Felix Salmon

Adventures with Greek restructuring math

By Felix Salmon
July 26, 2011

What does it mean, when a bank takes a 21% haircut on its Greek debt? With the release of Deutsche Bank’s results yesterday we have an interesting case in point: the bank took an impairment charge of €155 million on its Greek government bonds. That’s just under 10% of Deutsche Bank’s stated €1.6 billion in Greek sovereign exposure.

Deutsche Bank is well aware of the size of the haircut, of course — but not all of its Greek exposure comes in the form of securities classified as “financial assets available for sale.” Still, it would have been nice to have been told the size of the writedown in percentage terms, rather than just in cash terms.

Part of the problem here is that the exchange offer is rather opaque; Joseph Cotterill says that the structure of one of the discount bonds, in particular, is “still stumping analysts.” Here’s how the IIF describes it:

A Discount Bond Exchange offered at 80% of par value for a 15 year instrument. The principal is partially collateralized with 80% of losses being covered up to a maximum of 40% of the notional value of the new instrument. The collateral is provided by funds held in escrow. These funds are borrowed by Greece from the EFSF. The EFSF funding costs are covered by the interest earned on the funds in the escrow account so there is no funding cost to Greece of this collateral. The funds in escrow are returned to the EFSF on maturity, if not used, and the principal on the bond is repaid by Greece.

Let’s say you swap €1,000 face value of Greek debt into this new instrument. The first thing that happens is that you take a haircut: the new bond, which carries a coupon of 5.9%, will have a face value of only €800.

The second thing that happens is that Greece will borrow 40% of that sum — €320 — from the EFSF, the European bailout fund. It will put that money into an interest-bearing escrow account; the interest on the money will be enough to repay the interest that the EFSF is charging Greece.

And this is where things become unclear. In the event that there are “losses,” then 80% of your losses will be repaid out of that escrow account.

But what does that mean? My first thought was that it was something to do with a possible second haircut. Say bondholders were asked to take another 10% haircut on the new bond. That would be €80, and of that €80 the escrow account will repay you €64, which means your actual haircut would only be €16, or 2%. This continues up until the losses reach 50% of the value of the new bond, and the escrow account pays out in full. After that, there’s no money left and all further losses you suffer entirely.

That would be quite elegant, since it’s unlikely Greece would impose a second haircut of much more than 50%. But then I realized that coupon payments are just as important as the face value of any principal. If you reduce the principal by 10% and the coupon remains the same 5.9%, then you lose 10% of all your coupon payments, too. Let’s say there are 12 years left on the bond when the second restructuring happens — then a 10% haircut wouldn’t just knock €80 off your principal repayments, it would also knock €57 off your total coupon payments. Presumably 80% of that, or another €45, would also have to come out of the escrow account.

So now the escrow account is looking rather thin. Let’s say there’s another restructuring after 3 years with a more realistic 30% haircut, and the coupon staying at 5.9%. Then the escrow account loses €192 in principal and another €135 in interest, for a total of €327. We’ve already exceeded the amount of money in the account — it’s wiped out.

And the other question, of course, is when the money would be paid out of the escrow account. Would it pay out at the time of restructuring, even for payments which aren’t due for another decade or more? Or would it just sit there in escrow, paying out little €5.66 partial-replacement coupons every six months until it ran out of money? It’s a big difference, since €5.66 today is worth a hell of a lot more than €5.66 in ten years’ time.

I’m sure answers to these questions will slowly emerge — but for the time being, if you’re a bank tendering into the exchange, I think you can more or less write down your bonds by any vaguely plausible amount you like, and probably get away with it. If you’re marking to market, of course, it’s easy. But if you’re holding old Greek debt on your books at par, then — especially if you swap into another par bond — you probably have very wide latitude in where you mark the new debt. And if you swap into a discount bond, for the time being almost no one understands exactly what that’s reasonably going to be worth.

Update: Thanks to my commenters, especially Greycap, for pointing out that it’s just the principal, not the coupon payments, which are partially collateralized. But in a way this only makes things much worse. If Greece wants to restructure the bond, it will now have every incentive to push back maturities and reduce coupons to zero, rather than other options — because that wouldn’t trigger a payout of collateral. It seems very easy to game to me.

5 comments so far | RSS Comments RSS

“the exchange offer is rather opaque”

It’s not nearly as opaque as you are making it. The IIF document you linked to states several times, clearly and unambiguously, that it is principal that is collateralized. So your first train of thought was OK and your second one jumped the tracks.

“the other question, of course, is when the money would be paid out of the escrow account”

OK, that question is not explicitly addressed. But why should it not be paid out immediately? It’s hard to find a vested interest that would benefit by keeping the money sitting in escrow; though conceivably the issuer(s) of the collateral bonds might not want to see them hit the secondary market. Overall, the scheme looks like a very plain vanilla attempt to guarantee 40% recovery – a convenient number.

“you probably have very wide latitude in where you mark the new debt”

In economic terms, maybe you do. But once again, the IIF document tells you where debt is going to be marked, if you would just read it properly: “All instruments will be priced to produce a 21% Net Present Value (NPV) loss based on an assumed discount rate of 9%.”

No doubt the real mark-down ought to result in a bigger loss, judging from market prices of Greek debt. But not many players have an interest in higher mark-downs, and the IIF plan seems engineered to endorse 21%.

Posted by Greycap | Report as abusive

Felix, it’s a *discount bond* exchange. In other words, if you participate, you receive a zero-coupon bond, so there are no coupon payments (as such).

Posted by Kamekon | Report as abusive

Having reread the IIF financing offer, I think that my previous comment may be incorrect and that “discount” refers to the reduction of principal only. However, isn’t it true that coupon payments are not collateralized at all (so in your example the escrow account isn’t further reduced by EUR45) – only the principal is, and partially at that?

Posted by Kamekon | Report as abusive

Felix I just want to say thanks for your great coverage of the nuts and bolts of the deal. The internet is a big place but if there is a better source of analisis and informed debate I haven’t found it yet!

Posted by y2kurtus | Report as abusive

This article was -and continues to be, thanks to the commenters- a primary resource to me in digging deeper with the IIF article. Although I’m not clear as of now on all the aspects of the offer, I’d like to thank to all the contributors of this debate.

Posted by kahmet | Report as abusive

Post Your Comment

We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see http://blogs.reuters.com/fulldisclosure/2010/09/27/toward-a-more-thoughtful-conversation-on-stories/