Return Smoothing and its Implications for Performance Analysis of Hedge Funds

Jing‐Zhi Huang, John Liechty, Marco RossiSSRN 2009 · DOI

TL;DR

Hedge funds often report smoothed returns due to illiquid assets, which can significantly bias performance evaluations. This paper introduces a Bayesian framework to simultaneously account for return smoothing, time-varying factor loadings, and the short-lived nature of hedge fund histories, revealing that smoothing leads to an overestimation of hedge fund alpha and an underestimation of risk.

What Problem Does This Paper Address?

Evaluating the performance of hedge funds is notoriously tricky. Unlike traditional mutual funds, hedge funds often invest in illiquid assets, meaning their reported returns don't always reflect the true, underlying economic value in real-time. This phenomenon, known as "return smoothing," can artificially lower volatility and inflate performance metrics like alpha (risk-adjusted excess return). Furthermore, hedge funds can have dynamic strategies with time-varying risk exposures and often have relatively short histories, making traditional performance evaluation methods insufficient. This paper aims to address these challenges by developing a comprehensive framework that accounts for these complex features simultaneously.

Methods

The authors propose a dynamic linear Bayesian model to evaluate hedge fund performance. Key features of their approach include:

  • Bayesian Framework: This allows for incorporating prior beliefs and provides a richer understanding of parameter uncertainty, especially useful for short-lived funds.
  • Simultaneous Estimation: The model estimates return smoothing, time-varying factor loadings (betas), and time-varying alphas concurrently. This is a significant departure from previous methods that often treated these features in isolation or assumed them constant.
  • Return Smoothing Model: They model observed returns as a weighted average of current and lagged true economic returns, with the weights representing the degree of smoothing.
  • Time-Varying Coefficients: The model allows for alphas and betas to change over time, reflecting the dynamic nature of hedge fund strategies.
  • Shrinkage: To address the issue of short histories and improve efficiency, the model incorporates shrinkage, where individual fund parameters are pulled towards their category averages.

Datasets

The study uses data from the Center for International Securities and Derivatives Markets (CISDM) hedge fund database, covering the period January 1994 to December 2005. They filter the data to address common biases like incubation and survivorship, and to ensure statistical accuracy, they include funds with at least 24 consecutive monthly observations and exclude those with extreme returns. The final sample consists of 2,162 hedge funds across 21 categories.

Key Insights and Ideas

  • Smoothing is a Major Culprit: The paper strongly suggests that return smoothing is the most significant factor distorting hedge fund performance evaluations, more so than time-varying coefficients or shrinkage.
  • Upward Bias in Alpha, Downward Bias in Risk: Smoothing leads to an overestimation of a fund's alpha (abnormal performance) and an underestimation of its risk (volatility and factor loadings).
  • Illiquidity and Smoothing are Linked: The degree of smoothing is positively related to the illiquidity of a fund's strategy, which is intuitive as less liquid assets take longer to reflect true value.
  • Even "Liquid" Strategies are Smoothed: The study finds significant smoothing even in equity hedge funds, which are generally considered more liquid than other hedge fund strategies.

Key Results

  • A moderate level of smoothing can cause standard OLS alpha estimates to over-estimate equity funds' abnormal performance by more than 1% annually.
  • For equity hedge funds, smoothing causes an upward bias in excess performance measures (alpha) and a downward bias in risk measures (betas).
  • The Bayesian model, which accounts for smoothing, time-variation, and shrinkage, provides a more accurate assessment of hedge fund risks and rewards compared to simpler models.
  • The study demonstrates that ignoring smoothing can lead to substantial misinterpretations of hedge fund performance, even for strategies that appear relatively liquid.

Limitations and Future Work

  • Optimal Lag Structure: The paper acknowledges that the optimal lag structure for modeling smoothing was not investigated and suggests this as a direction for future research.
  • Fund-of-Funds Strategies: While the framework could be used for fund-of-funds trading strategies, this was beyond the scope of the current study.
  • Data Limitations: The authors note that for fixed income funds, the smaller sample size and fewer funds with low smoothing made it harder to reliably estimate the impact of smoothing on measured performance.
  • Managerial Discretion: The authors anticipate that the impact of smoothing might be even more pronounced for more illiquid strategies due to the discretion and difficulty managers face in marking positions to market.

Summary updated

BibTeX
@misc{huang2009return,
  author = {Jing‐Zhi Huang and John Liechty and Marco Rossi},
  title = {{Return Smoothing and its Implications for Performance Analysis of Hedge Funds}},
  year = {2009},
  doi = {10.2139/ssrn.1363957},
  howpublished = {SSRN},
}

Loading related content...

Figure