Working Paper: NBER ID: w24729
Authors: Adam M. Guren; Alisdair McKay; Emi Nakamura; J. N. Steinsson
Abstract: We provide new time-varying estimates of the housing wealth effect back to the 1980s. These estimates are based on a new identification strategy that exploits systematic differences in city-level exposure to regional house price cycles as an instrument for house prices. Our estimates of housing wealth effects are substantially more precise and smaller than recent estimates, though they remain economically important. Our time-varying estimates indicate that housing wealth effects were not particularly large in the 2000s. This contradicts a popular narrative that lax lending standards in the boom and skyrocketing loan-to-value (LTV) ratios during the bust elevated the housing wealth effect in the 2000s. We show, furthermore, that this narrative is inconsistent with a standard life-cycle model with borrowing constraints, uninsurable income risk, illiquid housing, and long-term mortgages. The housing wealth effect in the model is relatively insensitive to changes in the distribution of LTV for two reasons: First, impatient low- LTV agents have a high elasticity; Second, a rightward shift in the LTV distribution increases not only the number of highly sensitive constrained agents but also the number of underwater agents whose consumption is insensitive to house prices.
Keywords: Housing Wealth Effect; Business Cycle Dynamics; Instrumental Variables; Consumer Expenditures
JEL Codes: E21; R21
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| house prices (R31) | housing wealth effect (G59) |
| regional housing cycles (R31) | house prices (R31) |
| housing wealth effect (G59) | marginal propensity to consume out of housing wealth (E21) |
| housing wealth effect (pre-2000) (G59) | housing wealth effect (2000s) (G59) |
| no evidence of boom-bust asymmetry (E32) | housing wealth effect sensitivity (G59) |