The (Mythical?) Housing Wealth Effect

Abstract

Models used to guide policy, as well as some empirical studies, suggest that the effect of housing wealth on consumption is large and greater than the wealth effect on consumption from stock holdings. Recent theoretical work, in contrast, argues that changes in housing wealth are offset by changes in housing consumption, meaning that unexpected shocks in housing wealth should have little effect on non-housing consumption. We reexamine the impact of housing wealth on non-housing consumption, employing the Case-Quigley-Shiller data on U.S. housing wealth that have been used in prior studies to estimate a large housing wealth effect. Existing empirical work fails to control for the fact that changes in housing wealth may be correlated with changes in expected permanent income, biasing the resulting estimates. Once we control for the endogeneity bias resulting from the correlation between housing wealth and permanent income, we find that housing wealth has a small and insignificant effect on consumption. Additional analysis of time-series results provides further support for that view.

Revisiting the Relationship between Financial Wealth, Housing Wealth, and Consumption: A Panel Analysis for the U.S.

“…The importance of the stock market wealth on consumption is reinforced by the results derived from model III, which includes fixed effects, and by models IV and V, where changes in stock market wealth have still greater impact than the housing effect on changes in consumption. The results are in direct contrast to the findings of CQS, but in an agreement with Calomiris et al [13]. Also, Table 2 reports the t-statistics for the hypothesis that the coefficient of stock market wealth is equal to the coefficient of the housing market wealth.…”

“…While this specification addresses the non-stationarity issue, we understand that it does not take into account possible cointegration relationships. But we proceeded in order to compare our results with the ones obtained by CQS, Calomiris et al [13,17], and Bampinas et al [22].…”

“…Model II explores further the nature of estimated wealth effects and their robustness by including state-specific time trends, while model III includes time fixed effects. Please note that the above model specifications, as articulated by Calomiris et al [13], lead to inconsistent results, since the residual contains changes in permanent and current income, and these will likely be highly correlated with changes in housing and stock wealth. In order to correct for any correlation issues, we proceed with the estimation of three other models IV, V, and VI with the use of two-stage least squares and instrumental variables (The instrumental variables version takes account of possible endogeneity problems).…”

“…Calomiris et al [13] re-examined the impact of housing wealth by employing the CQS data. Following a method suggested by Hall [14], Auerbach and Hassett [15], and Campbell and Mankiw [16], they found that estimated housing wealth had a much smaller magnitude and less significant effect on consumption, compared with the financial wealth effect.…”

Consumption, Wealth, Stock and Government Bond Returns: International Evidence*

“…Muellbauer (2008) suggests that the positive effect on non‐housing consumption from an increase in housing prices is counterbalanced by a fall on housing consumption. Calomiris et al . (2009) emphasize that changes in housing wealth are typically correlated with changes in expected permanent income.…”