Wall Street and the Wages of Estimation

Last week, the New York Times blog Bits had a piece about Wall Street analysts "lying" to their computers. Written by a journalist who has covered both the Internet and trading and finance, it indicates how little many reporters know about the mathematical processes they write about -- otherwise there might have been much earlier warnings of the problems now facing the world's economy.

Yes, the current meltdown is supposed to be a "once in a hundred years" event, as the president of a consulting firm told the reporter. However, these black swan events, when considered together, happen far more frequently than every hundred years. That's because when you have potential events that are independent of each other, but still capable of creating financial cataclysm, the cchance of at least one of them happening is the sum of the separate chances of a given one happening.

So already experience and math suggest that the surprise at an unusual event should have the same tone as Captain Renault being "shocked, shocked" at the gambling at Rick's in Casablanca. The concept of adding probabilities for independent events has literally become math in grade school, or at least high school.

Now for the other revelation:

The people who ran the financial firms chose to program their risk-management systems with overly optimistic assumptions and to feed them oversimplified data. This kept them from sounding the alarm early enough.

Top bankers couldn’t simply ignore the computer models, because after the last round of big financial losses, regulators now require them to monitor their risk positions. Indeed, if the models say a firm’s risk has increased, the firm must either reduce its bets or set aside more capital as a cushion in case things go wrong.

In other words, the computer is supposed to monitor the temperature of the party and drain the punch bowl as things get hot. And just as drunken revelers may want to put the thermostat in the freezer, Wall Street executives had lots of incentives to make sure their risk systems didn’t see much risk.

I'm sure there's something to this, but let us get at a more fundamental issue: many complex problems in math and engineering and science are too tough to easily solve. What people do, then, is simplify. You put the world into models and approximations that you have a prayer of solving. You drop factors that seem tiny in comparison with the rest of the problem in an attempt to simplify the equations even more. You employ numeric methods to get closer and closer to the "real" answer ... as close as you need.

Unfortunately, your answer is simply an approximation, nothing more. It may be acceptible for your uses if the real world conditions are forgiving enough. But when things get hairy -- you're trying to predict the behavior of materials in the face of quantum mechanics effects or trying to understand how an incredibly complex system, such as the weather or global finance, will behave -- then approximate may not be good enough. If a chip fails, well, you head back to the drafting board. When an economy fails, then you end up with huge banks and investment companies going out of business, oil prices swinging by $25 in a single day as short-sellers have to cover their positions, and governments begging for the right to spend $700 billion of taxpayer money to get their croneys out of the frying pan.

Cisco Tries New Business Model

The Financial Times has an article (which seems to be publicly available, at least today) about Cisco's efforts at a more entrepreneurial approach to develop new business units, rather than just acquiring them. Some of the important lessons are:

• The company wanted to "create a sustainable long-term growth model."


• Use proven managers to make success more likely.


• Use a "free agent" employee model, where after a certain number of years at the same job, an employee could look for another opportunity within the company. That reduces fighting when someone is transferred between divisions.


• Make use of the enterpreneurs who joined the company through acquisitions because "[t]hey don't stop being entrepreneurs."


• Offer incentives like pay bumps when a venture is successful.


• Create an atmosphere where failure is acceptable, because new ventures are inherently risky, and you don't get breakthroughs by playing it safe.


• Innovate in business models, not just technology.

If you're in a large company, this article should be a must-read. One reason acquisitions generally do so poorly is that one corporation essentially pays a premium for a revenue stream - like giving someone a ten dollar bill for a five - and then has the additional costs of trying to make everything work together. When you innovate from the inside, the costs are far smaller, you get the full benefit of real success, and it's easier to make the fit work.

One thing that does seem to be missing is dealing with moderate success. Tigns that lose money you can drop and reassign the people involved. But what if a new venture is reasonably successful, yet not so profitable as to make managing it finacially worth the time of the parent corporation? They could sell the group to someone else, but lose access to the personnel who were involved. So, create a separate holding company that manages smaller ventures. That way there is still return on the investment, you don't need the involvement of the main management team other than looking at the holding company as it would an investment in a third party, and you can still bring someone necessary back and, should one of the small ventures suddenly hit on something big, the parent corporation can reap the benefit.