Document Type

Working Paper

Publication Date

2018

College/Unit

College of Business and Economics

Document Number

18-06

Department/Program/Center

Economics

Abstract

We develop a framework for using text as data in asset pricing models. We use the framework to test whether real estate agents exploit their informational advantage to sell properties they own for a premium. Consistent with the previous literature, baseline estimates that exclude textual information indicate agents sell their own house at a 3 to 4 percent premium in both Phoenix, AZ and Atlanta, GA. However, this premium dissipates when textual information is included. The results suggest that the baseline estimates suffer from an omitted variable bias, which previous studies incorrectly ascribe to market distortions associated with asymmetric information.

Included in

Finance Commons

Share

COinS