Interactions Between Housing Market and Stock Market in the United States: A Markov Switching Approach

Abstract

This study uses the Markov switching vector autoregressive model (MSVAR) model to examine dynamic relationships between stock and housing market returns in the United States covering the period from 1987 to 2017. The results show significant regime-dependent auto-correlations in stock and housing returns in both the high volatility and low volatility regimes studied. The feedback effects are stronger in the housing market than the stock market. We observe significant positive cross-market spillovers, consistent with the wealth story. Increases in stock returns in low volatility regimes create positive spillover effects into housing markets; likewise, positive spillovers in the reverse direction from housing market to stock market occur in high volatility regimes. We also find significant negative correlations between lagged stock returns and current housing returns in the high volatility regime, which implies that capital switching occurs as investors move their investments out of the stock market and into the housing market. In this manner, the housing market becomes a hedge against a volatile stock market.

Keywords:

• Markov switching
• regime-dependent relationships
• stock market
• real estate market
• wealth effect
• capital switching

Notes

1 A more technical review of the MSVAR model is provided by Krolzig (2013).

2 Chan et al. (2011) use “tranquil” and “crisis” represent the two states of the markets; these two regimes correspond to the “bull” and “bear” markets described by Maheu and McCurdy (2000), in which the bull market displays high returns with low volatility, and the bear market has a low return and high volatility.

3 The regime classification measure is to test whether the regime-switching model is the best fitted one. The test statistics is $\mathrm{RCM}=400\times 1/T{\sum }_{t-1}^T{p}_t(1-p_t),$ with $p_{1t}$ the smoothed (ex-post) probability of being in state 1 at time t. A value of 0 indicates a perfect regime classification and a value close to 100 reveals no regimes switch available information.

4 The Case-Shiller 10-City Home Price Index is a value-weighted transaction-based index, which is constructed using all residential property transactions in 10 cities that include Boston, Chicago, Denver, Las Vegas, Los Angeles, Miami, New York, San Diego, San Francisco, and Washington, D.C.

5 Source: https://fred.stlouisfed.org/release?rid=199 (Economic Research, The Federal Reserve Bank of St. Louis).

6 The quarterly panel consisting of 124 observations over the same sample period is also used in our analyses as robustness checks.

7 The delayed price adjustments, which are observed in the U.S. housing market (Case & Shiller, 1988; Gyourko & Keim, 1992), if unadjusted, could distort interactions between housing market and stock market.

8 The MSIAH(2,1) models were also estimated using the quarterly return series as the robustness checks, and the results are consistent. For brevity, the results are not shown in the Table.

9 Thank you for the comment by an anonymous referee. The spillover effects from and out of the housing market may be underestimated, if other policies encouraging housing purchases are not properly controlled for in the models. Some of the policies include the Tax Reform Act of 1986, the Federal Housing Enterprises Financial Safety and Soundness Act, the Taxpayer Relief Act of 1997, and the repeal of the Glass-Steagal Act.

Additional information

Funding

This article was funded by the Ministry of Science and Technology (105-2410-H-224-018-MY2).