Asymmetric or Incomplete Information about Asset Values?

Abstract

We provide a new framework for using text as data in empirical models. The framework identifies salient information in unstructured text that can control for multidimensional heterogeneity among assets. We demonstrate the efficacy of the framework by reexamining principal-agent problems in residential real estate markets. We show that the agent-owned premiums reported in the extant literature dissipate when the salient textual information is included. The results suggest the previously reported agent-owned premiums suffer from an omitted variable bias, which prior studies incorrectly ascribed to market distortions associated with asymmetric information. (JEL D82, G14, R00) Authors have furnished an Internet Appendix, which is available on the Oxford University Press Web site next to the link to the final published paper online.

Mitigating agency costs in the housing market

“…The cross‐sectional results reported here are consistent with this notion, but, as the results from Liu et al. (2020) imply, this cross‐sectional association should merit further scrutiny as we discuss below.…”

“…In contrast with salespersons and associate brokers, we find little evidence of moral hazard for principal brokers based on the cross-sectional evidence. However, as Liu et al (2020) point out, the cross-sectional analysis alone could be spurious, and that perhaps the crosssectional statistical relation we associate with agency cost is due to omitted variables. For identification, we then examine this prediction for within agent difference-in-differences (diff-in-diff or DD) model specifications, using the timing of changes in individuals' incentives to identify a causal effect.…”

“…Broadly, this empirical approach follows established conventions of measuring agency costs in the real estate market (e.g., Agarwal et al, 2019;Levitt & Syverson, 2008;Rutherford et al, 2005) by benchmarking an outcome when agency costs are absent (agents selling their own properties) relative to outcomes where agency costs are potentially present (agents selling their clients' properties). Given that recent studies like Liu et al (2020) have raised concerns of omitted variables bias with the cross-sectional evidence of agency costs, we also modify the standard hedonic approach with a difference-in-differences (diff-in-diff) research design, leveraging the timing of the change in licensure status within agents and changing the nature of the control group for identification across groups over time. Finally, we exploit a third difference that compares our diff-in-diff results across types of brokerages (franchise vs. non-franchise) in a triple difference (diff-in-diff-in-diff) design.…”

“…Given that recent studies like Liu et al. (2020) have raised concerns of omitted variables bias with the cross‐sectional evidence of agency costs, we also modify the standard hedonic approach with a difference‐in‐differences (diff‐in‐diff) research design, leveraging the timing of the change in licensure status within agents and changing the nature of the control group for identification across groups over time. Finally, we exploit a third difference that compares our diff‐in‐diff results across types of brokerages (franchise vs. non‐franchise) in a triple difference (diff‐in‐diff‐in‐diff) design.…”

“…To investigate these predictions empirically, our evidence draws from a Virginia real estate market, which replicates the agency cost results from Levitt and Syverson's Chicago market and other studies, despite being an altogether different market over a different timeframe. It contains detailed data on home transactions and the agents who take part in each of them, which allows us to exploit variation across (and within) real estate professionals who have varying levels of 1 One very recent study by Liu et al (2020) has called into question whether these agency costs actually exist at all, finding evidence that agency costs documented in the real estate literature may be the result of omitted variables bias. While our baseline results appear to be in line with prior literature, we acknowledge the omitted variables bias is an important concern.…”