Wall Street’s credit-derivatives traders, who before the financial crisis commanded $2 million of annual pay, are being replaced by machines as banks cut costs and heed new regulations.
UBS AG, Switzerland’s biggest bank, fired its head of credit-default swaps index trading, David Gallers, last week, with no plan to fill the position, according to two people familiar with the matter. Instead, the bank replaced Gallers with computer algorithms that trade using mathematical models, said the people, who asked not to be identified because moves are private.
UBS joins Barclays Plc, Credit Suisse Group AG and Goldman Sachs Group Inc. in using computer programs to trade financial instruments that once generated some of their biggest fees. With regulators preparing rules under the 2010 Dodd-Frank financial reform that will push swaps toward exchange-like systems to improve transparency, credit dealers are going digital as automated trading makes humans too expensive.
“It’s natural to push away from humans and large size to machines and small size,” Peter Tchir, the founder of New York-based TF Market Advisors, said in a telephone interview. “It’s been gaining momentum.”
UBS’s algorithm, which can trade as much as $250 million of the Markit CDX North America Investment Grade index and $50 million on the speculative-grade benchmark in one transaction, was introduced last month, the people said.
Megan Stinson, a spokeswoman for Zurich-based UBS, declined to comment, as did Gallers.
Automated trading of swaps marks a shift in a market where transactions historically have been negotiated over the phone after dealers, acting as a go-between for clients, send out indicative prices by e-mail. The dealers offer to buy a swap from a client at one price and sell the same contract to another for a higher amount, profiting from the gap, known as the bid-offer spread.
Outstanding contracts ballooned to more than $62 trillion at the market’s peak in 2007 from $632 billion in 2001 as the derivatives gained popularity as a way to wager on debt without owning bonds or loans, data from the International Swaps and Derivatives Association show.
As late as 2005, managing directors on credit-derivative trading desks were being paid an average $250,000 in salaries and $1.75 million in bonuses, Michael Karp, co-founder of executive-search firm Options Group, said in a 2006 interview with Bloomberg News.
Building an algorithm may cost a few hundred thousand dollars, said Tchir, a former credit-derivatives trader.
Elsewhere in credit markets, Dow Chemical Co., the largest U.S. chemical company by sales, is planning its first benchmark bond issue this year. Volkswagen AG, Europe’s biggest carmaker, sold 2.5 billion euros ($3.2 billion) in bonds that will automatically convert to shares at maturity to boost liquidity following the purchases of Porsche and Ducati.
Market makers have slimmed down as regulators have ordered them to raise capital to prevent a repeat of the taxpayer-funded bailouts that followed the 2008 collapse of Lehman Brothers Holdings Inc. Banks will hold more reserves against riskier assets under the rules, known as Basel III. Swiss capital rules, applicable to UBS and Credit Suisse, are among the most stringent.
“I don’t think it’s driven by a desire for efficiency as much as a desire to control costs,” Bonnie Baha, head of global developed credit at Los Angeles-based DoubleLine Capital LP, which oversees more than $45 billion, said in a telephone interview. “The cost of a major trading error which could possibly be avoided by having a real human person sitting and thinking about things will far outweigh the personnel costs they save by firing all these guys.”
Credit Suisse’s program, which started in early 2011, is “a natural fit with our other strong electronic-trading businesses in rates, FX, and commodities,” said Jack Grone, a spokesman in New York for Switzerland’s second-biggest bank.
Michael DuVally, a spokesman for Goldman Sachs in New York, didn’t immediately comment.
Barclays’s algorithm was designed to handle smaller trade sizes and began in April 2011 with the capacity to handle transactions as large as $25 million on the investment-grade index and $5 million on the high-yield benchmark, according to Drew Mogavero, head of U.S. credit-swaps trading. Those sizes have since doubled, he said.
For smaller trades in which there’s less at stake, “we want to automate that process as much as possible and free up the sales people and traders,” Fred Orlan, head of global credit trading, said in a telephone interview. “We want to spend our time driving ideas and solutions to things that have a bigger impact on clients’ overall returns, so that’s really what we’re here for.”
The algorithm is designed to respond to liquidity in the market, so the bid-offer spread widens and tightens according to flows, Mogavero said. In liquid markets, trading odd lots through the algorithm typically cuts down that spread, he said.
“It’s not hard to envision an environment given the growth and popularity of algo trading of indices where the sizes continue to increase,” Mogavero said.
The programs so far are primarily used when markets have a balance of buyers and sellers and are driven by dealers to make markets or hedge their own books, according to Nancy Davis, director of derivatives in New York at AllianceBernstein LP.
Barclays shut off its algorithm in Europe in May, deciding conditions and market structure weren’t yet suitable to support it, according to two people familiar with the decision.
Dealers are “definitely fighting for market share,” Davis said in a telephone interview. “Once you get plugged, it just becomes operationally easy to trade, so that’s what the rush is to get all these algos out. It’s kind of a race to say who has the best plug-and-play-model right now to gain market share. I don’t think there’s a clear winner or loser at this point.”
Clearing and margin required by Dodd-Frank will also change the cost structure of trading, and algorithms may be one area where traders will be required to post less capital relative to other types of transactions, she said.
Algorithms may also get a boost if CDX futures get traction, Davis and Tchir said.
“You’ll see more people do it, and as these products become more easy for people to trade electronically there will be more participants” from firms such as Citadel LLC, Susquehanna International Group LLP, or Knight Capital Group Inc., Tchir said. “They’ll be able to add their algos to it once it becomes more mainstream.”
The increasing popularity of algorithms is an example of how credit markets are becoming more like stocks, Tchir said, citing so-called E-mini S&P 500 futures, with a contract value of $70,513 as of yesterday and trading volume of as much as $200 billion a day with no real market maker, he said.
Banks will probably be successful in block trading or in systems resembling so-called dark pools where large orders are traded without identifying the brokers and institutions that buy and sell, he said. There will be “fewer market makers, but those that remain will provide very large-size block trades,” he said.
Computer-driven transactions and high-frequency trading have come under increased scrutiny after the so-called flash crash in May 2010, when a 20-minute plunge in stock prices temporarily erased some $862 billion of market value. A report by the U.S. Securities and Exchange Commission and the Commodity Futures Trading Commission pinned the decline partly on an algorithm employed by one firm trading stock futures.
“They have a bad rap on the street as driving the ’87 crash and they’re not considered by Main Street as friendly vehicles, but at the same time, they are liquidity providers and that’s the biggest change with Dodd-Frank,” AllianceBernstein’s Davis said. “Having more algos in the market in these products will help because it will give market makers a way to have more liquidity so when you call a dealer up they’ll have another outlet.”