Re-examining the impact of housing wealth and stock wealth on retail sales: Does persistence in wealth changes matter?

Case, Quigley and Shiller (2005, 2013) quantified stock versus housing wealth effects on quarterly state-level retail sales, which they interpret as an approximate measure of household consumption spending. We investigate the variation of these wealth effects with the persistence of each kind of wealth fluctuation in an estimated linear dynamic fixed-effects model, allowing for both cointegration and endogeneity. Retail sales respond most strongly to housing wealth fluctuations which persist for one to four years, whereas the response to stock wealth fluctuations is smaller and is concentrated on fluctuations with a persistence of either less than a year or more than four years. These differential persistence effects point to a need for a richer theoretical formulation in this area. A substantial literature has arisen that compares the wealth effect due to housing wealth fluctuations with the wealth effect due to financial wealth fluctuations (E.g., see Edison and Sløk (2001), Case et al. (2005) and Shirvani and Wilbratte (2011)). This issue is important because both of these kinds of wealth fluctuations have played major (albeit likely intertwined) roles in triggering and/or extending major macroeconomic episodes in the last few decades and because a fluctuation in each of these household wealth variables calls out for a different set of preventive and/or reactive government policies. Some studies argue that the transitory nature of the changes in stock prices causes them to have a smaller impact on consumption than changes of similar size in the value of other assets – e.g., Benjamin et al. find opposite results in modeling the seven provinces of the Australian economy, where fluctuations in financial wealth appear to have larger impacts than fluctuations in housing wealth. Belsky (2010) found similar consumption effects from real estate and corporate equity fluctuations, both at a magnitude of 5.5 cents on the dollar. Similarly, Carroll et al. (2011) find that the financial wealth effects grows to be more than four-to-ten cents on the dollar over the years following a shock. Their results also suggested that about 80% of the housing wealth effect is realized in one year, whereas a long run effect from the stock market takes five years to approach 80%. Engelhardt (1996) finds asymmetric wealth effects: changes in consumption are significantly associated only with drops in housing values. Case et al. (2005, 2013) used state-level panel data on retail sales, household financial wealth, and