By Paul Amery
How valid are the comparisons being drawn between ETFs and CDOs?
The comparison of exchange-traded funds (ETFs) with collateralised debt obligations (CDOs), made by Terry Smith a couple of days ago, is nothing new. London-based hedge fund and gold investor Hinde Capital made a similar argument last summer, saying that gold exchange-traded funds like State Street’s GLD are full of embedded structural risks, resemble CDOs and should be avoided.
In both cases—Smith’s and Hinde’s—one has to bear in mind that the arguments come from observers with an axe to grind. Smith has his own actively managed fund to sell, which comes with a total expense ratio of 1.25 percent a year, while Hinde offers you the opportunity to buy a fund that invests largely in allocated physical gold, but charges 1.5 percent flat a year, plus 20 percent of any gains, and with only monthly liquidity, to do so. That’s turned out to be much more expensive than the 30-40 basis point annual cost to own the average gold ETF or ETC.
But do these ETF critics have a point? And what should the average investor make of the comparisons between a fund type (ETF) that’s been collecting assets fast and a type of bond (CDO) that nearly brought down the whole financial system?
That there’s financial engineering in many ETFs is undoubtedly true, while ETF issuers haven’t exactly gone out of their way to highlight this. By some accounts, the issuers of physically backed ETFs are making as much in revenue from the lending out of the securities in their funds as they are from the fund fees, or possibly even more, while there are economic incentives for issuers of swap-backed ETFs—well described in the FSB and BIS papers released in April—to use the funds as a cheap financing vehicle for the parent bank, as well as possibilities for regulatory arbitrage.
Does this resemble what went on in the average CDO? In my opinion, ETF risk is not on the same scale.
CDOs had a fundamentally flawed structure, using bad maths to model possible risks, and this was made much worse by the misuse of credit ratings. If the ETF market had an “independent” ratings service rubber-stamping funds as offering effectively risk-free extra return, as was the case with CDOs, I’d be much more concerned.
CDOs based on mortgage-backed securities also had illiquid, poorly understood collateral, which relied on a fundamental assumption that the housing market would always rise. ETF collateral policies are not based on any such assumption. Nor do ETFs contain the difficult-to-model correlation risks between tranches that CDOs incurred as a matter of course.
It’s fair to say that synthetic ETF structures haven’t yet experienced the failure of a bank counterparty, and that therefore we can’t be sure what tail risks might occur. The scenario analysis I’m aware of suggests that the worst that could happen for investors in a counterparty failure is (a) that there could be a delay in accessing the collateral, and (b) that there is some risk that the collateral’s value may fall short of the amount investors are owed; but not that the collateral might not be there at all. A delay in getting their money back wouldn’t be pleasant, by any means, but is there a likelihood that an ETF’s collateral could turn out to be worth only 10 percent of its face value, as was the case with some "AAA"-rated CDOs?
I’ve heard stories in the past of certain synthetic ETF issuers holding a structured note issued by the parent bank as part of the collateral basket, potentially increasing counterparty risk from the 10 percent maximum allowed under UCITS rules to 20 percent. Having said that, the trend towards more widespread, daily disclosure of synthetic ETF collateral by many issuers, plus the maintenance of higher overall collateral levels, has been improving practices in this area.
Are collateral and counterparty risks difficult for the average investor to understand? Yes. Are ETF collateral risks on the same scale as those incurred in CDOs during the last decade? Not that I can see.
Regulators have been more circumspect than to make a direct ETF-CDO comparison, though Mario Draghi, FSB chairman and future president of Europe’s central bank, did say a couple of months ago that “ETFs are reminiscent of what happened in the securitization market before the crisis”. In its ETF paper, the FSB spoke specifically of the heightened risks of potentially dangerous financial innovations during periods of low interest rates.
Ironically, the near-zero interest rate policies that are creating skewed economic incentives across the board come from the same central bankers that populate the boards of the multinational regulators.
That aside, the debate over the structural risks in ETFs that the regulators have provoked has undoubtedly been a healthy one. I’ve lost count of the people from within the exchange-traded fund industry who’ve told me (off the record) that they felt the IMF, FSB and BIS raised several valid points of concern.
But are ETFs as risky as CDOs? I don’t think so.Original post