One of the first things you notice about Barclays Global Investors (BGI) is the magic markers. Brightly colored four-packs seem to be affixed to every glass wall and whiteboard at the renowned quant shop. They are a testament to the penchant of executives for scribbling out complex financial formulas when strategizing.
This devotion to mathematics doesn't stop at the trading room door. Buyside shops everywhere are charged with obtaining best execution, but at BGI transaction costs are forecast even before the traders arrive at work. And, at the end of the day, traders are judged by how close their executions matched their benchmarks.
Such strict controls are necessary when you're managing index funds. With about $300 billion in equity assets under management pegged to over 200 global stock indices, San Francisco's BGI is the world's largest index manager. It was also the first, having invented the index fund in 1971 when the unit of the giant U.K. bank was known as Wells Fargo. Some of its rivals in this penny-pinching corner of the money management world are State Street Global Advisors (SSGA) and Deutsche Asset Management.
BGI has been part of Barclays – a very tiny part – since 1995 when the bank bought Wells Fargo Nikko Investment Advisors. Although its $800 billion in assets under management place it in the top five money managers worldwide, BGI's profits account for less than one percent of Barclays' total. Chalk that up to the highly competitive, low-fee, commoditized business of indexing.
Higher Fees
Low fees are why transaction cost control is the mantra at BGI. Although a growing portion of BGI's assets under management now fall under the active' category and command higher fees, most investments are passive' and do not involve stock picking. BGI's performance is based solely on how close its portfolios track the indices. And close tracking means minimizing trading costs.
That puts the pressure squarely on BGI's traders. Because the value of an index component is based on its closing price, traders must buy the stock at a price better or equal to its close. On top of that, most trades at BGI are for baskets of hundreds or thousands of shares.
Traders also must keep a tight rein on the size of the commissions and fees paid to brokers. All such explicit costs are passed back to BGI's clients, mostly government and corporate sponsors of pension plans.
"Our actual trading costs are formulated and estimated daily for every single security we want to trade the next day," said Minder Cheng, managing director and global head of equities and currencies trading at BGI. "Our traders are expected to keep those costs in mind. I don't believe more than a handful [of buyside shops] do something like this."
Cheng has run trading for BGI since 1999. The 38-year-old manager, who earned a doctorate from the University of California at Berkeley, previously worked as a proprietary trader and strategist at Convergence Asset Management, Sumitomo Finance International and Salomon Smith Barney.
BGI has 27 traders, including Cheng, working out of seven offices worldwide. Eleven are based in San Francisco; seven in London; five in Tokyo; and two apiece in Australia and Canada. Many have sellside experience. Former domestic head Will Geyer came from Salomon. After returning there to run its program desk this spring he was replaced by an ex-Morgan Stanley trader, Mike Sobel.
That the crew in BGI's idyllic hometown by the bay is able to toil on a trading desk at all seems a miracle. San Francisco's natural beauty and its myriad distractions cry out for sick days.' But work hard they do because work hard they must. For traders at BGI, it all boils down to T-cost,' the distillation of BGI's scientific approach to trading.
T-cost, or trading cost, is the benchmark BGI traders must meet or beat when buying or selling a security. An amalgam of four components, the T-cost is expressed in terms of the number of basis points above or below the previous night's close.
Explicit Costs
Two explicit costs – commissions and bid-ask spreads – and two implicit costs – market impact and opportunity costs-are factored in. Market impact cost measures the extra a buyer pays or a seller gives up when the price of a stock moves adversely as a result of the trader's large order. Opportunity cost measures the same amount when the price move derives from trader inaction.
Cheng says both implicit costs must be taken into account. "We want to make sure the trader is not holding off on trades because they are too difficult – because they will hurt his performance," he said. "Because if he doesn't do it, it's a lost opportunity."
Such trading by the numbers may be stressful, but it is also lucrative. Trader compensation, according to Cheng, is more competitive than the average buyside shop. Pay packets aren't as high as those at the larger broker dealers, but they can be better than those at many smaller brokerages.
David Rothenberg, who heads up BGI's transition management group and is, in effect, a client' of the trading desk suggests BGI traders have to be good because they have little room to maneuver. "Trading in support of index management is a cost minimization exercise," he explained. "These are trades that need to be done. You have little flexibility. It is a challenge to get them done in the market for the lowest cost. That's the exercise run on the desk."
BGI doesn't just hold its own traders to the fire. It judges its broker dealer relationships in the same manner. Every day, Cheng checks a spreadsheet listing every trade done with every dealer. Columns itemize commissions paid; estimated bid-ask spreads; and actual total trading costs. If the actual T-cost diverges from the BGI estimate, the dealer gets a black mark. On the other hand, if the dealer executes the trade for 100 basis points (one percent of the previous day's closing price) when BGI expected to pay 120 b.p., the broker gets a gold star.
Reward Performance
If a dealer consistently disappoints BGI, it will find itself receiving fewer orders, says Cheng. Brokers that perform well, get more. Some dealers are now building similar cost-analysis systems, the executive adds. "When a broker comes in to visit us, we show him this [printout]," he said. "That's one reason they are going back and doing their homework. But it's also because they need to know how their traders are performing."
BGI would rather not use human brokers at all. "We have always been heavy users of ECNs," Cheng said. "Even before they were called ECNs." BGI, then Wells Fargo, was one of the original six buyside shops to trade in the Instinet cross in 1986. BGI was also an early user of ITG's POSIT.
Cheng cites three reasons why BGI loves ECNs: anonymity, low cost, and speed. The firm is connected directly to most, but not all, of the top six ECNs – Instinet, Island, REDIBook, Archipelago, BRUT, and Bloomberg Tradebook.
Volume sent to ECNs varies from five percent of total outflow on a light day to 50 percent on a particularly heavy day such as one might find in June during the annual reconstitution of the Russell indices.
Cheng is reluctant to discuss volume in absolute terms. As one of the largest stock traders in America, BGI does not want to tip its hand. "On extremely heavy days…our flow can be extremely heavy," Cheng joked. When pressed, he will only admit to trading "hundreds of millions of shares" on a typical day.
Those are shares leaving BGI and passing into the hands of dealers, electronic or otherwise. The figure does not include the tremendous amount of stock crossed in-house. Because BGI manages so much money for so many clients – approximately 2,000 – there is good chance that when Pension Plan A wants to sell, Pension Plan B will want to buy.
Rather than take both trades to the open market, a computerized crossing system dubbed FOX simply matches both orders at the next day's closing price. Although that strategy nixes any possibility of price improvement, it also eliminates any price dis-improvement due to market impact. But, most importantly, crossing is free or extremely cheap.
Often, no broker is involved. In that case, no commission is paid. If, for settlement and custody reasons, a broker is involved, the fee is nominal, according to Cheng.
Again, the executive is reluctant to divulge volume figures related to the daily cross. He will say that when both BGI's crossing and open market trades are totaled they range from one billion to three billion shares per month. (Or between 50 million and 150 million shares per day.)
Astounding Result
In a recent six-month period, FOX found matches for 15 percent to 30 percent of all orders, according to Cheng. This cross ratio can be as low as two percent or as high as 70 percent or 80 percent. At the high end, such a figure is astounding when one considers the success ratio of POSIT, the largest commercial crossing network. POSIT matches no more than five percent of its customers' orders. The reason for FOX's success goes beyond assets under management or number of customers.
Because the firm manages so many different index funds and because stocks frequently move between indexes, there is a good chance that one fund manager will have to buy a stock on the move and another will have to sell it.
Suppose, for example, Standard & Poors moves company ABC from its S&P Mid-Cap Index to its S&P 500 Index. The BGI manager running the Mid-Cap fund will have to sell ABC while the manager of the S&P 500 fund will have to buy ABC. The two orders can simply be crossed.
A common problem with crossing systems is that fund managers are often moving in the same direction at the same time. Most want to sell or most want to buy.
"Crossing is a big cost saver for us and our clients," Cheng said. "The bigger you are and the more benchmarks you manage, the higher the crossing ratio you can achieve."
Given those numbers, why hasn't BGI taken FOX to the Street? Why not compete with POSIT for order flow? After all, rival SSGA's brokerage affiliate markets a crossing system called Lattice. It's not as large as POSIT, but it's popular with money managers hoping to interact with SSGA-derived order flow.
Street gossip holds that BGI did once plan to commercialize FOX. To that end, it bought order management system vendor Longview in 1998 in order to offer its Landmark product as a front end. That it sold Longview after one year suggests BGI scuttled that plan. Other buyside shops were not comfortable trading on a competitor's system, according to reports. Cheng would not discuss the possibility of commercializing FOX.
Transition Business
FOX is being leveraged in support of another developing business, though-transition management. When a plan sponsor decides to move its funds from one money manager to another there is a brief time period in between when that money needs to reside in an interim portfolio. Neither the legacy' manager nor the target' manager is apparently able to fulfill the investment needs of the sponsor during the switch. And selling all of the stocks in the legacy fund and buying all of those of the target is needlessly expensive.
A transition manager such as BGI – or its competitors SSGA and Frank Russell Securities – will transition the plan sponsor out of its legacy portfolio and into an interim portfolio identical to the target portfolio for a few days or a week. The pension plan benefits from better investment performance and lower costs.
BGI, which has been transitioning clients to its own portfolios for years, is now pushing the service to those plan sponsors, transferring their assets to money managers other than BGI. Key selling points include a 33-person transitions department and savings through FOX and a cost-obsessed trading desk.
According to BGI, up to one-third of both a plan's legacy and target portfolios will, on average, find a match in FOX. Another third of the legacy portfolio overlaps with the new portfolio. The final third goes to the desk.
David Rothenberg, a former senior executive with agency broker ITG, was hired in 1999 to head up the transitions initiative. "Until just two years ago BGI was a very inward-looking firm," he explained. "We looked to service our portfolios and our clients, but we were not out marketing in some of those areas where we had created specialties. Now we've taken the transitions group and are offering it out."
Interestingly, all trades for the group are handled by BGI's affiliated brokerage subsidiary, Barclays Global Investors Services (BGIS). The electronic agency broker has been around for several years, but was registered with the NASD only in March 2000. Through Oct. 11 of this year, BGIS traded 650 million shares. (Cheng says between 10 and 20 percent of flow at the trading desk comes from transition management.)
BGIS' chief mandate is to trade securities involved in a transition, but it is also handling nearly five percent of other BGI trades. Department of Labor ERISA rules prohibit more.
Will it morph into full-fledged brokerage along the lines of arch-rival SSGA's captive State Street Brokerage? "We've seen an absolute acceptance of the value proposition of the broker dealer," said Rothenberg, who is also president of BGIS. "Whether or not we offer it out to other money managers is part of our strategic refresh question. The possibility is not something that is lost on us."
Rothenberg's worry is whether or not BGI's buyside competitors would even consider trading through its brokerage.
For the trading desk, the big difference between transitions and normal' trades is the time it takes to complete them. Most trades are executed within a few days, but a transition can take weeks or one or two months. A single trader is usually dedicated to the task.
The biggest headache, according to Cheng, is forecasting T-cost and tracking error over such long periods of time. "It's easier to see two steps ahead rather than 100 steps," he noted.
That's why BGI provides the magic markers.
