The answer is that Goldman never had much in the way of net mortgage exposure to begin with, and could therefore turn bearish quite easily. Banks which had tens of billions of dollars of illiquid mortgage bonds on their balance sheets, by contrast, had no real way to put on a bearish bet.
More generally, pure investment banks, like Goldman, always claim to be in the moving business as opposed to the storage business. Today’s NYT story shows that there are limits to how true that is — Goldman had significant long-term mortgage exposure which it hedged by building up a short position in synthetic CDOs. But at least its net position stayed very small.
Competitors like Merrill Lynch, by contrast, found themselves taking enormous amounts of mortgage-bond exposure onto their own balance sheets just because no one else was willing to buy the lowest-yielding tranches of their mortgage bonds. It was that exposure which ultimately doomed the bank. Merrill might have been talking the moving-not-storage talk, but in practice it was acting as a waste dump for all manner of risky paper that the bond desks couldn’t move.
And big commercial banks with investment-banking arms, like Citigroup and UBS, never even claimed to be movers rather than storers: they actively sought out high-yielding triple-A paper as part of their business model. So long as the yields were higher than their own internal cost of funds, they considered such a position to be inherently profitable — and the bigger the position, the more profitable it would be.
There was also a difference in the type of mortgage paper which each bank had in its long portfolio. While Merrill, Citi, et al were stocking up on the lowest-yielding debt, Goldman I suspect had mainly the equity tranches of the bonds it underwrote: the highest-yielding, riskiest paper. Because the equity tranche is inherently extremely risky, even in good times, it’s only natural for any bank owning such paper to want to hedge that exposure.
And this is where I think that Goldman might have gotten a little lucky. Because equity tranches can all go to zero very quickly, you need a fair amount of CDS insurance to hedge that exposure. But once you’ve bought that CDS insurance, it will continue to rise in value as the housing market implodes, long after your original equity tranche has been wiped out.
Let’s say the housing market is at 100, and you have $3 million of bonds which will get wiped out if the market drops to 97. So you hedge that exposure by buying credit default swaps which pay out $1 million for each point that the housing market drops. (This is massively oversimplifying, but work with me here.) If the market falls to 97, then you lose $3 million on your bonds, while making $3 million on your CDS — you’re even. But if the market falls even further, to 70, then you still lose only $3 million on your bonds, while making $30 million on your CDS — gravy! You don’t even need to buy any more insurance to make that extra money, you just need to keep holding the insurance you already own.
On the other hand, let’s say the housing market is at 100 and you have $50 million of bonds which will get wiped out if the market drops past 75. At that point, the temptation is to do nothing at all, since hedging costs money and your models tell you that the market will never fall that far. And by the time that the market starts falling, insurance is so expensive that you can’t afford it any longer.
More generally, if you’re starting from a market-neutral position, it’s easy to go short. But if you’re starting from a large net long position, it’s much, much harder to do that. Almost impossible, really, especially when the market seizes up and there’s no liquidity any more.
And while I’m on the subject, there’s one thing worth adding: that none of this would have happened if Goldman hadn’t been so mind-bogglingly enormous. Goldman could hold on to the short side of all the synthetic CDOs it was underwriting precisely because in some distant other arm of the Goldman empire, a mortgage desk had managed to make money by introducing lots of mortgage-backed bonds onto the bank’s balance sheet. So the logical thing to do was to create a new desk which could make money by introducing lots of mortgage CDS onto the balance sheet. That way Goldman made money on both trades, while its balance sheet was hedged and canceled itself out.
If however there was a cap on bank size, or punitive capital requirements on balance sheets above $100 billion, then those kind of shenanigans would be much harder to pull off. Goldman would then do neither the original mortgage deals nor the subsequent synthetic CDOs, and the world would be a better place.