Factor Modeling and Benchmarking of Hedge Funds, by Lars Jaeger and Christian Wagner

  • Beware of claims of alpha—alternative beta often mis-measured as alpha, as models separating beta from alpha are imperfectly specified.
  • Replicating Factor Strategies (RFS):
    • Long/ Short Equity: 1) convertible bond (models non-linear exposure); 2) small cap spread (Russell 2000 minus Russell 1000); 3) S&P 500 CPPI (Constant Proportion Portfolio Insurance, models a capital protection exposure). RFS R2 = 90%. Rapid alpha degradation.
    • Equity Market Neutral: 1) Fama-French momentum factor; 2) value spread (MSCI Value minus MSCI Growth). RFS R2 = 35%. Evidence of alpha persistence, low explanatory power of RFS.
    • Short Selling: 1) negative convertible bond (models negative, non-linear exposure); 2) value spread; 3) small cap spread. RFS R2 = 80%. Significant alpha likely due to alpha opportunities on the short side; however, handicapped by negative market beta.
    • Event-Driven: 1) broad market (S&P 500); 2) small cap spread; 3) high-yield bond (CSFB High Yield); 4) AR(1) (1-peiord lagged autocorrelation, models illiquidity and possible lagged pricing). RFS R2 = 80%. Highest alpha of all hedge fund strategies.
    • Distressed: 1) AR(1) (models high illiquidity and stale pricing); 2) broad market; 3) small cap spread; 4) high-yield bond. RFS R2 = 70%. High alpha strategy.
    • Merger Arbitrage: 1) short put (BXM Covered Call Writing Index, models limited upside, substantial downside short put exposure); 2) small-cap spread; 3) value spread. RFS R2 = 50%.
    • Fixed Income Arbitrage: 1) convertible bond; 2) emerging market s bond; 3) AR(1); 4) credit spread (BB minus AAA). RFS R2 = 40%.
    • Convertible Arbitrage: 1) AR(1); 2) convertible bond; 3) high-yield bond; 4) equity (S&P 500, models equity market and equity volatility exposures). RFS R2 = 60%. Low explanatory power due to two different types of convertible arb: gamma trading (credit-hedged), and credit-oriented.
    • Global Macro: 1) world bond (models global fixed income and currencies exposures); 2) convertible bond; 3) trend-following (sGFI, models trend-following momentum strategy); 4) emerging markets equity (models global/ emerging markets exposure). RFS R2 = 50%. Low explanatory power due to heterogeneous strategies.
    • Managed Futures: 1) sGFI; 2) world bond (models currencies exposure); 3) commodities. RFS R2 = 35%. Only hedge fund strategy with negative alpha.
  • The stability of the factor replication model is tested using a CUSUM test and rolling correlations. Results indicate model is stable.
  • RFS outperforms in all hedge fund categories (by a wide margin) except in distressed debt. Investable indexes (such as HFRX) are inadequate as benchmarks or investment vehicles, as they perform much worse than RFS and non-investable indexes (such as HFRI). However, their risk characteristics (risk factor betas) are retained, while the alpha is not.
  • Estimate that 80% of returns are due to beta exposures and 20% due to alpha (or not yet identified systematic risk factors).
  • Average alpha seen to degrade over time, likely caused by 1) higher competition for limited alpha opportunities, and/ or 2) increasing less skilled managers due to low barrier of entry.
  • Alternative betas, rather than alpha, should become increasingly central to hedge fund investing.

Finished: 5-Sep-08