Research

Job Market Project:

Information Intermediaries and the Distorting Effect of Incomplete Data

• Abstract: Financial data vendors intermediate the flow of information from firms to investors. I study frictions that arise in the context of this intermediation by focusing on one of the most prominent data vendors in the finance industry: Standard & Poor's ('S&P') Compustat database. Compustat provides subscribers with decades of 10-K and 10-Q data; however, it does not cover every public firm in every period. I show that a significant fraction of institutional investors do not invest in firms with missing data -- institutional ownership is over 36% below its unconditional mean for firms not covered in Compustat. A policy change instituted at S&P in the early 1990s provides a quasi-natural experiment to confirm a plausibly causal association between Compustat data coverage and institutional investor demand. In a battery of empirical tests, I then show that limited access to financial data is associated with lower informational efficiency of equity prices. This highlights the role that data vendors play in facilitating the flow of information within financial markets.

• Presentations: Louisiana State University, SFS Cavalcade (2024, scheduled), Virginia Tech 

Published and Forthcoming Papers:

Taking Over the Size Effect: Asset Pricing Implications of Merger Activity

with Jeffry Netter, Bradley Paye, and Michael Stegemoller

Journal of Financial and Quantitative Analysis, March 2024, vol. 59, no. 2, pp.690-726.

• Abstract: We show that merger announcement returns account for virtually all of the measured size premium. An empirical proxy for ex ante takeover exposure positively and robustly relates to cross-sectional expected returns. The relation between size and expected returns becomes positive or insignificant, rather than negative, conditional on this takeover characteristic. Asset pricing models that include a factor based on the takeover characteristic outperform otherwise similar models that include the conventional size factor. We conclude that the takeover factor should replace the conventional size factor in benchmark asset pricing models.

• Presentations: Aarhus University, University of Washington Summer Finance Conference (2022), Virginia Tech, Washington and Lee University

Working Papers:

High (on) Sharpe Ratios: Optimistically Biased Factor Model Assessments  

• • with Bradley Paye

  • Abstract: Estimates of maximum obtainable Sharpe ratios associated with prominent multifactor models are difficult to reconcile with risk-based economic models, even when such estimates are based on `out-of-sample' research designs. In this paper, we argue that Sharpe ratio estimates are subject to optimistic bias driven by the data-instigated nature of modern factor models. Common 'out-of-sample' research designs do not adequately address this bias. We propose alternative evaluation approaches that mitigate the bias, and show that Sharpe ratio estimates fall dramatically under these approaches for conventional models and for statistical factor models distilled from large sets of cross-sectional return predictors. Reassuringly, our reduced-bias Sharpe ratio estimates do not violate "good deal bounds.'' However, we also conclude that multifactor model improvements relative to the capital asset pricing model (CAPM) are more modest than suggested in the literature. 

  • Presentations: Financial Management Association Conference (2023), Portuguese Finance Network Conference (2023), Financial Management Association European Conference (2023), University of North Carolina at Charlotte, University of North Texas, University of Virginia, Virginia Tech

Works In Progress:

Why is Data Missing in CRSP and Compustat?

• Companion to Job Market Project

• Abstract: CRSP and Compustat are two of the most widely used databases in finance and accounting research. However, neither database provides complete information for all NYSE, AMEX, or NASDAQ firms at any point in time. I describe when data is missing in both databases for over 50 variables, which are used to construct a comprehensive selection of firm characteristics. Importantly, I identify why data is missing. Researchers employing either CRSP or Compustat should be wary of missing data. Whether missing data impacts empirical analyses relying on these databases depends critically on what data is used, and when and why that information is missing.