Fitch has warned that acceptance of low overcollateralisation (OC) of covered bond programmes could lead to market complacency, pointing the finger at lower OC sufficing for Moody’s triple-A ratings and a risk OC may not be topped up in harder times, in a rare public inter-agency dispute.

In a report published today, Fitch said covered bond market participants may be overlooking the role of OC in providing dependable protection for investors and suggested there may be too much reliance on issuers maintaining collateral levels in harder times to ensure their covered bonds remain low risk.

“We see evidence of this in the very low levels of OC that Moody’s deems to be compatible with its highest covered bond ratings and which reflect issuer default risk rather than credit and market risk,” said the rating agency.

Hélène Heberlein, managing director at Fitch, told The CBR that the rating agency felt the differences in the two approaches deserved explanation.

“We know that, in meetings between issuers and investors, some market participants ask why the same programme has, for example, next to 0% OC for Moody’s Aaa and 10% for Fitch AAA.” she said. “We felt it was up to us to explain the difference and where it comes from.”

Heberlein noted that the rating agencies in their analyses assess the same risks and that often Moody’s judgement of the risks which arise when recourse against the cover pool is enforced is more severe than that of Fitch.

“But because it is weighted by the probability of the issuer defaulting, the calculation leads to a low level of OC in Moody’s approach if that probability is low,” she said. “For Fitch, the breakeven OC for a given covered bonds rating is much less dependent on the issuer rating and doesn’t rely as much on issuers to top up OC over time.

“We think this is a less cyclical approach.”

Moody’s was unable to comment by The CBR’s deadline.

According to Fitch, across the 51 mortgage covered bond programmes rated triple-A by both Moody’s and Fitch as of the end of the first quarter of 2017, on average the amount of total cover pool losses factored into Moody’s analysis (17.9% of the cover pools) exceeded the Fitch AAA breakeven OC (14.7% of the covered bonds). However, Moody’s determines substantially lower OC to be consistent with its Aaa ratings, with an average of just 5% for the same sample.

On average over the sample, Fitch said that the OC considered commensurate with a Moody’s Aaa rating only covers about a quarter of the OC that would offset Moody’s estimated total cover pool losses. For almost 30% of Moody’s Aaa rated covered bonds programmes, 0% OC was declared consistent with the rating, while Fitch notes that only the covered bonds guaranteed by AAA rated entities have a 0% AAA breakeven OC under its approach.

In a summary of its fellow rating agency’s methodology, Fitch says that Moody’s OC requirements for a given rating are derived from the weighting of anticipated total cover pool losses by the probability that an issuer ceases to pay, calculated for each month and discounted to their present value. The probability of default is small for a highly rated issuer, and OC is expected to increase as the issuer’s credit quality deteriorates.

In contrast, Fitch’s OC requirements are largely independent of the issuer rating, it said. When the covered bond rating is significantly above the issuer rating – in practice, more than four notches – Fitch assumes that the issuer has a 100% probability of default when setting the breakeven OC requirement, which, it says, addresses the risk of an issuer “jump-to-default” risk.

Fitch added that while the vast majority of programmes currently have large OC buffers, there is no certainty that these buffers will be maintained over the life of the bonds in the absence of an issuer commitment.

“We believe this is the appropriate stance because, despite their best intentions, issuers may be unable to post more OC when facing financial stress,” it said. “Available collateral may increasingly be pledged for other purposes and institutions with already high encumbrance levels may be forced to raise other debt to finance additional OC when funding costs are rising.

“In short, we think that overreliance on issuer creditworthiness creates the risk that OC is not topped up before it is too late.”

An increased likelihood of undercollateralisation in programmes could affect the regulatory treatment of covered bonds in the event of a bank resolution, Fitch warns. Lower OC protection also leaves programmes more exposed to the impact of rising interest costs and property market corrections, it said.

While some jurisdictions feature legal and regulatory provisions that mitigate such risks, Fitch added that regulatory OC minimums, which range between 0% and 10% depending on the jurisdiction, with the exception of Spain, are not necessarily designed to sustain extreme stresses associated with high rating scenarios.

“We believe vigilance regarding OC is merited to avoid complacency that could ultimately weaken investor protection,” said Fitch.

Market participants were surprised at Fitch’s questioning of Moody’s approach, saying it is rare for one rating agency to criticise another’s approach directly and in public.

“It’s a bit unusual in my experience to see one agency calling out another by name on their approach,” said a banker.

Jörg Homey, covered bond analyst at DZ Bank, said the contrasting approaches represent the key difference between the rating agencies – that Fitch and S&P are more cashflow driven in their OC analysis, while Moody’s puts a stronger emphasis on the creditworthiness of the issuer.

“You could argue that Moody’s approach is super-linked to the issuer’s creditworthiness,” he said, “But from a historical perspective, 10 or 15 years ago Moody’s used a notching approach, so if you look from a continuity perspective they have simply stuck to their convictions.

“It is a matter of perspective – do you perceive a covered bond more from a structured finance perspective, in which case you can emphasize the cashflow aspect, or do you perceive it as a secured bank bond, in which case it is fair in my perspective to link it more strongly to the issuer’s creditworthiness. There are good arguments for both.”