Smart beta: Smart investing or smart trading?
S mart beta portfolios are really only different from the ‘dumb beta’ portfolio in three ways. First, by weighting their constituents in some way other than by market capitalisation, they potentially change the risk-diversification properties. Second, they may expose the investor to specific risks considered desirable, to a greater extent than the market portfolio does. And third, by breaking the link between portfolio weight and price, they all have to be traded to preserve the original weights.
When we ask which of these three potential sources of excess return is most important, surely we would expect practitioners to argue that, while diversification and trading might add something to geometrical returns, it is the explicit and idiosyncratic characteristics of the weighting scheme that really matter. After all, this is what they spent time working on and often what they use to label their ‘value’, ‘fundamental indexation’ and ‘low-volatility’ products.
“It would be strange for a practitioner that deliberately creates exposures to risk factors to explain excess returns in terms of some kind of rebalancing premium,” as Yazann Romahi, a portfolio manager in JP Morgan’s global multi-asset group, puts it.
Not a bit of it. While some outfits like INTECH Investment Management and Parametric Portfolio Associates have always emphasised the trading aspect of what they do, others, like Research Affiliates, are also moving in their direction. Instead of attaching greatest importance to the sales, value, cash-flow and dividend metrics that determine its fundamental index weights, Research Affiliates’ head of equity research Vitali Kalesnik emphasises the trading.
“By rebalancing periodically, you keep breaking the relationship between price and weight and you trade against mean reversion,” he says. “We believe this is the biggest reason why most smart-beta strategies outperform.”
To illustrate the point, he describes an experiment. Take the market portfolio of the year 2000, let it run for a year and then re-weight it back to those original ‘market 2000’ weights, and repeat that year after year until 2013: your return in excess of a market portfolio left to drift would be almost exactly the same as that to an annually re-balanced equal-weighted portfolio, implying that the excess return to both is due to the trading.
Paul Bouchey, managing director of research at Parametric, which attributes half of its excess returns to portfolio trading, describes the same experiment.
“You can exploit the energy in markets with a weighting scheme that is not linked to prices – and it almost doesn’t matter what that weighting scheme is,” he concludes.
At a glance
• Smart beta potentially differs from the market portfolio in three ways: in diversification; risk exposures; and in the necessity of trading.
• Which is the most important?
• Fuzzy boundaries and commercial imperatives makes for contention.
• But clarity comes when investors are clear about their own beliefs – particularly in the efficiency of markets.
That isn’t quite how Kalesnik sees it – he reminds us of the implicit value tilt in fundamental indexing by referring to the “overpriced” stocks in the market portfolio. Nonetheless, the shift of attribution, from a return to risks that are held to a return for trading, is significant.
And yet it is not as obvious as it seems that the ‘buy low, sell high’ element of re-weighting must always be the source of a return that exceeds the market’s.
“There’s a reason why we call this ‘volatility capture’ rather than a ‘rebalancing premium’”
A draft paper by Keith Cuthbertson, Simon Hayley and Nick Motson of the Cass Business School, seen by IPE, takes issue with the idea that trading can add anything to the excess return of a diversified portfolio, unless it takes place in markets that mean-revert. Kalesnik conceded as much in his own formulation.
Equity markets do exhibit mean reversion. It tends to dominate equity price movements out to about a month or two, but then begins to fade against price momentum, disappearing after 12 months. It returns again at around the 4-5 year point.
“If you believe that momentum exists, as we do, you certainly wouldn’t want to get in the way of it with your rebalancing process,” says Ronen Israel, a principal at AQR Asset Management.
A portfolio that systematically re-weights with a periodicity that coincides with the dominance of momentum in prices buys low and sells high relative its own last re-weighting, but not relative to the market, or the non-re-weighted version of itself, that both hang on to the trending winners. If its return is greater, it must have something to do with how the holdings differ from the market portfolio, rather than how the trading differs.
Eric Shirbini, global product specialist at EDHEC Scientific Beta, the smart-beta product centre of the French business school, goes even further. He relates how Edhec tested the exposure of a range of smart-beta strategies to a ‘contrarian factor’ (created by going long stocks that had gone down in price for the previous four years and short those that had gone up), finding that most showed a correlation of just 0.20-0.50.
And Research Affiliates’ products don’t re-weight every four years. They do so annually – smack in the middle of the time horizon over which momentum dominates prices. Moreover, Kalesnik recognises this as sub-optimal for capturing any supposed trading-related excess return.
“Uncontrolled allocations can quickly take you into a much-higher-risk portfolio”
“I totally agree that you will hurt yourself by trading against momentum,” he says. “Our fundamental indices would probably benefit from rebalancing less frequently, but we find that annual rebalancing also gives us enough time to maintain the index properly.”
The idea of an excess return to trading may have been better described in the ‘volatility-capture’ idea put forward by Robert Fernholz in the 1980s, and pursued by Intech, the firm he founded. Intech re-weights its portfolios monthly, not annually – but while that brings it closer to short-term mean reversion in stock prices, the important thing is that ‘volatility capture’ does not really rely on that.
“If a stock mean reverts you can obviously profit by buying when it is below the mean and selling when it is above it,” explains CIO and CEO Adrian Banner. “This is a form of rebalancing, but it’s not really the sort that we are talking about when we are doing it with hundreds of securities.”
As long as the way capital is distributed between stocks by the market is mean-reverting, he points out, it doesn’t matter if the stocks themselves are a long way from mean-reverting: the fact that the whole market is going up has no obvious bearing on the amount of capital allocated to any one stock relative to the rest.
“You can exploit the energy in markets and it almost doesn’t matter what that weighting scheme is”
“The number-one stock in the S&P 500 is Exxon,” Banner explains. “It has been Apple, Microsoft, IBM. But whichever stock it has been has accounted for close to 5% of the market for most of the time. Similarly, the deviation of the share of the market’s capital of the second-largest stock, the tenth largest, and so on, have also usually been remarkably small.”
It is the small deviations from these very stable mean levels of capital allocation that Intech tries to exploit in its ‘Alpha Capture’ range of systematic index products, by creating monthly-rebalanced ‘diversity-weighted’ portfolios. Banner is explicit that the excess return generated is a function of market volatility rather than any kind of mean reversion. This is why that return gets larger as the frequency of rebalancing is increased – the daily volatility of stock prices, annualised, is higher than monthly annualised volatility which, in turn, is higher than quarterly annualised volatility.
“There’s a reason why we call this ‘volatility capture’ rather than a ‘rebalancing premium’,” he says.
Bouchey at Parametric articulates a similar idea, referring to the “entropy” of the market pushing stock weights away from their means and the “energy” snapping them back again.
“It is analogous with the energy in the ocean and wind,” he explains. “If you just sit in a raft you will just get pushed around aimlessly. But put a keel down and a sail up and you can use this energy to propel yourself.”
Both Intech and Parametric re-weight as frequently as they can, compromising only in order to limit transaction costs: Parametric re-weights only those securities that have moved 20% away from their starting capital allocations.
‘Volatility capture’ owes a lot to diversification. The more independently the constituent stocks move, the better it works. Indeed, the contention in the Cass Business School paper is that all of the benefit comes from diversification and that trading adds nothing except the maintenance of that diversification. This is also the argument put forward by those smart-beta providers whose explicit strategy is to maximise diversification, such as TOBAM and Edhec.
“When we are talking about smart beta we are talking about risk premia, and the source of risk premia are the holdings of a portfolio, not the trading of a portfolio,” insists TOBAM’s president, Yves Choueifaty.
“What really matters for an as- set owner is his holding scheme, not his trading scheme”
To demonstrate, he ran his strategy with monthly and quarterly rebalancing next to one another for 10 years: return and volatility were the same – turnover, which had doubled, was the only differentiator. (For the record, like Intech, but obviously for a very different reason, TOBAM would re-weight as frequently as it could – and does so monthly.
“Trading never generates money for the buy-side – never,” Choueifaty adds. “It only generates money for the sell-side. What really matters for an asset owner is his holding scheme, not his trading scheme.”
The elevation of holding or weighting scheme over trading is common to those who seek diversification and, for the most part, those who seek specific exposures (with Research Affiliates being a major exception).
“The factor tilt matters for two reasons,” says Israel at AQR, a leading name in factor-based investing. “Breaking the link between price and weight introduces diversification; and we also believe that the factors we seek exposure to provide a long-term source of return to assets that fewer investors want to hold. Rebalancing is not as significant as these other two.”
Most of the active managers that have adopted systematic strategies from the smart-beta world fall into this camp.
“We believe that the risk exposures you choose count for most of the returns,” says Tim Gardener, head of institutional client strategy at AXA Investment Managers. “We are not rebalancing for the sake of rebalancing, but trying to keep our portfolio consistent with our investment beliefs.”
“The factor tilt matters”
AXA’s ‘Managed Volatility’ strategies systematically favour low-volatility stocks and those with low debt levels and sustainable earnings, while excluding any with “extremely high” P/E ratios. The weighting scheme for that universe is a version of diversity weighting that introduces further slight size and value tilts. AXA re-weights its portfolios quarterly but, like Parametric, only trades those securities whose weights have shifted substantially.
“It’s better to have some control over how movement in prices takes you into new risk exposures, especially if your initial exposures were to certain risks like small-caps or value, where uncontrolled allocations can quickly take you into a much-higher-risk portfolio,” says Lise Renelleau, director of managed volatility strategies with AXA Rosenberg – suggesting the familiar trade-off between transaction costs and the ideal of re-weighting as often as possible to maintain desired exposures. ‘Volatility capture’ is clearly not part of the thinking.
In this respect, then, the only apparent difference between ‘tilters’ like AXA and AQR and ‘diversifiers’ like TOBAM or Edhec is over whether diversifying can add any return in excess of what is achieved through the factor exposures themselves, as the diversifiers do not generally dispute the idea that the risk factors are sources of excess return.
So Edhec, for example, has chosen to tilt to value, small-cap, quality, momentum, illiquidity and low-volatility, but Shirbini argues that diversification adds more because those factors are uncorrelated. He then goes further, pointing to the unintended biases retained in the weighting scheme used to construct factor-tilted portfolios. Edhec’s solution is to diversify again, combining several waiting schemes to create each portfolio.
Where there is some dispute is in the share of excess returns between factors and diversification. Choueifaty at TOBAM, which achieves diversification by maximising the ratio of its portfolio constituents’ volatility to overall portfolio volatility, articulates the purest version of the pro-diversification argument.
“A fundamental indexer will want to hold positions that have exposure to value and small size and a minimum-variance practitioner will want to hold a low-volatility portfolio,” he says. “But my belief is that the diversification is the reason why these smart beta strategies add money, and that TOBAM builds in diversification by design while all of the others build it in incidentally.”
There is a problem with this view. To take fundamental indexing as an example, it is not clear that its weighting system introduces meaningful diversification relative to the market portfolio – which must mean that any excess return is due to its factor exposures (unless we still believe in some trading benefit).
“Some practitioners emphasise it, but smart betas benefit very, very little from this diversification effect,” insists Kalesnik at Research Affiliates. “While many remove the bias to large-caps they introduce a lot of small-cap bias instead.”
“Most smart-beta strategies do not add diversification because they select stocks according to a factor and then weight them by the same factor”
This statement is backed up by research done by the diversifiers at Edhec: Shirbini says that the concentration of Research Affiliates’ indices – the number of stocks they are effectively investing in – is almost the same as the market’s.
“The problem is that the value and small-cap tilts just deliver a whole new bunch of risks,” he says. “Most smart-beta strategies do not add diversification because they select stocks according to a factor and then weight them by the same factor.”
Of course, the practitioners don’t see that as a “problem”, but as maximising a desirable exposure. Ryan Taliaferro, a portfolio manager with minimum-variance specialist Acadian Asset Management, does not mince his words as he hits back at maximum diversification strategies.
“At best it is controversial, at worst just plain silly,” he insists. “There is absolutely no rationale for such a strategy. What, exactly, is the mispricing that maximum diversification exploits? Is ‘diversity’ not properly priced? If it is in any way successful, it will only be as the accidental side effect of picking up exposure to a characteristic like size or low-beta.”
“These patterns in returns are, in fact, mispricings”
This is precisely the opposite of Choueifaty’s formulation. But more significant is Taliaferro’s use of the term ‘mispricing’, which he quite deliberately prefers to the idea of an excess return to a ‘risk factor’. And here we come to our final contention – between those who, while they agree that portfolio weights are significant because they create specific exposures, disagree on how those exposures deliver excess return.
So, apart from reducing volatility, diversifiers might also argue that diversification generates excess return because the average investor does not, in fact, want to hold a diversified portfolio. It is an argument often heard for the ‘risk premia’ paid to holders of value and smaller companies, for example. Against this, Taliaferro believes that value, for example, outperforms on average because, on average, value stocks are “underpriced”.
“The risk-factor explanation creates awkwardness when the same people then want to gain exposure to those factors because of their higher associated returns,” he continues. “Then there is no advantage to holding, say, value stocks, since you’re just being paid for risk. I think the only position that describes the return anomalies in a way that also justifies investors’ interest in them is that these patterns in returns are, in fact, mispricings.”
This is the same contention about market efficiency that lay at the heart of 2013’s double Nobel for Eugene Fama (markets are efficient, and holders of riskier positions get paid a premium) and Robert Shiller (markets are inefficient, and those risk premia are really just mispricings). The reason it is of more than academic interest to us is that it suggests a way to formulate our smart-beta investment beliefs – and to cut through some otherwise confusing debate between smart-beta providers.
As an example, let’s compare TOBAM and Intech. For TOBAM, diversification is the be-all and end-all, but the idea that trading adds excess return is anathema. For Intech, trading is the be-all and end-all, but portfolio diversification is an integral part of that.
“Diversification and rebalancing are what I think are creating the effects we are concerned with,” as Banner puts it. “Our weights are not about picking up any risk premium – my attribution of excess return to that is close to zero.”
“You trade against the mean reversion, [and] we believe this is the biggest reason why most smart-beta strategies outperform”
We can square this circle by recognising that TOBAM’s process rests on a belief in efficient markets and risk premia, and that one of these premia is paid to holders of diversified portfolios; while Intech’s rests on a belief in inefficient markets in which changes in capital allocations are greater when diversification is higher.
To sum up, if you believe markets are efficient, and you believe there are some factors that present higher-than-market risk, you need only expose your portfolio to those factors and sit back. If you believe there are several such factors and that they are not highly-correlated, diversifying between them can improve things; and if you believe that there is a ‘diversification factor’, this improvement will exceed that of the simple volatility-reduction effect on geometric returns. In this worldview, re-weighting only maintains exposure to the original factor and generates no additional excess return.
If you believe markets are inefficient, you will need to trade in order best to capture the price (or capital-allocation) movements of the market. In this worldview, the types of exposure you have are irrelevant; the key is the nature of your fixed portfolio weights and the frequency with which you are able to trade back to them. Diversification matters, not primarily because it improves geometric returns and certainly not because it is some kind of rewarded risk, but because it increases the number and amplitude of the changes in capital allocation through the market.