Over the past half century or so, the U.S. government has supported homeownership with numerous policies. For example, it created the government-sponsored enterprises (GSEs) to develop a stable secondary mortgage market so that households could obtain the necessary financing, with particular focus on less advantaged households. The federal government has also chosen to subsidize homeownership through the tax code, implementing tax deductions for mortgage interest and property taxes. The result of such policies has been the upward trend in the homeownership rate from about 62 percent in 1960 to a peak of 69 percent in 2004. (The foreclosure crisis has resulted in a decrease in the homeownership rate to about 67 percent as of the first quarter of 2010.)1
One of the main rationales for the government's pro-homeownership stance is the social benefit that homeownership is believed to produce. Basically, many perceive that homeownership gives individuals a stronger incentive to improve their neighborhood and community. A fairly extensive literature on this subject has purported to confirm such beliefs. Evidence supports the notion that homeowners participate more in the political system than renters (DiPasquale and Glaeser 1999) and are more likely to become involved in community activism in general (Rohe and Stegman 1994). Alba, Logan, and Bellair (1994) and Glaeser and Sacerdote (1999) have found a negative correlation between homeownership and the incidence of crime. Other researchers have found some evidence that homeowners take better care of their homes than renters do (Mayer 1981).
Other factors may drive both homeownership and community activism
But there is a very difficult econometric problem that many of these studies either do not address at all or do not address completely: the possible existence of unobserved characteristics that are correlated with both homeownership and the tendency to participate in community activism. That is, the types of people who are likely to become homeowners may also be the same people who are more likely to participate in their community. If this is the case, then these studies are mistakenly identifying homeownership as a causal factor of these social outcomes and falsely concluding that homeownership yields positive social benefits. To avoid this econometric issue and truly identify the causal effect of homeownership on participation in these various activities, we need to find some way to create random variation in homeownership decisions that is not correlated with any unobserved characteristics of individuals.
Matching savings program facilitates study of homeownership's social benefits
A new study by Gary V. Engelhardt, Michael D. Erikson, William G. Gale, and Gregory B. Mills in the Journal of Urban Economics attempts to accomplish such a task. The authors performed a field experiment with low-income renters in Tulsa, Oklahoma, from 1998 to 2003 that subsidized saving for a home purchase through what is called an Individual Development Account (IDA). They started with a pool of individual renters interested in such a program and then randomly picked a sample of them to participate. (Participation after selection was optional.) The program matched participants' saving specifically for a future home purchase at a 2:1 rate for annual deposits of up to $750 for three consecutive years. Thus, counting both deposits and matched funds, at the end of the three years, a participant could accumulate up to $6,750. This may not sound like that much, but it is a non-trivial fraction (about 11 percent) of the median house value in Tulsa for a similar low-income population of homeowners during the same period. Indeed, the IDAs appear to have encouraged homeownership: the authors find that after four years, the individuals who were given the option to participate in the program (the treatment group) had a homeownership rate of 7–11 percentage points higher than the individuals not given the option (the control group).
The authors use participation in the IDA—more specifically, the ability to participate, which was randomly assigned—as an instrument for homeownership. They collected information for their sample of renters and homeowners on these attributes: the extent of interior and exterior home maintenance expenditures; political involvement (propensity to vote, amount of support in time and money given to political candidates, tendency to write to or call public representatives); neighborhood involvement (volunteering and fundraising for a church, school, or other neighborhood organization; time spent working on neighborhood projects; and time spent participating in community associations); and time spent giving to other community members (providing childcare or care for another adult, watching someone else's home or pet, and making calls or writing/reading letters for someone else). Thus, their empirical strategy is a two-stage regression in which the first stage uses IDA participation to instrument for the probability of becoming a homeowner, and the second stage regresses the various measures of social involvement on the component of the variation in homeownership decisions that is due to the IDA experiment.
Their first finding is that when they don't instrument for homeownership decisions, they find very large social benefits, which is consistent with the previous literature. For example, becoming a homeowner increases the probability of having called or written a public representative by more than 17 percentage points and of voting in an election by almost 24 percentage points! They also find that becoming a homeowner seems to significantly increase the amount of exterior home maintenance by 13 percentage points.
Controlling for ownership finds negative relationship, underscores need for more research
But the more interesting finding is that when they do instrument for homeownership, all these positive effects disappear. In fact, in some cases the estimated effects become negative. For example, becoming a homeowner makes one less likely to volunteer or help raise money for a church, school, or neighborhood organization, and makes one less likely to become involved in local politics. The evidence regarding maintenance isn't quite as definitive. The estimated effect of homeownership on the likelihood of performing exterior maintenance is not precisely measured (the point estimate is positive but is not statistically significantly different from zero). The estimated effect of homeownership on the likelihood of performing interior maintenance is positive and statistically significant, but interior maintenance is really an internal benefit of homeownership rather than a social benefit (what you do inside of your home does not really affect the value of your neighbors' homes).
In our opinion, this is a really nice piece of work, on a very topical subject. It emphasizes the need to revisit some of the findings of the early literature on the social benefits of homeownership, as many of the positive effects found in that literature appear to be the result of spurious correlation—unobserved characteristics that influence the likelihood of an individual both to become a homeowner and to participate in his or her community to a greater extent. The study has some issues, which the authors themselves point out, with the design and implementation of the IDA experiment that might not make it completely representative of the entire U.S. population. The experiment was performed on low-income, employed individuals in Tulsa, where housing prices are relatively low. In addition, by simply signing up for the program, the individuals were likely signaling that they were more motivated to save (and thus more patient) than others. The authors also point out another potential problem, which is the possible conflation of homeownership and wealth effects resulting from the IDA experiment design. The matching funds increased individuals' wealth in addition to making them more likely to become homeowners. If increased wealth has an effect on the various social outcomes studied in the paper, then IDA participation would be capturing both homeownership effects and wealth effects. In any event, at the very least, the paper is a nice starting point for future research on this important topic!
By Kris Gerardi, research economist and assistant policy adviser at the Atlanta Fed (with Boston Fed economists Christopher Foote and Paul Willen)