Information access and homeowners insurance purchases

Abstract

This paper examines how information access—proxied by three treatment variables: high speed internet connection at home, English speaking ability and difficulty getting out—affects homeowners insurance purchases. There are only a few studies on how information access to markets affects insurance purchases. To make causal estimates of the impact of access proxies on insurance purchases by consumers, we match treated (little access) to control groups (more access) using several alternative algorithms (propensity scores, Euclidean matching and singular value decomposition (SVD) matching). Access, as proxied by fast internet connection and ability to get out, increases the amount spent on homeowners multi-peril insurance, while English speaking ability has no impact in the U.S. We find that while all the matching algorithms in our U.S. sample from the American Community Survey generally provide the same quantitative estimates, SVD matching appears to match surprisingly well on ‘out of sample’ variables.