Sure, Goldman Sachs thinks it knows best or it wouldn’t have created its own digital system for analyzing financial risks it faces, aka its SecDB database and security pricing system.
But that is not what has protected it from making the kind of huge missteps in housing credit instruments that felled Bear Stearns and Lehman Brothers and are forcing Citigroup to break itself, to survive. (And made Bank of America want to break off its merger with Merrill Lynch).
“Almost everybody can get access to almost all the same models and there are very few advantages from proprietary models. So in this kind of game, everybody wants the fastest car,’’ said Robert Arvanitis, principal of Risk Finance Advisors on the ZDNet interactive discussion “Calculated Risk: Goldman Sachs’ Golden ‘Gut’ “ held Wednesday afternoon. “But in a race, you really bet the driver, not the car. You bet the rider, not the horse.’’
The drivers at Goldman include CEO Lloyd Blankfein, 10-year CFO David Viniar, who made the call to reduce exposure to housing payments in December 2006, Gerald Corrigan, a former head of the New York Federal Reserve Bank and Henry Paulson, Blankfein’s predecessor, who ran the U.S. Treasury Department during the height of the credit crisis in the fall – when even Goldman Sachs’ survival seemed at risk.
The company's use of SecDB and its thinkers' own analysis of market events was again evident this morning in this Bloomberg report, which noted that Morgan Stanley and Goldman Sachs share are gaining ground. And cautiously asserted “the worst of the writedowns may be behind them.”
Citigroup has taken $85.4 billion in writedowns, losses and credit provisions since the beginning of the financial crisis. That compares with $40.2 billion for Bank of America and JPMorgan’s $29.5 billion. By contrast, Morgan Stanley has written down $21.5 billion while Goldman Sachs has taken a $7.1 billion hit to earnings, according to Bloomberg data.
“You have to admit that in many, many, many ways, every firm is pretty much the same. They all use the same airplanes, they all use the same equipment, they all use the same services, they all wear the same clothes. In a whole bunch of different ways, they’re identical.,’’ said Charles D. Ellis, author of “The Partnership: The Making of Goldman Sachs” (Penguin, 2008). “So the visible differences are small, but, boy, are they decisive.”
He puts the decisive differences down as cultural, more than technological, discipline.
• Recruiting. Tougher, more rigorous, more disciplined hiring of exceptionally talented “natural leader type of personalities” than competitors, “by a substantial margin over time.” Looking at only the top 5 percent of recruits, by its measures – and then weeding out those that still don’t measure up in the first two or three years of employment.
• Ego-lessness. Even with all that talent, “no egos are allowed to show themselves at Goldman Sachs,’’ he said. “You better be comfortable inside your own skin. If you’ve got a need to be more important than the guy next to you, out you will go.’’
• Teamwork. You’re very likely to be the co-head of a committee or task. Get used to it. Make it work. And even so, at Goldman, you defer at all times to the person who knows the most about a subject, when making a decision.
• Communication. No hierarchy. No stopping. One-half hour of voice mail and email before you go to bed. One half-hour when you first wake up. Keep messaging on way to work (a driver helps). Expect to respond within one hour. Must respond within 24 hours.
• Speed. The more “nodes” in your network the faster you communicate.
Then there’s anticipation – and caution. And analysis of worst-case possibilities, before they happen.
“In ’87, when they had the BP (British Petroleum stock) offering and the market took a dive down 23% in a single day, every single dimension of that possible negativity had already been scoped out and put in a written report that went to that committee. The focus on what could go wrong and how it could go wrong, gets you in a mindset and a discipline that is different. That’s a tremendously big advantage,’’ said Ellis.
So, if you’re driving a risk analysis system, drive it hard, said Ron Papanek, head of RiskMetrics Labs. Maybe you didn’t really expect to see a lot of defaults, when the S&P/Case-Shiller Index of housing prices started to move down in the last half of 2006.
“Even if you did not foresee an extremely high default rate, it’s critical to stress the portfolio as if there was a high default rate,’’ he said. “And exceed the expectations in some of the stress tests because that gives you a very clear picture of what the potential losses can be in a portfolio.’’
But in the end, it may not take more than really gathering pertinent facts that are out in the open in the marketplace – or coming in from the “nodes” in your network.
“It’s an intelligence gathering that can really only be done at the end of the day in the front of the human brain and not on paper,’’ said Arvanitis. “Remember: Flight simulators don’t make you a pilot. War games don’t make you a general. And models don’t make you a financier.”
The complete discussion will be available here, for playback.
The original ZDNet report on how Goldman Sachs calculates risk can be downloaded here.