Liberals and Housing: A Study in Ambivalence

ABSTRACT

Do political liberals support or oppose zoning changes that allow more market-rate development? I use survey data from California and show that liberals are ambivalent. The ambivalence is explained in part by homeownership, which is associated with opposition to new housing of all kinds, even as it has little influence on attitudes about other policies. Even controlling for ownership, however, I find that self-identified liberals remain ambivalent about new development, never supporting it as much as they support more stereotypically liberal policies, and opposing it outright when reminded that enabling new housing might require less regulation, particularly environmental regulation. In contrast, liberals strongly and consistently support spending on subsidized affordable housing. The results together suggest that in supply-constrained cities with liberal electorates, the political calculus is unfavorable to new housing. Ownership injects some conservatism into development politics; liberal ideology could provide a counterweight to that conservatism, but that counterweight might be blunted if development also requires deregulation.

KEYWORDS:

• zoning
• deregulation
• liberalism
• public opinion
• homeownership

Acknowledgments

I thank Paavo Monkkonen and Matt Kahn, members of the UCLA Lewis Center Housing Initiative, and Adam Seth Levine for helpful comments. Miriam Pinski assisted with data cleaning.

Disclosure Statement

No potential conflict of interest was reported by the author.

Notes

1. A note on terminology: Localities have many ways to constrain the housing supply. Whereas some fall under what is legally considered zoning, many do not. For ease of exposition, throughout this article I will broadly refer to regulatory changes that allow more housing as upzoning, zoning reform, or zoning deregulation.

2. One might argue that these cities are not actually liberal, even if they are restrictive. Evidence from Tausanovitch and Warsaw (2014), however, suggests otherwise. These cities, and often their suburbs, are in fact dominated by Democrats, and lean further to the left than other United States jurisdictions do.

3. Recall that housing services refers not to development, but to the sale of housing (including owner occupancy, which in economic terms is an imputed landlord/tenant relationship). Thus homeowners are, in economic terms, producers of housing services, even if they do not physically produce housing (for more, see Meyerhauser and Reinsdorf, 2007).

4. Specifically, residential structures, the vast majority of which are detached single-family homes, represent 45% of all United States physical capital (Vollraith, 2020).

5. The 2014–2018 General Social Survey shows that self-identified liberals are over twice as likely as self-identified conservatives to agree that the government needs to improve the standard of living, needs to help Black people, and should do more to help the disadvantaged more broadly (GSS SDA Data Archive tabulations).

6. The typical stock purchase is rarely as leveraged as the typical home.

7. For example, Zillow Group (2018) reports that the average buyer searches for 4.5 months, then takes an additional 30–45 days to close.

8. More specifically, the hypothesis predicts that homeowners will support regulations that increase home values, and oppose regulations that reduce them. These regulations may or may not directly affect the housing supply.

9. Calculated from GSS data at the Berkeley survey data archive: https://sda.berkeley.edu/

10. See Pew Research: https://www.people-press.org/2014/06/12/appendix-a-the-ideological-consistency-scale/

11. This idea refers, broadly, to a filtering model of housing (see Rosenthal, 2014).

12. A newly built house in any given neighborhood is likely to be more expensive than a similarly sized older house nearby, both because the new house will have more amenities and because older houses can profitably sell for less than their construction costs, whereas new houses cannot.

13. Price changes might be still less salient if they manifest as a slowed rate of increase, rather than as an absolute decline.

14. Some liberals might, of course, value process over outcome, and balk at endorsing deregulation it because they see it as an ideological line that cannot be crossed. Imbroscio (2019), for example, argue that progressives should leave suburban exclusionary zoning intact, in part because undoing it would require zoning deregulation. Deregulating to achieve integration would lend legitimacy to market mechanisms, and thus set the overall progressive movement back, even if in the short term it advanced some progressive goals.

15. PPIC provides weights that correct for sampling error; all of the analysis in this article uses those weights.

16. The single-payer question stands out for its level of nonresponse and don’t know answers, making its sample size much lower than that for other questions.

17. County fixed effects reduce the sample size, because some small counties have few observations, and perfectly predict the outcome variable. Some specifications variously included controls for each county’s median home value, for its Zillow house price index, and for the ratio of its median home value to its median household income. I also interacted ownership with the Zillow index, and liberalism with the Zillow index. None of these variables was statistically significant, and none materially changed the other coefficients.

18. A similar approach that would preserve sample size, but might be harder to interpret, would involve keeping the full sample but interacting ideology and tenure. I do so in regressions not shown here, and the results are essentially the same.

Additional information

Notes on contributors

Michael Manville

Michael Manville is Associate Professor of Urban Planning at the UCLA Luskin School of Public Affairs, and research affiliate of the UCLA Lewis Center's Housing Initiative.