As a result of the relatively contained spreads in covered bonds, some analysts have recommended buying RMBS instead of covered bonds for certain jurisdictions. But is this like comparing apples and pears? William Thornhill reports.
Covered bond and RMBS both reference the same underlying mortgage assets. Yet with the former, investors have recourse to the underlying assets and the issuing bank. Investors in RMBS only have recourse to the assets.
While covered bonds are usually structured in fixed rate bullet format and are generally held on balance sheet, RMBS carry a floating rating, amortise as the mortgage assets pay down and are held by an SPV. Moreover, while covered bond spreads are usually quoted against mid swaps, RMBS are quoted against the three month Libor or Euribor.
While the additional protection offered by covered bonds means they should trade somewhat tighter than RMBS, it is debatable whether the spreads seen earlier this year were really justified.
In a report written in mid February, JPMorgan ABS research recommended clients switch out of Spanish five-year covered bonds, at the time typically yielding 20bp over mid swaps, and into five year RMBS, typically yielding 150bp – or possibly as much as plus 260bp, had the trade been executed in early April. JPM noted that given the significant differential, investors could also buy senior protection on the underlying names, "and still make the trade materially carry positive."
Buy buying RMBS and CDS protection on the issuing bank, investors would typically more closely match the lower risk of covered bonds (see accompanying graph for spread history). That trade would have worked well had it been possible to execute, but with liquidity in the secondary cedulas market almost as bad as it had been in the RMBS market, putting on such a trade may have been a challenge for bank dealers. It would, however, have been more executable for real money clients.
More recently, spreads have widened in the cedulas market, with two year primary issuance recently seen in the plus 50bp to 60bp range. Thus, adjusting for the curve, finding a bid for five year cedulas, would by mid May arguably be closer to plus 70bp or more, based on where the primary clears.
Another more recent report produced by ABN AMRO in early March also compared covered bonds with RMBS. Head of covered bond and ABS research, Christoph Anhamm, said spreads for Spanish RMBS and Cedulas discount a much higher probability of default than had been observable just two month's previously. "The conditions that could lead to a default of Triple-A tranches or Cédulas are substantially worse than everything seen in the Spanish market so far," Anhamm said.
For an 80% LTV RMBS, defaults would need to rise to 12.5%, a stress that would require a 42% fall in house prices before for Spanish Triple A RMBS with credit enhancement of 6% get hit. For 95% LTV deals that have 13% credit enhancement, 22.5% of borrowers would need to default and house prices would need to fall by 60%. Given the massive 150% overcollateralisation featured in Cedulas, default rates would need to rise to 50% and house prices fall by 73% before investors of the Triple A debt got hit.
Though all scenarios are highly improbable, the latter affecting Cedulas is the most negligible. But the risks for RMBS are marginally higher. In conclusion, Anhamm said that for 80% LTV deals, investors could lose an incremental 6%, whereas Cedulas investors would get all their money back.
Notwithstanding these highly unlikely events, the spread pick up currently offered for Spanish RMBS against Spanish Cedulas is still likely to prove alluring to those investors capable of arbitraging the two markets.
That universe may be growing but it is likely to remain very small in absolute terms. One of the main reasons for issuing banks to tap both the covered bond and RMBS market is that they target different investors. Were the securities’ buyers the same there would be no justification for issuing in both sectors.
In any case, with prime Spanish RMBS typically trading in the plus 150bp to 170bp range, there is still very good upside potential, irrespective of whether this is hedged against covered bonds or not.
But one factor that may put investors off is that a considerable part of the upside, in terms of the spread narrowing between the two asset classes, has already taken place. If recent primary issuance is anything to go by five year Cedulas is likely to trade closer to plus 70bp than the plus 20bp level originally highlighted by JPMorgan. On the other side prime RMBS has already tightened somewhat.
According to LehmanLive, prime Triple A Spanish RMBS widened from plus 150bp in mid February to plus 258bp in early April, before trading back to the mid February levels again. That suggests the spread between the two sectors has come in from around 230bp (258bp for RMBS and 20bp for covered bonds) seen in early April to just 70bp by mid May (150bp for RMBS and 70bp for covered bonds). Moreover, it is rapidly diminishing: the trajectory for RMBS is still tighter, but for covered bonds the present trend is wider.
Clause Tofte Nielson, of Norges Investment Bank, has looked at the trade but remains undecided. It is important to take into account the value of the covered bond issuing bank's rating, he said. Thus, although it may issue a Triple A covered bond, the bank could by be Single A rated. In that sense the incremental value of the bank's protection embedded within the covered bond is worth somewhat less than that of a higher rated bank; a factor that should be expressed through the issuing bank's CDS.
Another key factor likely to play a role in ascertaining the value of such a trade is the extent of financial engineering within the RMBS. There is a clear differentiation between UK prime RMBS and non-conforming. On the other hand, according to the Bank of Spain, a non-conforming RMBS market does not exist
Yet the plethora of more recent vintage high LTV RMBS deals, along with transactions backed by non-traditional obligors, such as those featured in the UCI and Credifimo originated loans that featured in several TdA RMBS, are to some minds more at the specialist end of their prime market.
Because of these idiosyncrasies, Tofte Nielson said it is important to stress test the actual collateral pool of the targeted issues before deciding. "Before doing the arbitrage you really need to make sure that the cover pool quality is broadly the same," he said.
For now it seems that the prospect for this trade looks limited, as spreads appear to have moved out of sync for both asset classes. However, these are volatile times, and it is quite possible the current RMBS levels could move wider again. Equally, given that the covered bond market never relied on leveraged investors like SIVs to the same extent that RMBS did, it would be natural to assume that real money demand will play a key role in driving covered bond spreads tighter from current levels.
The Spanish market is also likely to be hit by the sheer overhang of RMBS that has not been placed with third party investors, but with the ECB. The amount of assets repo'ed by Spanish banks has crept higher incrementally every month since the crisis began, rising to €48bn in April from €44bn in March.
These assets will at some point have to come back on the market suggesting Spanish RMBS in particular could suffer from more of a technically driven spread widening than any other region. That's not to say the trade would or could not work in other jurisdictions. In Spain it is conspicuous merely because spreads are so conspicuously wide, but there is little doubt that same concept could be applied to the UK.
For now the spread in UK RMBS is less appealing than it was just a few weeks ago. On March 20th LehmanLive suggested five year UK prime was at plus 198bp, but two months later the spread had come in by a massive 110bp to 88bp.
The fact that no UK issuer has yet tested the primary covered bond market suggests the theoretical levels being quoted in the secondary market are somewhat inaccurate. At some point UK issuers will test the covered bond market and, as in Spain, a spread widening cannot be ruled out.
Other jurisdictions such as Italy, Ireland and Portugal could also see some arbitrage activity as, in essence, the covered bond and issuing bank CDS is likely to offer significant carry versus their prime RMBS markets, which were indicated at plus 118bp, 148bp and 133bp respectively in mid May.