Does Owning a Home Make Us More Generous?

Abstract

We examine the multifaceted influence of homeownership on charitable giving through several channels: tax deductibility, household wealth, and mobility. We find that homeowners donate substantially more than renters, and tax deductibility, wealth, and mobility are important predictors of the owner-renter gap in donations. We also show that the owner-renter difference in donations cannot be fully explained by these three channels. After controlling for an extensive list of household characteristics and the three channels, homeowners still donate approximately 20% more than renters. Our results are robust to a variety of modeling and identification strategies as well as different measures of donations. Our study further reveals that the likelihood of donating correlates inversely with mobility but is insensitive to tax deductibility and wealth. In contrast, tax deductibility and wealth are important predictors of the size of contributions. Furthermore, we show the owner-renter difference in donations varies substantially by generational cohorts.

Keywords:

• charitable giving
• altruism
• homeownership

JEL CODES:

• D10
• D64
• R20

Acknowledgments

The views expressed are those of the authors and not necessarily those of Fannie Mae or the Federal Housing Finance Agency. We are grateful to an anonymous referee and the Managing Editor (Kimberly Goodwin), whose insights have markedly improved the paper. We would also like to thank Peng Liang, Zhenguo (Len) Lin, and the participants at the 2022 American Real Estate Society (ARES) Annual Meeting and the Financial Management Association (FMA) 2022 Annual Meeting for helpful feedback. All errors are, of course, our own.

Notes

1 See https://givingusa.org/giving-usa-2020-charitable-giving-showed-solid-growth-climbing-to-449-64-billion-in-2019-one-of-the-highest-years-for-giving-on-record/ (accessed on September 13, 2021).

2 For a more comprehensive review of the consequences of homeownership, see Dietz and Haurin (2003).

3 The PSID began to collect information on donations in 2001. However, in the initial year, the values were reported only for the first five categories. For the other six categories, the PSID asked only whether donations were made, to which the respondents answered yes or no. Since 2003, dollar values are reported for all eleven categories. For our donation variable to have a consistent interpretation across time and to better capture all donations, we chose 2003 as the starting point of our analysis. The 2019 survey was the latest PSID survey at the time we conducted our research.

4 The PSID classifies mobile home owners as homeowners. However, most mobile home owners pay rent for their lots. Due to the idiosyncrasy of their ownership arrangement, we eliminate from our sample the small fraction of mobile home owners.

5 Following previous literature, we henceforth refer to the after-tax price of donations as the price of donations.

6 To identify “endogenous itemizers,” we begin with the sample of itemizers (as reported in the PSID). We first feed their information into the TAXSIM program to determine whether they should itemize. We then set their charitable donations to zero and reestimate whether the household would have itemized without charitable donations. A household is an “endogenous itemizer” if it reported being an itemizer and is predicted to itemize, but only when donations are included among itemized deductions.

7 Housing equity is calculated as house value minus the total outstanding balance of mortgages. The seven asset categories are farm or business, checking and savings accounts, other real estate assets, stocks, vehicles, annuities and IRAs, and other assets. The eight debt categories are farm or business debt, other real estate debt, credit card debt, student loans, medical debt, legal debt, loans from relatives, and other debt.

8 Donations are zero for some household-year observations, and a logarithmic transformation would render those undefined. Following previous literature, we add $1 to the donation before taking the natural log.

9 See https://www.cdc.gov/ncbddd/childdevelopment/positiveparenting/index.html (accessed on January 22, 2023).

10 Two-sample t-tests are conducted on continuous variables, and two-sample ${\chi }^2$-tests are conducted on categorical variables.

11 Newman and Holupka (2016) applied a similar method to study the effect of homeownership on wealth accumulation. Additionally, Barker and Miller (2009) used the difference-in-difference method to study the effect of homeownership on child welfare.

12 Results of our overlapping and balancing tests are available upon request.

13 Gormley and Matsa (2011) used a similar approach to study corporate responses to the liability risk from workers’ exposure to newly identified carcinogens. As a robustness check, we also estimated the simple DiD specification with treatment and post indicators. The results are similar.

14 Henceforth, we use the term “boomers” to mean boomers and older. Similarly, we use the term “millennials” to mean millennials and younger.

15 See https://www.jec.senate.gov/public/index.cfm/republicans/socialcapitalproject (accessed on March 5, 2023).