The Housing Market “Contagion Effect”: Evidence for Time-Varying Volatility and Misperceptions of Risk Effects Between the U.S. Housing and Stock Markets Around the 2007-2009 Financial Crisis

Abstract

This paper addresses two main questions surrounding the change that recessions, particularly the one resulting from the 2007 to 2009 financial crisis, bring to the interactions between housing market returns and equity market returns. The first is whether or not the recession influences the long-run means of volatility processes for these returns that clearly exhibit time-varying volatility. The second is how badly the housing market’s impact in the larger market was underestimated in the total market before the beginning of 2008, a discrete time break around the financial crisis, relative to the period from 2008 onward. A GARCH(1,1) model, using varying levels of a discrete recession indicator, is performed on the housing market and stock market returns separately for both daily and monthly frequency. The coefficient of the recession indicator is found to be significant and negative for the monthly data, implying that the recessionary period causes up to an 8% drop in the long-run mean of the volatility process. Using several different model specifications added robustness to this result. A Structural VAR model was used to estimate the predicted miscalculation of the size of the “contagion effects” of the housing market downturn to the broader stock market in the most recent crisis in the United States. The study finds, across data frequencies and proxies for markets, that the post-crisis impulse response of the stock market to a housing market shock was in a range from 2 to 2.4 times the impulse response of the pre-crisis period, showing that the estimated effects of this contagion were too heavily discounted given all of the indirect effects present in a huge financial crisis. This points to evidence of a “threshold effect” for investor perceptions of housing risk.