Can we really know anything about US banks exposure to Europe?
A familiar epistemological question, which is being asked again in the wake of MF Global’s demise and Jefferies’ circuit-breaking slide.
We’ve taken a depressingly long look at some of the data and if you don’t want to read the splurge below, in sum, we don’t think we can confidently say very much ex ante about the extent and form of US banks’ exposure to European sovereigns, banks and corporations.
For Christmas, we’ve asked Isda and the DTCC for some more details around CDS exposures from their centralised data grotto but in the meantime we’ll have to make do with the lumps of coal provided by the banks and the Bank for International Settlements (Bis).
Of course, in the event of a default by Greece or a full-blown Italian crisis, it’s not just “actual” exposures that matter but perceived exposures — indeed, perceptions of perceived exposures. Hence there are several layers of information asymmetry to wade through.
______
The temptations of those Bis data (again)
Bloomberg’s Yalman Onaran on Monday took a look at the latest Bis data on consolidated bank exposures and spoke to several analysts about banks’ exposure.
U.S. banks increased sales of insurance against credit losses to holders of Greek, Portuguese, Irish, Spanish and Italian debt in the first half of 2011, boosting the risk of payouts in the event of defaults.
Guarantees provided by U.S. lenders on government, bank and corporate debt in those countries rose by $80.7 billion to $518 billion, according to the Bank for International Settlements. Almost all of those are credit-default swaps, said two people familiar with the numbers, accounting for two-thirds of the total related to the five nations, BIS data show.
The same banks report in their earnings statements that sales such as these are netted out using other hedges. Onaran notes that five banks write 97 per cent of CDS in the US and 74 per cent of global CDS trading takes place among 20 banks. Thus, for him, the problem becomes one of counterparty risk and margin calls as Frederick Cannon from KBW tells Onaran:
“Risk isn’t going to evaporate through these trades,” Cannon said. “The big problem with all these gross exposures is counterparty risk. When the CDS is triggered due to default, will those counterparties be standing? If everybody is buying from each other, who’s ultimately going to pay for the losses?”
In this post we pasted the following chart from Citigroup showing the gross exposure of US banks to Giips as of Q1 2011, according to Bis data (which is provided by the Fed based on data submitted by banks using this form):
Using the latest Bis data and our amateur Excel skills, we’ve made our own table showing US banks’ gross exposure to the GIIPS. It uses data from Q4 2010 through Q2 2011 — unfortunately this level of detail is unavailable before the end of 2010. Click to expand:
Some annotation:
– The first column describes the different types of exposures US banks have. “Foreign claims” can include “deposits and balances placed with other banks, loans and advances to banks and non-banks, holdings of securities and participations”. “Other potential exposures” is a mixture of derivatives where the reporting bank is in-the-money, CDS make an appearance in a couple of places, there are some contingent liabilities floating around, as well as irrevocable commitments to extend credit. Got that? More on this later…
– The other columns show the amount, in millions of US dollars, of exposure to each GIIPS country over three quarters: Q4 2010, Q1 2011 and Q2 2011. The highlighted box provides a total for the GIIPS. (You can see that the Q1 2011 GIIPS column is the same as in the Citi table above.)
For a second, let’s put aside the issues with the data. Looking at the trends, there are at least four intriguing patterns that could form the basis of juicy news stories:
1. Overall exposure, as defined by Bis, has increased quarter on quarter.
2. Foreign claims, as defined by Bis, have increased quarter on quarter.
3. Guarantees extended by US banks on entities in these countries have increased quarter on quarter. (This is the source of Onaran’s main factoid: the increasing number of CDS on GIIPS sold by US banks. However, this isn’t the only place in the data that CDS show up — we’ll discussion in excruciating amounts of detail below.)
4. Irish exposure, as defined by Bis, actually declined from Q1 2011 to Q2 2012.
______
The limits of those Bis data
Unfortunately, if you actually read the methodological notes, and the footnotes underneath the methodological notes, you soon fall into a nihilistic stupor, numb from self-inflicted ennui, cursing the day you ever read 111 pages of a Bis statistical paper. Or in other words, it’s borderline impossible to know the “true” extent of exposure from these data, and since it’s also borderline impossible to know the “true” extent of exposure from bank filings, we’re close to giving up.
Heck, we’re not even sure banks know the true extent of what’s going on. And we’re pretty sure no-one has a scooby about where all the concentrations of risk are at the moment. Ultimately, it comes down to trust: trust that the numbers are accurate, trust that the trades will settle, trust that Isda will call a credit event what it is.
We tried in this post to explain some of the problems banks say they have with the “foreign claims” part of the table.
In short, banks tend to argue that these exposure data are wild exaggerations because they do not include offsetting hedges; count foreign subsidiaries as being cross-border exposure; and do not offset cash collateral that is held in the country of exposure. Most importantly from the banks’ point of view, repo lending is counted but collateral is not counted as an offset. We have some sympathy with the banks here — if it’s US Treasuries then that should be pretty safe – but we don’t know what collateral is being posted so it’s hard to give them a full pass.
In any case, the foreign claims part of the table is a breeze compared to the ”other potential exposures” part of the table.
We first tried to understand that in this post. Returning to it now finds us none the wiser. Here is the Bis description of the table from the final page of the document:
Table 9 E As of current reporting quarter, this table provides a sectoral breakdown of consolidated foreign claims vis-à- vis individual countries by nationality of reporting banks on an ultimate risk basis. Data on other potential exposures through derivatives contracts at positive market value, guarantees extended and credit commitments are also shown. The grand total in the first column of the table comprises consolidated exposures of domestically owned banks in Austria, Australia, Belgium, Canada, Chile, Chinese Taipei, Finland, France, Germany, Greece, India, Ireland, Italy, Japan, the Netherlands, Norway, Portugal, Singapore, Spain, Sweden, Switzerland, Turkey, the United Kingdom and the United States. The total European banks comprises consolidated exposures of domestically owned banks in Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, Turkey and the United Kingdom. The total Non-European banks comprises consolidated exposures of domestically owned banks in Australia, Canada, Chile, India, Japan, United States, Chinese Taipei, and Singapore. Exposures of banking groups vis-à-vis the home country are not included, as these are not foreign exposures. Exposures of German banks on developed countries are on an immediate borrower basis, except claims vis-à-vis the Greek public sector.
(Incidentally, that last sentence might be worth looking into.)
We’ve emboldened the key words in the above paragraph. Now, if we use the statistical guide to check what Bis says it means, we soon realise that it’s both complicated and, unsurprisingly, dependent on what banks provide. So much so that it’s hard to know what to think.
Ultimate risk basis:
In line with the risk reallocation principle for measuring country exposure, the country of ultimate risk or where the final risk lies is defined as the country in which the guarantor of a financial claim resides and/or the country in which the head office of a legally dependent branch is located.
That makes sense: Bis is trying to follow the paper trail to the end. There are a couple of problems, though.
First, in the banks view this leads Bis to predict apocalypse in Excel format because not a lot of netting goes on here; everything that could go wrong is assumed could go wrong.
Second, the way some credit derivatives are reported on an ultimate risk basis may confuse which country ultimately accounts for a credit derivative:
Similarly, if credit derivatives are used to cover for the counterparty risk of financial claims in the banking book, the country of ultimate risk of these positions is defined as the country in which the counterparty to the credit derivative contract resides. In addition, in the case of holdings of credit-linked notes and other collateralised debt obligations and asset-backed securities a “look-through” approach should be adopted and the country of ultimate risk is defined as the country where the debtor of the underlying credit, security or derivative contract resides. However, it is recognised that this “look-through” approach might not always be possible in practice. Accordingly, reporting institutions might only be able to provide estimates for the allocation of claims to the country where the debtor of the underlying resides or to allocate the claims to the country of the immediate borrower, which is the country where the issuer of the securities resides.
Derivative contracts:
Derivatives contracts with a positive market value** have to be reported separately, regardless of whether the derivative contracts are booked as off- or on-balance sheet items
That makes sense: bring them all onto the balance sheet and them in one line.
**Derivatives with negative market values represent financial liabilities and are therefore to be excluded from the reporting of financial claims.
Actually, come to think of it, don’t net the derivatives contracts, but put the ones where you’re in-the-money in one place and out-of-the money in the other. This will give a clearer picture confuse the hell out of everyone.
And did we mention those credit derivatives? Well if you’re not trading them don’t put them here either:
Credit derivatives, such as credit default swaps and total return swaps, should only be reported under the item “Derivative contracts” if they are held for trading by a protection-buying reporting bank. Credit derivatives that are not held for trading should be reported as “Risk transfers” by the protection buyer…
Guarantees:
Guarantees are contingent liabilities arising from an irrevocable obligation to pay a third-party beneficiary when a client fails to perform some contractual obligation. They include secured, bid and performance bonds, warranties and indemnities, confirmed documentary credits, irrevocable and standby letters of credit, acceptances and endorsements. Guarantees also include the contingent liabilities of the protection seller of credit derivative contracts.
I.e. this is where US banks selling CDS on GIIPS book those sales. And why articles like Onaran’s can throw big numbers around. “But wait,” you can hear the banks saying, “we hedge these CDS sales”. Unfortunately for them, the FAQ section of the Bis statistical guide tells us that this won’t wash:
Q12: How should a bank report when it has sold a credit derivative and has then hedged this exposure?
A: A protection seller should report credit derivatives at face value under “guarantees” irrespective of the hedge.
Credit commitments:
Credit commitments are arrangements that irrevocably obligate an institution, at a client’s request, to extend credit in the form of loans, participation in loans, lease financing receivables, mortgages, overdrafts or other loan substitutes or commitments to extend credit in the form of the purchase of loans, securities or other assets, such as backup facilities including those under note issuance facilities (NIFs) and revolving underwriting facilities (RUFs).
This is relatively small fry but worth remembering as revolving lines of credit are often the first places to look when things turn sour.
______
The parameters of knowledge
The point in running through all that wasn’t some masochistic attempt by FT Alphaville to lose readers. It was to show that we think the Bis data may not be all that. It also shows that banks do have some case in the aggregate for arguing that these gross data do not reflect what would happen if Europe fell apart.
In fact, it’s worth remembering that there is only ever so much use in looking at “exposure” alone. MF Global was “exposed” to Europe but it was how that exposure was traded and the amount of leverage involved that determined its downfall.
However, the idea that their reported net exposure is the maximum they stand to lose if Europe goes all-out-crisis strikes us as fanciful. The counterparty risk is too big, as Onaran suggests. There will be concentrations of risk and we don’t know where those are. Contagion could prompt a firesale in risk assets. We don’t know the extent of any off-balance sheet exposure. The way banks actually report European exposure varies and isn’t crystal clear (to put it generously).
Thus, the best that we can probably do is provide some parameters.
At one end, these are the self-reported estimates of exposure provided by the big 5 US banks, net of nearly everything they can find:
At the other end, here is the aggregate exposure of US banks, including foreign affiliates, to all things GIIPS, as listed by Bis. In this case there’s hardly any netting going on, although it’s not fully clear what is included as per the boring discussion above.
Thus, “real” exposure lies somewhere between $46.4bn and $767.5bn.
Possibly. Because as Bloomberg’s Boris Groendahl reminds us there’s also an important footnote to the US data:
“(2)U.S. data are likely to be significantly revised”
Can’t possibly suggest the Fed and the Bis don’t understand or don’t trust what they see at first, can it?
Related links:
Will other banks go the way of MF Global? – Felix Salmon
How Do European Sovereign Debt Repos Work Anyway? – John Carney
Isda has a bone to pick with you (and so can we!) – FT Alphaville