Banks leapfrog the rating agencies

European banks are cutting rating agencies down to size with a new loan data pooling initiative. Dutch bank NIB is leading the initiative. By Nicholas Dunbar

The opaque world of European corporate lending is set to become far more transparent, with the announcement of a default data pooling initiative by a group of European banks. Called the Pan-European Credit Data Consortium (PECDC), the group plans to share data on loss-given default (LGD), exposure at default (EAD) and probability of default (PD) for their commercial loan portfolios.

So far, the only publicly confirmed members of PECDC are its management committee, namely Barclays Capital, Calyon, Royal Bank of Scotland, JP Morgan Chase and the Dutch specialised lending house NIB Capital. However, considerably more European banks are in the process of signing up, say the committee members. Although the PECDC plans to use rating agency Fitch to collate the default data, it has insisted upon retaining ownership of the data – its slogan is ‘by banks, for banks’ – which may represent a significant challenge to rating agency business models.

The consortium was the brainchild of Jeroen Batema, a market risk manager for NIB Capital. With a E21 billion balance sheet, the Dutch bank (which is jointly owned by pension funds ABP and PGGM) is considerably smaller than the other management committee members. However, a specialised focus in areas such as shipping, energy and transportation finance made NIB particularly dependent on rating agency decisions, which clashed with its own internal ratings judgements.

“We noticed that the rating agency statistics for bonds were not applicable for bank loans,” Batema says. “That’s intuitively understandable because bank loans are structured and actively managed, and because of the client relationship focus of banks: we grant loans through the economic cycle, whereas bond issuance mostly occurs at the top of the cycle. But our portfolio is not very big, so it’s very difficult to convince rating agencies, when you securitise bank loans, that the risk of structured loans is not as high as the risk of bonds with the same rating.”

According to Batema, this prompted NIB to first suggest data sharing between banks. In December 2003, following unfruitful discussions about the idea with other Dutch banks, NIB decided to go abroad and gauge interest among other European banks – and soon found willing participants. For the larger banks, the motivation would come from Basel II. In particular, as the regulatory capital advantages of adopting the standardised or foundation internal ratings-based (IRB) approach were whittled away by regulators, banks began to gravitate towards the advanced IRB approach.

But this proved data-intensive, as Scott Aguais, head of credit methodology at Barclays Capital, explains. “To get regulatory acceptance for advanced IRB, a substantial amount of data is needed. We wanted not only data, but also improved templates and enforceable data quality standards.” As a large US bank, JP Morgan Chase was not even given the option by its regulator of adopting a non-advanced approach to Basel II. An examination of the bank’s own portfolio suggested a similar data deficit. Jean-Pierre Lardy, head of European portfolio credit advisory at JP Morgan Chase, says: “We have sufficient US default data, but calculations showed that our non-US database wasn’t large enough.”

Rating agencies, which have large default databases of their own, have a number of default data pooling ventures already in place. In the US, the Loan Pricing Corporation (LPC, now owned by Reuters) set up such a system for US banks several years ago, and sold it to Fitch in 2001. In Europe, Standard & Poor’s has its own pooling initiative in the specialist areas of project finance and acquisition finance. However, an attempt by Fitch to replicate the LPC model in Europe has struggled to find supporters.

According to PECDC consortium members, there are several reasons for this. Martial Bienassis, head of credit analytics at Calyon, says: “The US is more homogeneous than Europe both in regulatory and cultural terms. There is a much larger array of banking facilities used in Europe, and it is more difficult to bring European banks together.” In addition, banking secrecy laws in some countries also prevent disclosure to third parties.

Equally significant, however, is the view among bankers that rating agencies are too commercially driven by their need for data ownership and control – which puts them closer to publishing companies than financial institutions. “Historically, the rating agency business model has been flawed,” says Barcap’s Aguais. With sophisticated portfolio credit models now widely available, banks are determined not to allow rating agencies any more leverage over what is perceived as a data bottleneck.

NIB’s Batema says this proved a catalyst in the PECDC’s first meeting in June 2004: “The interesting thing was that the banks wanted to make their own data templates instead of being dependent on what the rating agencies produced. The banks in Europe, instead of participating in rating agency data-pooling initiatives that are already in place, think it is better to set up a structure that is by banks and for banks.”

That is not to say rating agencies don’t have a place in the venture. “The goal is not at all to bypass the rating agencies,” says Lardy. But instead of their usual role of owning and controlling the presentation of data, rating agencies were invited to pitch for the role of collating and aggregating data – under the PECDC’s overall control. The sources say the consortium is close to signing an agreement with Fitch, which will collect raw data from the banks, and then weight and anonymise the data (to safeguard confidentiality) before distributing it back to consortium members.

However, according to sources close to the PECDC, the cost of paying a ratings agency for this data-managing service while withholding the right to data ownership was prohibitive. This obstacle was overcome in an interesting way, using a ‘sweetener’ arrangement with recently acquired Fitch subsidiary Algorithmics, which will be granted the right to calibrate its commercially available portfolio credit models using the data distributed to PECDC members. Other third parties will only be able to see an aggregated form of the data, which the PECDC plans to make available to everybody.

According to a PECDC letter of intent document dated September 2004, the consortium will begin by collecting LGD and EAD data for European large corporate and small and medium-sized enterprise (SME) borrowers, with a lower credit exposure limit of E1 million. Initially, the consortium will set up the data-sharing process for each country separately. Once a target of 60% of wholesale credit exposures in that market has been reached, and confidentiality issues resolved, then cross-market data sharing will be permitted. For specialised lending categories such as shipping, aviation, leveraged finance, trade and commodity finance, and project finance, cross-market data sharing will take place from the outset.

The first exchange of default data (extending back to 1998, to satisfy advanced IRB requirements) will take place in June 2005, and Fitch is expected to deliver its analysis by October. PD data sharing, which is more complex because of the need to normalise data produced by different internal ratings systems, will follow at a later date, according to PECDC.Risk