics, Moody’s Analytics, and other sources and offer the opportunity to study a wide range of potential drivers of house prices during the crisis.

A central goal of this paper is to shed light on the key drivers of house prices during a period of substantial turmoil and distress. Indeed, a number of substantial insights are produced. First among these is the importance of incorporating a measure of the distressed real estate inventory; a larger inventory of such real estate reduces the level of house prices and slows down the growth in house prices. The model also confirms the lingering impacts of excessive price appreciation and substantial amounts of subprime mortgage lending, during the boom and prior to the bust The estimated models also confirm the importance of a number of the traditional drivers of house prices like income and rent, though the empirical magnitudes of these estimates are sensitive to a variety of conditions. The analysis also confirms the presence of large and widely varying county fixed effects, which highlights the importance of local and difficult to quantify market conditions as a driver of house prices. Both the in-sample and out of sample predictions of the model for 2011 suggest that a full recovery is well into the future and also indicate a substantial number of outlier counties that have large and unexplained residuals both positive and negative.