ETFs under the spotlight amid high correlation

The sugar-fuelled riot that is typical of a child’s party becomes something quite different if the partygoers can be persuaded to play ‘Simon Says’: one of them calls out instructions to the rest, who obediently stand on one leg, turn around, or clap their hands as required. Where there was bedlam, order reigns. Something similar is happening in the world’s biggest equity markets – six-month realised correlation on the S&P 500 index breached the 60% level at the start of June and has remained above it ever since. But no-one knows who is playing the role of Simon, and some observers argue it might take a catastrophic collapse to bring the game to an end.

“Everyone is perplexed by these risk-on or risk-off days, where it feels like you can actually see the correlation increase because trades are all working or not working – and those days seem to come in droves rather than dribs and drabs like they used to,” says Nicholas Colas, chief market strategist at financial IT company BNY ConvergEx Group in New York. “People have theories about what’s happening and why, but there is no definitive explanation.”

One theory that has gained traction in recent months is that index investing – and specifically, the popularity of exchange-traded funds (ETFs) – is behind the move. Some research supports that conclusion, but some doesn’t – and the ETF industry is firmly on the side of the latter. “I’ve heard those arguments. I think they come from the same people who believe there was an extra gunman who shot Kennedy,” says one London-based head of ETFs at a major dealer.

“This theory makes no sense to me at all,” adds Christos Constandinides, ETF strategist at Deutsche Bank in London. “The ones promoting it are the same people who have been promoting active investing – I’m sure they’re quite disappointed to see money flowing into funds that charge people 20–30 basis points, instead of coming to them so they can be charged 4–5%.”

It is a touchy subject, because the ongoing high correlation episode is different from others the market has seen, and it is making investors and risk managers jumpy. In the past, correlation has tended to be high when volatility is also at elevated levels. There’s a common-sense explanation for this: if markets are moving dramatically, investors will jump on the bandwagon, buying or selling at the same time in response to the same big-picture development. But while correlation and volatility both rose as the eurozone debt crisis rocked financial markets in May and June, volatility has since drained away – the Chicago Board Options Exchange Vix index, which measures market volatility as implied by 30-day S&P 500 option prices, hit a 13-month high of 45.79 points on May 20, but had fallen back to 26.3 points by August 24.

This has had some analysts scratching their heads. Part of the explanation could be the ETF market, believes Maneesh Deshpande, head of US equity derivatives strategy at Barclays Capital in New York. “The simplest way to think about it is to imagine some money being allocated to equities. If it all ends up with active managers, who all have different views, it will flow through to a group of specific stocks. Those stocks will move while others will not, so the effect on the index as a whole won’t be that great. But if all the money goes into an ETF that tracks an index, then the amount of money going into each stock only depends on its weight in the index and the whole market will move together,” he says.  

In a client note published at the end of April, derivatives strategists at Royal Bank of Scotland (RBS) also drew a link between the popularity of ETFs and the rise in correlation – and argued the relationship, if there is one, would spring from the creation or redemption of ETF units.

“Once created, an ETF trades just like a share and is free to trade independently of other single stocks. But if someone comes along and orders 1,000 new ETF units, then the provider has to go into the market and buy all the shares in the index that is being tracked – either directly or synthetically – irrespective of whether one is good value and another is bad,” says Stephen Einchcomb, head of equity derivatives strategy at RBS in London. The same is true of ETF redemptions – if ETF units are being destroyed, the underlying equity has to be sold.

The analysts at RBS developed this argument in two ways. First, they compared S&P three-month realised correlation against the average turnover ratio (calculated as the daily turnover of index futures and ETFs versus the turnover of underlying shares). They found no obvious relationship between the two over the past decade – the turnover ratio climbs increasingly quickly before levelling off over the past couple of years, while the correlation of the S&P 500 ends high but endures serial peaks and troughs. Then Einchcomb and his colleagues compared identical correlation data with the change in the turnover ratio – a rough measure of the growth in index trading versus single-stock trading, and one that gets closer to capturing trends in ETF creation and redemption. This time, the two lines track each other closely.

The RBS report admits the apparent relationship could be a coincidence, but Einchcomb thinks there could be something deeper going on. “Investors have always been able to take index positions through futures and swaps, but my feeling is that futures don’t have the same kind of direct market impact investors get with ETFs through the creation and redemption mechanism. There isn’t a need to buy or sell instantaneously into the market in the same way – and it’s why we feel there is a high daily realised correlation,” he says.

The fear among some observers is that if correlation remains high, any new jolt to market confidence could produce a collapse more devastating and enduring than the May 6 ‘flash crash’, which saw the Dow Jones lose and then largely regain 9.2% of its value in a matter of minutes. “If ETFs drive the correlation up in normal times, wait for the impact when the market is stressed,” says the chief risk officer at one large hedge fund.

Others have less apocalyptic complaints. In its second-quarter report to investors, privately owned Canadian money manager Friedberg Mercantile complained that its equity hedge programme had suffered an unprecedented breakdown – the strategy involved taking long and short positions on 45 to 50 US stocks and staying market neutral. Friedberg claims the programme has now suffered five consecutive quarters of losses – the worst losing streak in the strategy’s 19-year history – and blames the rise of index trading.

“The growing popularity of exchange-traded funds, among other things, has levelled the returns of individual stocks and sectors. By maintaining (or freezing) over- and under-valuations, this phenomenon has become a nightmare for stock and sector pickers. And while it is true this phenomenon raises hopes that, at some point in the future, large profits will be able to be made by exploiting these inefficiencies, such may not be the case for quite some time,” the report claims. Friedberg has since scaled down the programme. The firm declined to comment for this article.

But does the ETF link stand up to scrutiny? Equity derivatives traders at Société Générale Corporate and Investment Banking (SG CIB) – when invited to comment for this article – decided to conduct some research of their own, and produced analysis for Risk they believe rebuts claims of a relationship between index trading and correlation. “This is a really hot topic, because there’s obviously something going on. But it’s my belief you can’t blame it on ETFs,” says Stéphane Mattatia, head of financial engineering and advisory in the global equity flow division at SG CIB in Paris (see box).

Others contest the theory from a more intuitive perspective. Because investors see stocks marching in step, they recognise there is no point betting on one name or one sector – it’s easier and safer to buy the index, they argue. “When correlation is high, ETFs are popular. It’s not the case that ETFs cause correlation to be high,” says Deborah Fuhr, global head of ETF research at asset manager BlackRock in London.

RBS’s Einchcomb accepts there is a chicken-and-egg element to the relationship between ETFs and correlation, but insists demand for index products is a causal factor. And even some in the ETF community accept their market could eventually drive correlations higher.

“As long as passive products are a relatively small proportion of the market, people will continue to allocate to them until we see a self-fulfilling effect and all the markets start following each other. At that point, active investing will start to look more interesting again – but we’re not there yet. I think we’re about three to five years away from that level of correlation,” says Michael John Lytle, marketing director at ETF provider Source in London.

The ETF relationship might be unproven, but high correlation will continue to stimulate comment – particularly about when and how the episode will end. RBS’s Einchcomb expects to see the return of fundamentals-driven trading at some point: investors will identify companies that shouldn’t be trading in step and will bet on greater dispersion. However, there’s a hitch with this scenario. The long/short pairs-trade has traditionally been conducted on a leveraged basis, but financial institutions have been reducing the amount of leverage in the system, which could make it harder for this strategy to have an impact, he says.

ConvergEx’s Colas also says he would expect correlation to return to lower levels almost naturally in ordinary circumstances, but warns the market’s self-correcting mechanisms may have been weakened. “This is not the way markets are supposed to work. You’d hope the anxiety would drain away and investors would start taking a view on individual stocks again. But the obvious risk factor is there’s less and less money allocated to equities every day, so there are fewer investors left to take those positions,” he says.

This is another of the current market’s mysteries – as of August 24, the S&P 500 had increased by almost 4% since the start of June, but traditional investors are pulling money out of the asset class. Coupled with the high correlation theme, this has some disturbing implications. According to Charles Biderman, chief executive of Sausalito, California-based equities research firm TrimTabs, it is impossible to say what is supporting the market’s recent rise. “We’ve had continual outflows from US equity mutual funds, from ETFs and from hedge funds. The performance of pension funds has been so good they haven’t needed to buy more equities. In short, we can’t find any money that came in. There is no bid. And what I’m expecting – ultimately – is a wave of natural selling. And if whatever buyers we do have then step away – well, you could have a market collapse that makes the one in May look like child’s play,” he says.

For now, the market’s game of copycat doesn’t look like ending and traders are prepared for a long haul – one senior equity derivatives dealer at a large European bank says he can foresee elevated levels of correlation lasting for months, if not years. Market participants will be hoping it doesn’t take a crash to finally burst the correlation bubble.

SG CIB tackles ETF theory

Stéphane Mattatia and colleagues in the equity derivatives team at Société Générale Corporate and Investment Banking (SG CIB) in Paris had heard arguments about the relationship between exchange-traded funds (ETF) and high correlations, but hadn’t spent time looking for evidence. When contacted for this article, they decided to get out the magnifying glass and deerstalker. “We wanted to do some investigative work,” says Mattatia.

The approach was simple – if ETFs were driving correlation, Mattatia felt it could be tested in three ways. First, an index widely tracked by ETFs might record higher levels of correlation than a less popular one. SG CIB chose to compare the Dow Jones Eurostoxx 50 and the Eurostoxx 600 indexes, finding that realised correlation on both climbed in a similar way. Strike one against the ETF link theory.

Second, if ETFs lead to higher correlations, it should result in similar correlation behaviour for all the sectors within an index. Mattatia compared 10 years of correlation data for six different sectors – autos, healthcare, technology, basic resources, oil and gas, and utilities. The first three showed no discernible relationship, although all have been higher for the past six months. The second three matched each other beautifully. Strike two, Mattatia argues.

“I love this graph – it’s really striking, and it shows you have a few sectors where a small number of macro variables drive the performance of all the constituents. That makes sense. If you think about oil companies or mining companies, they all depend heavily on continuing Chinese economic growth. The same is not true if you look at, say, pharmaceuticals, where idiosyncratic factors are far more important. But that’s not what you’d expect to see if ETFs were causing correlation,” he says.

Finally, if ETFs are causing higher correlation within an index, then new joiners to the index ought to become more correlated with the rest of the index, and those leaving the index ought to de-correlate. The results here were mixed. There was no evidence joining the Dow Jones Eurostoxx 50 resulted in companies becoming more correlated with the other index members, but firms leaving the index became less correlated – something Mattatia says could offer a scrap of evidence in favour of the ETF theory or could just be a sign that companies dropping out of the index tend to be experiencing some distress and are less likely to keep step with other companies. Strike three and – in Mattatia’s opinion – that’s out.