Goldman Would Be Happy To Take Some Correlation Off Your Hands


Goldman Sachs Portfolio Strategy Research has a fascinating research piece out today on equity correlation markets. It does good work as a piece of research because (1) if you like equity derivatives, it's got all sorts of fun charts and technical stuff and (2) if you don't, it's got a hard sell: trade equity corr with Goldman!

There are many market participants that are affected by the level of equity correlations but are not yet trading correlation actively. So far, the market has been primarily one-way; banks selling correlation and hedge funds and proprietary trading desks buying it for the positive carry. We believe that the correlation market can become a new area in which institutional investors could add alpha.

The equity correlation market, which is pretty niche-y, lets investors bet on how dispersed the returns of stocks will be in the future. And the trade to make now, Goldman thinks, is to sell correlation, which "appears attractive," meaning that current implied correlation is much higher than they think realized correlation will be in the future:

The rationale for doing the trade, other than just attractive price, is a little unclear. The analysts write that "long-only stock pickers, equity hedge funds or their investors (e.g. funds of hedge funds) are implicitly short correlation. They could hedge the risk of disappearing stock picking opportunities in ‘macro markets’ by buying correlation at low implied levels." But current implied correlations are not low - they're close to multi-year highs. And Goldman says the right trade is to sell correlation, not buy it.

But that trade also could make strategic sense for a long-biased hedge fund manager. If you can't make any money picking stocks because everything just moves in lockstep with the market, you need to juice your performance/justify your fees in other ways. One way to do that is to get paid premium for selling correlation - and with implied correlation very high right now, you can get paid a lot for selling correlation. Think of the message as being that correlation trading is countercyclical to the rest of your stock-picking business: in good times, you should buy correlation at low implied levels as a hedge against correlations going up and ruining all your careful stock-picking work, while in bad times you should sell correlation because your stock-picking isn't working anyway and you might as well make money somewhere.

So what's in it for Goldman? Well, to be fair they actually lay out their biases pretty clearly:

Implied correlation levels are driven both by expectations and supply and demand for correlation risks from structured product issuance and derivative flows. In equity derivative markets, on an aggregate basis there is usually excess demand for index volatility and excess supply of single-stock volatility. This leaves derivative dealers net short correlation and increases in correlations can pose a risk for them – with risk budget constraints and higher capital requirements banks are usually keen to reduce this correlation risk exposure. As banks are keen to hedge their short correlation exposure, implied correlation might trade at a higher premium compared to realised.

What this means is that dealers get most of their correlation risk, not from "correlation products," but from old-fashioned options on individual stocks and stock indices. Index vol markets tend to consist of "large institutions, pension funds and insurance companies that want diversified exposure to hedge equity risk." So they buy volatility (index puts, etc.). Single-stock vol markets mostly "generate yield by overwriting existing single stock positions. This generates more premium than selling volatility on the index, and allows investors to monetize their insight on single stocks." So derivatives dealers end up short a lot of index options and long a lot of single-stock options. If correlation goes up, index volatility goes up faster than single-stock volatility, because single-stock moves are less likely to offset each other than they would be in a low-correlation world. So banks get hosed on their volatility trades.

And they get hosed at exactly the wrong time. Correlations go up at exactly the time when everything else is going wrong for banks, making it a bad idea for them to be short lots of correlation in good times:

In other words, at normal times big banks tend to be long equities, long credit - and short correlation. When things fall apart, equities go down, credit gets wider - and correlation goes up. So they lose everywhere. As a matter of positioning, it makes a lot of sense for banks to find new customers to sell them correlation, since it's a good way to manage risk against future downturns. But the fact that they're advertising to buy correlation at a premium now - with markets screwy and volatility and correlation at historically high levels - suggests that Goldman research, at least, still sees some rough times ahead.


Greek Debt Management Guy Thought His Partners In Obscuring The National Debt Would Be The Last People To Rip Him Off

Bloomberg's story about the Greece-Goldman swap-debt-whatever kaboodle, so let's talk about the philosophy of derivatives for a minute. First the story: Greece’s secret loan from Goldman Sachs Group Inc. (GS) was a costly mistake from the start. On the day the 2001 deal was struck, the government owed the bank about 600 million euros ($793 million) more than the 2.8 billion euros it borrowed, said Spyros Papanicolaou, who took over the country’s debt-management agency in 2005. By then, the price of the transaction, a derivative that disguised the loan and that Goldman Sachs persuaded Greece not to test with competitors, had almost doubled to 5.1 billion euros, he said. There are at least three reasons to use derivatives. First you could be into some actual informed shifting of risks from those who want to pay to get rid of them to those willing to be paid to bear them, or from those who have Risk X and want Risk Y to those who etc. Boy are there a lot of textbooks that talk about this. And I suppose it even happens sometimes. You could imagine that a vanilla interest rate swap entered into by a corporation on its bonds or credit facility could qualify as this. I guess people who trade listed options to do covered-write strategies or speculate on takeovers or whatever fall in this category, maybe modulo the "informed." (Sometimes!) Then there's tax and regulatory arbitrage. This is time-honored and much of it, particularly the stuff with the best names, is focused on tax dodging, but there are also various other regimes - securities laws, accounting, whatever - that you might want to get around with derivatives. Paying $10 for CDS with a maximum payout of $10 purely to lower your capital requirements is a recent amusing/egregious example. The thing that wasn't mentioned in the CFA Level I derivatives primer is principal-agent arbitrage. This is ... first of all, let's say this isn't a derivatives issue, or a financial-industry issue, it's like a life issue. (Some would say it's why there's an M&A business, for instance.*) But it's also a derivatives issue! And you can see why if you're as baffled as I am by the Bloomberg story. So this: