Goldman's Libya Derivatives Were Fine
Selling derivatives is basically a miserable life. If you're selling a car, you show the customer the car, and you're like, look at this car. It's blue, it has some horsepower, it can make both left and right turns, the seats are soft, it has twelve cupholders. The customer gets in the car, drives it around the block, turns on the windshield wipers. He has an intelligible experience of the car that he can relate to other cars he's encountered. If he likes the car, he gives you some money, and you give him the car. He values the car more than he values the money. You value the money more than you value the car. Everyone wins.
If you're selling a non-expert customer a derivative, you show him a termsheet, and you're like, look at this derivative. It's a cash-settled forward on $600 million of Citigroup stock, it has an embedded put to floor your downside at zero, there's an upside cap at 140 percent on 90 percent of the trade, it has a nine-month lookback. The customer ... nods. "When do I get the shares," he asks. There is an awkward silence. He has no intelligible experience of anything that you are talking about. His goals are simple: He wants to give you a small amount of money to get back a larger amount of money, without much risk. Your goal is ... more or less to persuade him that he's getting that? Preferably without actually lying? While also making a lot of money for yourself? But you are limited by the constraints of economic reality: Upside comes with downside, leverage comes with risk, profit to you is cost to him. Eventually he gives you money, and you give him a derivative that is probably worth less than the money. Maybe he values it more than he values the money, but in expectation, that will change. He'll probably call you to complain when it does. Nobody exactly wins, though you have made a profit, which is not nothing.
On Friday, the Libyan Investment Authority lost its lawsuit against Goldman Sachs over some derivatives trades. We talked about that case a couple of weeks ago, and I should disclose that I worked at Goldman Sachs, selling derivatives, and in fact I know some of the characters who appear in the judge's decision. The basic issue is that Goldman sold the LIA $1.2 billion worth of call options on some stocks, mostly of big international banks (starting with Citigroup), in early 2008. The call options would only pay off if the stocks went up. The stocks went down -- it was 2008 -- and the LIA lost all the money it paid for the call options. So it sued, and now it has lost.
That decision is fascinating reading, and dredges up some memories for me. Here, for instance, is an e-mail that Goldman coverage banker Youssef Kabbaj sent to his colleagues, including the guys responsible for structuring and pricing derivatives trades for Libya:
We showed the structure to the chairman of LIA yesterday. He likes the idea of (i) getting exposure to Citi (ii) buying a lookback option on the first 6 months but he thinks that LIA has enough cash not to have to enter into any leveraged structure. On this, his [Chief Investment Officer and Chief Operating Officer] disagree. After some discussion, we agreed to meet this Sunday again to look at the two options: leveraged and unleveraged. GS is to prepare a two pagers in plain English to explain the structures to the chairman. They also specifically asked to have at least 50% participation on the upside. They are flexible on the 5 time leverage so we can do 4 times if needed.
I suppose that to a lay audience that sound like finance guys talking about finance to other finance guys, but as a former derivatives structurer my reaction was: arrgghh. Those words all sort of mean things, but it's not clear what they mean, particularly in that combination, or what the client's actual economic goals could be. Kabbaj was a coverage banker, not a derivatives structurer, and he was talking to the chairman of the LIA, also not a derivatives structurer. "What about a lookback," asked Kabbaj, I assume. "Don't mind if I do. How bout some leverage?" "Sure, leverage it up!" His colleagues were nonplussed. What, exactly, was he talking about with the client? How would those vague conversations about the LIA's hopes and dreams translate into an actual deal?
But eventually a Citigroup trade was agreed, as far as anyone could make out. The LIA would pay Goldman $200 million for three-year call options on $600 million of Citigroup stock, with some bells and whistles. The LIA team got its board to approve, with a presentation that ... didn't really describe the trade? I mean, it tried to -- it may or may not have been prepared with help from the Goldman team -- but it just didn't quite. It waved in the direction of the actual trade. It got the economics more or less right. I think. It was kind of gibberish. It did, however, come with an amazing -- and accurate -- payoff table, prepared by Goldman Sachs:
The Citigroup Board Memo then sets out a payoff table which has been cut and pasted from another document. This is said to show a summary of the possible outcomes of the trade for guidance purposes. The table shows the internal rate of return that would be achieved by different percentage increases in the value of the shares. The lowest figure on the table shows what would happen if there were a 30% increase in price so that the price of the shares at maturity was $32. The return in that situation would be $182,370,821 which, since the initial premium is $200 million, would give a negative cumulative and internal rate of return. The posited share price increases shown in the table then rise in increments of 10% to 280% which would result from a share price of $93 and generate a return of over $1.7 billion.
This is not exactly a surprise. If you pay $200 million for an at-the-money call option on about 24.9 million shares of Citigroup stock, which was then trading at about $24.45, then you'll only get any money back if the stock goes up. If Citigroup finishes at $24.45 you will get back zero. (And be out $200 million.) You need the stock to go up to about $32.50 to break even: Then your call options will finish up in the money by $8.05 per share, or about $200 million total, earning back your premium. If the stock is up only 30 percent, you will lose money. If it's up 20 percent, you will lose more money. (About $79 million.) If it's up 10 percent, you'll lose even more. (About $139 million.) If it's flat, or down, you will lose the full $200 million. (You can't lose any more than your premium.) Sometimes stocks go down! This table started with the stock being up 30 percent, which would be a small loss for the LIA, and ran all the way to up 280 percent -- two hundred eighty percent -- which would be a huge gain for the LIA. The rest of the range from up 30 percent down to, you know, down 100 percent, was omitted from the chart. Three years after that board meeting, Citi's stock was down 81 percent. The LIA lost its entire $200 million, and repeated that feat on all of its other disputed option trades with Goldman.
You can see why the table cut off at up 30 percent! Everything to the left of that was depressing; why talk about it? But it doesn't seem like Goldman was the one trying to bury the bad news. Here is a passage in which the judge, Dame Vivien Rose, questions a junior member of the LIA team, Gamal El Harati, about Mustafa Zarti, the deputy chairman and main decision-maker on these trades:
MRS JUSTICE ROSE: When you said, "If we put negatives in the report, Mr Zarti would say, 'You have to remove that'", what do you mean by "negatives in the report"?
A. The, like, pros and cons of a certain investment. So say, for example, that we believed that the stock price -- one of the risks of this investment is that the stock price, for example, might go down or we are going to be giving up the dividends, he's like, "You don't have to put that in; I mean, I will explain it".
The LIA was full of optimists. And Goldman let them be optimistic. A lot has been written about the efforts of Kabbaj, the Goldman coverage banker, to woo the LIA through the traditional techniques of wining and dining, but his most impressive marketing move may be this e-mail:
Early in January 2008, Mr Kabbaj sent Mr Matri (a member of the Equity Team) an email with the subject heading ‘Why it may be a good time for the LIA to buy some Financials?’. He attached to the email an article from the Financial Times on 4 January 2008 describing what the author calls “the latest lemming-like craze to hit Wall Street: the repairing of many firms’ badly depleted capital accounts by taking money from deep-pocketed, state-owned foreign-investment funds”.
Kabbaj tried to get a sovereign wealth fund to invest in bank stocks by telling them that sovereign wealth funds investing in bank stocks was a "lemming-like craze." And it worked! That shouldn't work.
Last time we talked about this case, I said that the essential problem -- the basis for LIA's complaint -- was that Kabbaj was too good a salesman. He had "undue influence" on the LIA, it now claims; its staffers couldn't resist the allure of his derivatives. He befriended them, they trusted him, and without any independent ability to judge the merits of these trades, they just did whatever he said. The judge didn't buy it:
In cross-examination the [LIA] Equity Team members accepted that they understood at all times that Mr Kabbaj was a salesman for Goldman Sachs and that his job was to sell investments to the LIA from which Goldman Sachs would make money. Some of them gave evidence that they were a little guarded about Mr Kabbaj’s obvious attempts to ingratiate himself with the team
Even the fact that Kabbaj spent so much time and money on entertaining the Libyans is an ambiguous fact: "Indeed the perceived need to keep providing expensive entertainment in order to maintain the relationship rather negates the idea that the relationship had grown into one where Goldman Sachs could exercise undue influence."
But beyond her factual skepticism, the judge's decision reflects doubt that the law really concerns itself with salesmen who are too good -- that a salesman can exercise "undue influence" just by doing a really good job of befriending his customers. She describes an earlier effort by Goldman to guilt the Libyans into finalizing the deal:
They also complain that Mr Vella suggested to Mr Kabbaj that he play what the LIA referred to as ‘the emotional card’. The LIA say that Goldman Sachs took advantage of the close friendship that had developed between Mr Kabbaj and the Equity Team by telling Mr Kabbaj to say to them that he might get into trouble if they did not commit to a deal. Again, I do not consider it is fair to criticise Goldman Sachs for being concerned that the LIA might simply walk away from any deal leaving Goldman Sachs unable to recoup the costs of its weeks of effort. As to the ‘emotional card’ I am sure that it would not have helped Mr Kabbaj’s career with Goldman Sachs if, after all that had happened, he did not manage to bring a deal to fruition.
This is all very clear-eyed. Goldman had spent time and money building a relationship, and wanted to be repaid with a profitable deal. Kabbaj was the actual person most responsible for that relationship, and so had the most at stake in actually completing a profitable deal. The relationship was about money, and he wanted his money. It's a bit rude to put it like that, as you're actually building the relationship. But once the relationship is gone -- once you're in court -- that's really the only way to look at it.
Besides claiming that Kabbaj was too good a salesman, the LIA also claims that, in effect, it was too good a client. It had a huge pile of money that was burning a hole in its pocket, with a mandate to invest quickly and no particular idea of what it was doing. It was incapable of understanding the complicated trades that Goldman showed it, and so it was just misled by Goldman into doing terrible trades. This seems ... mostly wrong. There is definitely evidence that the LIA didn't quite understand what it was doing. That memo to its board about the Citigroup trade was nonsense, and reflected the LIA's most fundamental misunderstanding here: Apparently it thought that it was buying shares, and was outraged to find out that it was only buying a derivative.
But all in all, the record looks like LIA wanted to get leveraged long exposure to financial stocks, and Goldman tried to get it that exposure in the most plausible possible way. Just lending the LIA money to buy a bunch of financial stocks was not really feasible: If the stocks went down, Goldman would have trouble enforcing its loan. So it came up with a derivative structure to get as close as possible to what the LIA wanted. (Also to make the most possible money for Goldman, though the judge has no problem with that: "The fact that a bank prefers the client to do a deal which generates a higher profit does not show that something unusual or inappropriate was happening.") And it explained the structure to the LIA. And the LIA ... sort of understood? In broad strokes? It was apparently missing a big element -- that it would never actually get the shares -- but that doesn't seem to have been Goldman's fault, and more importantly, it just didn't matter that much. Even if the LIA had gotten the trade it thought it was getting -- borrowing money to buy shares on a levered basis, and buying a put to limit its downside risk -- it would have ended up in the same place: rapidly losing all of its money. The particular form of the trades was not the problem; the problem was that LIA really wanted levered long exposure to financial stocks just before the global financial crisis hit. In some ways the LIA's continued obsession with actually getting the shares shows just how naive it was: It's an irrelevancy, but it's the only thing that the LIA focused on. "What drove Mr Mustafa crazy was the fact that we had no shares in the trades that we did with Goldman Sachs," recalls one LIA employee.
The other naive thing that the LIA did is that when Goldman told it that its derivative would cost $200 million, it just ... paid the $200 million. You're not really supposed to do that either. The bespoke over-the-counter derivatives market allows for a certain amount of haggling. You could say, for instance: "What about $190 million?" You could engage an independent consultant to model up the trades and come up with a fair price. Or you could just go to another bank, show them the term sheet, and ask them to give you a better price. (This is rather rude!) Just accepting whatever price you get, with no check on its fairness, is a bit silly. But it's not that silly. It happens. "Even if the profits were greater because of the LIA’s failure to challenge the prices it was offered," writes the judge, "that would not be unusual and so not something from which an inference of undue influence could arise." There's a lot of discussion about how much money Goldman actually made on the trade, and whether that profit was so crazy as to itself prove undue influence, and I will summarize it by saying: The profit was very large, but not obviously inappropriate.
This case boils down to the question: If you are a bank, and you are smart and good at finance, and you come across a customer with a gigantic pile of money, and the customer isn't especially sophisticated at finance, and you rub your hands together in glee about how much money you can make, and you wine and dine the client, and you get the client to do some big trades, and the customer doesn't exactly understand every detail of the trades, and they work out badly for the client, and they look in hindsight kind of stupid, but they make you a lot of money -- is that bad? There is a long tradition, in the financial industry and the law, of thinking that it's just fine. It's unseemly, sure; you don't talk about it quite so explicitly. But the life of a derivatives salesman is hard; lots of potential trades never work out, and lots of customer are tough and sophisticated and wring every penny of profit out of you before agreeing to a trade. When you find a customer who will do huge trades at huge profit margins without complaint, that makes up for a lot of troubles elsewhere, and you will buy that client all the fancy dinners it can eat. And you will trust in the longstanding assumption that the customer is an arm's-length counterparty, responsible for looking out for its own best interests just as you are responsible for looking out for yours. That assumption has been questioned a lot recently: After the financial crisis, it is no longer entirely satisfying to say that customers dealing with banks should look out for themselves, and that banks have no responsibility for protecting customers from themselves. But, in the U.K. at least, it is still the law.
That is: If someone pays you $100 million for a derivative, and your profit margin is 3 percent, then he's getting back a derivative worth, in expectation, $97 million. You can critique this math as a matter of derivatives math -- what does "expectation" mean, anyway? -- but it has some force.
Mr Magnifico is currently an Executive Director in the Equity Derivatives team which is part of the Investment Banking Division of Goldman Sachs. As a junior analyst at the time of the Disputed Trades, his role was primarily one of co-ordinating the various tasks necessary to propose, revise and execute the trades.
Some time after that, he came to New York and worked on the same desk as me. Dmitri Potishko, who was at the time of the trades an equities trader in New York responsible for pricing exotic equity options, also appears in the opinion; he was a trader who worked with my desk for much of the time I was there. I probably met one or two of the other characters in the opinion too. I don't recall ever discussing Libya with any of them.
From the decision:
The response to this was some confusion within Goldman Sachs about which of the several different iterations that had been discussed the LIA were in fact interested in. Mr Devenish, a senior person at Goldman Sachs who had been copied into the exchange, said in response “I really cannot keep up with these iterations and personally I think we should call a timeout on this stuff until Andrea has met them face to face”.
Andrea is Andrea Vella, a Goldman partner with derivatives expertise. Later requests from Kabbaj for more prices and structures "brought a rebuke" from a derivatives structurer, "reminding him that it was part of his role to moderate the demands placed by the client on the Goldman Sachs trading team."
Here's another fun e-mail that Kabbaj sent:
“Just spoke with Abdulfettah. On EdF, LIA would like to compare three strategies:
- delta one with lookback paid by dividends
- leveraged similar to Citi
- caesar or any other one we recommend
They are looking to see expected payoffs for 100 m euros and [internal rates of return] as functions of final price.
Enrico, should we meet tomorrow to prepare these?”
The judge comments:
Mr Enaami said in his witness statement that this could not be an accurate record of what they discussed or what he had requested because the only ‘caesars’ of which he was aware were Roman emperors and salads. That does not mean that Mr Kabbaj fabricated the discussion. There is nothing unusual in a banker having a discussion with a client using layman’s terms and then reporting it back to his colleague using terms of art with which he knows his colleague is familiar. The fact that Mr Kabbaj may not have referred to a potential structure as a ‘Caesar’ when talking to Mr Enaami does not mean that he did not explain what it actually was to Mr Enaami or that Mr Enaami did not understand it, even if he did not know it was called a Caesar among the banking fraternity.
A "Caesar" is a silly name for a pre-paid put. (I believe it's a quasi-acronym for "cash enhanced share repurchase.") I mean, presumably the client didn't say "delta one" either. That e-mail boils down to "let's show them three choices: Just buying the stock, buying calls, or selling puts."
Those really are three ways to get exposure to a stock! But it seems a bit weird to be indifferent among them. And I would not want to explain the difference. Not because it's hard -- though I suspect that you make it harder on yourself by using terms like "Caesar" and "delta-one" -- but because some nuance is going to go missing in the explanation, and later it's something they'll fixate on.
Principally a lookback feature that "gave the LIA the advantage of any temporary dip in the price in the early months of the contract and conversely generated a greater risk for Goldman Sachs that there would be a larger positive difference between the strike price and the maturity date price."
It also made the trade weirder, more expensive, and harder to compare to simple derivative pricing models.
Oh, by the way, I say "call options," though they were documented as forwards plus puts. Those are the same. Also some of the trades -- not the Citi one -- were actually call spreads, where LIA's upside was capped at some level well above the initial price.
Here you go:
“Therefore, we would like to inform you that investment in this Bank may be carried out by dealing with Goldman Sachs, while applying the following strategy:
1. The Libyan Investment Authority pays USD 200 million.
2. Goldman Sachs Bank is to be used in designing an investment portfolio to acquire stocks in Citigroup by attaining leverage at and equivalent of USD 607,902,736 for three years.
3. Dividends distributed on Citigroup stocks are to be paid to Goldman Sachs in advance to offset part of the cost of borrowing (LIBOR 3.34%), considering a settlement to be made after three years.
4. Loan surety is to be paid through purchasing a put option from Goldman Sachs.
5. The lowest price possible per share during the next nine months will be secured through purchasing a financial derivative called “Look Back Option” at a 90% barrier.
This strategy will enable the Libyan Investment Authority to acquire the equivalent of 0.5% of Citigroup.”
At a high level, that accurately describes the economics, but it's sort of a goofy way of putting it. It also seems to describe the trade -- as far as I can tell -- as a purchase of the underlying stock, with a loan from Goldman, and the purchase of a put option to protect the downside. That is more or less economically equivalent to buying a call -- if you draw the payout profiles you'll get the same results -- but it is not quite the same trade. For instance, if you borrow money to buy the stock, you have the actual stock.
These numbers tie to the presentation but not, apparently, to the actual trade, which was on "about 22 million shares" -- presumably because it was executed after Citi had rebounded a bit from that $24.45 price. But my math is the $607.9 million notional divided by the $24.45 price used in the presentation, which makes the numbers described in that quote make sense.
There's a delightful excerpt from a January 2008 presentation by an LIA official to its board:
"We consulted with a number of financial analysts specializing in equity markets, who confirmed that the current decline in shares witnessed by markets was not founded on factual bases, and that it was only temporary and caused by illogical panic that adversely affected the value of shares. They further noted that the value of shares will increase again due to the new economic conditions in USA, remaining industrial states, and developing states, which are not expected to suffer recession in the foreseeable future."
Oh, random fun sentence about him:
He has also been described by his former colleague as being more concerned about his remuneration package and his annual bonus than most other bankers.
That's saying something!
The judge summarizes the contention: "They allege that Goldman Sachs crossed the boundary of the usual counterparty relationship between a bank and its client and became in effect the LIA’s adviser or ‘man of affairs’ to adopt a rather outmoded expression used in the case law."
"Man of affairs" would be a fun title, if it didn't come with extra legal duties.
Besides the entertainment expenses, Goldman also gave the brother of Mustafa Zarti (the LIA deputy chairman) an internship. The judge does not think that this created any undue influence, but it is potentially awkward for Goldman as a matter of, say, U.S. law.
The judge also writes:
Many banks and financial institutions were eager to do business with the LIA and many succeeded in obtaining business worth millions of dollars. The fact that Goldman Sachs were prepared to go an extra mile when competing for this business by installing Mr Kabbaj in Tripoli to help the LIA does not mean that they were in a different relationship from the relationship that existed between the LIA and its other counterparties, even when looked at together with the other factors. The Goldman Sachs internal commentary on the relationship is striking. But it must be seen against a background where they could see a hugely lucrative and long term line of business available from the LIA and realised that they needed to fight their way to the front of the queue of other institutions lining up to provide the LIA with services. One must also bear in mind that the customer-facing Goldman Sachs personnel knew that their individual remuneration might be influenced not only by the value of business achieved but by the bank’s perception of their individual role in winning that business for Goldman Sachs. Having considered all these factors, I am not satisfied that the relationship that developed between the two parties crossed the line from being a strong, cordial business relationship between a buyer and a seller of financial services to being the kind of relationship of trust and confidence giving rise to a duty of candour and fairness on the part of the bank to its client.
There's a sad paragraph about the junior team's deep and various misunderstandings:
I accept as truthful their evidence that they remained under a misapprehension until Ms McDougall arrived that the Disputed Trades somehow would result in the LIA acquiring some shares in the underlying companies. They did not all believe the same thing about the nature of the trades and their thoughts about how the trades would play out were confused. They seemed to differ as to whether shares would be acquired at the start of the trade or only on the maturity date; how many shares would be bought (was it an amount represented by the premium or the total notional amount or some other number); how would the purchase of the shares be paid for (given that they knew that the premium was expected to cover the cost of the put option and the risk that Goldman Sachs undertook that the share price might rise). They were also very unclear about how the vast amount of money borrowed by the LIA from Goldman Sachs to finance the trades would be repaid if the share price dropped. I am satisfied that this inconsistency in the evidence of the LIA witnesses was not the result of their fabricating this evidence consciously or unconsciously but genuinely reflected their confused state of thinking at the time. I find that at least some members of the Equity Team had in their minds throughout the relevant period the idea that some shares would be acquired under the trades though by what mechanism was unclear.
"The evidence from Mr Vella and supported by the contemporaneous documents was that Goldman Sachs would not have taken on the credit risk of lending money to the LIA even with the shares posted as collateral for the loan because of the problems it would anticipate in enforcing such a loan against a sovereign entity in Libya."
The LIA eventually did that, for this litigation, but that is much too late!
The judge sort of concludes that there was an "overall mark up of 2.5% across the Disputed Trades," but that is the markup taken by the sales team; the trading desk had probably already built some profit into the walk-away price it showed to sales. (Probably; I am to some extent guessing here.) Overall the trades seem juicy but not crazy. If the LIA had shopped the trades around and tried to generate some competition, it could probably have paid quite a bit less for them. But it still would have lost everything it paid!
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 email@example.com
To contact the editor responsible for this story:
James Greiff at firstname.lastname@example.org