What are pros and cons of extending the definition of financial index beyond conventional market capitalization (buy-and-hold) weighting? In the October 2015 draft of his paper entitled “What Is an Index?”, Andrew Lo proposes that any portfolio satisfying three properties should be considered an index: (1) transparent (public and verifiable); (2) investable (realistic and liquid benchmark); and, (3) entirely rules-based (allowing no judgment/discretion). He calls indexes that are not weighted by market capitalization dynamic indexes (requiring frequent rebalancing). He distinguishes between active investing and active risk management. He also addresses the elevated risk of snooping bias as dynamic indexes proliferate. Based on a broader perspective on indexes, he concludes that:
- Dynamic indexes may contain more subtle risks (tail, illiquidity or credit) than traditional indexes.
- Traditional linkages of active risk management with active investing and passive risk management with index investing can and should be severed. In other words, index investors should actively manage risk. For example, a simple scheme that each day weights the capitalization-weighted U.S. stock market by the ratio of a fixed volatility target of 16.9% to realized volatility over the past 21 trading days improves net annualized Sharpe ratio (assuming cash earns the one-month U.S. Treasury bills yield, and trading friction is 0.05%) from 0.36 to 0.48 during 1925 through 2014.
- The single biggest challenge in a dynamic index environment is backtest bias, whereby testing isolates not the highest actual Sharpe ratio but instead the largest favorable estimation errors (good luck). Investors in new products must, by definition, rely on simulated returns that are inherently more biased than actual returns of established products. To mitigate backtest bias:
- Be skeptical of all investment performance records, acknowledging that stellar track records include luck.
- Separate track records into market timing, security selection and other sources of value-added.
- Conduct live out-of-sample tests over reasonably long periods.
In summary, investors who do not adapt to a dynamic index environment with appropriate caution may suffer amplified, short-term consequences.
Cautions regarding conclusions include:
- Different dynamic indexes may have different costs of implementation due to disparate turnovers and liquidity tendencies. Differences between net and gross index performance will be greater for dynamic indexes than for traditional indexes.
- Management fees for new dynamic index tracking funds may be higher than those of traditional index funds.
- In the volatility weighting example, assumptions regarding speed of data collection/data processing/trade execution and costs of trading/leverage appear extremely optimistic for most of the sample period. Availability of speedy implementation and low costs early in the sample period may have altered market behavior. Results for recent subperiods are mixed.
- Discussion and examples are equity-centric.
