Selected Presentations and Press Attention 


Published Papers


Working Papers

Abstract: We introduce non-myopic investors into the standard conditional Capital Asset Pricing Model. In equilibrium, the intertemporal hedging demand of non-myopic investors leads to a two-factor CAPM in which risk premiums are determined both by the market (myopic) beta and by the “non-myopic beta,” with respect to the future return on the mean-variance efficient portfolio. We identify this efficient portfolio non-parametrically as a solution to a fixed-point problem, and use it to estimate the non-myopic betas. We show that non-myopic betas are indeed priced in the cross-section of stock returns, and the relationship between expected returns and non-myopic betas is monotone increasing and economically significant. Using U.S. mutual fund data, we find that non-myopic betas of mutual fund returns are negatively related to their long-term Sharpe ratios, in agreement with theoretical predictions. In the presence of funding constraints, our model predicts that a low non-myopic beta is associated with a higher alpha. We confirm this prediction by constructing a “Betting Against Non-Myopic Beta” factor and showing that it generates superior performance over and above a number of factor models.

Abstract: Alternative assets, such as private equity, hedge funds, and real assets, are illiquid and opaque, thus posing a challenge to traditional models of asset allocation. In this paper, we study asset allocation and asset pricing in a general-equilibrium model with liquid assets and an alternative risky asset, which is opaque and incurs transaction costs, and investors who differ in their experience in assessing the alternative asset. We find that the optimal asset-allocation strategy of the relatively inexperienced investors is to initially tilt their portfolio away from the alternative asset and to hold more of it with experience. Counterintuitively, a decrease in the transaction cost for the alternative asset increases the portfolio tilt at the initial date, and hence, the liquidity discount. Transaction costs may induce inexperienced investors to hold a majority of the illiquid asset at later dates, even if they are pessimistic about future payoffs, and produce a sizable liquidity discount. During periods when the alternative asset is illiquid, investors trade the liquid equity index instead, leading to strong spillover effects.

Abstract: In this paper, we study the effect of proportional transaction costs on consumption-portfolio decisions and asset prices in a dynamic general equilibrium economy with a financial market that has a single-period bond and two risky stocks, one of which incurs the transaction cost. Our model has multiple investors with stochastic labor income, heterogeneous beliefs, and heterogeneous Epstein-Zin-Weil utility functions. The transaction cost gives rise to endogenous variations in liquidity. We show how equilibrium in this incomplete-markets economy can be characterized and solved for in a recursive fashion. We have three main findings. One, costs for trading a stock lead to a substantial reduction in the trading volume of that stock, but have only a small effect on the trading volume of the other stock and the bond. Two, even in the presence of stochastic labor income and heterogeneous beliefs, transaction costs have only a small effect on the consumption decisions of investors, and hence, on equity risk premia and the liquidity premium. Three, the effects of transaction costs on quantities such as the liquidity premium are overestimated in partial equilibrium relative to general equilibrium.

Abstract: We show how to extract the expected risk-neutral correlation between risk-neutral distributions of the market index (S&P 500) return and its expected volatility (VIX). Comparing the implied correlation with its realized counterpart reveals a significant index-to-volatility correlation risk premium. It compensates for the fear of rising and enduring volatility due to market crashes and measures a new dimension of risk not covered by other variables. The correlation risk premium asymmetrically focuses on tail risk, unlike the variance risk premium. Incorporating information from both equity and volatility markets, it predicts future index returns and changes in both future returns and volatilities.

Abstract: Motivated by extensive evidence that stock-return correlations are stochastic, we analyze whether the risk of correlation changes (affecting diversification benefits) may be priced. We propose a direct and intuitive test by comparing option-implied correlations between stock returns (obtained by combining index option prices with prices of options on all index components) with realized correlations. Our parsimonious model shows that the substantial gap between average implied (38% for S&P500 and 44% for DJ30) and realized correlations (31% and 34%, respectively) is direct evidence of a large negative correlation risk premium. Empirical implementation of our model also indicates that the index variance risk premium can be attributed to the high price of correlation risk. Finally, we provide evidence that option-implied correlations have remarkable predictive power for future stock market returns.

Abstract: We study whether option-implied conditional expectation of market loss due to tail events, or tail loss measure, contains information about future returns, especially the negative ones. Our tail loss measure predicts future market returns, magnitude and probability of the market crashes, beyond and above other option-implied variables. Stock-specific tail loss measure predicts individual expected returns and magnitude of realized stock-specific crashes in the cross-section of stocks. Investor, especially the one caring about the left tail of her wealth distribution (e.g., disappointment-averse), benefits from using the tail loss measure as an information variable to construct managed portfolios of risk-free asset and market index. The tail loss measure is motivated by the results of the extreme value theory, and it is computed from observed prices of out-of-the-money put as the risk-neutral expected value of a loss beyond a given relative threshold.

Abstract: Our objective in this paper is to examine whether one can use option-implied information to improve the selection of mean-variance portfolios with a large number of stocks, and to document which aspects of option-implied information are most useful for improving their out-of-sample performance. Portfolio performance is measured in terms of four metrics: volatility, Sharpe ratio, certainty-equivalent return, and turnover. Our empirical evidence shows that using option-implied volatility helps to reduce portfolio volatility; option-implied correlation does not improve any of the metrics; and, expected returns estimated using information in option-implied volatility and option-implied skewness increase substantially both the Sharpe ratio and certainty-equivalent return, even after prohibiting shortsales and accounting for transactions costs.

 Abstract: Using data on all U.S. exchange-traded individual stock options, we show that the currently observed option-implied ex ante skewness is positively related to future stock returns. This contrasts with the existing evidence that uses historical stock or option data to estimate skewness and finds a negative skewness-return relation. We compute the ex ante skewness using the model-free implied skewness (MFIS) as in Bakshi, Kapadia, and Madan (2003) and show that high MFIS stocks outperform low MFIS stocks by 45 basis points per month after correcting for systematic exposure. We find that the positive MFIS-return relation stems from the ability of the current MFIS to identify the deviation of a firm’s value from its fundamental value, and the most overvalued stocks have the most negative ex ante skewness. We further find that the speed of the value correction process depends on the arbitrage risk faced by arbitrageurs trying to exploit the observed inefficiencies. Our results have implications for the segmentation of equity and options markets as well as for limits of arbitrage in equity markets.


Computer Science/ IT