![Sample our Health and Social Care journals, sign in here to start your FREE access for 14 days]({"type":"image" , "src" : "/sda/5938/TOC-Subject-Sample-2021-Health-Social-Care.jpg"})
Abstract
We use psychological theory to investigate how attitudes toward homebuying relate to first-time home purchases over the past decade. Homeownership rates in the US have dropped to 20-year lows, but whether views toward homebuying shifted due to the financial crisis is not known because studies have not compared attitudes for the same respondents pre- and post-crisis. We address this gap with 2004–2014 panel data from low-income renters. We find that a negative shift in homebuying attitudes is associated with a decline in first-time home purchases. Older renters aged more than 35 years at baseline report the greatest declines in homebuying intentions. Younger renters aged 18–34 also report diminished homebuying intentions, yet express highest overall levels of homebuying intentions pre- and post-crisis. Blacks report greater homebuying intentions although their odds of home purchase are 29 per cent lower than whites. Homebuying norms and favorability are associated with homebuying intentions but not with actual purchases, while perceived control over homebuying influences both outcomes.
Notes
1. The income threshold was equal to 80 per cent of the AMI if the percent minority population was less than 30 per cent or equal to 115 per cent of the AMI if the percent minority population was 30 per cent or greater. Area median income is calculated according to the Core Based Statistical Area (CBSA) or the state median income (when CBSA does not apply).
2. Details of the renter’s eligibility criteria and the original pool of baseline renters are available (Akin et al., Citation2004). Additional information about the CAP program, the surveys, and database is available from Riley, Ru, and Quercia (Citation2009).
3. A total of 792 renters completed the final interview in 2014. The number of completed interviews and response rates for each of the 11 years of surveying are available in the online technical report by Riley (Citation2015).
4. Relative income equals annual household income divided by area median income as indicated by the CBSA or the state median income (when CBSA does not apply).
5. Response options were reverse coded after data collection so that higher responses indicate stronger agreement. All respondents answered these items at baseline. In follow-up surveys, these items were anchored with the phrase “within the next three years,” and only those participants who agreed that they intended to buy a home were administered the questions about attitudes, norms, and perceived control. The latter two items were not assessed after 2011. Baseline responses were used when these items were missing in later years.
6. This item wording began in the first follow-up of 2005 and continued through 2014. A slightly different wording was presented at 2004 baseline: “I eventually intend to buy a home.” Model results show non-significant differences by year when baseline 2004 Intentions are compared to those in 2005. Regardless, the 2004 baseline wording does not drive the changes observed over time because that item is modeled as a control variable, whereas 2005 serves as the comparison group reference category for 2006–2014. Model results show that, when compared to 2005 (and while controlling for 2004 baseline Intentions), the downturn in Homebuying Intentions gathers strength in 2007 and then intensifies over the remainder of the panel.
7. Robustness checks indicate that model results do not change whether or not this “living rent-free” group is excluded from the analysis.
8. The proportion of renters purchasing homes between 2005 and 2007 is similar to the 6–8 per cent reported for this same time period in an earlier analysis of CAPS data (Cohen et al., Citation2009). However, the sample size and number of home purchases that we show in Table are higher due to (1) a data retrieval effort in the 2008 survey that captured the housing history of the respondents, (2) manual cleaning to correct address/tenure inconsistencies over time, and (3) the addition of new rows to include data from 2007 to 2014. These updates improve the quality of the data and increase statistical power.
9. Additional information about this issue is provided on pages 246–247 of Allison’s (Citation2010) manual on survival analysis.
10. Whereas homebuying intentions are estimated as a conditional mean, first-time home purchase is estimated as a non-repeating event history survival analysis that has been shown to produce unbiased estimates for right censoring (Allison, Citation1984, Citation2010). Thus, we would not expect home purchase estimates to be biased by the fact that, over time, the sample contains a higher proportion of renters with less positive attitudes toward homebuying. Our robustness check in Specification 9 of the Appendix 1 supports this empirically in showing that the time-invariant baseline covariate homebuying intentions is not a significant predictor of home purchase when time-varying intentions are also entered in the home purchase model.
11. Variance inflation scores fall below 3 for all predictors, indicating an acceptably low level of empirical overlap to proceed with multivariate analysis.
12. These generational names come from popular culture, according to The Pew Research Center (Citation2010), which notes that Generation X, or Gen-Xrs, refer to those people born from 1965 to 1980. The Millennial generation, or Millennials, refers to those people born after 1980 (through 1999).
13. Results shown in the Appendix 1.
14. A test of the interaction between gender and marital status (not shown) indicates that single males are more likely to purchase a home than married/partnered females.
15. Robustness checks in the Appendix 1 indicate that alternate samples (i.e. balanced and unbalanced panels of participants who stayed renters) provide highly consistent results.