Which immigrant and minority homeownership rates are gaining ground in the US?

Abstract

This paper investigates post-2000 trends in homeownership rates in the US by immigrant status, race, and ethnicity. Homeownership rates for most groups examined rose during the housing boom of the early and mid-2000s but fell during and after the housing bust. By 2015 homeownership rates had fallen below year 2000 levels for most groups but not all. In particular, some Asian immigrant groups experienced sizable gains in overall homeownership rates and in regression-adjusted differences relative to white non-Hispanic natives. Some other immigrant and minority groups also made gains relative to white non-Hispanic natives. We document and discuss these changes.

Appendix: Results with heterogeneous income effects

As noted briefly in Section 4.4, we also conducted additional analysis in which we re-estimated the full specifications in Tables 1 and 3 but also allowing log household income to have differing effects for each immigrant and minority group. To do so, we first centered log household income at the mean log household income for the base group, native non-Hispanic whites. We then interacted the centered log household income with dummy variables for each immigrant and minority group and included these in the models for the additional analysis. Results corresponding to Tables 1 and 3 are reported in Appendix Tables 5 and 6, respectively. Columns 1, 2 and 3 for each Appendix Table are for years 2000, 2007, and 2015, respectively. These include all of the socioeconomic characteristic controls and MSA fixed effects in the full specification of those tables and also add the income*group interaction terms.

The coefficients on the group dummies in the Appendix Tables are overall very similar to their counterparts in Tables 1 and 3 estimated without the interaction terms. Many of the group dummy coefficients and standard errors are exactly the same as their counterparts in Tables 1 and 3. None of the group dummy estimates go from statistically significant to not statistically significant and none of the significant coefficients change signs. Thus, all of the main conclusions above are qualitatively robust to this alternative specification with income*group interaction terms.

The coefficient on log household income (not interacted) is statistically significantly positive in all 3 years with coefficients increasing from 0.040 in 2000 to 0.050 in 2007 and then falling to 0.038 in 2015. This captures the homeownership-income responsiveness of the base group, native non-Hispanic whites. The increase in 2007 indicates that homeownership responsiveness to income was especially strong in 2007 for the base group.

The income*group interaction coefficients are mostly negative and many are statistically significantly negative. In Appendix Table 5, the income*group interaction is only statistically significantly positive (at the 10 % level) for Hispanic immigrants in 2007; all other interactions in Appendix Table 5 are negative or not statistically significant. A negative group*income interaction term indicates that the homeownership decisions for members of the group are less responsive to household income than that of the omitted group, native non-Hispanic whites. The negative income interaction effect in Appendix Table 5 is largest for Native Americans in all 3 years, implying that Native Americans have a much smaller homeownership response to income than native non-Hispanic whites. We can compute the total effect of log income on homeownership for each group and year by summing the log income coefficient with the group*income interaction term. For example, for Native Americans in 2015, the total effect is 0.038 + −0.021 = 0.017, which is still statistically significantly positive based on the joint standard error. In fact, all of the total effects obtained in this way are statistically significantly positive at least at the 10 % level. Thus, the total effect of income on homeownership varies in magnitude across groups, but the direction is always positive. Income*group interaction results in Appendix Table 6 exhibit somewhat more heterogeneity. The coefficients are still largely negative, but the estimates are significantly positive in at least some years for Filipino immigrants, Vietnamese immigrants, other Asian natives, Cuban immigrants, and Puerto Rican born persons. These patterns suggest a nuanced story about how different immigrant and minority group homeownership outcomes respond to income. Future research should consider this nuance in more detail.

We also considered examining group interaction effects for other factors like age, education, and family structure. However, our main analysis controls for these factors using a large number of dummy explanatory variables. The main goal of our paper is to estimate and discuss differences in mean homeownership rates for a large number of immigrant and minority groups relative to native non-Hispanic whites, both unconditionally and conditional on a detailed set of socioeconomic characteristics. Including many dummy control variables is a useful approach for our main analysis because it allows for a flexible non-linear relationship between the explanatory factors and the dependent variable. However, it would be difficult to accurately and concisely report interaction effects for the interactions of many socioeconomic characteristic dummies with our group dummies because there would be so many interaction terms. Thus, we suggest that future research should consider examining group interaction effects for other factors like age, education, and family structure, but doing so may require adopting a different approach to measuring these socioeconomic factors than the approach taken here that uses a large number of detailed dummy variables. Alternatively, future research looking at interactions with detailed dummy variables could focus on one group or possibly a few immigrant and minority groups to limit the scope but increase the depth of analysis. Our analysis offers useful insights, but there is much more to be learned.

Table 5 Heterogeneous income effects on immigrant and minority homeownership gaps (Broad Groups)

Table 6 Heterogeneous income effects on homeownership gaps for Asian and Hispanic detailed groups