Full Citation

Title: Supplement to "Do households use home-ownership to insure themselves? Evidence across U.S. cities": Appendix

Citation Type: Miscellaneous

Publication Year: 2014

ISBN:

ISSN:

DOI: 10.3982/QE346

NSFID:

PMCID:

PMID:

Abstract: S.1. Robustness checks for home-ownership and LTV effects Here, we check for robustness in the impact of price levels and volatilities on ownership (Table S.1) and LTV (Table S.2). We report results for both OLS and instrumental variables (IV) (with land scarcity as the instrument) for volatilities measured using different windows (5 years, as in the main text, and 10 years also), for different year cross sections (2000, as in the main text, and 1990 also), and for different data sets for LTV (AHS and MIRS). For home-ownership, the observed patterns are robust to choice of cross section, volatility window, and year. In each case, we see the strong negative relationships described in the main text. Looking at the OLS specification, the effect of price is very similar across years: a doubling of house prices is associated with a fall in the ownership rate of between 21 and 24 percentage points. The IV results are very similar. With regard to volatility, changing the window of measurement has little effect on the coefficients for each cross section. But the results do vary across years: the OLS effects are larger in 2000 and the IV effects are larger in 1990. Still, the reduced-form effect of the land-scarcity instrument (in the final row) is very similar across years. In Table S.2, the estimated effects are remarkably similar in magnitude across the AHS and MIRS data sets for the 2000 cross section for all variables. In each case, there is a strongly significant negative effect, consistent with the main text. The results are not very sensitive to the chosen volatility window either. However, the effects on LTV in the 1990 cross section (as measured by MIRS), while negative, are all statistically insignificant. This is a result of smaller coefficients, rather than larger standard errors. In Figure S.1, we explore this further: we plot the estimated coefficients from reduced-form regressions of LTV (from the MIRS data) on land scarcity separately by year (over 1978-2008). The dashed lines are 95 percent confidence intervals. The effect has always been negative, though it was small and insignificant until the mid-1990s (averaging about −05). It has grown steadily since though, reaching almost −2 in 2008.

Url: https://pdfs.semanticscholar.org/da10/2822fe1ecb22c8ff65b550a41df770593d8d.pdf

User Submitted?: No

Authors: Amior, Michael; Halket, Jonathan

Publisher: Centre for Economic Performance

Data Collections: IPUMS USA

Topics: Housing and Segregation

Countries: