💰 quantitative finance

Robust dividend policy: Equivalence of Epstein-Zin and Maenhout preferences

This paper proves that an investor's Epstein-Zin preferences for dividends are mathematically equivalent to a firm executive adopting a robust dividend policy under Maenhout’s ambiguity-averse preferences, resulting in a threshold-based dividend strategy.

Original authors: Kexin Chen, Kyunghyun Park, Hoi Ying Wong

Published 2026-04-27

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Kexin Chen, Kyunghyun Park, Hoi Ying Wong

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are running a successful lemonade stand. You have a certain amount of cash in your register (your "surplus"), and every day, you sell more lemonade (your "earnings"). You have a big decision to make: How much of that cash should you take home as a personal treat (dividends), and how much should you leave in the register to make sure the business stays afloat?

This academic paper is essentially a high-level mathematical guide on how to make that decision when the future is uncertain.

Here is the breakdown of the paper using everyday concepts:

1. The Two Characters: The Investor vs. The Manager

The paper looks at this lemonade stand from two different perspectives, and it turns out they are actually looking at the same problem through different lenses.

The Investor (The "Epstein-Zin" Perspective): Imagine you are a person who loves two things: having cash right now to buy things, and having a stable future. You don't just want a lot of money; you want a steady stream of it. You are sensitive to "risk"—you hate the idea of the lemonade stand going bankrupt and leaving you with nothing.
The Manager (The "Maenhout" Perspective): Now imagine you are the person running the stand. You want to pay out dividends to keep the owners happy, but you are worried about "ambiguity." Ambiguity isn't just risk; it’s the "I don't even know what I don't know" feeling. You aren't just worried that sales might be low; you're worried that your entire understanding of how many people like lemonade might be wrong.

The Big Discovery: The paper proves that these two perspectives are mathematically identical. The Investor’s desire for a "smooth, steady life" is the exact same thing as the Manager’s desire to "protect against being wrong."

2. The Strategy: The "Safety Threshold"

The paper finds that the best way to handle this is a Threshold Strategy.

Think of it like a "Full Cup" rule. You decide that you won't take any money out of the register until the cash reaches a certain level—let's say $50.

If you have $40, you leave it all in the register to protect against a rainy day (bankruptcy).
The moment you hit $50, you take everything above $50 and put it in your pocket.
If the next day you have $60, you take $10. If you have $45, you take nothing.

The paper uses complex math (called "Hamilton-Jacobi-Bellman equations") to find exactly where that "$50 line" should be drawn.

3. The "Signaling Confidence" Metaphor

This is perhaps the most interesting part for the real world. The paper suggests that the dividend policy is a secret language between the manager and the investor.

Imagine the manager says, "I think we'll make $100 tomorrow." But the manager might be lying or just guessing.

If the manager then sets a very high "Safety Threshold" (e.g., "I won't take any money out until we have $200 in the register"), they are effectively saying: "I am so confident that we will make money that I am willing to leave a huge pile of cash in the register to prove it."

By choosing a specific threshold, the manager is "signaling" their confidence. If they set the threshold too low, it looks like they are nervous and just trying to grab cash before the business fails. If they set it high, they are signaling that they are "ambiguity-averse"—they are playing it safe because they truly believe in the business's strength.

Summary in a Nutshell

The Problem: How much cash to take out of a business when the future is a mystery?
The Math: It connects how much an investor wants "stability" to how much a manager fears "being wrong."
The Solution: The best move is to wait until the business hits a specific "safety level" before taking any cash.
The Takeaway: Watching how much a company pays in dividends isn't just about the money; it's a way to read the manager's mind and see how much they actually trust their own business.

This paper, titled "Robust Dividend Policy: Equivalence of Epstein-Zin and Maenhout Preferences," provides a rigorous mathematical bridge between two distinct frameworks in financial economics: recursive utility (investor perspective) and robust optimization (executive perspective).

Below is a detailed technical summary of the work.

1. The Problem: The Dividend Signaling Puzzle

The paper addresses a fundamental tension in corporate finance: Why do firms pay dividends?

The Theory Gap: While the Miller-Modigliani theorem suggests dividend policy is irrelevant to firm value in perfect markets, empirical evidence suggests dividends serve as a "signal" of future earnings (Dividend Signaling Theory).
The Conflict of Perspectives:
- Investors want dividends to satisfy their specific utility preferences (e.g., wanting to separate risk aversion from intertemporal substitution).
- Executives face uncertainty regarding the firm's future earnings (ambiguity) and must decide on a dividend policy that maximizes value while managing the risk of bankruptcy.
The Mathematical Challenge: Existing literature often treats these two perspectives separately. The authors seek to prove that a dividend policy optimal for an ambiguity-averse executive is mathematically equivalent to a policy optimal for an Epstein-Zin (EZ) investor.

2. Methodology: Stochastic Control and BSDEs

The authors employ advanced stochastic calculus and control theory to model the firm's surplus process.

Modeling the Surplus: The firm's uncontrolled surplus is modeled as a diffusion process (SDE) with a stochastic drift and volatility. Dividends are modeled as a singular control (a non-decreasing process $D_t$ that can involve jumps or continuous flows).
Investor Framework (Epstein-Zin): They use Epstein-Zin (EZ) recursive utility, which allows for the decoupling of the coefficient of relative risk aversion from the elasticity of intertemporal substitution (EIS). This is formulated as a Backward Stochastic Differential Equation (BSDE) with a non-Lipschitz aggregator.
Executive Framework (Maenhout): The executive is modeled using Maenhout’s robust preference, which accounts for ambiguity aversion. The executive does not know the true drift of the surplus and acts to maximize the "worst-case" expected discounted dividend stream.
Mathematical Tools:
- BSDEs with Random Terminal Times: To handle the risk of bankruptcy (the random time $\tau$ when surplus hits zero).
- Hamilton-Jacobi-Bellman (HJB) Variational Inequalities (VI): To characterize the optimal threshold strategy.
- Shooting Methods: To solve the free-boundary problem associated with the optimal dividend threshold.

3. Key Contributions and Results

A. The Equivalence Theorem (The Core Result)

The paper's primary contribution is Theorem 2.10, which proves that the EZ singular control utility and the Maenhout robust dividend criterion are equivalent. Specifically, if the risk-aversion parameter in the EZ model matches the ambiguity-aversion parameter in the Maenhout model, the value functions are related by a power transformation:
$V_{EZ} = (V_{Maenhout})^{1-R}$
This proves that the "uncertainty" an executive fears is mathematically identical to the "preference structure" an investor seeks.

B. Characterization of Optimal Policy

The authors prove that the optimal dividend policy is a threshold strategy.

When the firm's surplus reaches a certain level $b^*$ , it pays out all excess cash as dividends to keep the surplus at that threshold.
The threshold $b^*$ is the "free boundary" of an HJB variational inequality.

C. Well-Posedness and Sensitivity

Existence and Uniqueness: They provide the first proof of existence and uniqueness for EZ utility in a singular control setting with bankruptcy risk, using a sequence of Lipschitz aggregators to handle the non-Lipschitz nature of the EZ aggregator.
Sensitivity Analysis: They show that as ambiguity aversion ( $R$ ) increases, the optimal dividend threshold $b^*$ and the value function $V$ change continuously. Specifically, higher ambiguity aversion leads to a different optimal payout structure.

4. Significance and Implications

Economic Significance: "Signaling Confidence"

The paper introduces the concept of "Signaling Confidence."

In traditional signaling theory, dividends signal information.
In this model, dividends signal confidence. An executive setting a specific dividend threshold is essentially communicating their level of certainty (or lack of ambiguity) regarding the firm's ability to generate future earnings.

Mathematical Significance

The paper advances the technical frontier of singular stochastic control. By successfully integrating non-Lipschitz aggregators, non-decreasing controls (singularities), and random terminal times (bankruptcy), the authors provide a robust toolkit for solving complex recursive utility problems in continuous time.

Practical Application

For investors, the model suggests that by observing a firm's dividend policy and financial statements, they can "reverse-engineer" the executive's level of ambiguity aversion, thereby identifying firms that align with their own risk/utility preferences.