Multi-level orchestration – multi-factor indices explained

by , 21st August 2017

When multi-factor indices were first launched back in 2010, the implication from much that was said at the time by the providers about their origination was that it marked another staging post in the evolution of smart beta.

In short, if smart beta is passive investing that has gone to university, then smart beta is well on its way to gaining a doctorate.

As its name suggests, multi-factor indices have grown out of the more long-established single-factor products that seek to assess stocks according to certain criteria or factors which, according to a substantial body of research that goes back nearly 50 years, have been proven to offer returns that can beat the market.

The most well-established and recognised factors are as follows:

  • Value – stocks with high versus low book-to-market value
  • Momentum – stocks with high versus low returns over the past 12 months, but omitting the most recent month
  • Quality – stocks that are characterised by low debt, stable earnings growth and other financial quality metrics
  • Low size – stocks with low versus high market cap
  • Low volatility – stocks based on the estimate of their volatility and correlations with other stocks
  • Yield – stocks that appear undervalued and have demonstrated stable and increasing dividends

When it comes to the multi-factor indices, depending on the index provider the multi- version will encompass all or some of these factors in combination in order to provide diversification and an adjusted weighting to suit the investment needs and objectives of investors.

All of the major index providers have opted to construct multi-factor indices including MSCI, EDHEC-Risk, FTSE Russell and S&P, though there are key differences in how these are constructed. These differences bear some examination along with some of the issues that affect single-factor indices which multi-factor attempt to address.

Bottom up versus top down

An important issue with single-factor index returns is that this type of exposure is highly variable. Each individual factor will offer different return patterns depending on which underlying trend it seeks to exploit. In particular, single factors can suffer extended periods of underperformance as they are driven by very different market factors which can pay off at different times in the economic cycle.

S&P has provided research with regard to the S&P 500 Low Volatility Index which shows that though the index outperformed the benchmark over the 20 years between 1995 and 2015, it also underperformed the benchmark for protracted periods within that timeframe.

The research points out that in the 69-month period between December 1995 and August 2001, the index suffered long cyclical downturns. The cumulative return peaked in September 2002 but it took another 72 months to reach the same height in August 2008.

The research goes on to add that similar extended periods of over- and under-performance can be seen in other single factor indices.

Such is the variability that it has engendered much academic debate regarding factor timing and factor rotation. Yet another approach, though, has been to construct fixed combinations of factors that can address the problem of cyclicality but in a lower cost way. Hence, we get to the concept of multi-factor indices.

Another index provider, FTSE Russell, provides an examination of three routes it identifies as approaches that can be taken towards constructing a multi-factor index.

Composite index

This is the simplest multi-factor index. At its most basic it takes the weighted average of just two single-factors indices – FTSE Russell suggests as an example a 50/50 split between value and quality. The advantage of this approach is its top-down simplicity. In principle, this is no different from replicating single-factor indices in their chosen weights but by having both factors together the index provider maintains the fixed weights, this relieving the customer of having to adjust index-replicating products.

Composite factor

The next step in the evolving landscape of multi-factor indices has been to combine a weighted average of the individual factors in a bottom-up approach. This takes better advantage of the interaction between factors and offers potential trading economies: if a stock is eliminated from inclusion in one factor but added to another, then no trade needs to take place to maintain index replication.

Tilt-tilt

This is the most recent evolution in multi-factor indices and the most sophisticated. The approach sees the provider construct the index according a tilt towards one factor or another rather than looking to average out. This is also known as a multiplicative approach or sequential tilting.

Behind such simple definitions lies a whole mathematical library of equations and calculations which underpin both the single factor indices and the multi-factor versions that seek to address issues around weightings and exposures.

There are also basic disagreements between the various providers over which is the best and most consistent approach but the basic tenet of all multi-factor indices holds true for all. Namely, any multi-factor approach – however it is reached – will provide diversification of the specific risks and lead investors away from concentrated benchmarks that are exposed to often undesired and unrewarded risks.

It is also true that within funds notionally covering the same factors, the differing approaches to index composition can lead to differing sector exposures.

Multi-factor performance and research weaknesses

Much as there is plenty of evidence that stretches back 50 years to support the theory behind single-factor indices, there is simply not the same body of data backing up the investment theory around the multi-factor versions.

Index provider MSCI – which takes four of the factors to go towards its Diversified Multi-Factor Indices dropping low volatility and yield – said its MSCI World DMF Index substantially outperformed the MSCI World Index over the 16-year simulation period before launch in 2015.

However, as critics would point out, this outperformance only exists on paper and doesn’t come with data on turnover, trading costs or capacity. What is certain is that investors will desire to see more data over time.

The EDHEC-Risk European ETF and Smart Beta Survey of 211 investors and fund managers for 2016 showed that when it came to factors, the respondents were primarily concerned with the existence of extensive empirical literature documented factor premium.

The future – more factors

The popularity of multi-factor products has already been proven. According to the 2017 annual FYSE Russell Global Institutional Smart beta survey, multi-factor combinations have become the most popular smart beta index construction utilised.

It is a certainty that more multi-factor ETFs will emerge. The EDHEC-Risk European ETF and Smart Beta Survey showed that the enthusiasm for more complex smart beta ETFs showed no signs of abating.

A third of respondents said they would like to see more multi-factor ETFs designed in the future while aggregated together with the desire for more single-factor and smart beta indices generally, more than half of the respondents wanted further developments in at least one of these categories. “The development of ETFs based on advanced forms of equity indices is now by far the highest priority for respondents,” concluded the survey authors.

Smart beta will indeed continue to evolve and proliferation of hundreds of ETFs based on multi-factor indices that has been seen in the past few years points to their popularity with investors.