U.S. Housing Market Volatility

Author

Xibin Zhang, Southern Illinois University CarbondaleFollow

Date of Award

5-1-2012

Degree Name

Doctor of Philosophy

Department

Economics

First Advisor

Gilbert, Scott

Abstract

The first chapter considers the determinants of U.S. housing market volatility. With volatility defined as the sample variance of home value monthly returns during a given quarter, a model of volatility from equilibrium in the housing market is derived. Supply and demand effects on housing return volatility for the U.S. as a whole and for a panel of 16 cities are tested, during the period 1985:1-2009:4. There are some statistically significant links between volatility and fundamentals, but significance generally plummets when fundamentals are allowed to be endogenous. These results are consistent with time plots, which show no clear historical link between housing return volatility and basic market factors. This enigma may be due to data limitations and/or short-run disequilibrium in home prices. In Chapter 2, GARCH(1,1) model family are utilized to describe dynamics of U.S. national and metropolitan housing market volatilities ranging from the first month of 1985 to the last month of 2009. ARCH effects of housing return exist in U.S. national and 15 MSA markets. Housing return volatility positively influences housing return in some MSAs and negatively in some other MSAs. Only in Detroit does housing return negatively affect volatility, which makes Detroit is the only market with a two-way effect (return and volatility). A couple of MSAs show leverage effects. Almost all selected markets present long memory of volatility. All the conditional volatilities estimated by GARCH(1,1) model are much bigger than unconditional ones in selected MSAs except U.S. national market. Even though component GARCH(1,1) model generally does the best job in forecasting U.S. housing market volatility compared to other members of GARCH(1,1) model family, the forecasting accuracy is far from being satisfied. The last chapter discusses the housing return volatility spillovers across U.S. metropolitan markets. House return volatilities in 15 MSAs ranging from the first month of 1985 to the last month of 2009 are estimated by the standard GARCH(1,1) model. The results of the vector error correction model and the vector autoregressive models show that there are unidirectional and bidirectional housing return volatility spillover effects among not only contiguous MSAs but also noncontiguous MSAs. More volatility spillovers happen among the MSAs that share the same economic characters. The long-run housing return volatility convergence exists in U.S. market.