Factor or anomaly? Analysing ETF flows

by , 31st July 2018

A research paper issued by quant strategists at Deutsche Bank in the US has been gaining attention in the ETF world for what it says about fund flows at the time of every quarterly portfolio rebalancing.

It’s what would appear to be an arcane area but stick with us. The analyst note – entitled ‘What happens when the world goes passive?’ – comes up with what would appear to be a relatively simple proposition.

Every quarter, the major indices around the world perform a rebalance to reflect the ebbs and flows on the share prices of their constituent companies. The strategy is to identify the companies whose shares will be exiting indices – and hence will be sold by the passive funds – and buy these while also selling the ones they are buying.

It sounds counter-intuitive but of course there is nothing intuitive about passive investing. Indeed, the Deutsche team suggest they have found a new investment anomaly, a way of profiting from the distortions caused by the sheer scale of passive strategies in the US market.

The analysis starts by making the essential the clam that like active fund strategies – which definitely move markets and to some extent are designed to do just that – passive strategies in theory should do no such thing.

In order to back up their claims, the Deutsche team do some pretty serious number crunching, looking at ETF composition, descriptive information and constituent level stock data for each ETF starting in 2005.

To calculate ETF ownership, the team then aggregate positions across each US stock held by US-listed equity ETFs. International stocks held by US-listed ETFs or US stock holdings from non-US-listed ETFs are not included in our database.

Seasonal measure

Their first conclusion is that ETF flows display seasonality in lie with index changes. Such, at least notionally, would seem obvious. But the findings from the research into the data comes up with some surprising news; stocks with large positive ETF flow reliably underperform by 9% in the year following an index rebalancing event; stocks with large negative ETF flow reliably outperform by 11% in the year following an index rebalancing event.

Deutsche Bank put forward their own reasoning for the moves. “Our story is simply market participants not being able to discriminate between information-led active and information-less passive flows,” they suggest.

“Prior to the index event, arbitrageurs including broker-dealers hold inventory in anticipation of forced buying and selling by passive asset managers. The inventory build-up increases prices prior to the index change – following the index change these positions are unwound.”

They then turn to what they suggest is a contrarian ETF long-short trading strategy which takes as its premise the idea that ETF flows affect the price of securities and then subsequently reverse in the following months.

As they say: “Specifically, we conjecture that market participants do not realize that passive investors are buying or selling securities. Hence, an ‘information-less’ buy order by a passive investor due to flows into an index or an index reconstitution is construed the same way as an ‘information-led’ buy order from an active investor.

“Thus, market participants overreact to passive flows and eventually learn there is little price discovery associated with passive flows and prices revert.”


Among the intriguing elements of this research is that until 2009 this strategy didn’t turn a profit (according to Deutsche Bank’s backtesting) which suggests it might just be a bull market phenomenon.

However, the theory was backed up by a piece of research issued by S&P in the middle of July which also suggested that price fluctuations caused by ETF flows were only short-term.

The model built by S&P quantitative research analyst Daniel Sandberg showed a single-day impact of up to 370 basis points a day on an individual security and up to 250 baiss points impact on the index itself. “Analyses indicate the effect is transitory and reverses over a period of 3-5 trading days,” Sandberg wrote.

But looking back earlier this year to the February volatility bump when the market correction was accompanied by a $25bn outflow of assets from ticker SPY, the SSGA S&P 500 Trust ETF, Snadberg’s modelling suggests that as much as one- third of the pullback was due to price pressure from ETF trading and that securities more sensitive to ETF flow underperformed.

Talking to Bloomberg, analyst Ronnie Shah from Deutsche Bank was cautious about his findings and shied away from suggesting he had uncovered a new factor. “Whether or not it’s a factor is debatable,” Shah said. “But this is an anomaly that we found and it’s worth exploring further.”