residential demolition and new construction data, and votes to renew local government tax levies
Description
2,264 votes 1995 - 2017 representing 254 unique local governments in Ohio, mostly villages and cities but also members of joint fire districts and a few townships, matched to dollar value of residential new building construction and demolition from 1993 to 2022. Identifier data includes subdivision name, subdivision type, county, any joint district membership, and year. Voting data contains tax type (income, property), tax purpose (EMS, cemetery, fire, etc.), millage or percent of tax, duration of tax levy, votes for, votes against, % in favor of tax. Demographic controls include population, % poverty, % with children, % single parent households, highest educational attainment (% less than high school, % high school only, % some college, % bachelors, % graduate degree), unemployment rate, median family income, % owner occupied housing, % aged under 5, % aged 5 to 17, % aged 18 to 64, % aged 65 and above, % White, % Hispanic, % married, % separated, % divorced, labor force participation rate. New construction data includes dollar value of new buildings constructed in current year as well as two prior years (lags) and five years after the vote (leads). New demolition data includes same for dollar value of demolished structures. Data can be used to replicate "Housing demolition, local government spending, and the types of cities that consistently demolish the most," by David M. Brasington, which has been revised and resubmitted 5/27/2026 to Journal of Real Estate Research. Data was gathered from the Ohio Department of Taxation, the Ohio Secretary of State, and the U.S. Census Bureau. Data was used with rdrobust Stata command to show that otherwise similar cities that renew tax levies hae more demolition of residential structures than cities that barely vote to cut local tax funding. Data also shows no difference in new construction between cities that barely renew vs. cut local taxes.
Files
Steps to reproduce
open the dataset. In Stata, perform the following code: rdrobust resdemolished_lead2 pctfor, c(0.5) p(1) q(2) kernel (tri) bwselect(mserd) vce(nn 3) all change the leads and lags for the other years before and after the vote. change to the new construction outcome variables to replicate those (e.g., resnewconstruction_lag1) like so: rdrobust resnewconstruction_lead2 pctfor, c(0.5) p(1) q(2) kernel (tri) bwselect(mserd) vce(nn 3) all
Institutions
- University of CincinnatiOhio, Cincinnati