Abstract
This paper investigates a continuous-time intertemporal consumption and portfolio choice problem for a robust investor utilizing Epstein-Zin type stochastic differential utility (SDU). Addressing the risk of model misspecification, the authors provide a verification theorem for the Hamilton-Jacobi-Bellman-Isaacs (HJBI) equation under a non-Lipschitz condition. This framework allows for the derivation of explicit closed-form solutions for optimal robust consumption and portfolio strategies within a Heston volatility model. Numerical comparisons demonstrate that robust decision rules align with financial intuition, showing increased conservatism in the face of uncertainty.
Introduction
In classical financial modeling, investors are often assumed to have perfect knowledge of the underlying probability distributions governing asset returns. However, in practice, these models are merely approximations. Model misspecification—the risk that the chosen model does not accurately reflect reality—can lead to suboptimal investment and consumption decisions.
In "Robust Consumption Portfolio Optimization with Stochastic Differential Utility," Jiangyan Pu and Qi Zhang address this challenge by developing a rigorous mathematical framework for the "robust" investor. This investor does not take their baseline model at face value; instead, they seek strategies that perform well even if the model is slightly inaccurate, effectively planning for a "worst-case" scenario within a defined neighborhood of uncertainty.
Theoretical Framework: SDU and Robustness
A key feature of this research is the use of Stochastic Differential Utility (SDU) of the Epstein-Zin type. Standard utility functions often conflate two distinct psychological dimensions: risk aversion (an investor's dislike of uncertainty) and the elasticity of intertemporal substitution (EIS) (an investor's willingness to shift consumption across time). By employing SDU, the authors decouple these traits, allowing for a more nuanced analysis of investor behavior.
To incorporate robustness, the problem is framed as a Stochastic Differential Game. In this zero-sum setup, the investor acts as the "maximizing player," attempting to optimize their utility, while a hypothetical "Nature" acts as a "minimizing player," introducing the most damaging possible distortions to the model. The investor’s goal is to find a strategy that is resilient against these adversarial distortions.
Mathematical Contributions and the Verification Theorem
The integration of SDU with robust control theory introduces significant mathematical complexity, specifically resulting in non-linear partial differential equations (PDEs) that do not satisfy standard "smoothness" or Lipschitz conditions.
The HJBI Equation
The authors' primary technical contribution is the formulation of a Verification Theorem for the Hamilton-Jacobi-Bellman-Isaacs (HJBI) equation under a non-Lipschitz condition. This theorem provides the necessary mathematical justification to ensure that the solutions derived from their PDEs are indeed the optimal strategies for the investor. To achieve this, they utilized a Stochastic Gronwall-Bellman inequality to manage the monotonicity of the utility aggregator, extending existing proofs to accommodate a two-control game between the investor and Nature.
Solving the Heston Model
The Heston volatility model is a staple in finance due to its ability to capture the "volatility clustering" observed in real markets, but it is notoriously difficult to solve analytically. The authors achieved a breakthrough by applying a technique known as Maenhout’s approach, where the robustness parameter is scaled by the value function. This scaling allows the complex non-linearities in the HJBI equation to cancel out, reducing the problem to a system of solvable linear ordinary differential equations (ODEs).
Key Insights and Results
Through their theoretical derivation and subsequent numerical simulations, the authors identify several critical behaviors of robust investors:
- Explicit Optimal Strategies: The paper provides closed-form formulas for the optimal stock allocation ($\pi^*$) and the consumption-to-wealth ratio ($c/X$).
- Increased Conservatism: Robust investors consistently allocate a smaller proportion of their wealth to both consumption and risky stocks compared to non-robust investors. By accounting for potential model errors, they maintain higher levels of "dry powder" in riskless assets.
- Sensitivity to Distrust: While the optimal stock allocation is relatively insensitive to initial volatility, it is highly sensitive to the investor’s level of distrust in the model (the robustness parameter). As the robustness parameter increases, the investor acts as if expected returns are lower and market volatility is higher than the baseline model suggests.
- Validation of Intuition: The numerical results confirm that robustness acts as a form of mathematical pessimism, leading to more cautious financial behavior that aligns with common-sense risk management.
Conclusion and Future Directions
This research bridges a significant gap in mathematical finance by proving that dynamic programming remains a viable approach for SDU even when the utility aggregator lacks perfect smoothness. However, the authors note several avenues for future exploration. The current model assumes a "complete" market where all risks can be hedged; extending these findings to incomplete markets with trading constraints remains a challenge. Additionally, the authors suggest investigating whether the penalty term for model uncertainty could be formulated as a recursive utility itself, potentially deepening the integration of robustness and utility theory.
Summary updated