Index providers clash over evolution of multi-factor products

Institutional investors are flocking to smart beta strategies referencing new indexes that capture multiple types of equity factor premia. But providers disagree on the best means to design these indexes and client interest is similarly scattered

The complexity of multi-factor strategies can be confusing

'Smart beta' is getting smarter. The overwhelming interest among institutional and retail investors in investible equity indexes that promise above-benchmark returns at bargain prices is spurring an arms race among index houses and investment banks alike. Now providers are seeking to enhance their respective arsenals with sophisticated multi-factor offerings. These are complex indexes that combine a variety of smart beta strategies with so-called intelligent weighting mechanisms in a single package, providing investors with the opportunity to reap a bumper harvest of risk premia without the need to maintain a large stable of individual products.

Single-factor smart beta products seek to beat their chosen benchmark by emphasising stocks that exhibit certain well-defined characteristics, such as low volatility or sustained momentum in a directional market. Multi-factor strategies are a more recent concept and consequently there is less live performance data available. Therefore each provider issuing a multi-factor product suite is betting that their interpretation will prove the most robust in all market conditions, provide the most outperformance versus their respective benchmark, and yet not be so complex as to alienate potential investors.

"There are two ways to combine factors in multi-factor indexes," says Alain Dubois (pictured), head of new business and product development in the index unit at MSCI, which has been active in the multi-factor arena since 2010. "You have a way that is top-down and a way that is bottom-up. The top-down approach is when you take different single-factor indexes and combine them into one. The bottom-up approach involves combining factors not at the level of the index but at the level of the single security. The latter approach, by definition, will optimise the exposure to these factors to a greater degree than the former," he adds.

In May this year, MSCI put this bottom-up principle into practice when it launched the MSCI Diversified Multi-Factor (DMF) index family. These combine four well-established equity factors – value, momentum, size, and quality – with a weighting strategy designed to keep volatility in line with the underlying benchmarks.

Put simply, MSCI assigns each security in the underlying benchmark – say the MSCI World ex-USA index – a score for value, quality, size and momentum, averages out the four scores as one score and then calibrates the weighting of that security in the index accordingly. Those stocks with the highest score are assigned the highest weightings, adjusted to ensure overall volatility matches the underlying benchmark. Dubois says the MSCI DMF family is gaining traction among institutions seeking to replicate their performance in their own portfolios.

It is also making waves in the exchange-traded fund (ETF) market among both retail and institutional investors. Blackrock rolled out the first range of ETFs tracking the MSCI DMF index family in September under the iShares FactorSelect brand, having partnered with MSCI when developing the indexes. Tom Fekete, head of Europe, the Middle East and Africa products at iShares, says the bottom-up equity-selection approach it took with MSCI "is more effective at maximising factor exposure".

MSCI does not have the field to itself. Edhec-Risk Institute is competing with its own range of multi-factor products under the Scientific Beta Multi-Strategy brand, one of which – the Scientific Beta Developed Multi-Beta Multi-Strategy index – was licensed for use in a range of Amundi ETFs last summer, which today have combined assets under management of $470 million. Total assets tracking Edhec strategies across institutional and retail stand at $8 billion.

FTSE Russell is also making strides in this area with its Global Diversified Factor Index Series, developed in partnership with JP Morgan Asset Management to serve as underlyings for its own US ETF product suite last year. "It will be a few years before the market decides on the best way to capture specific factors and then combine them appropriately in a single index. The more competition we have the better," says Gareth Parker (pictured), senior director for index research at FTSE Russell. It is the same sentiment he has voiced in relation to the nascent industry for multi-asset indexes.

This mushrooming of strategies is leading to increased interest among institutional investors. FTSE Russell's 2015 smart beta global survey of 214 asset owners found that 47% were evaluating multi-factor combinations last year. However, actual allocations are still lagging behind better-known single-factor offers. Only 29% of respondents are currently using multi-factor products, compared to 51% using value types and 54% using low-volatility types. Nevertheless, there is evidence of a potential shift. Data shows there is greater interest in multi-factor strategies among those yet to invest in smart beta strategies than there is among those already using such strategies, suggesting newcomers to the smart beta party could help close the gap.

Complexity concerns

One potential barrier to take-up is the lack of transparency – perceived or otherwise – of the underlying multi-factor methodologies used by different providers. This is an obstacle bedevilling all participants. Edhec-Risk Institute published the findings of its 2014 ETFs survey in June this year with responses from 222 investment decision-makers – ranging from chief investment officers to portfolio managers – and found that 88% believe full transparency on methodology and risk analytics is a precondition when it comes to investing in smart beta indexes.

The complexity, and lack of comparability, across multi-factor offerings is therefore a key stumbling block. This complexity comes in two layers. The first concerns the underlying factors used in the indexes and to what extent they drive their returns. While the elders of the factor tribes – value, quality, size and momentum – are well-known and reinforced by a battery of academic research, younger pretenders are less proven, says Jason Hsu, vice-chairman at California-based indexing house Research Affiliates (Rafi).

"There's meaningful disagreement as to what the ingredients ought to be before putting them into this factor blender. I tend to err on the side of using the tried and true. I tend not to favour new, exotic factors that we haven't heard a lot about," he adds.

Hsu cites profit margin, low debt, and earnings growth as three factors that have not yet passed his test. Rafi itself is yet to issue a multi-factor product.

Even within the established factors, different providers offer different interpretations. For example, the MSCI/Blackrock definition of value in the DMF family takes each stock's value score as being determined by three descriptors: forward price to earnings, enterprise value/operating cash flows, and price to book. The traditional Benjamin Graham definition – named after the twentieth-century economist – focuses solely on earnings per share and book value per share.

In Edhec's survey, moving from factor definitions raises concerns. Respondents voted "factor premium should be documented in extensive empirical literature" as the third most important requirement when considering a set of factors in their investment approach – just behind the need for the factors to be related to a rational risk premium and the need for the strategy to be easy to implement.

According to Felix Goltz (pictured), head of applied research at Edhec-Risk Institute, using so-called enhanced factors, as MSCI and others do, requires data-mining – basing factor premia on the strength of historic datasets rather than empirically verified characteristics.

"Our value definition is book-to-market. Our momentum definition is the recognised academic one using 12-month data. [In contrast] MSCI and FTSE use highly complex and proprietary factor definitions," he says.

Parker is quick to rebut this allegation. "FTSE Russell aims to develop factor-link benchmarks without over-complication. Under some market conditions this may result in index returns being reduced, but we're comfortable with this because it's a trade-off with complexity. It is of the utmost importance that users get the right factor exposure," he says. "We would argue FTSE Russell has the most transparent methodology in this area. We don't hide our intellectual property. FTSE has been writing transparent index ground rules since 1984 and we publish them for all our indexes, including the Global Diversified Factor Index Series," he adds.

The second layer of complexity is the weighting methodology. Traditional indexes are weighted based on market capitalisation: larger companies account for a greater portion of the index. Factor indexes can be weighted this way too, though this allows statistical noise from risks not related to the chosen factors to skew the index output. A more targeted methodology seeks to diversify risks across the chosen equity universe rather than concentrating them in a small subset, while maximising the influence of those underlyings exhibiting the greatest exposure to the chosen factors.

Providers have been experimenting with different weighting schema. Edhec's so-called smart weighting approach offers five choices of weighting methodology – including equal weighting and an efficient volatility weighting – as well as a diversified multi-strategy approach that allocates stock weighting based on the average of these five separate methodologies.

"It gives you a kind of double-diversification effect," says Goltz. "If you use any of these weighting schemes, you will have a well-diversified portfolio, but using the multi-strategy is superior because each individual scheme relies on different assumptions and none of them is perfect."

Others are not so sure. Rafi's Hsu says his findings suggest that complex weighting optimisations often do no better than a simple equal-weighted approach.

"Weighting optimisers tend to assume the inputs you supply are perfectly accurate, and we know that's not the case," he warns. "Statistically, you just don't have enough reliability or model structure to say that, so when you put it in an optimiser there's overconfidence being magnified by an opaque computer programme. When you have imperfect information, an optimiser is actually a really bad thing to use."

Goltz argues that the solid theoretical underpinning of the smart weighting methodologies endows them with reliability that other schema lack – and he is not afraid to name names. "The MSCI optimisation is very complex as there are lots of constraints on the factor exposures and it is not clear why these constraints are set the way they are," he says.

But Dubois says the MSCI DMF weightings are designed to reflect the risk weightings of their respective benchmarks, so that investors know when they allocate to the MSCI World DMF index, for example, they are taking on a similar level of risk as if they invested in the plain vanilla MSCI World index.

Each weighting methodology has its pluses and minuses. Goltz says the lack of comparability is fine as long as providers disclose how returns would differ using different schema. "If I have a lot of constraints in an index, it would only be fair as a provider to show what the results would look like if I took out some of these constraints, then they could see how much they affect the outcome," he says. For Goltz, transparency is king. 

Clients' conundrum

Potential investors – both retail and institutional – have to juggle numerous variables when it comes to multi-factor investing. Their own take on what makes a viable factor, how it should be scored, and to what extent a strategy should be weighted either to maximise exposure to the chosen factors or match the characteristics of an underlying benchmark, needs to be known in advance of making a selection.

Already a clear divide is opening up between large, sophisticated institutions such as pension funds and smaller firms such as family offices and wealth managers. The former crave control over their factor allocations, and hence lean towards investment banks, rather than index houses, which can offer a variety of bespoke single-factor plays they can cycle between at will.

"Our clients have their own view on how they want to put these indexes together," says Jay Watson, the head of multi-asset index structuring for Europe, the Middle East and Africa at Barclays. "You could go down the route of combining the different factors together in one grand single index. There are instances where we have done that, but what we're finding is that clients want us to leave open the possibility of combining them in different ways," he adds.

Certain pension funds go one step further by shaping not only the allocation strategy but also the underlying factor selection. "We know exactly what algorithm we want to use to extract factor premia, and find it very comforting to play a large role in the construction phase. But with certain factors it's preferable for us to implement such a strategy with a bank, since they have a much better trading infrastructure," says Søren Grooss, portfolio manager at Danish pension fund PKA. "We develop the algorithm in collaboration and then finally have the bank implement and run the strategy. Most of these strategies are very labour intensive on a day-to-day basis, and since we have very few portfolio managers it would not be optimal for us to implement these strategies ourselves." As yet, PKA does not use pre-packaged strategies such as those offered by the main index houses.

Other firms take a different approach and prefer to leave the underlying factor methodology to the index providers. "When you go to large family offices or wealth managers, some of these investors may not have strong views about which factors they want to invest in and when, so they look to [an all-in-one] multi-factor strategy," says Blackrock's Fekete.

It will be a few years before the market decides on the best way to capture specific factors and then combine them appropriately in a single index – Gareth Parker, FTSE Russell

There is a third category of investors: hedge funds. These institutions have long made use of custom baskets of equities divined from proprietary correlation matrixes to act as hedges against positions on vanilla indexes. The countercyclical nature of certain factor strategies – such as low volatility and quality – can provide a similar service without the need to construct a bespoke basket.

"Segmenting the equity market has been a key area of focus from a hedging perspective. The next logical thing for investors looking to do so is to use these multi-factor indexes as they provide the correlation exposure they are looking for. The challenge is that these index methodologies are based on past correlations and they may change in the future. That's something to be aware of," says Chris Rossbach, co-founder of J Stern & Co, a long-term-value-focused family office.

Then there are more prosaic considerations to take into account – ease of implementation and cost being chief among them. Twelve per cent of respondents to the FTSE Russell smart beta survey that had not allocated to a smart beta strategy cited cost of implementation as a barrier, while just over a quarter of total respondents rated cost as a "concern" when evaluating whether to invest or not.

Expenses depend partly on method of implementation. Multi-factor strategies wrapped in ETFs have their costs expressed as a total expense ratio on returns and are generally competitive with their single-factor cousins. iShares' FactorSelect range, for example, currently charge between 35 basis points and 50bp, close to the 25–30bp charged for its single-factor quality, momentum, size and value products. MSCI's Dubois says US institutions are increasingly seeking to harvest multi-factor premia through ETF allocations.

The majority of large allocations, though, are handled by investment banks, which provide exposure through excess return swaps, and large fund managers, which physically replicate the index constituents in-house. Both wrappers attract variable costs depending on the complexity of the strategy and the counterparty's own balance sheet constraints.

Simon Midgen, head of index funds strategy at Legal and General Investment Management, which has roughly £20 million ($30.3 million) of assets allocated to smart beta strategies, says that some fund wrapper formats may offer investors more flexibility than the ETF format and may therefore be more suited to complex multi-factor plays. "There are clearly factors that make some indexes easier to implement than others. For example, an index with a high turnover poses a high opportunity cost. One of the keys of success in managing index funds is balancing the trade-off between close index tracking and minimising costs. The investor wants us to track an index closely, but they don't necessarily want us to track it so closely that we're mimicking every tiny change in the index calculation and incurring transaction costs when it may have no impact on the tracking error of the portfolio," he says.

By agreeing with the investor the right to alter the mandate within strict risk limits, fund managers such as Legal and General are able to balance costs and tracking error and hold out the possibility of greater outperformance over the long term.

It is not just multi-factor strategies themselves that are evolving but also the way in which they are packaged for investors. Barclays' Watson says: "We've put big efforts into our transaction documentation to give our clients flexibility."