The effect of macroeconomic uncertainty on housing returns and volatility: evidence from US state-level data

In this chapter, we first estimate a dynamic factor model with time-varying loadings and stochastic volatility (DFM-TV-SV) using Bayesian methods to disentangle the national and local factors affecting real housing returns and volatility in the 50 US states and the District of Columbia. We then use panel data regressions with heterogeneous coefficients to relate the first and second-moment of the local factors to corresponding state-level uncertainty. The latter is estimated using the average forecast error variance of a range of regional variables and 248 national-level data series in a factor augmented forecasting regression with stochastic volatility in the regression residuals and the error term for the factor dynamics. We estimate uncertainty at a forecasting horizon of one to four quarters over the periods 1977Q2 to 2015Q3 and 1991Q1 to 2015Q3, depending on model specifications. We find that all but three states register a positive and significant spillover effect from macroeconomic uncertainty to house price stochastic volatility, with Hawaii and Michigan ranking highest in terms of spillover effects. The majority of the most severely impacted states are from the Midwest region, as well as a number of states in the Southern region, known to be lower income states. A negative impact of macroeconomic uncertainty on house price returns is recorded in some states, notably from the Midwest region. Our results have important implications for homeowners, mortgage lenders and investors.