One service that I like to provide is pointing my readers to ridiculous derivatives trades, and so a while back I pointed you to a ridiculous "snowball" swap that Portuguese state-backed railroad Metro do Porto did with Santander. Santander agreed to pay MdP a fixed rate, and in exchange MdP agreed to pay Santander a variable rate that started at zero, stayed at zero if Euribor stayed within certain bounds, and grew enormously and cumulatively if it went outside those bounds. It went outside those bounds. So the payments snowballed, MdP ended up paying Santander more than 40 percent a year, teeth were gnashed, etc.
Today I learned that this snowball was no aberration at Metro do Porto, which seems to have spent years accumulating an arsenal for the nerdiest and most expensive snowball fight imaginable. Here is an Independent article that sounds familiar: MdP entered into a snowball swap where it agreed to pay a bank some horrible accumulating goulash of interest rates,1 and that swap quickly and horribly went against it. But it's a different swap! This one was with Goldman and happened in early 2008, about a year after it entered into its snowball swap with Santander. And after the Goldman snowball went horribly awry, MdP decided to get out of it by entering into another swap with Nomura in 2009, in which Nomura agreed to pay MdP the goulash that it owed Goldman, in exchange for MdP paying Nomura a totally different goulash so outlandish that it reduced the Independent to incoherence.2
The best part is probably this:3
Sources close to Goldman say Metro do Porto knew what it was buying and that it was the result of a “request for proposals” -- in effect, a tender inviting banks to pitch.
That's amazing. A year after doing its first (?) crazy snowball, Metro do Porto was back in the market asking banks to throw snowballs at it. It just loved snowballs so much! Obviously the banks obliged: If you are a customer, and you love crazy derivatives, you will be very popular with your bankers.
What do you think the RFP said? There are two possibilities. One, virtually unthinkable, possibility is:
- Here is our formula:
- [Gives crazy formula ultimately found in the trade]
- Please fill in the numbers.
The other, far more likely, possibility is:
- We would like to swap the interest payments on our bonds into much lower interest payments.
- We understand that "we give you less money and you give us more money" is not a trade that you will want to do.
- But please sing us a song that sounds like that.
Right? A swap is a simple thing: I give you some money, and you give me a different amount of money, over and over again for several years. If I'm giving you a lot more money than you're giving me now, then you will probably end up giving me a lot more money later.4 The complexity of the formulas doesn't change that simple arithmetic.
What is the difference between now and later? Well, now is now. Later is later. Now, you are the treasurer of Metro do Porto, and you're paid and praised and promoted based on how much money you can save the company. Later, you won't be the treasurer of Metro do Porto, and you won't care at all how much money the company is paying on its swaps. I mean, that's the theory, though Metro do Porto's snowballs snowballed so quickly that the guy who did them was probably still around to get fired.5
This is a general rule to keep in mind when reading about governments and companies that were victimized by swaps that they didn't understand: Governments and companies don't understand anything. Governments and companies don't have brains. Governments and companies do have human agents, and those human agents have brains, and they are capable of understanding many things. Different agents might or might not be capable of understanding the particular formulas at issue here. But most agents are capable of understanding (1) that getting a low low teaser rate now probably means paying a high high rate later, (2) that there are ways of saying that that don't sound like that, (3) that their interests and the interests of the government or company they work for are not perfectly aligned, and (4) that those interests are likely to drift further apart over time.
That paragraph is sometimes abbreviated "IBGYBG."
If you're a bank, of course, your customer is supposed to be the institution: Whatever duties Goldman and Nomura and Santander had, they were to Metro do Porto, not the guys who wanted to put these goofy swaps on. On the other hand, you never meet the institution. You meet the humans who work for the institution, and you butter them up, and you try to win their trust and their business. And they want what they want, and they tell you, and you basically try to give it to them. I mean, you try not to bribe them, mostly, but who are you to second-guess their judgment about what's best for the institution? They work there. If they want a crazy derivative, who are you to tell them no?
Metro do Porto may have needed saving from its banks, but it probably also needed saving from its employees. And it's a bit too much to ask of the banks that they protect Metro do Porto from its own employees.
1 The Independent charmingly published a chunk of the termsheet; basically MdP paid Goldman
- 1.665 percent if both 12-month Euribor and 12-month Libor were below 7.75 percent, and otherwise 7.75 percent; plus
- The amount by which the greater of Euribor and Libor exceeded 7.75 percent, snowballed (so it adds up each time the index exceeds 7.75 percent), plus
- some stuff based on the 2s10s euro swap rate differential (this formula looks to me to have a wrong sign in it somewhere, so I'm not going to describe it in detail, but perhaps I am just dense), also snowballed, plus
- something based on the difference between the GBP and euro 10-year swap rates, also snowballed.
The snowballs are quarterly, and are all subject to caps, but the caps are averages, not absolutes, and can creep up.
Also the rate that Goldman paid MdP on its leg wasn't fixed either, so that adds to the goulash.
2 I have no idea what this means:
[T]he folk at Nomura decided to make things a bit more interesting. They added a new bet involving payments based on a “black box” index invented by Nomura. “The black box index is a real piece of work,” said one swap expert. It is in effect an algorithm devised by Nomura’s whizz-kids that responds to an array of movements in the financial markets. An investment in their picks is supposed to perform well whatever the financial weather, although in the past other algorithms have blown up spectacularly when unexpected events, like the sub-prime mortgage disaster, have hit.
Basically I guess this leg of the swap was indexed to a bunch of stuff, but who knows what stuff.
3 Though a close second is that MdP went to Goldman and Nomura and asked them to unwind the legs of the swap that were offsetting (where MdP was paying Goldman some formula and Nomura was paying MdP the same formula):
Imagine the look on the subwaymen’s faces when the two banks said they would only cancel the supposedly identical components of the two bets, in return for €26m to reflect the difference in their valuations of these identical parts.
Now in theory there should be a cost: Each bank should want to get out at its side of the market. But the original notional amount of this trade was 126 million euros (though with the snowballing that may not mean much), and a bid-ask of 20 percent of notional suggests that something had gone horribly wrong. As does this:
In 2010, Metro do Porto finally sought some independent financial advice and, as a result, confronted the banks about their behaviour. Perhaps fearing a public scandal, from demanding €26m to end the identical components of the two bets, the valuation models of the banks magically changed. The new valuations - funnily enough - meant Metro do Porto would get a windfall of about €20m to cancel the identical parts out.
Paying MdP 20 million euros is even more suspicious than demanding 26 million euros. The right number is, like, MdP pays each bank a small amount, maybe a million give or take, for bid-ask. If the banks are willing to unwind a zero-value aggregate position by paying 20 million euros, that sure sounds like they took too much at the front end.
4 Now here I exaggerate a bit. There's probably a term premium in swap rates such that a vanilla swap to floating should, more often than not, mean that the customer ends up paying the bank less than the (hedged, of course) bank pays the customer, though this is uncertain and modest. But as a general rule for interacting with banks, you should assume that if a bank is giving you $X, then in expectation you are giving it back more than $X.
5 I actually have no idea who approved this or what happened to them, though Risk has reported that, at the time of the Santander swap,
The financial manager was Coutinho dos Santos, now a lecturer in finance, while the director responsible for financial matters was Oliveira Marques, now a professor of finance.
To contact the writer of this article: Matt Levine at firstname.lastname@example.org.
To contact the editor responsible for this article: Tobin Harshaw at email@example.com.