The Siren's Song of Numbers

The world of finance is built upon a foundation of complex formulas, dense spreadsheets, and super computers churning complex algorithms 24/7. Yet it became painfully clear over the past 9 months that this foundation is unsound at best. With all these rocket scientists (yes many are in fact real rocket scientists) wielding massive computing power, how can such a thing possibly happen?

Here's how: pseudoscience, seductive details, and garbage in = garbage out.Societe Generale's strategist, James Montier, recently published a fascinating paper on this topic entitled "Pseudoscience and finance: the tyranny of numbers and the fallacy of safety". In it he points out that most of us are easily fooled by anything that sounds even vaguely scientific. People are easily distracted by "seductive details". Anything that sounds complex or scientific becomes instantly more believable. He concludes, and I agree, that the scarcest commodity in this world is not in fact oil, copper or corn, but critical thinking and skepticism. Mr. Montier points to a variety of experiments which demonstrated that test subjects thought that explanations were better if they contained scientific information, even if it was meaningless. In other experiments, subjects had difficulty recalling important information if passages they read contained interesting, but irrelevant information. The "seductive details" apparently captured the subjects' attention , distracting them from the important details. As he said, "suddenly the world of analysts is starting to make some sense to me!"I read a lot of analyst reports, and usually the only thing I learn is how useless analysts really are. Pick up any report and it will be full of details about a particular company. These seductive details are 99.9% of the time completely worthless to anyone trying to make an intelligent investing decision. One simply needs to listen to a quarterly conference call as these analysts ask executives for meaningless minutae to plug into their massive spreadsheets. They then pull a variety of inputs out of the air to build their "valuation" models and choose a "target price". All of it arbitrary meaningless pseudoscience full of false precision. So why do I read them? In the hopes of pulling out a useful tidbit of factual information, rare as they are.Discounted cash flow models are a pillar of stock valuation, yet it's hard to think of anything more pseudo-precise than a dcf model. At the very least, the analyst must predict the free cash flow for the next 5-10 years, and he must choose a discount rate. Finally, he must choose a perpetual growth rate (the rate the company will grow cash flows forever beyond year ten) or a "terminal multiple" (the multiple of cash flow a buyer would pay in year ten). Small changes in each of these assumptions can make HUGE differences in the valuation, yet analysts treat their derived value, no doubt calculated to two decimal places, as THE correct value.

Now, a confession. I in fact use a very large self-built spreadsheet, complete with XBRL (which allows it to load historical and real time financial data into it on demand and create several models on the fly), 23 worksheets and 7 valuation models when looking at most stocks. I don't know exactly how many calculations this thing does, but it's probably somewhere around a bazillion.

Yet, I pay only moderate attention to the values it spits out. I pay much more attention to the "mosaic" it creates which shows me how a company has performed over the past ten years, and how it might roughly look in the future. There are tons of assumptions built in, and I endeavor to make each more conservative than the last, in the hopes that any valuation range which appears is likely to be very conservative.

Then, I mostly ignore it and do a few simple calculations in my head. As opposed to most of Wall Street, this is the preferred methodology of the real investors. Buffett doesn't have a computer in his office (unless he has one now to play bridge online), or even a calculator. My guess is his analysis of a business goes something like this, after reading through several annual reports:"I've known about company X for along time. Has a great brand. People will be buying X-widgets long into the future. The have generated free cash of $1B for the past several years, and it looks to grow some into the future. The manager is a great guy, family business, and he's already rich. Doesn't need to work anymore, but does it because he loves it - he should be around for a while, plus he has a deep bench of managers.

Doesn't matter anyway, the brand and products are so universal a monkey could run the business. I figure on average, once I own this company, that it will pay me $1.5B in cash pretax with no end in sight. If it was a 30 year bond yielding 5% risk free, it would have a par value of $30B. I would like to earn 10%, so it's would be worth $15B (assuming zero growth). I expect it can reasonably grow those cash flows at 4% per year, so that would up the valuation to $25B. I'll pay 60% of that to build in a margin of safety. My bid then will be $15B."

No calculator required, yet Buffett has built his incredible legacy on such models. Monish Pabrai, a Buffett disciple and very successful hedge fund manager, has stated that once you need a spreadsheet to make an investment decision you're in trouble. Adding layers of calculations and computing power simply does not improve on reality - it just obscures it. When Buffett says that many investments are "too complicated" for him, he really means that they cannot be modeled simply with any degree of confidence.

This of course has been completely lost on Wall Street. We've all been and continue to be witnesses to the spectacular failure of "risk management" as a precipitator of the credit crisis. As Montier says, risk management is "clearly pseudoscience of the highest order". Risk management, an entire industry unto itself, is simply used as a justification for the ridiculous risk-taking antics that led to the credit crisis.

"Value at Risk" is the drug of choice for risk junkies, because it rationalizes away the most dangerous risks. A portion of the CFA curriculum is devoted to the theory of VaR. I can remember thinking what a complete waste of time it was. Much like a DCF, a VaR model can tell you anything you want it to. By it's very nature it cuts off the "tails" of the probability distribution of possible outcomes.

Yet "fat tails" and "black swans", ignored by VaR models, are just the events that lead to financial ruin. Just ask the LTCM rocket scientists. Yet, hordes of "risk analysts" obediently spit out VaR numbers so that financial institutions and hedge fund managers can feel safe in levering up their risky investments 30x. The moral of the story is this: numbers without critical thinking are useless, even dangerous. Complexity should not engender comfort; simplicity should.

Montier:"Blind faith in anything containing numbers is the curse of our industry... The artificial deployment of meaningless numbers to generate the illusion of safety is something we must all guard against." Too late for Wall Street (they'll never listen anyway). You and I however can and should take heed.

Comments

[[e-mail]]

Do you remember Long Term Capital? Do you remember a Ponzi scheme structure? What is the common design in all financial scams? Do you remember the Wizard of Oz?

Borrow short and invest long, attract lots of succor investors, siffon off money into your own account with large fees. Hire lots of unsuspecting pitch persons. In the age of computers and ostentation get a giant computer to run a giant program in a fancy new and big building.

Wear fancy clothes. Move out of town just before the collapse.

Look to new PhD. economists for new cons.

The inoculation requires an old fashioned financier like Baron Rothschild.