After all the euphoria in late February about the European Commission’s decision to elevate covered bonds to tier one assets in the liquidity coverage ratio, it’s time to consider the practicalities.
Following the EC’s change of heart — the commission had considered leaving covered bonds as level two assets — the European Banking Authority has started to consult the market about how to define the eligibility criteria for all asset classes, not just covered bonds.
As with all good consultations, the EBA tries to give an indication of what it expects to happen. But it fails to do so.
The consultation is biased towards empirical data. Historic turnover and bid-ask spreads tell you plenty about one idea of liquidity but very little about what liquidity ought to mean in this context: the ability to sell bonds in a crisis.
For example, there is no meaningful liquidity in high quality jumboliño Pfandbriefe (Pfanbreife of €500m), as measured by the EBA’s preferred metrics. No market maker will go short, no investor will sell. But such paper is still recognised in the market as a high quality liquid asset.
The tests look arbitrary, too. If you care to, you can generate data showing that UK covered bonds are as liquid as Pfandbriefe. Then again, the same market will show you that they are nowhere near as liquid. The difference? The date range: liquidity only really emerged in the UK as the covered bond market matured, so data from 2006 will tell you a very different story to now.
Cliff risk
The EBA consultation also says: “It could be that... bonds are defined as liquid only if they are rated above a certain grade and for issue sizes above a given threshold.”
There is an obvious risk that comments like this undo all of the good work that the market has done to challenge the prejudice against small or lower rated deals.
More importantly, having a rating or size threshold creates liquidity cliff risk. When a covered bond is downgraded below that “certain grade”, every bank treasury will have to sell it simultaneously. Which is pro-cyclical.
It’s a similar story when the EBA tackles the issue of different national markets. It says that finding evidence that an asset class is liquid in a specific jurisdiction does not imply that the same asset would be liquid in all jurisdictions. No risk of any controversy there, then.
Valuable diversity
And how about correlation? Is, to take an example at random, the liquidity of a covered bond from Italy more closely correlated to the liquidity of an Italian government bond or to a Spanish covered bond? The recent delinkage of govvie and covered spreads suggests that covered bonds add valuable diversity to predominately government bond portfolios.
How to explain the relatively limited moves in Italian covered bonds (10bp-15bp wider) in a week when Italian government bonds widened by 50bp-70bp?
Clearly covered bonds are a lower beta product: they trade anything from 50-150 through the government curve. But they are not lower beta by a factor of five.
They are, however, a less liquid product, which, it turns out, can be a good thing. Italian covered bonds are generally very well placed with end investors, and the recent widening was nothing to do with real flows and everything to do with market makers thinking that bonds should be marked a bit wider.
The EBA has been mandated to judge liquidity using a quantitative approach but it must overlay this with a qualitative sanity check. In the final analysis there is a big risk that a strict adherence to statistical data does not tell you anything about liquidity. In times of stress, a bank looking to access liquidity will only be looking for a good bid — it is really that simple.
Thanks to the covered bond blog: http://coveredbondblog.blogspot.co.uk/