Global X Research TeamMuch of the attention around smart beta or factor-based investment strategies has focused on the potential for outsized returns by accessing historically well-rewarded factors such as value, momentum, low volatility, and size. However an equally important part of the equation when analyzing these strategies is their risks and how they (may or may not) seek to mitigate those risks through diversification. Diversification is often considered one of the few free lunches in finance, because when properly implemented it can reduce risks without necessarily sacrificing returns. Scientific Beta, the indexing arm of the EDHEC-Risk Institute, has taken a unique approach to smart beta by focusing equally on harvesting outperformance from well-rewarded factors, while maintaining a well-diversified portfolio to potentially improve risk-adjusted returns.
It’s easy for ETF users to take diversification for granted. After all, one of the core benefits of ETFs is that they allow investors to access a portfolio of hundreds or even thousands of stocks in a single trade. But while market cap weighted ETFs, such as ones that provide exposure to the S&P 500, are significantly more diversified than buying a single stock, they can still suffer from concentration issues. For example, the top 10 holdings in the S&P 500 make up nearly 18% of the portfolio’s weight1. The smallest 295 stocks in the S&P 500 make up the same 18% weighting, which poses an important question: although a broad market cap weighted index can provide access to hundreds of stocks, is it an optimally diversified portfolio if it is so top-heavy?
Academics often address this question by calculating the effective number of stocks in a portfolio, based on the Herfindahl Index’s2 measure of concentration. The effective number of stocks equates a particular index’s concentration to that of an equal weighted index. For example, the S&P 500, despite holding 500 companies, has an effective number of stocks of 1423. This means that the S&P 500 has the same level of concentration as an equal weighted index with 142 components. Therefore according to this measure the diversification, 358 stocks, or over 70% of the index, are wasted due to the top heavy weighting scheme.
In addition to the concentration issues related to cap-weighted indexes, there is a potential performance setback as well: market cap weighted indexes inherently provide more exposure to large cap companies and growth stocks, since these companies tend to have higher market capitalizations than small caps or value companies. Fama and French, in their seminal factor research, however, showed that small caps and value companies tend to outperform large caps and growth companies. Therefore constructing an index that has better diversification characteristics not only holds the potential to reduce concentration risks, but also improve performance by moving away from the concentrated large cap growth names in a market cap weighted benchmark. The S&P 500 Equal Weight Index demonstrates this effect: by equal-weighting its components it takes a simple approach of de-concentration and shifts its exposures away from the large cap growth names that dominate the regular market cap-weighted S&P 500 index. Since 2006, the equal weight index has outperformed the regular S&P 500 by an annualized 112 basis points, largely attributable to the inherent size and value tilt of the index4.
Unfortunately, many smart beta or factor indexes are more focused on maximizing factor exposures and reaping their potential rewards than diversifying away idiosyncratic risks associated with concentration. In some instances, this can ultimately lead to even greater concentration issues than a cap weighted benchmark. For example, while the S&P 500 has 18% of its weight in the top 10 holdings, the S&P 500 Value Index, by nature of being a smaller subset of its parent index, has over 24% of its weight in the top 10 holdings and less than 100 effective components5.
Scientific Beta believes that chasing factor performance at the expense of increasing the idiosyncratic risks in a portfolio is not a worthy tradeoff. Instead, they believe that a well-constructed smart beta strategy should both filter for stocks that exhibit well-rewarded factors while improving upon the concentration issues of a cap weighted index, ideally resulting in a much improved Sharpe ratio6.
To incorporate this risk-adjusted approach into their index construction process, they developed a methodology grounded on a few key principles:
- Select stocks based on their factor exposures, but use weighting schemes to effectively diversify the components
- While there are a variety of weighting schemes that can achieve enhanced diversification over a cap-weighted benchmark, each has inherent model-specific risks, such as relying on historical data or assumptions. They may also have incidental factor tilts or sector biases.
- Therefore incorporating multiple weighting strategies simultaneously is necessary in order to both properly diversify stocks as well as to effectively mitigate the individual risks and biases associated with each weighting scheme
Based on these principles, Scientific Beta developed a process whereby a stock’s final weight in the Scientific Beta United States Multi-Strategy Index is the average of 5 different weighting schemes:
Maximum De-concentration: similar to an equal weighted methodology, it seeks to maximize the effective number of stocks.
Diversified Risk Weighted: similar to a risk-parity weighting scheme, it seeks to equalize the volatility contribution of each stock in the index.
Maximum Decorrelation: seeks to reduce the volatility of an index by tilting weight towards stocks with lower correlations to the other components.
Efficient Minimum Volatility: seeks to minimize the overall volatility of a portfolio by considering both the volatility and correlations of its components.
Efficient Maximum Sharpe Ratio: seeks to maximize the portfolio’s risk-adjusted returns, based on each component’s volatility, correlations, and expected returns.
While it may seem excessive to depend on five different weighting schemes, consider the drawbacks of using only one approach. The Efficient Minimum Volatility strategy, for example, over-weights stocks with lower volatility and low correlations to the rest of the portfolio. In order to do so, it depends on historical estimates which could change in the future which introduces model risks into the methodology (i.e., historical volatility and correlations may not continue in the future). The strategy also results in a tilt towards the low volatility factor, which can reduce the overall portfolio’s factor-level diversification. Last, it often suffers from sector concentration in certain stocks like utilities. Therefore combining Efficient Minimum Volatility with other strategies, which have their own model risks, factor tilts, and sector biases, can ultimately mitigate the risks and biases inherent in each weighting scheme.
Since 2013, the diversification-focused Scientific Beta United States Multi-Strategy Index has outperformed the S&P 500 by 107 basis points and with 56 basis points lower volatility, showing that diversification can potentially both enhance performance by reducing large cap growth concentrations while reducing volatility. Adding a multi-factor stock selection element to this weighting strategy further increased returns by an additional 99 basis points, for a total of 206 basis points of outperformance versus the S&P. These results demonstrate the additional return potential from factor-based investing7. We believe that Scientific Beta’s approach, which marries a diversification-minded weighting scheme with factor-based stock selection, is a superior smart beta strategy for investors’ core equity allocations because of its focus on improving risk-adjusted returns.