The rise in popularity of factor investing strategies and products raises questions about crowding, writes Paul Amery
At a glance
• Valuations on many forms of factor investment are high.
• A key concern is not how many people own a particular security or factor but whether those owners share other common properties.
• Industry participants argue that the inherent diversity of factor approaches helps ease concerns about index-related crowding.
With smart beta exchange-traded fund (ETF) assets forecast by BlackRock to reach $1trn (€941bn) by 2020 and institutional smart beta assets to probably rise to a multiple of this sum, some market participants argue that the weight of new money might eat into the returns promised by factor-investment strategies.
To what extent do the assets under management in factor strategies matter for return expectations? Could implementation costs hit returns, and is there hidden crowding in the indices and risk models used to access factors?
In 2016, index provider and investment strategist Research Affiliates made a splash by warning of a potential smart beta crash.
“Valuations of many of the most popular factors and smart beta strategies are well above historical norms,” wrote Rob Arnott, Noah Beck and Vitali Kalesnik in a research paper for the firm.
“These high valuations are indicative of lower future returns,” they added, pointing to low volatility and quality-factor strategies as particularly expensive and arguing that the value factor was the only relatively cheap smart-beta strategy.
Research Affiliates’ warning came after a period of sustained heavy inflows into exchange-traded funds (ETFs) tracking low volatility and minimum volatility indices, as well as heavy interest among institutional investors in such strategies.
If Research Affiliates’ timing call was correct – value-factor portfolios have outperformed low-volatility and quality-factor strategies by some margin since last summer – there has been nothing so far resembling a crash.
Nevertheless, anyone involved in the smart beta business is mindful of the so-called ‘quant crisis’ of 2007, when several hedge funds following similar momentum strategies were caught in a vicious market reversal.
Dan di Bartolomeo, president and founder of financial risk modelling firm Northfield Information Services, is keen to distinguish between what he sees as normal rotations between competing equity strategies and situations threatening a potential market dislocation.
“Every security is owned by someone,” says di Bartolomeo. “So, if a certain set of securities, say, those representing a factor, is concentrated in a few people’s hands, by definition that set is not concentrated in other people’s hands.
“What we should be concerned about is not how many people own a particular security or factor, it’s whether those owners share other common properties, such as margin leverage or higher-than-usual liquidity needs. A factor doesn’t become crowded – what may become crowded is the type of investor holding that particular factor,” di Bartolomeo continues.
“Different investors have different preferences and things come in and out of fashion. I wouldn’t describe that per se as crowding.”
Manuela Sperandeo, European head of specialist sales at ETF provider iShares, disputes the suggestion that the volume of assets invested in low and minimum-volatility ETFs could be a cause for worry. “Low volatility and minimum volatility US equity ETFs represent just over 0.2% of the overall US equity market,” says
Sperandeo. “In terms of assets under management, we’re quite far from having crowding concerns.”
Harindra de Silva, president and portfolio manager at Analytic Investors, puts the overall interest in smart beta and factor strategies into perspective. “Global equity markets are worth about $20trn and there probably isn’t $1trn yet in factor-based or smart-beta strategies. It’s early to talk about crowding, but we should maybe start worrying about this when smart-beta assets approach $3-4trn,” says de Silva.
Ronen Israel, principal at AQR Capital Management, plays down factor valuation concerns and points out that assets in factor portfolios may simply have shifted from similar actively managed funds.
“Currently, some factors look a little more expensive than their historical valuations, others are a little cheap. That said, on average, valuations are about normal. We don’t see any evidence of factor investing in the aggregate looking very expensive,” says Israel.
“There’s a lot of talk about money coming into smart beta and factors, but the question we should be asking is where it’s coming from. For example, if investors are moving from traditional active value managers into smart-beta value strategies, then that’s not necessarily new money chasing the cheap stocks.”
Dimitris Melas, global head of equity research at MSCI, adds that some factors are more prone to valuation concerns than others.
“The value, size and yield factors have a built-in protection mechanism against overvaluation,” says Melas. “As stocks become more expensive, larger or lower-yielding, they drop out of the universe. On the other hand, naïve implementations of quality and low volatility do not have a valuation anchor.”
It is well known that most equity and bond indices are constructed without taking trading costs into account. As a result, higher-turnover factor strategies, such as momentum, risk disappointing investors if rebalanced too frequently.
Index firms typically impose turnover constraints on the more active factor and smart-beta strategies, diluting the purity of the investment approach but recognising the reality of the frictional costs associated with buying and selling constituents.
But other implementation costs associated with factor investing may also rise with the increasing popularity of these strategies, says Research Affiliates’ head of equity research, Vitali Kalesnik.
“Implementation costs fall into two categories,” says Kalesnik. “The first is the cost of replicating a particular index, reflected in the expected performance slippage against the index when an external asset manager is awarded the mandate of tracking the index. Often this is close to zero.”
“But a second, less visible cost is potentially more important but, unfortunately, often gets ignored by investors. On the rebalancing date the prices of stocks entering the index get pushed up, and those of stocks exiting get pushed down. The effect of these temporary price changes then gets baked into the index.”
There are studies dating from the early 2000s aiming to quantify this implicit rebalancing cost for standard, capitalisation-weighted indices.
Joop Huij, head of factor investing research at asset manager Robeco, aimed to quantify the rebalancing costs for smart beta benchmarks in a paper he recently co-authored with PhD candidate Georgi Kyosev.
“We estimate that factor indices, which are concentrated in small and mid-cap stocks, face much larger rebalancing-related costs than cap-weighted indices,” says Huij.
“For example, we have found strong evidence of abnormal trading patterns around the rebalancing events for fundamental and minimum volatility indices. These led to strong price pressures on the stocks involved, up to a 100 basis point price increase on average for stocks entering the index.”
Huij, whose firm manages factor strategies based on proprietary risk models and indices, therefore warns of the dangers of following publicly-disclosed indices when buying factor strategies.
“Generic indices are cheap because they are scalable, but when the same index is sold to an ever larger number of investors the crowding effects related to index rebalancing increase,” says Huij.
Other industry participants argue that the inherent diversity of factor approaches helps ease concerns about index-related crowding.
“Just because there are more factor products on offer doesn’t necessarily mean there’s a significant overlap in the names that are being captured by different strategies,” says Ana Harris, European head of equity portfolio strategy at State Street Global Advisors.
Some argue that the risk models used by investors in quantitative portfolio construction may incur greater crowding concerns than factor or smart beta indices. Analytic Investors’ Harindra de Silva notes the level of concentration in factor-risk modelling software, which is used by active and index-following asset managers alike.
“The two most popular industry models, Barra and Axioma, probably have 90% market share and using them leaves investors more exposed to crowding and model-related turnover,” says de Silva.
A representative of one of the pioneers of smart beta argues that crowding questions are now central to whether non-standard index approaches live up to their promise in future.
“In our opinion, many investors do not pay enough attention to potential crowding issues,” says Research Affiliates’ Vitali Kalesnik.
“We think we’re just at the beginning of the trend towards factor investing. As assets increase, questions of capacity and trading costs will determine whether investors actually benefit from these strategies.”