Published and Accepted Papers

We derive lower and upper bounds on the conditional expected excess market return that are related to risk-neutral volatility, skewness, and kurtosis indexes. The bounds can be calculated in real time using a cross section of option prices. The bounds require a no-arbitrage assumption, but do not depend on distributional assumptions about market returns or past observations. The bounds are highly volatile, positively skewed, and fat tailed. They imply that the term structure of expected excess holding period returns is decreasing during turbulent times and increasing during normal times, and that the expected excess market return is on average 5.2%.

We also derive closed-form expressions for any physical moment of the excess market return (e.g., mean, variance, skewness, kurtosis, etc.) when the functional form of the utility is specified. We provide closed-form expressions for the SDF obtained when a representative agent has CARA, CRRA, and HARA utilities. In these cases, we also derive closed-form expressions for physical moments of the excess market return. Bounds are not needed. Although we derive these closed-form expressions, our bounds are for the general case when the utility function and SDF are not known.

We develop a methodology to decompose the conditional market risk premium and risk premia on higher-order moments of excess market returns into risk premia related to contingent claims on down, up, and moderate market returns. The decomposition exploits information about the risk-neutral market return distribution embedded in option prices but does not depend on assumptions about the functional form of investor preferences or about the market return distribution. The total market risk premium is highly time-varying, as are the contributions from downside, upside, and central risk. Time series variation in risk premia associated with each region is primarily driven by variation in risk prices associated with the probability of entering each region at short horizons, but it is primarily driven by variation in risk quantities at longer horizons. Analogous decompositions implied by prominent representative agent models generally fail to match the dynamic risk premium behavior implied by the data. Our results provide a set of new empirical facts regarding the drivers of conditional risk premia and identify new challenges for representative agent models.

I use a novel decomposition to estimate information and bias components from the returns implied by analyst price targets and provide evidence that prices simultaneously under-react to information and over-react to bias. Price reactions to information are permanent, and prices drift in the direction of their initial reaction for up to 12 months. Price reactions to bias are transitory, and prices reverse their initial reaction after about three months. Price reactions are relatively efficient. Approximately 85 percent of the total price reaction to information occurs during price target announcement months. Market participants are able to mostly (but not fully) debias analyst-expected returns before incorporating them into prices, with the announcement-month reaction to bias being relatively weak at about 15 percent of its reaction to information. A trading strategy analysis implies that mispricing induced by bias is only about one-third of that implied by prior research. 

Prominent factor models are based on tradable factors that do not represent theoretically relevant risks. To address this issue, we develop a factor model that captures the risks to long-term investors present in the Intertemporal CAPM (ICAPM). Empirically, we construct intertemporal risk factors as long-short portfolios based on stock exposures to dividend yield and realized variance. These tradable factors mimic news to long-term expected returns and volatility, and they offset part of the marginal utility increase in recessions induced by wealth declines. Our intertemporal factor model estimation implies significant risk prices that are consistent with the ICAPM restrictions under moderate risk aversion. Moreover, our model performs well relative to previous factor models in terms of its tangency Sharpe ratio and its pricing of key test assets, including single stocks, industry portfolios, and portfolios sorted on risk exposures and lagged anomalies.

Working Papers

We use the long-term Capital Market Assumptions of major asset managers and institutional investor consultants from 1987 to 2022 to provide three stylized facts about their subjective risk and return expectations on 19 asset classes. First, there is a strong and positive subjective risk-return tradeoff, with most of the variability in subjective expected returns due to variability in subjective risk premia (compensation for market beta) as opposed to subjective alphas. Second, belief variation and the positive risk-return tradeoff are both stronger across asset classes than across institutions. And third, the subjective expected returns of these institutions predict subsequent realized returns across asset classes and over time. Taken together, our findings imply that models with subjective beliefs should reflect a risk-return tradeoff. Additionally, accounting for this subjective risk-return tradeoff when modeling multiple asset classes is even more important than incorporating average belief distortions or belief heterogeneity in our setting.

In addition to a dominant level factor, stock market index returns contain an “idiosyncratic financial factor” (IFF) unrelated to macroeconomic aggregates. We argue the IFF contaminates tests of the risk-return tradeoff in the time series and cross section, then we reevaluate these tests using an alternative index unaffected by the IFF. Our index generates a stronger relation between its risk premium and conditional variance. It also generates larger cross-sectional variation in market betas, and these exposures explain more variation in expected returns. Our index prices size portfolios and eliminates the pricing power of size factors across many standard factor models. 

Work in Progress

"Idiosyncratic Labor Income in a Production General Equilibrium Model" (with Miguel Palacios and Lawrence Schmidt)

We develop a highly tractable, general equilibrium model with production and incomplete markets. In the model, agents can invest in physical capital and human capital, where the latter investment technology is subject to uninsurable, idiosyncratic disaster risk. The quantity of both inputs is time-varying and endogenously determined in equilibrium, subject to aggregate adjustment costs. We demonstrate that the presence of uninsurable risk has first-order implications for the riskiness of human capital; in particular, the risk premium on human capital and the share of total wealth in human capital are considerably larger and smaller, respectively, relative to the complete markets benchmark. Moreover, the presence of state-dependent, idiosyncratic risk increases the equity risk premium and has important implications for agent's optimal investment behavior.