The ongoing debate on smart beta strategies has led to a number of misconceptions. In a recent paper in the 2016 Journal of Index Investing, Smart Beta is not Monkey Business, Noel Amenc, Felix Goltz and Ashish Lodh analyse some of these misconceptions.
In this article, we provide a summary of selected results concerning the sources of outperformance of such strategies.
Some have argued that the limitations of cap-weighted indices are so strong that any alternative index construction, including randomly generated portfolios (so-called monkey portfolios), will do better.
Moreover, it has been claimed that smart beta strategies would not be different from such monkey portfolios and their actual product design would not matter.
This is sometimes supported by saying that when inverting the strategy of popular smart beta approaches one gets similar or better outperformance.
Here we summarise results from Amenc, Goltz and Lodh, who empirically assess the validity of such claims for a range of test portfolios which employ stock selection to obtain a given factor tilt and different weighting schemes.
Such strategies correspond to common offerings in the area of smart beta indices.
The results show that inverting the strategy not only turns the weights upside-down, but also changes the performance.
For example, while a score-weighted value-tilted strategy leads to a positive outperformance of 3.94 per cent annualised, the inverse of this strategy leads to -2.07 per cent returns relative to the cap-weighted reference index. Similar results hold for other factor tilts.
These results of contrasted performance between factor-tilted strategies and their inverses contradict the claim that anything will beat cap-weighting.
Indeed, designing exposures to negatively rewarded factors (such as growth, low momentum or large cap) moves away from the cap-weighted reference index but does not lead to outperformance.
Thus, rather than relying on a supposedly automatic effect that moving away from cap-weighting will deterministically improve performance, investors in smart beta strategies need to analyse the factor tilts and diversification mechanisms employed and identify which smart beta strategies correspond to their investment beliefs and objectives.
Some argue that once we deviate from selecting and weighting stocks by their market value, as is done in cap-weighted market indices, this will necessarily lead to value and size exposures and these tilts suffice to explain outperformance.
While this may obviously be true for some smart beta strategies which – by design – will lead only to small-cap and value exposures, this notion is inconsistent with evidence on a wide range of smart beta strategies.
In particular, Amenc, Goltz and Lodh show that typical factor-tilted smart beta strategies can have exposure to factors other than small-cap and value.
This finding may not be surprising, and is fully consistent with the academic literature, which has documented the importance of various equity risk factors beyond value and small cap.
Additional relevant factors include most notably the momentum and low-risk factors, as well as additional factors related to the so-called “quality” dimension of firm fundamentals such as the profitability and investment factors.
Amenc and colleagues show in particular that the low-volatility and momentum-tilted portfolios, irrespective of the weighting scheme, derive a large portion of their performance from their exposure to low-beta and momentum factors, respectively.
The contributions of factors other than value and size to portfolio risk and return invalidates the claim that there is nothing beyond size and value exposure in smart beta strategies.
Moreover, they show that many smart beta strategies present a considerable portion of unexplained performance, which suggests that the portfolio construction of these indices captures effects that cannot be explained fully by the relevant factors.
Possible explanations of this unexplained part of performance are that the improved diversification scheme allows value to be added beyond the explicit factor tilts, or that yet other additional factors, which are omitted from the factor model, are at work.
However, while the findings in Amenc, Goltz and Lodh are in line with this literature, they stand in stark contradiction to the claim that there is nothing beyond value and small-cap exposure in smart beta strategies.
Instead, these results suggest that different smart beta strategies derive performance from different exposures to several factors that may go beyond size and value.
In fact, the claim that all smart beta strategies lead to size and value exposure ignores the past few years of product development in the smart beta space.
Many of the first generation approaches (Smart Beta 1.0) – by deviating from standard cap-weighted indices – may indeed introduce implicit factor exposures (such as value and size, and potentially others).
However, the more recent Smart Beta 2.0 approach allows the issues with such uncontrolled implicit exposures to be addressed.
In fact, research on the Smart Beta 2.0 approach shows that methodological choices can be made independently for two steps in the construction of alternative equity index strategies: the constituent selection and the choice of a diversification-based weighting scheme.
They show that, even though some argue that the risk and performance of diversification-based weighting schemes are solely driven by factor tilts, it is straightforward to correct such tilts through the selection of stocks with appropriate characteristics while maintaining the improvement in achieving a risk–return objective that is due to a diversification-based weighting scheme.
Such a Smart Beta 2.0 approach provides controls over deviations in terms of factor exposures, which invalidates the claim that all strategies simply tilt to value and small-cap, and also goes beyond simple Smart Beta 1.0 approaches in allowing for additional flexibility and explicit risk control.
In a smart beta strategy, rebalancing takes place at regular intervals to ensure the weights are in line with the strategy objective.
This has led some to argue that “rebalancing” is the main driver of smart beta outperformance.
To assess this claim, it is useful to look at two separate questions.
A first question is whether a positive performance effect necessarily arises from rebalancing.
A second question is whether smart beta strategies necessarily capture such an effect.
On the first matter, empirical research has shown that rebalancing effects are highly dependent on the time horizon.
There is ample evidence not only of return reversal effects, but also of return continuation momentum effects.
It is well known that rebalancing effects typically occur at a frequency which is much higher than typical index rebalancing frequencies.
One should recognise also that there is no consensus in the literature on the existence of a positive rebalancing effect.
Whether a rebalanced portfolio will outperform a buy-and-hold portfolio or underperform it will depend on the behaviour of the component assets.
Given this dependency of a rebalancing bonus on specific conditions, it is perhaps not surprising that standard asset pricing models such as the widely used models of the Fama-French-Carhart type do not include any rebalancing factor.
A second question is whether smart beta strategies gain exposure to such rebalancing effects.
On this matter, it can be noted that no convincing attribution of smart beta performance to rebalancing effects has been provided to date.
In their 2016 working paper, Ten Misconceptions about Smart Beta, the authors provide an illustrative assessment.
They draw on empirical finance research that has come up with a range of “reversal” factors, which simply move out of stocks that had strong price appreciation and into stocks that had weak returns relative to the average stock, and can thus be seen as related to rebalancing effects.
In particular, researchers have documented that there are positive returns to tilting to past one-month loser stocks (short-term reversal) and past five-year loser stocks (long-term reversal or contrarian) strategy.
We investigate the explanatory power of such reversal factors when omitting more standard factors.
In particular, we look at unexplained average returns (alpha) in a model that only includes such reversal factors in addition to the market factor, but excludes the more standard size, value and momentum factors.
We attempt to capture the returns to several smart beta indices.
Our results suggest that the different smart beta indices show high and significant alpha when accounting for the reversal factors.
Thus, the reversal factors do not fully capture the high average returns of such strategies.
This result suggests that the performance of these strategies is not primarily driven by the reversal factors and the associated rebalancing effects.
Also, when including the reversal factors on top of standard factors (market, size, value and momentum), such reversal factors have hardly any marginal explanatory power.
Overall, there are serious uncertainties concerning the existence of a positive performance effect from rebalancing in general.
Moreover, there is no evidence suggesting that smart beta performance is mainly driven by the mechanics of rebalancing.
Given these doubts on the relevance of rebalancing effects for smart beta performance, it is unreasonable to expect guaranteed outperformance of smart beta from a deterministic rebalancing effect.
Instead, factor exposures and diversification properties of such strategies need to be analysed carefully.
Towards a differentiated understanding of performance drivers
A careful analysis shows that, more often than not, superficially convincing claims about smart beta performance drivers stand on shaky foundations.
In fact, all too often, claims about performance drivers of smart beta abstract from the large variety of approaches that exist.
In a nutshell, our analysis cautions against oversimplification and calls for a detailed analysis of smart beta strategy performance taking into account the specific properties of the respective strategy.
Felix Goltz, head of applied research, EDHEC Risk Institute, research director, ERI Scientific Beta
Jakub Ulahel, quantitative research analyst, ERI Scientific Beta