The Mean-Variance Anatomy of Behavioral Portfolios. (with Roméo Tédongap)
Abstract: We study portfolio choice under generalized disappointment aversion in a static, modelfree, multi-asset setting with short horizons. Despite nonlinear and kinked preferences, optimal portfolios admit a closed-form mean-variance representation with endogenously distorted moments. Risk-taking decomposes into standard and downside-driven components, revealing that disappointment aversion increases risk avoidance with horizon—reversing standard predictions. Ignoring downside asymmetries leads to a substantial welfare loss. Our framework delivers tractable portfolio design under behavioral preferences, offers testable implications, and highlights how psychological frictions reshape risk attitudes at short horizons before gradually fading as Sharpe ratios, skewness, and kurtosis regain dominance over longer horizons.
International Asset Pricing with Heterogeneous Agents: Estimation and Inference, Journal of Empirical Finance, Volume 75, January 2024. (with Roméo Tédongap)
Abstract: This paper empirically validates (Constantinides and Ghosh’s, 2017) heterogeneous-agents consumption-based asset pricing model for predicting expected returns in international equity markets. Using the model’s implications, we proxy the unobservable state variable driving income shocks with the principal component of consumption growth cumulants across agents. We confirm that both the level and changes in this cross-sectional consumption risk serve as pricing factors, emphasizing the importance of higher moments like skewness. The estimated structural parameters obtained from the Euler equations are statistically significant and plausible, while the factor risk premium estimates align with theoretical expectations. Our approach effectively explains the emerging versus developed premium, outperforming traditional methods reliant on cross-sectional variance. Our findings, robust across different model specifications and asset menus, highlight the imprecision of consumption-based factor risk premia estimates when limited to developed markets, a limitation mitigated by including emerging markets. The model demonstrates a 60% explanatory power, surpassing the global Fama–French model.
Portfolio Optimization and Asset Pricing Implications Under Returns Non-Normality Concerns, Finance, Volume 43, Issue 1, January 2022. (with Roméo Tédongap)
Abstract: We investigate the implications of non-normality for asset allocation and pricing. Asset returns non-normality is captured through a multivariate normal-exponential model; we develop an estimation procedure based on a generalized method of moments. Investors' nonnormality concerns are introduced by adding a linear non-normality constraint to an otherwise standard mean-variance framework. The optimal portfolio solution is obtained in closed form and can be reformulated as a three-fund separation strategy. Suboptimal portfolios that ignore non-normality or are naive in terms of diversification may result in important welfare costs as measured by the certainty equivalent, notably for the most risk-tolerant investors who target large non-normality ratios. In equilibrium, expected returns admit a two-beta representation in which the most important beta in explaining their cross-sectional variation is the one capturing non-normality (more than 60%) while the CAPM beta explains less than 12%.
Consumer Heterogeneity, Firm Characteristics, and Risk Exposures in Cross-Sectional Asset Pricing. (with Samuël Nelemans)
Abstract: We extend the novel approach to measuring firm-level risk exposures and costs of equity developed by Dittmar and Lundblad (2017) in a representative consumer setting to a heterogeneous consumer framework. Our measure of firm-level risk exposures is multidimensional and relates to higher-order cross-sectional moments of households idiosyncratic consumption growth shocks. We find that our multi-factor consumption-based asset pricing model better explains the variation in average returns across anomaly portfolios than the single-factor consumption-based model, and the firm-level risk exposures to cross-sectional consumption growth moments are related to the business cycles. The use of our model to estimate the cost of equity shows that it generates more variation in the cost of equity across assets compared to the single-factor model, and it better keeps track with the business cycles.
Learning from the Wisdom of Mutual Fund Managers (with Roméo Tédongap)
Abstract: We define Stock Active Share (SAS) as the degree to which a stock in a benchmark index is actively weighted by mutual funds relative to its index weight. We analyze the risk-return characteristics of portfolios ranked by SAS values, finding that the top quantile portfolio delivers significant monthly risk-adjusted returns, highlighting mutual fund managers' capital allocation proficiency. However, due to the delayed disclosure of fund holdings, SAS is unobservable in real time, making the strategy unfeasible for typical investors. To address this, we apply machine learning models to historical fund holdings and stock characteristics to predict future SAS and sort portfolios accordingly. These models demonstrate substantial out-of-sample accuracy, and the feasible top quantile portfolio consistently outperforms the benchmark across risk-adjusted measures. Our findings illustrate the enduring value of fund managers' stock-picking skills, challenging the view that technological advancements diminish their importance. Furthermore, the feasible strategy not only outperforms traditional analyst recommendations but also aligns with sustainability goals by favoring stocks with lower carbon intensity.
Macro Uncertainty and the Term Structure of Risk Premium (Online appendix)
Abstract: Leading frictionless consumption-based asset pricing models (Long run risks and Habit formation) predict that the expected return on assets whose cash flows appear in the distant future are higher than or equal to the expected returns on assets which pay-off in the near future. Contrary to that prediction, some recent empirical studies have found that short-term assets earn a higher expected return than long-term assets. Here, I show that allowing the cash flows to be negatively affected by volatility shocks, as observed in the data (“leverage effect”), could make the short-term assets riskier than long-term assets. This modification gives more flexibility to those models in capturing various shapes of the term structure of equity returns while still matching the observed level of the equity premium and the risk free rate.
GMM estimation of the Long Run Risks model (with Nour Meddahi)
Abstract: In this paper, we propose a GMM estimation of the structural parameters of the Long Run Risk model that allows for the separation between the consumer optimal decision’s frequency and the frequency by which the econometrician observes the data. Our inference procedure is robust to weak identification. We find that the Long Run Risk model is not rejected in the data and using the estimated parameters to simulate the model enables us to improve some quantitative predictions of the model. However, we show that the commonly used statistical inference methods such as the bootstrap (parametric or block bootstrap) might be misleading in providing parameters' confidence intervals since they imply an under-coverage of the true confidence interval.