Research (Click here for a C.V.)

Current Working Papers

Extrapolative Expectations and Corporate Risk Management, 2024, with Haibo Jiang, Yuhang Xing, and Yifan Zhang. What are the corporate finance implications of extrapolative expectations? Using a novel dataset of detailed hedging positions of gold miners, we show that past gold returns are an important determinant of hedging behavior. Gold hedging declined by over 90% as gold prices rose in 1990s. We show that analyst expectations, gold ETF investor flows, and manager statements in annual reports about why they reduce hedging, all exhibit extrapolation.

Can the Representativeness Heuristic Explain the Asset Growth Anomaly?, with Miao He and Sheri Tice. We show that the time-variation in returns of four prominent growth-related anomalies—asset growth, sales growth, book-to-market, and analyst expected long-term growth—can be predicted by a single variable. This variable is constructed to estimate the time-variation in the representativeness heuristic of Kahneman and Tversky (1972), as formalized in models of diagnostic investor expectations (e.g., Bordalo, Gennaioli, La Porta, and Shleifer, 2019).

Publications

Firm-specific risk and equity market development (with Gregory W. Brown), 2007, Journal of Financial Economics 84, 358-388. The changing composition of publicly listed firms explains the increase in idiosyncratic volatility from the 1960s to the late 1990s.

Tracking down distress risk, 2011, Journal of Financial Economics 102, 167-182.  Exposure to aggregate distress risk explains the size and value premiums. HML and SMB hedge increases in aggregate defaults. A single factor chosen to optimally predict aggregate defaults works as well as SMB and HML in pricing size and b/m sorted portfolios. The key innovation in this paper is to measure distress risk using covariance with economy-wide defaults rather than the firm-specific measure of default probability.

Death and jackpot: why do individual investors hold overpriced stocks? (with Jennifer Conrad and Yuhang Xing), 2014, Journal of Financial Economics 113, 455–475. A potential for ‘jackpots’ (a small probability of really high  returns) explains the low average returns of stocks with high default risk shown by Campbell, Hilscher, and Szilagyi (2008).

Davids, Goliaths and business cycles (with Jefferson Duarte), 2017, Journal of Financial and Quantitative Analysis 52, 2429-2460. A cool new predictor of market excess returns, bond excess returns, GDP growth, investment growth, SMB, and, HML that is intuitive, grounded in theory and works out-of-sample. Old version (focuses on predicting market returns and includes a calibration of the Menzly, Santos, and Veronesi (2004) model)

Safe Minus Risky: Do Investors Pay a Premium for Stocks that Hedge Stock Market Downturns? (with Barbara Ostdiek, James Weston, and Morad Zekhnini), 2019, Journal of Financial and Quantitative Analysis. Internet Appendix.   Stocks that hedge against sustained stock market downturns -- peaks to troughs in S&P 500 index levels at the business cycle frequency -- should earn low returns, but they do not. A tradeable Safe Minus Risky portfolio that delivers returns of 4.6% per month during stock market downturns, also earns large unconditional mean returns and alpha. There appear to be no costs to hedging 'bad times', only benefits. Sentiment predicts returns to SMR.

Do Idiosyncratic Jumps Matter? (with Morad Zekhnini), 2019, Journal of Financial Economics). The entire annual return of a typical stock accrues on the four days (on average) on which its stock price experiences jumps, or large idiosyncratic movements relative to its volatility, Stock prices drift down by about 2% before jumps. We find evidence consistent with the hypothesis that these patterns are due to a premium for idiosyncratic jump risk. A trading strategy that buys stocks with high ex-ante jump probability earns high average returns and alphas. Returns to this strategy are related to limits to arbitrage.

On the information content of ratings and market-based estimates of credit risk, 2022 (with Oleg Gredil and Junghoon Lee,  Accepted, Journal of Financial Economics), 2022. We find that ratings complement market-based measures and are not redundant in predicting defaults across horizons. Market-based measures differ from ratings in that they respond to both cash-flow and discount-rate news, while ratings respond primarily to cash-flow news, which is more informative of future defaults. Ratings ignore transitory shocks to credit risk, while market-based measures do not. 

One vol to rule them all, 2022 (with Mathew Linn and Bradley Paye). Accepted, Journal of Financial and Quantitative Analysis. We show that a common component governs volatility dynamics across a wide range of traded equity factors  including standard characteristics based long-short factors, statistical factors, tracking portfolios for macroeconomic   variables, and industry portfolios.

Working papers

Estimating the cost of equity: Why do simple benchmarks outperform factor models? 2015:  Compares the performance of ‘naive’ estimators of cost of equity such as the historical market mean with plug-in estimators from factor models. One key insight of the paper: even under extremely favorable conditions for the CAPM (most notably, the CAPM is true), the historical market mean is more accurate than the standard plug-in CAPM estimator for a little over 50% of the cross-section of stocks! Results extend to multi-factor models and industry portfolios.

Testing Factor Models on Characteristic and Covariance Pure Plays (with Kerry Back and Barbara Ostdiek), 2015: We find that returns are associated with characteristics rather than covariances with risk factors for the new Fama-French and Hou-Xue-Zhang factor models. We generate test assets from Fama-MacBeth regressions that are bets on a particular characteristic or covariance and are neutral to all others. Our methods rely on the portfolio interpretation of Fama-MacBeth regression coefficients to construct tests that are unbiased in Monte-Carlo simulations despite errors-in-variables. 

Slopes as Factors: Characteristic Pure Plays (with Kerry Back and Barbara Ostdiek), 2013: We examine the performance of factors formed using Fama-MacBeth regressions instead of sorts.

The next Microsoft? Skewness, idiosyncratic volatility, and expected returns, 2007