Publications

Recent Publications

We empirically document that the propensity of high real-investment stocks to underperform their counterparts, that is, the investment anomaly, is attributable to firms building additional production capacity. We develop a real options model to rationalize that finding based on the premise that newly-built capacity often embeds innovative technologies characterized by a high initial uncertainty, as typically assumed in the literature. Since that uncertainty is idiosyncratic, it drags down the expected return of constructing firms until it is eventually resolved. Further empirical evidence based on profit sensitivities to aggregate conditions; analyst forecast-error volatilities; and high-vs.-low tech industry subsamples offers support for our uncertainty-based explanation.

Conditionally Accepted November 2023 - Accepted version available here.

We study the asset pricing implications of being able to optimally early exercise plain-vanilla puts, contrasting expected raw and delta-hedged returns across equivalent American and European puts. Our theory suggests that American puts yield less negative raw but more negative delta-hedged expected returns than equivalent European puts. The raw (delta-hedged) spread widens with a higher early exercise probability, as induced through, for example, moneyness, time-to-maturity, and underlying-asset volatility (variance and jump risk premiums). An empirical comparison of single-stock American puts with equivalent synthetic European puts formed from put-call parity supports our theory if and only if we allow for optimal early exercises in our return calculations. More strikingly, allowing for optimal early exercises significantly alters the profitability of 14 out of 15 option anomalies, with the average absolute change equal to 32% and five anomalies becoming insignificant.

Accepted October 2023 - Accepted version available here.

While recent theoretical and empirical work suggests that the physical skewness of a stock’s future discrete return distribution prices stocks, it does not tell us over which return horizon(s) that physical skewness is priced. Developing a novel block bootstrap estimator that allows us to calculate realized return skewness over arbitrary horizons, we aim to identify those return horizons. In doing so, we first show that our block bootstrap estimator produces more accurate realized skewness estimates than other recent estimators do. Next, we report that the existing skewness proxies used in the empirical asset pricing literature differ in how well they predict skewness over short or long return horizons. Finally, we reveal that the skewness pricing evidence documented in the empirical asset pricing literature is mostly driven by skewness over short (and not long) return horizons.


Prior studies show that, in models in which the value of an asset is exogenously determined, the asset's volatility has an unambiguous effect on the expected returns of European options written on it. Using a stochastic discount factor model, we show that this is not the case when the value of the asset is endogenously determined. In our model, the effect of volatility on expected option returns hinges on whether volatility is systematic or idiosyncratic and on moneyness. While an increase in idiosyncratic volatility only affects option elasticity, leading idiosyncratic volatility to be unambiguously priced, an increase in systematic volatility also has an oppositely-signed effect on the expected underlying asset return, with option moneyness determining which effect prevails. Our empirical analysis supports these prediction, suggesting that idiosyncratic volatility unambiguously prices European call options, while systematic volatility positively (negatively) prices in-the-money (out-of-the-money) European call options.


In agreement with a representative agent economy with recursive preferences in which the mean and volatility of consumption follow independent Markov chains, we show that exposures to both consumption growth as well as expected consumption volatility price the cross-section of delta-hedged single-stock calls. While the consumption growth premium is significantly positive, the volatility premium is significantly negative. Our evidence reveals that consumption risks offer rational foundations for why moneyness, option skewness, and idiosyncratic underlying stock volatility condition delta-hedged option returns. Remarkably, the consumption premiums estimated by us also help to explain a large number of stock anomalies. Our evidence finally suggests that, in our representative agent economy, investors prefer early resolution of uncertainty.  

We study whether a recent French collateral reform, Ordonnance 2006-346, led to a democratiaztion of credit access across France. The reform updated security laws put into place by Napoleon in 1804, enabling firms for the first time to pledge hard assets without dispossession, to pledge fungible and/or future assets, and to utilize rechargeable security interests. The reform was, however, undermined by non-codified laws from the 1970/80s, allowing firms to pledge liquid assets to factoring companies. Using differences tests, we show that firms operating mostly hard assets and located far away from factoring companies significantly raised their leverage after the reform, with the proportion of zero long-term leverage firms among them dropping from 90% to 30%. We further show that small, profitable, and start-up firms benefitted the most. Spatial points analysis suggests the reform reached firms in rural areas, reducing capital access inequality across the country.

While recent studies have shown that exogenous distress risk increases can prompt industrial firms to take on more risk, they offer only limited evidence on whether this behavior hurts creditors, leaving it unclear whether the risk-taking translates into risk-shifting. In our paper, we show that moderately, but not highly, distressed firms increase the risks of their operating segment portfolios in response to exogenous distress risk increases induced through hurricane strikes. The higher risk is facilitated through closing down low-risk segments with high growth opportunities, boosting ex-post failure risk. We also show that creditor control facilitated through covenant violations keeps the most highly distressed firms from raising their risk. Our paper is first in showing that firms' risk-taking behavior in high distress risk situations can amount to risk-shifting.

Real options models of the firm often suggest that the difference between a firm's installed production capacity and its optimal capacity ("capacity overhang") conditions stock returns, although it is unclear what the exact shape of the relation is. Models with highly or completely irreversible investments, for example, often suggest a positive or U-shaped relation, whereas models with more reversible investments often suggest a negative relation. In our paper, we employ a stochastic frontier model to estimate stock-level capacity overhang. Using the capacity overhang estimate in portfolio sorts and Fama-MacBeth regressions, the data suggests that the stock return-capacity overhang relation is monotonically negative, supporting real options models of the firm with more reversible investments. Further supporting these real options models, capacity overhang helps explain momentum and profitability anomalies, but not value and investment anomalies. 

An updated capacity overhang estimate is available from the "Data" section on this website.

Prior empirical studies, such as Dichev (1998), Garlappi et al. (2008), and Campbell et al. (2008), report a flat, negative, or hump-shaped relation between default risk and the cross-section of U.S. stock returns. Constructing a novel dataset of bankruptcy filings for firms in 14 developed non-U.S. countries, we use Campbell et al.'s (2008) logit/hazard model approach to recursively estimate default risk for the firms in these countries. Using the default risk estimates in asset pricing tests, we report a monotonically positive default risk-stock return relation outside of the United States, which is statistically and economically significant. Decomposing default risk into its systematic and idiosyncratic components, we show that the systematic component is responsible for the positive default risk-stock return relation. We also show that the default risk-stock relation is more positive in countries in which bargaining power is skewed towards creditors and away from shareholders.

Older Publications

We use a statistical leverage method to identify those stocks driving the size, book-to-market, and momentum premia in Fama-MacBeth (1973) regressions of stock returns on the stock characteristics. We document that a surprisingly small number of stocks (often less than 1%) are responsible for the anomalies. We further show that the responsible stocks are often volatile, perhaps signalling limits to arbitrage. Persistence tests suggest that the identity of the stocks driving the anomalies changes rapidly over time.  

We study the risk-shifting behavior of firms approaching bankruptcy, investigating whether these firms raise managerial compensation, increase shareholder payouts, and speed up investments in response to positive uncertainty shocks. Our results suggests that only those firms not known to be in distress before their bankruptcy (i.e., those with a high Z-score or distance-to-default) engage in such activities. Studying why the firms known to be in distress do not engage in them, we offer evidence that creditor control facilitated through covenant violations keeps them from doing so. Studying how the risk-shifting firms hide their distress, we offer evidence that the firms pursue more aggressive accounting policies and manipulate their earnings.

We study optimal forecasts under a target-zone loss function, motivated using the example of a central banker in an inflation zone. We derive the optimal forecast of a non-normal (Rayleigh) random variable under that loss function, showing that the optimal forecast is biased. We also show that the optimal bias does not necessarily monotonically increase with the forecasting horizon, giving examples in which the optimal bias either decreases or first increases and then decreases with the forecasting horizon.

We use the method of Marsh and Pfleiderer (1997) to decompose changes in firms' default risk, extracted from Merton's (1974) model or credit default swaps (CDS) data, into global, country, and industry components. In line with intuition, we show that changes in default risk are dominated by global, and not domestic, factors. We, however, further show that domestic factors become more important -- and often dominant -- in expansion periods. Interestingly, cross-sectional asset pricing tests suggest that stock markets tend to price the domestic component of changes in default risk, while not pricing the global components.

Ferguson and Shockley (2003) show that omitting debt claims from the market portfolio proxy used in empirical stock pricing tests induces downward bias in a stock's market beta estimate, with the bias rising with the stock's leverage and distress risk. While the positive downward bias-leverage/distress risk relation appears promising to explain several well-known stock anomalies, we offer both theoretical and empirical evidence suggesting that the relation is likely negligible in the real world. In particular, our theory suggests the relation is diversified away in large economies, while our empirics reveal that economy-wide distress risk is simply too low for the omission of debt claims from the market portfolio proxy to generate a meaningful market beta bias.

We use the methodologies of Elliott, Komunjer, and Timmermann (2005) and Patton and Timmermann (2007) in combination with a block bootstrap to re-assess whether stock market return expectations retrieved from the Livingston Surveys are rational under asymmetric loss preferences (i.e., the differential weighting of positive and negative forecast errors). Assuming homogenous asymmetric loss preferences, our results align with the prior literature in strongly rejecting the hypothesis of rationality. However, allowing asymmetric loss preferences to vary across survey participants, we are no longer able to reject the same hypothesis.

We present a comprehensive overview of the theoretical and empirical literature on corporate hedging. In our overview of the theory, we outline well-established positive rationales for how corporate hedging can contribute to shareholder value in the presence of market imperfections, such as financial distress costs, costs of external financing, asymmetric information, and taxes. In our overview of the empirical literature, we discuss the results from a broad set of empirical studies testing the positive rationales. 

We offer evidence that exposures to the Fama-French-Carhart benchmark factors SMB, HML, and WML price the cross-section of stock returns because the factors are timely and efficient proxies for macroeconomic risks, such as shocks to economic growth expectations, aggregate default risk, and the term-structure of interest rates. The factors reflect the macroeconomic risks since the long and short legs of the factors are differentially exposed to the macroeconomic risks. Setting up a linear factor model directly based on the macroeconomic risks, we show that this macroeconomic-risks model can compete with the Fama and French (1993) and Carhart (1997) models in pricing portfolios univariately or double-sorted on popular firm characteristics.   

Please see my CV (available on the landing page) for several even older publications.