Experimental Design
We field a survey experiment embedded in an online survey of US residents aged 25–70 who own or rent their home and live in metropolitan ZIP codes. Respondents are recruited through an online panel invited from a list of urban and suburban ZIP codes, with homeowners and renters each making up half the sample and quotas on age, gender, region and household income. After demographic and screening questions, respondents answer open-ended questions about housing, then report their perceptions of local home values and rents, the affordability of housing for a hypothetical lower-income family, the prevalence of housing cost burdens and the causes of high housing costs, and how fast the number of homes in their area has grown. They then read a hypothetical scenario in which a state law removes some local restrictions on housing development and the number of homes in their ZIP code becomes 10% larger in five years, and are randomly assigned to one of three framing arms (Control, Broad, Narrow). Respondents report their expected home value or rent under the baseline and alternative scenarios, expected effects on metro-wide prices, property taxes, neighborhood amenities and new residents, and their support for and perceived fairness of the policy; the order of these two sets of questions is randomized. A random half also report the lowest and highest values they consider possible, to capture uncertainty in their expectations. The survey then asks about other housing policies, zoning and local versus state control, the respondent's housing situation, political views, trust in government, zero-sum thinking and local political participation.