The U.S. interest rates complex, centered around trading U.S. Government debt, used to be the province of primary dealers (mostly large banks). There were 46 primary dealers in 1988, at the peak of their influence. Nearly every single U.S. Treasury transaction involved a primary dealer on one or both sides of the trade.
“We were kings,” a trader from Salomon Brothers, the legendary bond trading firm, told me as he was recalling scenes that could have been straight out of Liar’s Poker.
Three decades later, we are now down to 23 primary dealers. Market share is concentrated in the hands of five of the largest banks, which are responsible for over 60% of customer orders. High-frequency trading (“HFT”) firms provide all the liquidity in the interdealer markets. Treasury futures, which have a broader base of participants, have volumes that rival the cash Treasuries market. Fixed income revenue, a historical driver of bank profitability, never recovered from the mortgage crisis. By the time I started trading U.S. rates, there were no more BSDs left on the trading floor.
Two weeks ago, CME announced the purchase of NEX for $5.5 billion. CME is the world’s largest futures exchange and has a U.S. Treasury futures business that trades over $300 billion notional per day. BrokerTec, a NEX subsidiary, is the largest interdealer platform with over 80% market share. The BrokerTec acquisition gives CME a near monopoly of the U.S. interest rates complex.
For banks, whose influence has been waning for years, the CME-NEX merger marks the end of an era. Or, as Risk.NET put it last week, “… some fear CME’s ownership of BrokerTec could be the final nail in the coffin of bank influence over the largest interdealer liquidity pool in the US Treasuries market.”
How did we go from “Masters of the Universe” to whatever role it is banks play today? In one word, technology. Electronification democratized access to the market and reduced the influence of gatekeepers on market structure. Along the way, dealers made a number of moves that maximized their control of the market in the short-term but caused them to lose greater control over the long-term. In today’s Trading Places, we trace the twenty-year history of the dealers’ waning influence on the U.S. Government debt market.
In the mid-1990s the U.S. Treasury market was entirely phone-based. Customers negotiated trades with dealers telephonically. Dealers traded with each other through brokers. Market data was available on “screens,” but none of it was actionable in the way we think of modern trading platforms.
Meanwhile, progress was rapidly changing the U.S. equities market. Small trades began to be executed automatically by Nasdaq market makers. Modern exchanges, like Island and Arca, were springing up. Small proprietary trading firms could participate on an even footing with the largest of banks. Online brokers like eTrade opened up instant execution to the retail market.
Cantor Fitzgerald had a 90% market share in the interdealer voice market and was not blind to the threat that an equity-like exchange could pose to its franchise. To get ahead of the curve, Cantor launched eSpeed, a fully electronic order book for U.S. Treasuries in 1999. Joe Noviello, who was featured in Trading Places last week, was the Chief Technology Officer. The platform replaced human voice brokers with a piece of software that allowed traders to place orders directly.
More importantly, eSpeed democratized access to the interdealer market. A small group of voice brokers at Cantor, and by extension, their customers who were the largest of the primary dealers, no longer had a monopoly on information. Now, everyone had access to the same markets. Even small broker-dealers (not even primary dealers) could participate on a level playing field with the big boys.
With the launch of eSpeed, the old boys club that the primary dealers and voice brokers had built began rapidly changing. Suddenly, a piece of software was providing direct access to a market to which the primary dealers once were the sole gatekeepers. Banks were justifiably afraid that eSpeed might one day let buy-side participants participate, effectively disintermediating them from their customers.
Fourteen banks, sensing this danger, banded together to form BrokerTec, a direct competitor to eSpeed. Dealers waged a full-scale war to wrest control of the market away from Cantor. “We were told to leave our orders on BrokerTec even if there was more liquidity available on eSpeed,” a trader told me of the time. One of the original BrokerTec salesmen recalled how he used to “take over traders’ keyboards and mice to place orders for them.”
BrokerTec was slow to gain traction. Then 9/11 happened, and Cantor lost 68% of its workforce. eSpeed was up and running quickly and arguably saved the company, but BrokerTec was able to use the loss of salesmen and support staff to take market share. BrokerTec also entered into agreements with its dealer-owners to send a minimum level of volume to the venue. Eventually, BrokerTec’s anti-competitive practices caught up with them and attracted the attention of the Bush Justice Department, which forced a sale to ICAP in May 2003. The dealers had lost direct control of their cudgel for determining the direction of the interdealer market.
eSpeed and the newly-independent BrokerTec were locked into a vicious war for volume, with market share seesawing back and forth, but never moving much beyond a 55/45 split. In late-2003/early-2004, both venues quietly allowed two high-frequency trading (“HFT”) firms, Getco (now a part of Virtu) and Citadel, onto the platform to supercharge liquidity in the hopes of gaining an edge in market share
The old saying goes, liquidity begets liquidity, and the HFT firms delivered liquidity in spades. While dealers were still keying in orders by hand, Getco and Citadel relied on fully automated algorithmic trading systems to place orders. The effect was immediate. Spreads tightened, liquidity increased, and volumes on the platforms soared.
A decade later, HFT firms make up 65% of the volume in the interdealer market. When Risk.NET leaked the top BrokerTec participants in 2015, dealers were shocked to learn that HFT firms took 8 of the top 10 spots. Today, it is 9 of the top 10.
The dealers, which started BrokerTec to regain control of the interdealer market, had found themselves out of the driver’s seat only five years later. However, as the Treasury collusion lawsuit details, dealers were effective in bullying the interdealer brokers to prevent buy-side firms from accessing the market.
Nasdaq intends to create a dark pool for Treasuries within the eSpeed platform, CEO Robert Greifeld said at a financial services conference held by Goldman Sachs. … The Treasuries dark pool offers Nasdaq the opportunity to attract more customers and more trades to its platform, thus increasing revenues.
At eSpeed, leadership was shocked by Greifeld’s comments. They knew this would not play well with dealers. Dealers reacted in force, pulling much of their order flow from eSpeed. According to the Treasury collusion lawsuit, Morgan Stanley did not trade on eSpeed for two years. RBS never traded significantly on the platform again. By the close of the merger, eSpeed’s market share was down to 35%. eSpeed would steadily lose market share from there. Today, eSpeed has an 11% market share according to Greenwich Associates. Nasdaq officials disagree with Greenwich Associates’ findings and claim their market share is 19%. It is difficult to independently verify either number due to the opaqueness of the U.S. Treasury market.
eSpeed learned its lesson. When PIMCO asked for direct access, Nasdaq executives pulled the plug on the initiative according to the collusion lawsuit. Nasdaq is convinced that superior technology will turn eSpeed’s fortunes around according to a source familiar with one Nasdaq executive’s thinking.
Dealers had the power to destroy eSpeed, but in doing so they concentrated too much market share in BrokerTec. Without a viable alternative to BrokerTec, dealers gave up their only point of leverage. Launching a new interdealer broker, without the volume agreements that supported BrokerTec or 9/11, which disrupted eSpeed’s operations, was an impossible mountain to climb.