Synthetic CDO's and Bubbles

There has been a chorus lately arguing that synthetic CDO's fueled the bubble and made the subprime debacle much worse. For example, Felix Salmon writes:

Then, after AIG exited the market, everything should have ground to a halt. But it didn’t, because banks continued to build synthetic subprime CDOs out of the credit default swaps which were being bought by Greg Lippmann and others. The demand for those CDOs from investors like Wing Chau was enormous, and helped to ratify the valuations that everybody else was placing on their own subprime assets. Remember that this is a market with almost no pricing transparency in the secondary market: because all securitization deals are unique, the only way to get a feel for the health of the market is by looking at where primary deals are pricing. Whenever anybody said that the marks being put on subprime assets by banks and hedge funds were delusional, it was easy to point to the booming market in synthetic subprime CDOs to prove them wrong. No one, of course, remarked on the irony that the synthetic subprime CDO market was only booming because John Paulson and others were providing a huge amount of demand for bearish bets.

Or Roger Lowenstein in the NYTimes,

While such investments added nothing of value to the mortgage industry, they weren’t harmless. They were one reason the housing bust turned out to be more destructive than anyone predicted. Initially, remember, the Federal Reserve chairman, Ben Bernanke, and others insisted that the damage would be confined largely to subprime loans, which made up only a small part of the mortgage market. But credit default swaps greatly multiplied the subprime bet. In some cases, a single mortgage bond was referenced in dozens of synthetic securities. The net effect: investments like Abacus raised society’s risk for no productive gain.

And the famous ProPublica article on the Magnetar trade makes the same point too. Yves Smith at Naked Capitalism I think is the originator of this argument, or at least heralded as so. I presume one can also find it at Baseline Scenario, since it follows that bankers are evil. Or Felix Salmon again.

But I cannot understand the argument. When CDO's are created that finance mortgages and other asset purchases one gets price appreciation in the underlying asset that creates a bubble. When the asset price collapses there is a destruction of wealth. With a synthetic CDO, on the other hand, there is no underlying asset appreciation. The accusers all claim that synthetic CDO's are just bets. But if they are just bets then they lead to a transfer of wealth, not a destruction of wealth. Those on the long side who lost transferred wealth to those on the short side who won. But no amount of betting like that can cause a financial crisis.

If betting on the Kentucky Derby multiplied ten-fold would that cause a financial crisis? Hard to see how. Then how does the proliferation of synthetic CDO's make the underlying losses any bigger? I cannot see that. Perhaps the losses were concentrated in some banks that were too big to fail, but that is not what caused the crisis.

I don't understand why this argument has taken off, except that it makes banks look evil I suppose.