Not actually a Portuguese metro.

Photographer: Timothy A. Clary/AFP/Getty Images.

Portuguese Train Companies Can't Escape Snowballs

Matt Levine is a Bloomberg View columnist. He was an editor of Dealbreaker, an investment banker at Goldman Sachs, a mergers and acquisitions lawyer at Wachtell, Lipton, Rosen & Katz and a clerk for the U.S. Court of Appeals for the Third Circuit.
Read More.
a | A

One major purpose of investment banks is to take financial risks away from customers. A customer has some risk that keeps her up at night, she goes to an investment bank, and the bank writes her a contract that takes away her risk in exchange for a nice fee.

But you can't just make risk disappear. Once the investment bank has taken the risk, where does it go? To the bank's creditors? To its depositors? To the taxpayers? That isn't ideal. The best answer, from the bank's perspective, is often to find another customer to pass the risk along to. Sometimes this works out neatly, because a risk for one customer is a windfall for another, so you can get them to sell each other their risks. (This is the standard story of airlines and oil companies hedging oil prices with each other. ) Sometimes you can find customers who just like risk and want to take more of it: Aggressive hedge funds, say, or pension funds with long-term outlooks, might be quite cheerful bearers of various kinds of risk, and if a bank can pipe risks from worried corporate clients to eager hedge funds, it can make money while making the world a better place.

But the world is a fallen place, and another thing that investment banks sometimes do is look around for corporate or government clients who don't particularly want more financial risk, but who can be talked into taking it for a fee. Sometimes this works out great, but when it doesn't -- when the risks come to pass -- it is always a bit embarrassing. Why, people ask, was a sophisticated investment bank selling kooky financial risk to an unsophisticated, oh say, public-transport company?

There is a further embarrassment. If you are taking risks from clients, they are paying you. In fact, they are necessarily overpaying you: You're charging them more than the fair value of the risk, because that's how you make money. If clients are taking risk from you, on the other hand, you are paying them. But not enough! In a world of bid/ask spreads, you just as necessarily underpay them for the service. Which, again, is awkward later.

We have talked a couple of times about a Portuguese train company that did some ridiculous "snowball" interest-rate swaps with Banco Santander, which one expert called "a contender for the worst trade of all time." That company, Metro do Porto, was one of four Portuguese public-transport companies that did these silly swaps with Santander, which have moved against the companies to the tune of 1.3 billion euros ($1.43 billion). The companies stopped paying 272.6 million euros ago, in 2013, and have tried to get out of their swaps, with their basic theory being, you know, come on, these swaps are ridiculous. "Come on, these swaps are ridiculous" is honestly not that bad of a defense: The swaps were so dumb that their own dumbness is prima facie evidence that the train companies were in over their heads, and that Santander knew it and took advantage of them. But the train companies lost last week, as a U.K. court ruled in Santander's favor and told the companies to pay up. 

The court's judgment is 163 pages long, and if you are interested in how investment banks sell risk to customers, it is a delightful read. To review, here is roughly how the swaps worked :

  1. Santander would pay the companies a floating rate;
  2. the companies would pay Santander a fixed rate;
  3. the companies would also pay Santander an extra "spread" if the floating rate was above some cap (say 6 percent) or below some floor (say 2 percent); and
  4. each quarter's spread would be added to the previous quarter's: If the floating rate was 6.25 percent one quarter, the spread for that quarter would be 0.25 percent; if it was 6.50 percent the next quarter, the "spread" would be 0.75 (0.25 + 0.50), and so on. (That's the "snowball.")

You can sort of see the transport companies' reasoning. They had a lot of debt, and they had to pay a lot of interest, and they didn't want to. So they were in the market for a transaction that would reduce their interest costs. The promise of the snowball was that, in normal circumstances, it would be much cheaper than regular fixed-rate swaps.  

The downside is that, in abnormal circumstances, it would be horrible. Abnormal circumstances obtained: All of these swaps were entered into between 2005 and 2007, at a time when three-month Euribor, the floating rate used to compute some of the spreads, had never been above 6 percent and had almost never been below 2 percent. Then 2008 happened, and the rate fell below 2 percent and stayed there. (It's been below zero for almost a year now.) The rate that Metro do Porto had to pay on its swaps got above 40 percent, and the other companies are in similar boats, or I guess trains. 

So two obvious questions are: Were the companies dumb, and was the bank evil? There are some indications that Santander was maybe a little evil. Sir William Blair, the judge in the case, writes:

It needs to be said at the outset that there are documents that, though not typical of the overall documentation, do the bank no credit. A presentation of November 2005 from Global Treasury (so not emanating from Portugal) aimed at boosting the sale of "exotic products" encourages employees to "Think Big: don't criticise ourselves. Reward Aggressiveness." 

You gotta criticize yourself a little bit. Santander was also pretty annoying: "One internal BST email refers to the necessity to keep the 'siege very tight' on the finance officer concerned." (BST is Banco Santander Totta SA, Santander's Portuguese bank.) And even within Santander, it is clear that there were some worries about the credit and reputational risk of doing these silly swaps.  

Similarly, there is some evidence that the companies were a little dumb. "They were not in the court's view at a particularly high level of financial sophistication," says Mr. Justice Blair, diplomatically. As evidence of their own dumbness, the companies noted that one Metro do Porto official said that "Many banks proposed snowballs swaps -- they were not proprietary to BST -- and it appeared to be a commonly used instrument in the financial market." It is ... not. It's a crazy trade! It's "a contender for the worst trade of all time"! But they thought it was totally normal.

But mostly Mr. Justice Blair concluded that the trade was neither stupid nor evil. He concluded that "there cannot have been any real doubt on the part of those responsible within the Transport Companies for entering into the swaps about the scale of the risk that the swaps entailed": They knew it was a "very attractive pick up" in current interest expense, and a "very, very aggressive risk profile." And he said of the senior Santander salesperson:

The court does not consider that she (or her colleagues) set out to persuade Transport Companies to enter into contracts that she thought would be contrary to their interests. The 20 June 2007 report supports the conclusion that the derivatives were seen at the time as advantageous to the bank but also to the clients. The court finds that these swaps were entered into because they suited both parties at the time: the bank got to enter into very profitable transactions, and the Transport Companies got appreciably lower rates than they would otherwise have been paying.

"However," the court adds, "this was at a price in terms of risk, and much of the dispute in this case turns around this equation."

But it doesn't really. There's just not much to say there. The transport companies said it all, succinctly: "The Transport Companies were, as they put it, selling volatility to the bank in return for low interest rates."  Santander paid the transport companies a lot of money; in exchange, the transport companies took a lot of risk -- the risk of very high or, in the event, very low interest rates -- off of Santander's hands. Then the risk came true.

Where do you go with that? The transport companies argue, in essence, that they shouldn't have been allowed to do that: that it is illegitimate for companies, or at least Portuguese public transport companies, to get paid to take risk off of banks' hands. 

So for instance the companies argued that "under Portuguese law, each company lacked the capacity to enter the swaps" because they were "speculative transactions"; the theory is that it's fine to use derivatives to hedge (to pay to offload risk to investment banks), but not to use them to, um, anti-hedge (to get paid to take risk from investment banks). But there is no particular authority for that. "The experts agree," says the court, "that hedging is only one of a number of purposes of swaps": Reducing interest expense is also a legitimate purpose, and "the fact that a swap is not a hedge does not mean that it is necessarily speculative."  

People sometimes attach legal and moral significance to terms like "hedging" and "speculation," but the swap doesn't care. Were these companies doing these snowball swaps to "hedge" (or at least manage) their current interest rates, or to "speculate" on the future of interest rates? A little of both, I guess, but the companies' (managers') state of mind isn't particularly relevant to how the swaps worked.  The companies got paid some money, and then later they had to pay some money, though they might not have had to if things had worked out differently. They'd have preferred that outcome, but anyone would have; that doesn't necessarily make them speculators.

There are other arguments in this vein, including that speculative swaps are a "game of chance" that are illegal under Portuguese law. The court disagreed, but there's some intuitive appeal to that argument. When a bank helps a company reduce its risk, that seems like what the bank (and the company) should be doing. When a bank helps a company increase its risk, that seems like, you know, gambling. It is much the same to the bank -- it has its own accounting of risk, and when it takes risk away from one client prefers to pass it on to another -- but for the company it looks a little suspicious.  

Other arguments are less about the derivatives themselves, and more about the bank-customer relationship. The companies argued that Santander should have avoided "conflicts of interest" and given "preference to clients' interests" over its own. This argument sounds nice, but it cannot be right; it's a complete misreading of what derivatives trades are. The companies weren't clients; they were counterparties, and the whole transaction was, in its essence, a conflict of interest. As the court said:

The key point is a functional one -- when a bank is acting on its own account, its duties to the counterparty under EU legislation (or its national implementation) are commensurate with that relationship. There is no room for applying a rule relating to conflicts of interest in this situation. Where a bank is dealing on its own account, plainly it cannot be expected to give preference to the counterparty's interests over its own. In the case of an interest rate swap, by definition, a movement in interest rates that is positive for one party, will be negative for the other.

Cut out and save that paragraph; it will be useful in any number of conversations about whether investment-bank salespeople ought to be fiduciaries for their customers. If the bank and the customer are on opposite sides of a trade, the trade is itself a conflict of interest, and the only way to avoid a conflict is never to trade.

There's one last argument, which is that the swaps should be void because of an "abnormal change of circumstances": Because of the financial crisis, interest rates were lower for longer than anyone anticipated in 2007, and since the parties didn't expect the crisis, the contracts should be torn up. Of all the arguments in the case, this one seems the craziest to me. The point of a derivative contract is to manage unexpected risks. This is a contract in which the transport companies agreed to pay Santander a lot of money if, among other things, interest rates were low for a long time. It's right there in the formula. The fact that interest rates were in fact low for a long time isn't a reason to tear up the derivative: It's a sign that it did what it was supposed to do, and hedged Santander against a scary risk. If the trade had been the other way -- if Santander had hedged the client against a scary risk that came true in a huge way -- no one would be likely to complain.

Weirdly, though, this is also the argument that did the best with the court. Mr. Justice Blair found that "the fact that since 2009 policy and swap reference rates have been at unusually low levels for an abnormal length of time, by itself amounts to an abnormal change of circumstances" and isn't covered by the contracts. But he also concluded that this whole argument -- the ability to get out of the swaps for an "abnormal change of circumstances" -- is a feature of Portuguese law that didn't apply to these (English-law) swaps, so he didn't do anything about it. Still it is very strange to think that an interest-rate swap could be void just because interest rates have moved. What is the point of a swap, if not to protect you when rates move unexpectedly?

Or, I mean, the opposite. That's how swaps work: One side is long each risk, and the other side is short, and the more you are paid to take risk, the riskier that risk will probably be. At some level, Santander's customers had to have known this, and the English court was probably right to hold them to their deal. But you can understand why they'd feel a bit run over by their snowballs.

  1. To its shareholders? Investment banks are pretty levered, so the shareholders get just a little first-loss slice.

  2. Yes I know that investment banks don't have depositors, but most major derivatives-dealing investment banks are part of larger bank complexes that do. And for a reason!

  3. One thing to say about this flavor of risk transfer is that it is pretty linear. Sometimes a bank has some customers who want Quantity X to be high, and other customers who want it to be low, and so they can hedge with each other.

    But this works less well with nonlinear risks, what we call "options," or "volatility": Lots of customers want Quantity X to be stable or predictable, but fewer customers have a natural desire for it to be erratic. So they come to banks to hedge volatility, and natural customers on the other side are harder to find.

    The rest of the text is more about volatility than it is about linear quantities.

  4. Companhia de Carris de Ferro de Lisboa SA, which "operates the bus, tram and funicular (elevadores) system in Lisbon"; Metropolitano de Lisboa EPE, which runs the Lisbon metro; Sociedade Transportes Colectivos do Porto SA, which runs buses and trams in Porto; and MdP, which runs the Porto metro. They are state owned, "or state and local authority owned (in the case of MdP)."

  5. This oversimplifies. Some of the contracts were fixed-for-fixed with only the "spread" changing; some involved different interest rates for the floating leg and the spread; many involved leverage in which the spread was a multiple of the difference between the floating rate and the floor/cap. There are termsheets in the Annex to the judgment (pages 138-163), and a summary in paragraphs 150 and 151 (pages 29-32).

  6. See paragraphs 138-139 (page 27) of the judgment. The range is 77 to 334 basis points per year cheaper, versus "the equivalent 'vanilla' swap."

  7. There is also a crazy story about a swap that Santander sold to one of the companies, STCP, in a deal that also involved Société Générale somehow:

    Société Générale had done a sensitivity analysis in relation to the transaction. It emailed Ms Antunes figures stating that in a "stress scenario you can lose up to 389,368,711 USD. Please confirm us your customer has received all this information".

    In fact, it was not passed on to STCP, the BST response being, "What we can guarantee is that we are aware of the transaction we want to close".

    (Cristina Antunes was a senior Santander salesperson for these trades.) That is not a good response. 

  8. For instance, Santander's Credit Committee expressed its "concern for the reputational risk associated with these operations," and required "stress testing so that the client understood the risks, making alternative options available to the client, and making sure the client was not making a 'merely speculative transaction.'" There's a lot of discussion of stress testing, which seems to have been emphasized more at committee meetings than in marketing; one committee memo "gives what the court regards as a not fully accurate view of the sales process, in particular a reference to stress testing for 'extreme scenarios.'" That didn't always happen; for instance, with the Lisbon Metro, "BST's proposals did not include any forward or stress testing because they knew from their discussions that MdL was undertaking its own internal analysis."

  9. Really: The court quotes this, and says "The Transport Companies cite this as evidence of naivety."

  10. "The court finds (and it is not seriously in dispute) that these instruments were not in fact common in the swaps market as a whole." Santander seems to have sold 14 of them in Portugal, all to public-sector enterprises, and knocked it off in June 2008.

  11. Paragraphs 72 and 73: 

    The Transport Companies say that the possibility of accumulating interest rates are not the only risks inherent in these swaps. They say that the risks can be illustrated by likening the snowball spread to a series of options. The Transport Companies were, as they put it, selling volatility to the bank in return for low interest rates.

    BST does not accept the option analysis, but the court thinks that it is broadly correct. There were not of course true options in the swaps, which appears to be Mr Evans' objection, but there was similar functionality, and there is some evidence that BST itself recognised this functionality as part of the price (see below).

    That "not of course true options" objection sounds absurd if you are a derivatives person. Interest-rate floors and caps are the canonical examples of interest-rate options; the fact that no one has to "exercise" the options is an irrelevancy. Of course the snowball was a series of options!

  12. As the court summarizes it:

    The Transport Companies' case is that they took on the risk of paying the snowball spreads in return for obtaining lower fixed rates of interest payable under the swaps than the fixed rates that would have been payable under plain vanilla swaps. If interest rates stayed within the barriers the bet was a good one, if interest rates went outside of the barriers, the bet was a bad one. The swaps involve speculation in future interest rates with the object of making a profit in order to reduce borrowing costs.

  13. And:

    It could be said that all swaps contain an element of speculation in this sense: in the case of vanilla swaps used for hedging against rises in interest rates, if interest rates fall, the counterparty paying the fixed rate will have turned out to make the wrong call. If rates rise, the swap will be an instrument with an inherent value that can be realised if desired. The same could be said of most investments.

  14. There is a weird amount of talk about the expected probabilities that the swaps would end up being profitable for the companies. For instance (paragraph 140): "Overall, the experts agree that the swaps had an approximately 72% chance of providing these potential benefits to the Transport Companies over the life of the swap." 

    Of course! That seems low, to be honest. Again, take the MdP swap we discussed previously. It would be profitable for MdP as long as Euribor stayed between 2 and 6 percent, which it almost always had for the entire history of the euro. The ex ante probability that that swap would remain profitable would, you'd think, be pretty high.

    The issue isn't probability but expected value. The problem with these swaps isn't that they were unlikely to work out for the transport companies. The problem was that, if they didn't work out, they wouldn't work out in a giant way. They were small deep holes. If they worked out, they'd save the companies a few hundred basis points a year. If they broke, they'd cost a few thousand basis points a year. Or as I put it earlier:

    There's a series of dials to turn: The more money you save now, the longer you save for, and the less likely you are to pay more later, the bigger the amount of money you'll end up owing if the trade moves against you. The bank knows exactly how those dials work, and the math is no doubt hard, but anyone can figure out the gist: The lower the probability that you have to write the bank a check, the bigger that check has to be.

  15. "Even if it were established that the swaps were 'purely speculative', the recent cases support the view that speculation is not an illegitimate activity and that entering into interest rate swaps, being a regulated activity, is not gaming or betting."

  16. One expert argued that where "(i) swaps created [risk], rather than operated as an antidote to risk and (ii) that the margin of risks assumed by the parties was dramatically imbalanced," they are games of chance. I don't quite know what that means, though clause (i) relates to my formulation.

  17. Santander also "underpaid" the companies for taking on this risk, in some Platonic sense: The amounts that it paid the companies on its legs of the swaps were less than the model-derived expected value of the payments that the companies agreed to make to Santander.  No one makes too much of that here, perhaps because the initial profitability of the trades is so trivial compared to the later moves against the companies. So the first Lisbon Metro swap was singled out for praise at Santander because it booked an initial profit of $7,562,500, meaning roughly that Santander "underpaid" Metropolitano de Lisboa by that amount. That swap had a mark-to-market value against MdL of 185.5 million euros as of March 2013, and has probably gotten worse since. The initial profits -- the pricing of the swap, as opposed to the contingent outcome -- are pocket change.

    Incidentally, paragraph 92 has a list of the initial mark-to-market values for these swaps, which were all negative (for the companies -- i.e. profitable for the bank), as you'd expect. But paragraph 93 says:

    It should be noted that, to the limited extent that BST statements are available that far back, the numbers differ, and are positive. This is a reflection of the fact that the calculation of the mark to market value of a swap involves a good measure of judgment.

    Hmm, yes, but hmm.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

To contact the author of this story:
Matt Levine at mlevine51@bloomberg.net

To contact the editor responsible for this story:
James Greiff at jgreiff@bloomberg.net