Copying and distributing are prohibited without permission of the publisher.


Label mania grips covered bonds, but investors should look deeper

02 Apr 2012

Covered bonds are moving towards a system of labelling, designed to act as a kitemark of quality. But investors should remember: there is more to credit analysis than ticking boxes. Weak deals can pass through labelling systems, while great deals can get excluded.

The term ‘covered bond’ is supposed to be a stamp of quality, and that is the principle underlying the European Covered Bond Council’s initiative to set up a formal covered bond labelling system. But investors should not get hung up on labels. Credit risk can lurk in many corners, and unlabelled bonds can sometimes offer great value.

In the world of covered bonds, particularly, investors need to look out for asset-liability mismatches, the likely rating migration of deals, timely payment of coupon and principal, the precise make-up of collateral pools, and the size of collateral.

While complying with covered bond legislation gives investors some minimum standards on all these issues, it does not necessarily give them clarity or adequate protection.

These factors were vividly illustrated this week when two insurance companies launched dual recourse bonds secured by residential mortgage-backed securities.

Axa’s €1bn European deal is a soft bullet covered bond. It is Ucits-eligible and falls into line with the covered bond definition of Basel’s capital requirements directive – so it should qualify for the covered bond label.

Prudential America’s $1bn deal, on the other hand, will never qualify for the covered bond stamp – even though it is a hard bullet.

In Axa’s deal the biggest anomaly is the potential 45 year asset-liability mismatch. The security is backed by self-originated RMBS assets with legal final maturities of 2049 and 2050, but the bond liabilities sold to investors mature in just five years.

Prudential’s deal, on the other hand, is massively overcollateralised by third party subprime RMBS from Pru’s books. The three year bond’s underlying cashflows are so large they should ensure that there is no asset-liability mismatch.

If Axa goes bust and the regulator can’t find another issuer willing to manage its pool, investors could be waiting a long time to get their money back.

A default seems unthinkable in the French market, but not in Spain, where banks have loaded up on real estate which is plummeting in value. Spanish covered bonds face similar asset-liability mismatch issues to French ones.

A Spanish systemic banking crisis is not unthinkable. If, heaven forbid, an issuer was to go down, finding a replacement cédulas issuer to take over the collateral and ensure seamless payments in such circumstances might prove impossible. However, cédulas are all expected to qualify for the ECBC’s label.

At 21.4%, Axa’s voluntary overcollateralisation (OC) is larger than that on most covered bonds, but this pales into insignificance beside Prudential’s pool, which is permanently fixed at nearly three times the size of the deal.

As with all other covered bonds, Axa is only obliged to size its OC to the legal minimum. This is less than what it has voluntarily given the deal at launch. Theoretically, therefore, Axa could later allow the OC to decline, without breaking the legal threshold.

Axa’s deal, like all other covered bonds, is therefore arguably more at risk of rating downgrades and potential mark to market losses than Prudential’s.

Furthermore, Axa’s pool, like all other covered bond pools, can vary in composition over time. In contrast, there is full certainty over Prudential’s pool – which is static.

Yet Prudential gets no rating benefit above its senior unsecured rating from its almost absurdly large collateral pool. And investors still get 100bp more on the coupon compared with the issuer’s unsecured corporate bonds.

As covered bonds go, Axa’s deal is actually one of the strongest, but it would be a mistake to think that just because a bond has a label it is safer than others that don’t. Spare a thought for labelless Pru – perhaps a bargain?

02 Apr 2012