Last Updated: 23 May 2021
According to analysts Jon Maier and Elias Lanik of Bank of America Merrill Lynch, it doesn’t really matter what a synthetic ETF holds as collateral.
“It is our opinion that the issue of collateral is only relevant in the event of a counterparty default,” said Maier and Lanik in their recent report, A better understanding of synthetic ETFs.
“While we are a proponent of full transparency of the underlying collateral basket, we are not necessarily of the belief that it is problematic to have a mismatch of the underlying collateral versus the fund’s benchmark,” the analysts continued.
In our investigation of Eurozone government bond ETFs, we found that different providers of synthetic ETFs, tracking indices that were almost equivalent, used quite different portfolios for their substitute baskets.
db x-trackers’ Eurozone government bond fund owned a portfolio of government bonds that resembled, give or take a few percentage points in country weightings, the index itself. Lyxor’s fund had a lot of agency, municipal and covered corporate bonds, plus some sovereign debt. Amundi’s ETF had more government bonds than Lyxor’s, but with a lot of volatile, long-dated, zero-coupon securities among them, plus a big overweight in Italy. And Comstage used a basket of Eurozone equities, plus a few German government bonds, to collateralise its swap.
One astute observer has pointed out to me today that, if db x-trackers is using a substitute basket that resembles the index being tracked, isn’t this effectively physical replication? That would be an unexpected consequence of synthetic ETF providers taking a more conservative approach to collateral policy.
But if you look at the four ETFs I’ve just described, in the opinion of Maier and Lanik there’s nothing much to choose between them, as long as their substitute baskets are UCITS-compliant. You’d then, I suppose, choose between these funds on the basis of total expense ratio, tracking performance and secondary market liquidity.
But that can’t be right. An ETF’s make-up does matter, and here’s why.
First, there’s a cost involved in sourcing the assets that go into the substitute basket. It’s more expensive for the issuer (or its parent bank, often the same entity) to put in AAA-rated bonds than junkier ones, for example. Even if the cost is invisible to the end-investor, it’s still there.
Where is this cost, if you can’t see it? It’s in the increased probability of: (a) something going wrong; and (b) there being a larger, rather than a smaller, loss if things do go awry.
Let’s put this another way. In the summer of 2008 there was a steady rise in the cost of insuring against the default of AIG. At the time, I pointed out that this was a hidden cost of investing in some of ETF Securities’ exchange-traded commodities, something that earned me a very unpleasant conversation with the firm’s then head of sales.
You couldn’t see that there was a cost of holding the ETCs from their tracking, which continued to be accurate until AIG did approach default. Then, of course, and without warning, the ETCs were suddenly trading at ten cents in the dollar, until the US taxpayer rescued all of AIG’s counterparties, ETC investors among them. They got lucky.
Now I know that ETFs are not the same as the formerly AIG-backed ETCs (most of ETF Securities’ commodity trackers are now collateralised, by the way). ETFs have an asset basket behind them and are funds.
But if the quality of assets in the basket of a synthetic ETF is lower than that of the index securities, you’ve simply increased the risk of loss in the future. And that risk has a cost attached to it, the cost of insurance.
The story doesn’t end there. As regulators have pointed out, the more an issuer is able to raise finance at subsidised rates via an ETF (by the provision of lower quality collateral), the more it is likely to resort to this mechanism if other sources of funding get tough. If you’re an ETF investor, incentives for your synthetic issuer to abuse the structure increase, just when you don’t want them to, in other words.
For all these reasons I’m surprised that Maier and Lanik conclude that synthetic ETF collateral mismatches are unimportant. We’re less than three years from the worst of the credit crisis, a cataclysmic event that was all about the abuse of misaligned incentives, particularly in the credit ratings system. It appears that some people have short memories.
ETF structure does matter, and people should pay attention to it.