Correlation: weirdest of the weird sisters

Correlation: weirdest of the weird sisters

“Like volatility, correlation can itself be treated as an asset class and traded in its own right.”
JP Morgan

“Investing in ‘implicit’ assets such as correlation has become an accepted addition to traditional investments within a portfolio, such as equities and bonds.”
Barclays Capital

“‘Hidden assets’ (such as correlation) are usually uncorrelated with traditional assets, and therefore provide an attractive diversification potential.”
Société Générale

As you can read, the geeks are unanimous.

Correlation is the next big, obscure thing for which they will mould a compelling narrative. As I write, the hedge funds are being administered the spiel. They are told correlation holds “attractive diversification potential”, “large arbitrage opportunities” and “good value in the current environment” – all faithful quotes from banks active in this burgeoning little market.

Also on today:
'Is correlation killing metals too?' by Sian Williams

Correlation is the third of a spooky sisterhood, with dividends and volatility. Where correlation is today, dividends were in 2007, and volatility was in 2003. Namely an OTC swap-based asset class looking for its big break – a move to exchanges.

Whether you call them partial, component, implicit or hidden asset classes, one thing is clear – these things are becoming big business.

We trade conventional asset classes – equities.

We then trade derivative asset classes – options on those equities.

We now increasingly trade the partial derivatives of those derivatives – volatility, correlation and dividends on those options on those equities.

The breathtaking sophistication of the market is such that one wonders if the investors can keep up.

In the summer blockbuster film Inception, I confess to having got a little lost once the third dream – the dream-within-a-dream-within-a-dream – began. The way each pre-existing, prerequisite dream level impacted on the next was a little complex for me. Thankfully, with the aid of the director’s careful hand-holding and exposition, it all made sense in the end.

As a consequence, a film with a complicated premise, with “art house flop” written all over it, has turned in more than $750m.

Banks, however, seem to be doing just the opposite. Rather than unravelling the complexity, they just portray these assets-within-assets-within-assets as simple instruments. No more than absolute and relative value opportunities.

Let’s see if their latest offering, equity correlation, stands up to scrutiny.

The perfect moment

In September 2008, volatility (using the Vix as a proxy) rose 200% while all else fell or failed. Money can’t buy advertising that good.

Today, volatility is discussed in the same breath as credit or commercial property – just another asset in the mix. As has been demonstrated countless times before, an asset that has risen during the most recent stockmarket crash is often seized on by sell-side opportunists.

Enter correlation. The backdrop is less dramatic than that of the Vix, but nevertheless the correlation idea is beginning to catch on as coordinated movement between options, and between underlying stocks, becomes more pronounced.

Russell Investments told us 2010 was going to be “the year of the stock picker”, where stocks and options with compelling fundamentals would outperform the pack.

It hasn’t worked out that way. The S&P Implied Correlation Indices (tickers: JCJ and KCJ) show that, as John Authers recently put it in the Financial Times: “The world trades in unison. Many bright people spend many long hours pondering which stocks, sectors or countries to buy but, of late, it scarcely seems to matter.”

The near date index (JCJ, tied to options with a January 2011 maturity) rose to a record high of 81.09 last month, suggesting an unprecedented level of correlation between stocks in the S&P 500.

If everything is moving together, one only needs to pick the right sectors, regions or indeed asset classes. Bottom-up fundamental analysis has never been less effective – indeed, some have declared the death of stock-picking.

Why is this happening?

Wagging the dog

The unflagging popularity of exchange-traded funds hasn’t been without consequence – the tail now wags the dog. ETFs and other ‘delta one’ trading approaches have begun to move the markets they are intended to track.

See JP Morgan’s recent global equity derivatives review: “Most players have high ambitions in delta one products, ETFs in particular, where volume expected to grow further by 20% per annum.”

This isn’t a bad thing. ETFs, as Morgan goes on to say, offer broad market access, liquidity, low costs and tax efficiency.

But as they grow in power, fundamental differences between companies may be less keenly reflected in their stock prices.

Imagine a big fall in assets under management (AUM) for State Street Global Advisors’ SPDR S&P 500 Index ETF (known as Spy after its ticker).

Spy is the most popular ETF, and the most heavily traded security in the US. It reaped more than 10% of total domestic equity trading in August according to data from Abel/Noser, with over $19.5tr of trades a day – more than five times as much as Apple, the next most popular stock.

Yet a big withdrawal of money from Spy would not hit the most overvalued stocks worst. The ETF would sell all stocks in proportion to their index weighting.   

What is correlation?

The realised correlation of a pair of stocks measures how much the stock prices tend to move together. The realised correlation of an index, such as the S&P Implied Correlation Index, is simply an average across all possible pairs of constituent stocks.

Correlation trading is a strategy in which the investor takes exposure to the average correlation of an index. Dispersion trading is betting on low correlation.

The key to correlation trading is understanding the principle of diversification – that the volatility of a portfolio of securities must be less than the average of the volatilities of all the individual securities in that portfolio. Some of the individual stocks’ movements will cancel each other out.

However, this does not happen if all the stocks are perfectly correlated. In that case, the portfolio’s volatility would be identical to that of the components.

Normally, an option on an index is cheaper than a portfolio of options on its component, because its volatility is lower. But the extent to which it should be cheaper is less if the stocks are more correlated.

Therefore if you go long the volatility of the index and short individual volatilities, by buying a call on the index and selling calls on the stocks, you are long correlation.

If correlation turns out higher than the market expected, the index volatility should be lower than those of the components by less than expected, so the undervaluation of the index option relative to the single stock options should decrease. Quids in!

Naturally, going long single stock option vol and short index vol is a bet on short correlation.

It’s important to remember, as Nassim Nicholas Taleb explains in Dynamic Hedging, that correlation is not asset X moving 1% for every 1% move in asset Y. The two assets can be 100% correlated and asset X moves 2% for every 1% move in asset Y if asset X has twice the volatility of asset Y.

Hence stocks tend to rise and fall together when funds like this are most popular.

Barclays Capital believes that “ is perhaps driven by the corresponding increase in popularity of index funds and ETFs.”

So, is going long correlation really a bet that Spy’s AUM will hold up? Hmm...

With JP Morgan forecasting ETFs to keep growing, it’s not unreasonable to expect equity correlation to get stronger. Sounds like an interesting wager, no?

While it’s tough to say whether correlation will top 81.09 any time soon, no one I have spoken to disputes that correlations in general are rising. And sustainably so.

With this in mind, how exactly does one take advantage of this rising ‘delta one demographic’ trade?

The death of pairs trading?

In today’s market, the cruel truth may be that the simplest way to go long correlation is to short market-neutral (pairs trading) funds. These funds have the most to lose if correlation is high.

Equity market-neutral funds aim to short overvalued stocks and buy their undervalued counterparts. The hope is that fundamental value will win out over time. However, if the market is content with stubbornly ignoring value, these funds suffer.

For example, as commentators including Tyler Durden have pointed out, the market-neutral portion of the $400m Friedberg Global-Macro Hedge Fund has been losing money for the past five quarters.

This, for reasons discussed in the previous section, is to be expected, as Friedberg concedes in its second quarter 2010 report: “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 undervaluations, this phenomenon has become a nightmare for stock and sector pickers.”

So, long correlation equals long Spy AUM equals short the constituents of the Hedge Fund Research Equity Market-Neutral Index? An index which lost value in 2008, 2009 and is nearly 2% off so far this year? Hmm...

Fish not flesh

But before I delve too far into realised equity correlation, the experts remind me that implied correlation is a derivative, not an equity trade.

Put simply, to trade implied correlation proper, investors can do one of two things.

1 Buy a correlation swap

2 Buy a portfolio of options on the index and sell a portfolio of options on the individual constituents of the index (see box above)

Of the two alternatives, the first is over-the-counter and the second isn’t.

There used to be a third option in variance swaps trading. One would buy variance on the index and sell variance on individual components. But it proved difficult to hedge. As Sebastien Krol commented in Derivatives Week: “During the crisis, static hedging assumptions failed spectacularly, as the difficulty of practically hedging the variance swaps and the lack of understanding of the risks associated with the product signed its death penalty.”

How does implied correlation trade relative to realised? In other words, does it reflect reality? The narrative is not dissimilar to that of implied and realised volatility.

Implied volatility tends to trade too high (although of late, both realised volatility and realised correlation have been coming in higher than implied).

On the whole, implied correlation is also often mispriced by structural buyers and hedgers who care little for value.

As Derivatives Week has pointed out, structured equity product investors are often long correlation, without even knowing it. While they rarely bother hedging this exposure, the structured product issuers, who are short correlation, do seek to clean this risk off their books.

As they are all trying to buy correlation exposure, the market is one-sided and the price of correlation too high.

So the banks would love to persuade hedge funds and prop trading desks to start “recycling” their short correlation exposure. And the opportunity seems compelling, as there is a supply-demand imbalance.

But these narratives that say asset classes have got to thrive because there is a natural buy and sell side only remain true up to a point. In cases like this, as the banks’ argument that there is an imbalance becomes accepted and more people start to trade the asset class, it ceases to be a one way bet.

Volatility is so heavily traded now that the easy money mean-reversion trades – straddles and strangles – are harder to come by.

And so it may be with correlation. If it follows the same trajectory as vol, the juicy opportunities touted by the banks will soon be gone as hedge funds crowd the trade.

Where have all the cowboys gone?

But there’s a twist in the tale. Just at the moment, the banks pushing correlation to the masses are finding it tough to make sales.

“According to JP Morgan,” related the Financial Times in September, “there are now only about 20 hedge funds still actively trading correlation.”

Well, that won’t do at all. Correlation, it seems, is struggling to be recycled. Despite clear signs of correlation rising, and real changes in the market microstructure with the rise of ETFs, the hedge funds aren’t biting.

Could it be, for once, that investors have realised their own limitations and have decided correlation is just too spicy for them?

Alternatively, complacency on the banks’ part could be a factor. Jakob Bronebakk of structured product firm Jubilee Financial Products believes they ought to take this risk more seriously. He draws a contrast between correlation and its sisters, volatility and dividends.

“Most of the larger banks in the market,” Bronebakk says, “have active enough market making desks on the flow side to hedge their dividend exposures, and most do enough structured flow to get rid of their vega even without using the new volatility futures, i.e. just using variance swaps.

“Correlation, on the other hand, gets treated – or certainly has in the past – in a much more haphazard way when pricing, even for relatively sensitive structures.”

He points out some high profile casualties of correlation over the years, notably Lehman Brothers’ $300m-$400m whack in 2007. This is a serious concern to banks struggling with too much unhedged exposure. Were correlation trading more popular, such episodes might be less common.

Risk is so last cycle!

However, it may be a little ivory towerish to discuss the niceties of appetite for correlation when the big thing happening out there is a raging, market-wide hunger strike from taking risk. Just look at the government debt rally.

It may be, therefore, that the correlation game is merely taking a break. Investors have not decided to stay on the straight and narrow path, shunning dangerous new asset classes – they are just sitting down for a bit.

When risk appetite returns, the great correlation sale may be back on.

If so, the warning lights should flash on too. One of them says ‘model risk’ – the very real fear that these swaps are too clever for the fund managers. A fund with inadequate models may misprice or mishedge a correlation position.

The man in the street – even the hedge fund man in Berkeley Street – simply doesn’t have the infrastructure or relevant education to manage the complex risks associated with correlation.

It is worth remembering that one of the hothouse flowers of the mid-2000s structured credit mania was trading correlation on debt portfolios, using tranched CDOs. The air was thick with talk of “long mezz, short equity” trades and the like. Then came the flood...

It might be tempting to hope that the present lull in correlation trading means investors have hit their intellectual ceiling and will stop there. However, experience suggests that, sooner or later, they will plough on.

In that case, it will be better for everyone if the banks make an effort to genuinely and dispassionately educate investors about the risks and complexities of correlation, rather than just plugging it.

This might help the banks themselves – the renaissance is likely to come sooner if they do it.

With equities, when a new stock is issued, there is a stabilisation period. For a while, the bookrunning bank will buy the stock of the company to settle the price, so that investors can consider the opportunity in the safety of a managed, adequately liquid market.

A similar commitment to correlation and other such exotic asset classes might just help their adoption rates.

That’s the kind of hand-holding and exposition that might just take this asset class from art house flop to blockbuster status.

Theo Casey, a MoneyWeek managing editor who is writing a book about equity volatility, is a Futures & Options Intelligence columnist. The opinions expressed are his own.