A few years ago, Goldman Sachs Group’s leaders took a hard look at how the bank carries out initial public offerings. They mapped 127 steps in every deal, then set out to see how many could be done by computers instead of people.
The answer so far: about half.
Just 21 months after the firm disclosed its plan to re-engineer one of Wall Street’s most lucrative businesses, the project has found ways to eliminate thousands of hours of work long performed by humans. A computer-based interface called Deal Link has replaced informal checklists that were once tended and passed down between generations of rainmakers. It now arranges and tracks legal and compliance reviews, fills in forms and generates reports.
The initiative’s progress — described by senior Goldman Sachs executives in recent interviews — shows how quickly big investment banks may be able to automate tasks once beyond the reach of computers. The industry is under intense pressure to improve profitability, while contending with young workers less willing to put in 18-hour days. At Goldman Sachs, managers say they’re looking to new technology to free up junior bankers in particular, letting them focus on more satisfying work. That could help slow an exodus of talent to private equity firms, tech titans like Google and hot startups.
The review is the brainchild of Goldman Sachs banker George Lee. Long an adviser and confidant to the Silicon Valley elite, Lee became chief information officer for the investment banking division three years ago. Since then, he’s focused on two key goals: digitizing bankers’ workflow and using technology to enhance their advice to clients.
The firm has expressed interest in automating parts of the IPO process for years — Martin Chavez mentioned it publicly as early as September 2015, when speaking as chief information officer — but it has never before detailed the efforts publicly in such depth.
In the beginning, Lee and his team took aim at the most obvious steps for disruption: the routine phone calls, emails and tasks that young bankers plow through at the beginning of every IPO. That included phoning the compliance department to look for potential conflicts, contacting legal to assign lawyers and compiling an organizational book for meetings.
Now that’s all done with the click of a mouse in the new application, which features a step-by-step guide replacing ad hoc training materials and word-of-mouth advice. Hot links trigger processes or fill in forms. Information used across multiple forms is populated automatically.
The firm also streamlined the way that it delivers updates to clients about IPO orders, sending instantaneous details about pricing, size and timing to a phone or tablet application, Lee said in an interview. In the past, that would have been a static report that Goldman Sachs emailed or faxed to the client’s hotel at the end of the day.
The bank soon concluded the focus on IPOs was too narrow, and it turned attention to other deals, such as corporate mergers and bond sales. To date, more than 150 steps have been mapped across various investment-banking transactions handled by the 2,500 bankers in the division. Hundreds of hours initially saved on IPOs swelled into the thousands as the project grew.
“What we are trying to do is slowly but surely pick off the ones that are the most redundant, the most repetitive, the most labor-intensive, and automate them so that you save time,” Lee said.
Goldman Sachs has a history of innovating to earn more from IPOs. In the 1940s and ’50s, senior partner Sidney Weinberg elevated relationship banking to a new level, eventually working personally with the Ford family for years to take their carmaker public, running what was then the biggest-ever IPO. That built the bank’s franchise. In 1984, when Eric Dobkin was asked to improve its ninth-place ranking in stock underwriting, he pushed to sell large blocks of new shares to institutional investors. The strategy replaced a longstanding model relying on thousands of regional brokers to peddle stock to retail investors. The modern IPO market was born.
Goldman Sachs ranks No. 6 this year among managers of global IPOs, its worst position in the league tables since 2012, according to data compiled by Bloomberg. For banks, it’s a particularly desirable business, with fees averaging about 7% on midsize transactions in the U.S. in recent years, according to research by Jay Ritter, a professor at University of Florida’s business school. The average fee earned for selling U.S. investment-grade bonds last year was half a percent.
Goldman’s latest effort has mainly helped to erase grunt work for analysts and associates, the bottom rungs. In the lofty realms of Wall Street, they’re relatively “cheap,” said Brad Hintz, a former CFO at Lehman Brothers and top-ranked research analyst.
(Associates working in equity capital markets at top Wall Street banks typically earned about $326,000 last year, according to a survey by recruiter Options Group. But that compares with $494,000 for vice presidents and $860,000 for directors.)
Terminating a significant number of bankers from the lower tiers to save money could create new problems in the long run, Hintz said, because banks replenish their senior ranks by promoting from that pool. It would risk “ossification,” he said.
Instead, the new system lets those employees pursue more productive work, Lee said. They can help draft filings used to communicate with investors and register with the Securities and Exchange Commission, shape marketing strategy or spend more time talking with clients. There’s been no impact on headcount or hiring, according to the firm.
Goldman Sachs’s approach may seem obvious, but it’s cutting-edge for Wall Street. The conventional wisdom has long been that investment banking was too reliant on human-to-human interactions, and that trading was easier to automate. Banks also lacked investment dollars for new projects as they battled fines, investigations and revenue pressures after the financial crisis.
That’s beginning to change. JPMorgan Chase has used machine learning to predict when clients might need to raise capital through a secondary equity offering. And startups abound. Kognetics, for example, uses a similar system to find and catalog data to identify attractive acquisition candidates in the tech industry.
“There is an ecosystem pushing toward more automation,” said Matthew Dixon, an assistant professor at the Illinois Institute of Technology’s business school.
As a result, Wall Street’s online chat rooms are rife with young workers worrying about their future — and rebuttals from optimists who predict relationship banking will largely be immune to automation.
Lee insists that, at least for the foreseeable future, there’s plenty of demand for uniquely human ideas. For anyone at Goldman Sachs who may have such a concern, he has a simple answer. “Our strategy is to elevate the activity and impact of bankers,” he said. “Not replace it.”