A Look at US House Price Bubbles from 1980-2010 and the Role of Local Market Conditions (Working Paper)

Author(s): Follain, James R. and Seth H. Giertz
Publication Date: July 2011

69 pages; Inventory ID WP11JF1; English

FREE DOWNLOADS BELOW


A Look at US House Price Bubbles from 1980-2010 and the Role of Local Market Conditions	PDF	2.20 MB

Abstract

We investigate the relationship between housing prices and several explanatory variables for a large panel of U.S. MSAs in order to gain insight into the ex ante likelihood that price bubbles are present. We estimate a vector error correction (VEC) model for a number of different time periods and subsets of MSAs. Estimated coefficients from these models are inputted into a Monte Carlo simulation model in order to project future house prices, as well as the distribution of potential house-price paths. Special attention is paid to both the median projection and the severity of extreme or stressful outcomes. These projected house price (and employment) distributions are generated for each MSA, which allows us to test the role of local market conditions in the formation of price bubbles. The model’s predictions are then compared to actual outcomes – focusing primarily on the house price bubble and bust of the Great Recession.

We find that the estimates of the coefficients of the lagged covariates in the three equation VEC model are very similar for various MSA groupings and time periods. Despite this first finding, our simulation model projects great variation in house price appreciation across MSAs for each time period for each MSA grouping. The variation in projected house price appreciation appears to be driven by differences in (initial) local market conditions, which include the three year histories of the growth rates in real house prices, employment, real income per capita, and a variety of fixed effects. This conclusion is particularly true for estimates of extreme events, which also vary dramatically among these local housing markets. Our projections (at the median) for 2008 to 2010, when using data through 2007, prove to be highly correlated with the actual house price changes. However, the magnitudes of the actual house price declines substantially exceed the predictions, especially among those MSAs in which house price declines in 2008-2010 were especially severe. More broadly, our analysis underscores the complexity of forecasting bubbles. It especially highlights the difficulties of doing so in the framework built upon the notion of a national housing market and relatively simplistic equilibrium conditions. Even in recent years, housing markets are heavily influenced by local supply and demand conditions.