Banks Wanted to Keep the CDS Market to Themselves

The job of a dealer in financial markets is to buy low and sell high. Of course that's the job of an investor too. That's the obvious, only job. What distinguishes dealers from investors is, loosely speaking, that dealers are not making their buying-and-selling decisions based on some analysis of the value of what they're buying and selling, but rather based on information about who else wants to buy and sell, and at what price. Their job is to know their customers, not their securities. And the information they have about potential customers is pretty much the value that dealers add. It's what they get paid for: Sellers come to dealers to find buyers, and buyers come to find sellers. So the dealers try to make sure that they have information that other people -- like the customers -- don't have.

This explains like half of the scandals in modern finance. We talked earlier this week about the former Nomura mortgage bond traders who were accused of going to baroque and possibly illegal extremes to hide information from customers. In the mortgage-backed securities market, trades were not publicly reported, and the only way for customers to find out the price of a bond was to ask a dealer. The Nomura traders were accused of lying to the customers about the trades they were doing, to obscure the market price from the customers. In other markets, trades are publicly reported, but dealers can still make nefarious use of customer order information. That's sort of what the foreign exchange fixing scandal looks like: Customers gave orders to dealers, and dealers used those orders to manipulate prices in ways that the customers did not expect. In the equity market, both trade and order information are public: Anyone can see how many buyers and sellers there are for a stock, and at what price. But electronic market makers race to see it faster, so they can make a fraction of a penny on each trade, and the exact mechanics of how they do that are subject to endless controversy. Meanwhile, spoofersare trying to deceive market makers about the supply and demand for stocks, so they can profit from the market makers' incorrect information.

Today, a lawyer told a judge that "Some of Wall Street’s biggest financial institutions -- including Goldman Sachs Group Inc., JPMorgan Chase & Co., Citigroup Inc. and HSBC Holdings Plc -- have agreed to a $1.87 billion settlement to resolve allegations they conspired to limit competition in the lucrative credit-default swaps market," and that is a pretty big number! The other scandals -- mortgage bonds and FX and high-frequency trading and the rest -- are mostly about dealers exploiting the structure of their markets to hide information from customers and make some extra money at their expense. But this one is about the dealers conspiring to structure the market itself to hide information from customers, and to make, in the words of the CDS customer plaintiffs, "billions of dollars in supracompetitive profits" at their expense.  

The idea is that the CDS market, like a lot of markets, was one where dealers were pretty good at keeping information to themselves. Customers could trade only with dealers, and trades were not reported publicly in real time, so customers didn't know what the market price was unless dealers told them. So dealers had a lot of flexibility to charge whatever they could get away with. As CDS became more popular and liquid, you would expect this dealer advantage to dissipate. That's just sort of how markets evolve. When the first person invents a financial product and sells it to the first customer, the dealer owns pretty much all of the market information: The dealer is the market. Eventually other dealers start trading the product, and the customers get a bit more leverage: They can check prices with multiple dealers and take the best one. As the product gets more liquid and standardized, customers start to expect more transparency. If a thing trades once a month, the price is whatever the dealer says it is, but if it trades once a minute, you can meaningfully ask what the "market price" is, and expect to pay that price. So customers start to expect -- and/or regulators start to demand -- transaction reporting, so that customers can just look at the market price rather than relying on the dealer to tell them. And if it trades once a second, then calling up a dealer to trade will be too slow, and some sort of electronic trading system will have to develop. Electronic trading tends to cut out middlemen: If the customers are all on the same electronic trading system, they can just trade directly with each other, without going through a dealer.

The accusation in the CDS case is that dealers didn't like this erosion of their informational advantage, so they put a stop to it. They took care to keep information to themselves:

To protect their positions of prominence, Dealer-Defendants restricted pre- and post-transaction price transparency. Before a transaction, Dealer-Defendants strove to keep investors in the dark about both the volume of supply and demand and the real price at which CDS were trading. For instance, investors could not see Dealer-Defendants’ solicitations of bids and asks.

And after a CDS contract traded, the trade information went to Markit, another defendant in the case. But:

In exchange for receiving this data, Markit agreed to DealerDefendants’ condition that Markit not provide pricing information to its subscribers in real-time. Instead, Markit would delay before circulating information, allowing Dealer-Defendants to quote different prices in the interim and to disavow as stale the information that Markit eventually released.


With increased standardization, the market was ripe for alternative means of CDS trading, such as an electronic exchange. Such alternative means would have diminished the buyside’s dependence on the over-the-counter trading services offered by Dealer-Defendants.

And so Citadel and CME Group got together to set up the Credit Market Derivatives Exchange to provide electronic trading of CDS through a central limit order book where anyone could post orders to buy and sell CDS and trade with each other directly, and where trades would be reported in real time. "CMDX would thus have excluded Dealer-Defendants as intermediaries in many CDS transactions and made real-time pricing information available to investors."

But, according to the plaintiffs, the dealers got together to shut this down. They refused to invest in CMDX when asked, refused to participate in the exchange, cleared transactions through a different clearinghouse from CMDX, and convinced Markit and ISDA not to license price data or CDS agreements to CMDX, making it impossible for CMDX to actually trade CDS. And they did this all in classically conspiratorial fashion:

They reached their agreement at secret meetings and through telephone and email communication. Some of their gatherings took place in midtown Manhattan on the third Wednesday of the month during the Fall of 2008 and were masked as board or committee meetings -- some of them for Markit and ISDA.

One thing to say about this is that it is not hard to understand where the dealers were coming from. CMDX wasn't just trying to compete with them. It was trying to use their own information to compete with them: It wanted to license their documents and use their trade data to build a product to replace them. The dealers had built up all that informational advantage, which was their livelihood, and CMDX wanted to put them out of business. So they said no.

Roughly speaking, it would be fine for them to say no, individually: If CMDX had gone to Goldman Sachs or whoever and asked it to report its trades in real time  data to CMDX, Goldman could have said no. Every dealer could have said no. The problem isn't that they said no; it's that they -- allegedly -- coordinated to say no. That's an antitrust problem. But it's also a result of the need for market coordination. We talked the other day about how financial markets need some sort of coordination among participants so that everyone can agree on who owns what and what traded where and how transactions work. Maybe that's through the blockchain, or maybe it's through information providers such as Markit and trade organizations such as ISDA. The banks need to get together and figure out how trades will be processed and how information will be shared; they need to coordinate with each other to make the market function properly. The theory of this case is that they spent some of their time at those meetings plotting to keep the market from functioning better. 

Another thing to say about it is that the plaintiffs have pretty high counterfactual expectations for how much better the market could have functioned. Mmmmmaybe CMDX would have led to cheap electronic trading of CDS where dealers couldn't overcharge clients. But there are reasons to doubt that. People keep trying to launch cheap electronic trading of corporate bonds, but the corporate bonds keep stubbornly resisting. Perhaps bond dealers are conspiring to prevent that too, but more likely the corporate bond market is still pretty illiquid. (As people keep worrying.) Corporate bonds don't trade every minute, so investors who want to sell their bonds can't easily find other investors who want to buy, and have to rely on dealers to intermediate across time. CDS has similar problems. And dealers' informational advantages aren't worth that much these days anyway: "After 2008, the popularity of single-name CDSsdropped dramatically and new regulation effectively reduced the ability of big banks to deal in the products." The market is a fraction of the size that it was before the crisis, and banks are getting out of the business. It is hard to see where the billions of dollars of supracompetitive profits came from. Really, you'd think that the dealers would want to move to electronic trading if it was viable: It might compress their margins, but a thin margin in a functioning market is better than a fat margin on no trades.

So if you'd asked me yesterday, I would not have put all that high a dollar value on this case. (The Justice Department looked at it and passed, and the European Union launched an antitrust investigation a while back that remains open.) But here we are. The lawyer who announced the $1.87 billion settlement, Daniel Brockett of Quinn Emmanuel, said:

“In a case where neither the DOJ or the EC was able to make any charges stick against the banks in their investigations, we came in and took over the case and they will pay almost $1.9 billion,” said Brockett.

“All of the discovery is under a protective order, so I can’t tell you what was in the e-mails or what was in the documents or what they testified to,” he said. “We felt we had developed, brick by brick, a very powerful case on the liability.”

So one lesson is the usual one: Don't put it in e-mail. But another lesson might be that customers and regulators and courts are just less tolerant than they used to be of dealers' desires to keep information to themselves. It used to be that everyone grudgingly conceded that dealers should be able to profit from their information. But in a new age of transparency -- and of banking scandals -- that's a less popular view, as banks keep unhappily learning.

  1. Speaking extremely loosely! Obviously, a good dealer-side bond trader knows stuff about bonds, and lots of investors look at flows and things. This is a rough conceptual division, not like a strict either/or. 

  2. Of course, as a dealer, Nomura was allowed to ask whatever price it wanted. The problems came when the traders held themselves out as just brokering trades between customers, but (allegedly) lied to one customer about the price the other one wanted. That's arguably material information -- the customer wants to know where the market for the bond is -- and lying about it is bad.  

  3. And also in ways that customers did expect, and do, and that are fine. 

  4. That's from the January 2014 amended class action complaint (in two parts hereand here). Also useful reading are the March 2014 motion to dismiss the complaint, and Judge Denise Cote's September 2014 opinion more or less denying that motion. 

  5. I'm using the past tense here because this case runs through 2013. There is much discussion about increasing transparency and centralizing trading of CDS, but this description isn't particularly wrong for the current market either:

    New rules stemming from Dodd-Frank force most index swaps to be traded openly on regulated platforms and processed by clearinghouses that take fees to guarantee the trades. The Securities and Exchange Commission hasn’t yet mandated that single-name credit swaps be subject to those trading and clearing requirements, making them less liquid. 

  6. Though middlemen are a persistent lot, and the experience of electronic market makers in equity and Treasury markets suggests that electronification doesn't kill them off. 

  7. I take this story more or less from the recitation of the facts on pages 7-18 of Judge Cote's opinion. She in turn took the facts from the complaint, so they're not necessarily facts, just accusations, but they "are accepted as true for purposes of this motion." 

  8. From the dealers' motion to dismiss (citations omitted): 

  9. See, e.g., Bondcube, or BlackRock's never-ending proselytization. Disclosure: Bloomberg LP, the parent of this news organization, also has a platform for trading corporate bonds. 

  10. Or maybe worse:

    One dynamic easily lost in the euphoric headlines ushering in a new era of transparency for OTC derivatives is that certain types of derivatives are, and will likely continue to be, illiquid. The US corporate bond marketplace outpaces the global volume of daily trading in single-name CDS by five times. The most actively traded CDS names traded only 10 times per day on average. 

    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

    To contact the editor responsible for this story:
    Tobin Harshaw at

Download File