The Fallacy of US Bank EU CDS Exposure

Have been reading countless articles lately on US banks sovereign CDS exposure to the EU. I won’t call anyone out, do a google search. Here is the problem with every single one of them. They have no idea how that exposure is reported. The issue lies with the fact that a bank, that writes a CDS on $1B of Greek debt and then subsequently buys $1B of CDS on Greek debt from a 3rd party would ONLY report the $1B they wrote as their “exposure”. The 3rd party that wrote them the $1B of CDS would then report their exposure as $1B.

Because of this, when we read “Bank A has $30B of CDS exposure to Greece” (be it $BAC or $JPM or anyone else), that number is more likely than not, highly bloated as what we do NOT know is how much CDS exposure to Greek debt they have bought. Their net exposure may very well be zero OR even in a plus position…..

Just another reason to dig for information folks and not simply believe what you read or are told. This is especially true of the banks and financials now. They are VERY hard to digest and that makes them ripe for rumor mongering and false claims either intentional or not.

Emphasis mine:

(Extract from page 22 of BIS Quarterly Review, September 2011)

Stefan Avdijev

Recently, there has been a substantial amount of interest in the extent to which the category “guarantees extended” 1 of the BIS consolidated banking statistics on an ultimate risk basis could be used as a proxy for the credit default swap (CDS) exposures of various banking systems to individual countries. Several important caveats apply to such an approximation.

First, while the contingent liabilities of the protection seller of credit derivatives contracts are a part of the category “guarantees extended”, they are not the only item included in it. In addition to CDS contracts sold by BIS reporting banks, this category also includes secured, bid and performance bonds, warranties and indemnities, confirmed documentary credits, irrevocable and standby letters of credit, acceptances and endorsements. Therefore, the fact that US banks, for instance, had $37 billion worth of guarantees exposures to Greece as of the end of Q1 2011 (Table 9E in the BIS Statistical Annex) does not imply that US banks had sold $37 billion worth of CDS protection on entities located in Greece.

Second, banks are not the only institutions that buy and sell CDS contracts. Other financial enterprises, such as insurance companies and hedge funds, also actively participate in the CDS market. As a result, not all CDS written on entities located in a given country are included in the category “guarantees extended” of the BIS consolidated banking statistics. Thus, US banks’ $37 billion worth of guarantees exposures to Greece from the above example is not the correct ceiling on the total amount of CDS written on Greek entities by US institutions.

Third, in the category “guarantees extended” of the BIS consolidated banking statistics, CDS sold are reported at notional values, not at fair values. In order to illustrate that point, suppose that a French bank sells a CDS to a Spanish bank on $1 billion worth of securities issued by the Greek government. Suppose further that, at the time of reporting, the CDS has a positive fair value of $100 million from the seller’s perspective (ie the French bank). According to the Guide to the BIS consolidated banking statistics, the French bank should report $1 billion (ie the notional amount of CDS sold) worth of “guarantees extended” to Greece.

Fourth, in the category “guarantees extended” of the BIS consolidated banking statistics, CDS sold are generally reported at gross (not net) values. To illustrate this, suppose that the French bank from the above example sells a CDS to a Spanish bank on $1 billion worth of securities issued by the Greek government and simultaneously buys a CDS on the same set of securities from an Italian bank. If these were the only two transactions the French bank engaged in during the period, it would report $1 billion (ie the gross notional amount of CDS sold) worth of “guarantees extended” to Greece, despite the fact that it has also bought a CDS on the same contract from a third party (in this example, from the Italian bank).

Finally, CDS bought by banks are not reported in the category “guarantees extended”. Their treatment in the BIS consolidated banking statistics depends on whether the reporting bank that purchased the CDS contract owns the underlying security or not. Suppose that the CDS contract that the French bank bought from the Italian bank in the above example has a positive fair value of $100 million from the buyer’s perspective (ie from the perspective of the French bank). If the French bank does not own the underlying security, it should report $100 million (ie the positive fair value of CDS bought) worth of “derivatives” exposures to Italy. If, on the other hand, the French bank owns the underlying security, it should report a risk transfer of $1 billion out of the Greek public sector into the Italian banking sector (ie on an immediate borrower basis, the French bank will report $1 billion worth of foreign claims on the Greek public sector; on an ultimate risk basis, it will report $1 billion worth of foreign claims on the Italian banking sector).