How Smart is ‘Smart Beta’?

The efficiency of alternative index approaches

Worldwide, investors are increasingly keen on “smart beta” investing. By this we mean passively following an index in which stock weights are not proportional to their market capitalisations, but based on some alternative weighting scheme. Well-known examples of smart beta include fundamentally weighted and minimum-volatility indices.

In this article, we first take a critical look at the pros and cons of smart beta investing in general. After this, we discuss the most popular types of smart indices that have been introduced in recent years. The added value of smart beta indices has come from systematic tilts towards classic factor premia that are induced by their weighting schemes. We will argue that investors should be aware of the potential pitfalls of smart beta indices, which arise because they are not specifically designed for harvesting factor premia in the most efficient manner, but primarily for simplicity and appeal. Although passive management can be used to replicate smart indices, it is important for investors to realise that, without exception, smart indices themselves represent active strategies.

Smart beta investing in general
The argument which is typically used to motivate smart beta investing is that the capitalisation-weighted index is inefficient, and that a more efficient portfolio can be constructed by applying some alternative stock weighting scheme. Investors should understand, however, where the added value of such weighting schemes really comes from. Research has shown that the weighting schemes used by alternative indices result in structural tilts towards stocks which score high (or low) on certain factors, and that the premia which are known to be associated with these factors are driving performance.¹ For example, when compared to the capitalisation-weighted index, fundamental indices have a systematic tilt towards value stocks. These exposures enable the strategy to benefit from the well-known value premium, which, in fact, turns out to fully explain its performance. Similarly, a minimum-volatility index captures the low-volatility premium by tilting the portfolio towards low-volatility stocks. Although this may seem obvious to some, many smart beta index providers are still reluctant to acknowledge factor exposures drive their performance, and that their weighting schemes are merely a novel way of establishing exposures towards classic factor premia.

We are often asked whether smart beta investing is a form of passive investing. It is important to realise that it is not. Although passive management can be used to replicate smart indices, smart indices themselves are essentially active strategies. The only truly passive investment strategy is the capitalisation-weighted broad market portfolio, which represents the only buy-and-hold portfolio that could, be held in equilibrium by every investor. Smart beta indices are fundamentally different because they require various subjective assumptions and choices. Their active nature is also illustrated by the fact that they require periodic rebalancing to maintain their profile. Smart beta indices may bear some resemblance to true passive investing, for example by investing in a large number of stocks with relatively low turnover, but their deviations from the capitalisation-weighted index, which are the key to their added value, represent active investment decisions.

Smart beta investing is a way to tilt a portfolio actively towards certain factor premia. As we are proponents of factor investing, this makes smart beta investing a potentially promising investment approach. For example, in a recent paper we argued that equity investors should strategically allocate a sizable part of their portfolio to the value, momentum and low-volatility factor premia.² Smart beta investing represents one way in which this could be implemented.

Our view on smart beta investing can be summarised as follows: although smart beta investing may be a good start,investors can do better. The reason is that the main appeal of smart beta indices, namely their simplicity, is at the same time their biggest weakness. Specifically, the simple tilts towards factor premia provided by smart beta indices often involve significant risks that are undesirable. In addition, smart beta strategies can be inefficient from a turnover perspective, or can have unattractive exposures to factor premia other than the one(s) specifically targeted.

Another concern with smart beta indices is that they are often based on back-tests which only go back 10 or 15 years in time. Investors should therefore be careful to avoid chasing recent performance. To properly understand the behaviour of a smart index in different environments, we recommend analysing its performance over long historical periods, covering multiple economic cycles. Investors should also carefully think about whether the factor premia driving historical smart beta index returns are likely to persist in the future.

In the following sections we will elaborate on these points by discussing the pros and cons of the most popular types of smart beta indices.

Fundamental indices
In a fundamental index, stocks are weighted in proportion to their fundamentals, such as book value or earnings. In other words, instead of letting the market decide on a stock’s appropriate weight, one might say that fundamental index investors prefer to rely on the assessment of accountants. The differences in weights between a traditional, capitalisation-weighted index and a fundamental index are, by definition, entirely due to differences in valuation ratios of individual stocks. Compared to the capitalisation-weighted index, a fundamental index is tilted towards stocks which are cheap on such ratios, i.e. value stocks. Studies have shown that the added value of fundamental indices is, in fact, entirely attributable to this tilt towards the value premium.³ For a long time, Research Affiliates, the inventors of fundamental indexation, denied that the success of fundamental indexation is critically dependent on a value premium. They argued that random mispricing causes capitalisation-weighted indices to be biased towards overvalued stocks, resulting in a structural drag on performance.⁴ Nowadays, however, Research Affiliates acknowledges that the value premium explains most, or all of their indices’ performance.⁵

Our main concern with straightforward value strategies such as fundamental indexation is that they tilt towards financially distressed firms. The share price of a company in financial difficulty falls, and its weight in the cap-weighted index drops correspondingly. Initially, the same happens in a fundamental index. At a certain point, however, a fundamental index rebalances back to the weight based on past and current fundamentals, which have typically not (or only partly) adapted to do the new situation. This exposure to distressed firms might not be a problem if distress risk is the source of the value premium. Studies have shown, however, that the stocks of companies in difficulty underperform and that the tilt to distressed firms of naïve value strategies increases risk.⁶ This implies that the value premium can be captured more efficiently by avoiding cheap stocks of financially distressed firms.

A related concern is that, since rebalancing involves buying stocks which have recently experienced a large price drop, fundamental indices go against the momentum premium. As the momentum premium is as strong as the value premium, the return of a value strategy may be enhanced by avoiding its natural tendency of going against the momentum premium.

Another concern with fundamental indices is their sensitivity to settings choices. For example, in certain calendar years, the arbitrary choice of the annual rebalancing moment of the FTSE/RAFI fundamental indices can make the difference between an outperformance of 10 percent or a small underperformance.⁷ The more recently launched fundamental indices of MSCI, called MSCI Value Weighted indices, address this concern by rebalancing every six months, while those of Russell rebalance a quarter of the portfolio every quarter. In light of these developments, FTSE has decided to provide a staggered quarterly rebalanced variant of the FTSE/RAFI indices in 2013, although these will not replace their current indices but will coexist with them.

Fundamental indices represent a low-conviction approach to capturing the value premium. To understand this, note that a fundamental index is not concentrated in stocks with the most attractive valuation characteristics. For example, the FTSE/RAFI US and Developed ex-US indices each invest in 1,000 stocks, and the MSCI Value Weighted indices invest in all the stocks in the regular MSCI indices. In other words, stocks with the least attractive valuations are still included in these indices, only with smaller weights.

Low-Volatility indices
Low-volatility indices are designed to benefit from the low-volatility premium: the empirical finding that low-risk stocks have similar or better returns than the market average, with substantially lower risk. Minimum-volatility indices use optimisation techniques to create a portfolio with the lowest expected future volatility. The resulting portfolio consists mainly of stocks with low past volatility, although it may also include some higher-volatility stocks if these help to reduce volatility through low correlations. A drawback of optimised low-volatility indices is their lack of transparency. For example, the most popular minimum-volatility index, the one provided by MSCI, uses the proprietary Barra risk model and optimisation algorithm, and many investors regard the index to be a ‘black box.’ Another concern is that the raw turnover of minimum-volatility strategies is very high. MSCI addresses this concern by imposing turnover constraints⁸, but this causes a new drawback, namely path-dependency. This means that today’s composition of the MSCI Minimum-Volatility index depends on its past composition; a feature which is undesirable for investors who are interested in a fresh minimum-volatility portfolio because they wish to invest in the strategy from scratch.

A more transparent alternative is provided by the S&P 500 Low Volatility index, which simply invests in the 100 stocks in the S&P 500 index with the lowest volatility over the preceding 12 months.⁹ Empirical studies have shown that this simple ranking approach results in a very similar risk-return profile to more sophisticated optimisation approaches.¹⁰ The added value of both approaches comes from their tilt towards low-volatility stocks, which enables them to capture the low-volatility premium.¹¹ We believe, however, that both represent a sub-optimal way of benefiting from the low-volatility premium.

Our first concern with low-volatility indices is their one-dimensional view of risk, focusing mainly on past volatility and correlations. Risk cannot be captured by a single number, and our research confirms that a multi-dimensional approach, which also includes forward-looking risk measures, is able to reduce risk—in particular tail risk—further.¹² A second concern with low-volatility indices is that they completely ignore expected return considerations. There is, in fact, a large dispersion in the expected returns of stocks with similar volatility characteristics.

We also observe some significant differences in the composition of different low-volatility index portfolios. The S&P 500 Low Volatility index does not constrain sector weights, resulting in a huge sector concentration. For example, at the time of writing around 60 percent of this index invested in only two sectors (utilities and consumer staples). The MSCI Minimum Volatility index, on the other hand, does not allow sector weights to deviate more than 5 percent from their weight in the regular, capitalisation-weighted index. In our view, both approaches are too extreme. The MSCI Minimum Volatility index is overly constrained, while the S&P 500 Low Volatility index is overly concentrated. Our assessment is that the optimum lies somewhere between these two approaches.

Russell recently launched its so-called “defensive” equity indices, which can be regarded as a “low-volatility light” alternative. This is because the weight of low-volatility factors in these indices amounts to only 50 percent. The other 50 percent is based on “quality” factors, such as earnings stability, profitability and leverage. The reason for blending in these other factors is not entirely clear. The back-tested index returns indicate that these factors increase, rather than reduce volatility. So if volatility does not improve, the benefit should probably come from improved returns. Thus, investors should be convinced that the incremental return from tilting towards quality more than offsets the higher volatility induced by these factors.

Maximum Sharpe ratio indices
We next discuss two closely related smart beta indices, namely the FTSE/TOBAM Maximum Diversification index and the FTSE/EDHEC Risk Efficient indices. Both approaches essentially try to maximize the expected Sharpe ratio, i.e. the ratio of expected return to expected risk. Although the way in which expected risk and return are defined is not identical, the differences are relatively small. For example, the Maximum Diversification index assumes that expected returns are proportional to volatility, while the Risk Efficient index assumes that expected returns are proportional to downside volatility.

The Maximum Diversification and Risk Efficient indices are often regarded as alternative low-volatility approaches. To understand this, note that lowering portfolio volatility helps to maximize the Sharpe ratio, which has volatility in the denominator. However, the indices actually go against the low-volatility premium by assuming that expected returns are proportional to (downside) volatility, which makes high-risk stocks more attractive in the numerator of the Sharpe ratio. These two opposing forces, i.e. a preference for low-volatility stocks from a risk perspective versus a preference for high-volatility stocks from a return perspective, can cause the indices to have either a low-volatility or a high-volatility profile. In the long-term, the high-volatility profile actually dominates.¹³ Compared to the capitalisation-weighted index, the indices also appear to load on the small-cap and value factor premia.¹⁴

To sum up, classic factor premia fully explain the added value of the Maximum Diversification and Risk Efficient indices. Unlike fundamental and minimum-volatility indices, however, the tilt towards factor premia is less direct and more dynamic in nature.

Momentum indices
Historically, the momentum premium has been at least as large and consistent as the value and low-volatility factor premia. Momentum indices are much scarcer though, probably due to the fact that momentum struggled during the most recent decade (while value and low-volatility strategies showed very strong performance over this period) and because the relatively high turnover of momentum strategies fits less well with the idea of a “passive” index strategy. Momentum deserves more attention, if only because it does well when value and low-volatility struggle simultaneously, such as during the tech bubble of the late 1990s.

Although momentum strategies have shown impressive long-term average returns, they can show a large underperformance over shorter periods of time. For example, the generic long-short momentum strategy shows a return of -83 percent over the year 2009.¹⁵ In our view, the main challenge involved with harvesting the momentum premium is how to control the high risk involved with the strategy. AQR, which recently introduced the first serious momentum indices, does so by limiting the tilt towards momentum stocks. Specifically, they invest in a relatively broad set of stocks (the top 33 percent based on a ranking on return over the past 12 months, excluding the most recent month) and they weight these stocks in proportion to their market capitalisation. Although these choices are indeed effective for controlling a momentum strategy risk, they also prevent investors from benefiting from the momentum premium’s full potential magnitude.

To earn the momentum premium it is not necessary to be exposed to the large risks involved with naïve momentum strategies. Specifically, a more sophisticated momentum strategy is highly effective at eliminating precisely those risks that are not properly rewarded, thereby resulting in significantly better risk-adjusted returns.¹⁶ The essence of our approach is to adjust the momentum of each stock for the part that is driven by its systematic risk characteristics (for example, high-beta stocks are expected to outperform the market in proportion to their beta). By ranking stocks according to their remaining, idiosyncratic momentum we obtain a more sophisticated momentum strategy, which is much less sensitive to systematic risk, such as a broad market reversal. This enables us to create a portfolio which is tilted more aggressively towards the momentum premium, whilst staying within the same risk budget.

Turnover is also a major concern with momentum strategies, which have relatively high turnover by definition. From this perspective, the AQR momentum indices are clearly not entirely optimal, because they may involve buying a stock ranked just above the selection threshold and selling it at the next rebalancing, three months later, if its rank has dropped to just below the selection threshold. More sophisticated buy-sell rules may be able to avoid such unnecessary turnover.¹⁷

Equally weighted indices
Several index providers, including MSCI and S&P, have introduced equally weighted indices. These are typically regarded as a means to harvest the small-cap premium. However, we believe that a word of caution is appropriate here. The evidence for a small-cap premium mainly concerns the smallest, least liquid stocks. Equally weighted indices do not actually invest in these stocks, but continue to invest in large and medium-sized firms. For example, the S&P 500 Equal Weight index still invests in the 500 largest US stocks, while the total number of US stocks is well over 5,000. Thus, equally weighted indices are better described as strategies that try to exploit a possible difference in return between large stocks and even larger stocks. Equally weighted indices are thus able to profit only partly, at best, from the small-cap effect.

Another concern with equal weighting is that portfolio weights move continuously away from their target levels, so frequent rebalancing is required to maintain equal weights. As this rebalancing involves selling recent winners and buying recent losers, this goes against the momentum effect. A nice anecdote in this regard is that back in the early 1970s, when the concept of passive investing was conceived, some of the early adopters chose equally weighted portfolios, but soon abandoned this approach.¹⁸ In our view, a traditional capitalisation-weighted (buy-and-hold) index of true small stocks is a more appropriate and a more efficient way to capture the small-cap premium.

Summary
In smart beta indices, such as fundamental and minimum-volatility indices, stock weights are based not on their market capitalisations, but on some alternative formula. The added value of smart beta indices comes from systematic tilts towards classic factor premia that are induced by these alternative weighting schemes. Smart beta indices are not specifically designed for harvesting factor premia in the most efficient manner, but primarily for simplicity and appeal. For a number of popular smart beta indices we have discussed the main pitfalls, and how investors may capture factor premia more efficiently by addressing these concerns. Finally, it is important to remember that although passive management can be used to replicate smart indices, investors should realise that smart indices themselves always represent active strategies.

References

1. See, for example, Chow, Hsu, Kalesnik & Little (2011), “A Survey of Alternative Equity Index Strategies”, Financial Analysts’ Journal, Vol. 67, No. 5, pp. 37-57.
2. Blitz (2012), “Strategic Allocation to Premiums in the Equity Market”, Journal of Index Investing, Vol. 2, No. 4, pp. 42-49.
3. See Asness (2006), “The Value of Fundamental Indexation”, Institutional Investor, (October), pp. 94-99 and Blitz & Swinkels (2008), “Fundamental Indexation: an Active Value Strategy in Disguise”, Journal of Asset Management, Vol. 9, No. 4, pp. 264-269.
4. Moore (2005), “Fundamental Indexation”, Financial Analysts’ Journal, Vol. 61, No. 2, pp. 83-99.
5. See Chow, Hsu, Kalesnik & Little (2011), “A Survey of Alternative Equity Index Strategies”, Financial Analysts’ Journal, Vol. 67, No. 5, pp. 37-57.
6. See de Groot & Huij (2011), “Is the Value Premium Really a Compensation for Distress Risk”, SSRN working paper no. 1840551.
7. See Blitz, van der Grient & van Vliet (2010), “Fundamental Indexation: Rebalancing Assumptions and Performance”, Journal of Index Investing, Vol. 1, No. 2, pp. 82-88.
8. We note that although MSCI aims for a one-way turnover of no more than 20% per annum, they have, on several occasions, relaxed this constraint. For example, a methodology change implemented at the end of 2009 caused a turnover of 45% at that moment.
9. Stock weights in this index are set inversely proportional to their volatility, so the lowest volatility stocks get the highest weights.
10. This paper was recently published as Soe (2012), “Low-Volatility Portfolio Construction: Ranking versus Optimization”, Journal of Index Investing, Vol. 3, No. 3, pp. 63-73.
11. For a discussion of the low-volatility premium we refer to Blitz & van Vliet (2007), “The Volatility Effect: Lower Risk Without Lower Return”, Journal of Portfolio Management, Vol. 34, No. 1, pp. 102-113.
12. See Huij, van Vliet, Zhou & de Groot (2012), “How Distress Improves Low-Volatility Strategies: Lessons Learned Since 2006”, Robeco research note.
13. See Clarke, de Silva & Thorley (2011), “Minimum Variance, Maximum Diversification, and Risk Parity: An Analytic Perspective”, SSRN working paper no. 1977577. In their Table 2 they report a volatility of 19.0% for a Maximum Diversification strategy applied to U.S. equities over the 1968-2010 period, which compares to a volatility of only 15.6% for the cap-weighted index over the same period.
14. See Chow, Hsu, Kalesnik & Little (2011), “A Survey of Alternative Equity Index Strategies”, Financial Analysts’ Journal, Vol. 67, No. 5, pp. 37-57.
15. Returns for this strategy are publicly available on the website of Prof. Kenneth French: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html.
16. See Blitz, Huij & Martens (2011), “Residual Momentum”, Journal of Empirical Finance, Vol. 18, No. 3, pp. 506-521.
17. In all fairness, AQR also acknowledges that mechanically following their momentum indices would be a suboptimal approach and recognizes the need for a more efficient implementation strategy.
18. Quoting Eric Falkenstein: “[…] It should be noted that there were several missteps among the index founding fathers. John McQuown and David Booth at Wells Fargo, and Rex Sinquefield at American National Bank in Chicago, both established the first passive Index Funds in 1973. These were portfolios targeted at institutions. The Wells Fargo fund was initially an equal-weighted fund on all the stocks on the NYSE, which, given the large number of small stocks, and the fact that a price decline meant you should buy more, and at a price increase sell more, proved to be an implementation nightmare. It was replaced with a value-weighted index fund of the S&P500 in 1976, which eliminates this problem. […]” See http://falkenblog.blogspot.nl/2011_09_01_archive.html.