Basel II has finally arrived: yet in spite of the long lead time prior to its implementation, participants in the covered bond market are still grappling with its implications. William Thornhill reports.
The impact of Basel II depends to a large extent on the approach adopted by investors rather than that of issuers. It is essentially a question of the extent to which paper is placed with Basel II-sensitive bank investors, versus non-bank investors. If paper is placed mainly with banks, the question is which approach they intend to take.
The risk weighting for underlying residential mortgage loans is expected to fall from 50% under Basel 1 to 35% under Basel II’s Standardised Approach (SA) and 13% under the Internal Ratings-Based Approach (IRBA). The risk weighting for commercial loans under Basel II’s SA is 50% versus 100% under Basel 1.
Basel II is geared to incentivise banks to adopt to adopt the IRBA, with the key objective being to bring regulatory capital in line with economic capital. The IRBA is more advanced and requires better risk management and banks that adopt it are rewarded with a lower capital requirement versus under the standardised approach.
Qualitative impact studies carried out by the EU commission along with local regulators indicate that the IRBA does indeed lead to a lower capital requirement than standardised approach.
Standardised approach
Under Basel II’s SA, regulators can assign either of two options. Under the first, the risk weighting is based on the rating of the sovereign entity. For all eurozone banks, excluding Greece, the sovereign rating is higher than Double A minus, so the risk weighting is 20%. But, for Single A rated Greece, the risk weighting would be 50%.
Under the second option, the risk weighting is assigned based on the credit risk of the bank itself. Thus, for banks rated higher than Double A minus, the risk weighting is 20%. It is 50% for Single A through to Triple B minus rated institutions and 100% for those from Single B minus to Double B plus. Deutsche Bank covered bond analyst, Bernd Volk, reckons that countries with high average bank ratings, such as the UK and Sweden, are likely to favour option two.
The large potential jump in the risk weightings of bonds as a result of Basel II is particularly pertinent in Germany, where a number of issuers are lower rated. As such, bank investors in the UK, Spain and Italy could apply a risk weighting of 20% for German covered bonds, up from 10% currently.
Internal ratings-based approach
On the proviso banks receive their local regulator’s approval, they can instead choose to follow the Internal Ratings Based Approach (IRBA). Again, there are two options: the Foundation Approach (FIRB) and the Advanced Approach (AIRB). Around 70% of banks are likely to adopt the FIRB approach.
Under FIRB, banks estimate their own probabilities of default (PD) while local regulators assign the Loss Given Default (LGD), Exposure at Default and the Effective Maturity. The PD is expected to start from 0.03%, though empirical evidence suggests a level of closer to zero or 0.01%. This reflects the fact that few Western European banks have defaulted and default data is therefore very limited.
In practice, this means that most covered bonds placed within the EU will be assigned an LGD of 11.25%, while in the case of bonds placed outside the EU the LGD jumps to 45%.
Under the AIRB the Effective Maturity is used instead. Whether the AIRB or FIRB approach is used, the local regulator may require the investor to apply the Real Effective Maturity for durations between one and five years (anything longer will be considered equal to five years). Bankers generally assume a 2.5-year duration.
LGDs must also be estimated and based on five years’ data which, in the case of Germany, is 7% according to research conducted by the German Mortgage Bank Association (vdp). However, under the AIRB approach, investor banks are not allowed to rely on external data such as the vdp’s.
Deutsche Bank’s Volk says banks in the EU may use the same LGD under both AIRB and FIRB approaches. Losses attributable to segregated cover pools are significantly below the average mortgage loan portfolio losses, suggesting low LGDs for countries like Germany, France, Luxembourg, Ireland or the UK.
In contrast, Spanish loan assets have to-date, not been segregated, so the average loss rate of the portfolio has been higher, leading to higher LGDs and PDs as a result. However, with this in mind, the Bank of Spain is set to introduce a Cover Register for Cedulas which “will also have the formal benefit of a segregated pool,” says Volk.
So, based on the approach used and component inputs taken, any one of a number of risk weighting for covered bonds, ranging from 2.4% through to 18.5%, can be applied (see chart).
Under the Capital Requirement Directive (CRD), the EU’s interpretation of Basel II, a number of covered bond criteria are specified before certain LGD levels can be applied.
So, for example, under the FIRB approach, the LGD would be 11.25% for bonds meeting all the CRD’s defined criteria. When combining that with the lowest eligible PD of 0.03% (meaning the issuer is rated Double A or higher), under the FIRB approach, an investor could well reach a risk weighting of 3.6% for covered bonds. This is well below the 10% that currently prevails under Basel I for covered bonds with a legal backing, let alone the 20% that applies to structured covered bonds. The large potential drop in risk weighting is likely to bring this asset class significantly closer to the supra-sovereign sector, implying scope for spread convergence.
Despite phenomenal growth in covered bond supply over the last few years, it is still dwarfed by the size of the sovereign sector. Thus, it would only take a small percentage of investors in sovereign debt to switch into covered bonds to result in a relatively large percentage shift in demand for the smaller covered bond market.
But, according Ralf Grossman, covered bond analyst at Natixis, not all countries comply with the CRD definition of a covered bond. France recently adjusted its legal framework to comply with the definition, so all Obligations Foncieres automatically benefit.
And the situation is different for most countries. The UK is still working on a framework that complies with the definition; Luxembourg will not modify its law which does not comply; there is no law in place in the Netherlands, so Dutch bonds will also fall outside the definition; Italy’s CdP is technically not a bank, so its covered bond issues will fall outside the definition. However, the new Italian law for private-sector banks will comply with the definition. As such, some covered bonds will be in line with definition and others will not, which is probably a recipe for confusion.
Adding to confusion is the possibility that covered bond risk weightings could be significantly higher than the 20% level many had assumed would be the ceiling. In November, Fitch published a report regarding the CRD’s view of covered bonds. It suggested that, depending on what PD is applied by investors, the risk weighting could go to as high as 120%. But bankers say this study was based on unrealistic assumptions, not seen in the last 40 years. Nevertheless, the report acted as a wake-up call to any issuers complacent over the impact that PD levels might have on the risk weightings their investors apply.
And, says Natixis’s Grossman, the Fitch study also threw up another previously unconsidered aspect. According to its interpretation of the CRD, it is possible that covered bonds that do not comply with the CRD definition could get a low risk weighting in any case. This is based on the premise that an issuer’s programme rating can also be used to determine the PD. If the programme rating is Triple A, then a PD of 0.03% can be used, which would, in turn, lead to a low risk weighting. Whether Fitch’s interpretation is realistic or not remains to be seen and will in any case need to be revisited and discussed with the local regulatory authorities.
Given the many approaches and exclusions, it seems that experienced European bankers are not yet up to speed. Indeed, even on the most fundamental aspects, some players are still not quite there. Once banker said he had been taken by surprise that one of his peers at a major institution had continued to use the PD of the covered bond itself and not the PD of the issuer. This is a fundamental mistake, as Basel II does not recognise the double default approach that favours covered bonds.
The upshot is that syndicate desks will have to carefully assess which is the most apt investor base and plan the roadshow for each issuer if a significant jump in risk weighting and therefore spreads is to be avoided. In essence, the marketing of an Aa2/AA rated entity will be completely different from one rated A3/A-.
Though many of the major countries are expected to have released the CRD transposition sometime this year, the final implementation of the new regime will ultimately be unilateral. Most bank investors are expected to start on the Standardised Approach and then switch to IRBA, which will need local regulator approval.
Spain, France, Germany and the Netherlands have models more or less in place for the IRBA, although it is not clear that their local regulators have yet approved these models.