Thursday, February 26, 2009

Momentum Caused the Credit Crisis

Wired magazine had a great article explaining the causes behind our credit problems. Unfortunately, they blamed math as the root cause of the problem, which is silly. It’s like blaming a hammer for smashing your thumb.

A decade ago, most investors didn’t like to put their money into mortgage pools (unless they were backed by the U.S. federal government) because they couldn’t quantify the risk. Then along came a clever guy named David Li who developed a formula to measure the amount of risk.

This formula came to be used in finance all over the world to measure the risk of baskets of mortgages and other types of debt. The main problem before Li came along was that although financial markets could measure the risk of an individual mortgage, they couldn’t measure correlation: the degree to which debtors tend to default at the same time. Li had a solution.

However, Li didn’t really work out the correlation himself. He examined the movement in prices of individual debts (in the credit default swap market) to estimate what the market thought the correlation was. Li just trusted the market to get the correlation right.

Li’s formula just assumed that the recent history of actual prices gave a good measure of the likelihood that multiple debtors would default at the same time. So, as long as markets ticked along well and had few defaults, Li’s formula predicted that the world would continue to move along smoothly.

This is essentially the same as believing that the future will always be the same as recent history. If it hasn’t rained for a month, then it’s safe to set up priceless paintings outside because it’s never going to rain, right? When the credit crisis finally hit, it became obvious that Li’s formula hopelessly underestimated the correlation of debt defaults.

I see little reason to blame Li for all this. No doubt his formula is useful in some areas and less useful in others. Those who use a formula without understanding its limitations should blame themselves for a bad outcome.

Of course there were people who understood the limitations of Li’s formula, but it was hard to get the attention of bankers who were raking in millions of dollars. They were guilty of momentum-based thinking as well. They had been using Li’s formula and making money consistently. Why would they want to stop? Picking up coins on a highway is profitable until that terrible road kill moment.


  1. It is very convenient to blame a formula for a credit crisis. These financial geniuses simply find new pistols to shoot themselves with. I have no doubt that they'll find new formulas and models to blame in the future (after a decent interval has passed recovering from the current one).

  2. CC: I agree that blaming a formula makes little sense. Just because Li developed a formula and wrote a paper on it doesn't mean that he was given control over trillions of investment dollars. Those who controlled the money had ample resources to evaluate the formula and decide whether to use it. The idea that they had to trust it and were duped is ridiculous.

  3. I blame Li entirely, even though I'd never heard of him until now.

    You say: "Picking up coins on a highway is profitable until that terrible road kill moment." Wouldn't you blame the guy who first suggested, "Let's pick up coins on the highway"?

    The bankers should have known better, probably, but they're just bankers. Li, with a PhD in statistics (according to the article you linked to), should have known not to apply the statistics of the Gaussian bell curve to a reality that doesn't fit the bell curve (investment returns/variations).

    The article quotes Taleb, and you could do worse than read his work and watch his recent interviews. Taleb calls Benoit Mandelbrot his mentor, and you may be interested in reading Mandelbrot's book The (Mis)Behaviour of Markets.

    It's not like blaming the hammer for hitting your thumb. It's like using a hammer to turn a screw.

  4. Robert: Blaming Li makes no sense. He had an idea and wrote a paper. The financial industry employs hundreds of people with the background and intelligence to evaluate Li's formula and decide where is can be applied safely and where it can't. Thinking of bankers as the low-level employees at a bank branch is misleading. The bankers who control the trillions of dollars in the financial markets are organizations that include many very talented technical people. If they made mistakes using Li's formula, then they are to blame.

    As it happens I have read Taleb's books. Their main value is that they show that certain ideas about randomness that are obvious in some contexts are also applicable in some less obvious contexts. Unfortunately, his books are very repetitive, contain many pointless meanderings, and contain off-topic vague insults that he doesn't justify. What sane person accuses all French people of smelling bad in such a book?

  5. Today's blog post reminded me of Taleb too. I just heard an interview with him, and he says that the incentives are wrong for bank executives. If they make a ton of money on their investments, they get huge bonuses. If they lose a ton of money, they get bailed out and live to fight another day.

    He thinks that the retail banking functions (credit cards, deposits, withdrawals, basic lending) should be highly regulated utilities, and protected by government. All the risky investments and such should be separate, and allowed to fail if they screw up. Fragile companies should be allowed to break, he asserts. By keeping the risky functions separate from the essential consumer business, it allows them to fail without huge repercussions.

  6. Gene: Separating the two sides of banks is interesting. I'd be interested in opinions on how feasible this would be.

  7. Good post,

    Wired has always been an edgy magazine, and this atricle shows one authors obscure point of view.

    I don't see how anyone can put the blame on a theoretical formula for this recession, this blame should be put on the investment bankers who incorrectly interpreted it.

  8. I don't think bankers should be let off the hook. For example, the article states: "...when financial markets began behaving in ways that users of Li's formula hadn't expected." Markets haven't changed over the centuries. The bankers should have known better. That's where Taleb's commentary on incentives becomes interesting.

    But Li definitely should have realised the misapplication of statistical tools he used in his paper. Anyone who's worked in real world finance (as opposed to academia) should know that correlation is not constant, hence Gaussian math simply doesn't apply. I'm only arguing that he should take responsibility. Maybe the way Alfred Nobel felt guilty after inventing dynamite?

    You don't have to enjoy the "prose" to realise there are ideas of merit in Taleb's book. What sane person dismisses interesting ideas, because they are presented alongside the insinuation that all French people smell bad? I'm sure it was meant tongue-in-cheek, but have you ever been to France?

    You didn't seem to be afraid to disagree with the opinion that I wrote. Don't be afraid to disagree with the opinion that the Wired journalist wrote, either.

  9. Robert: If you blame the bankers, I'm not sure why you said that you blame Li entirely in your first comment. In any case, Li's mistake is minor compared to the bankers who lined their pockets while risking trillions of dollars of other people's money.

    Actually, Li did realize that his tools could be misapplied, and he told people about it, but they were making too much money to care (see the article at

    Yes, I've been to France several times, not that this is particularly relevant. You seem to believe that I have rejected all of Taleb's ideas. You are wrong. As I said before, Taleb showed that certain ideas about randomness that are obvious in some contexts are also applicable in some less obvious contexts. This is valuable. The other drivel in his books is not valuable. So, I'm happy to learn from him and think little of him simultaneously.

    I did disagree with the Wired journalist. The tone of his article put the blame primarily on Li, even though he pulled his punch at the end allowing that maybe the bankers should have known better. As is often the case when something goes wrong, multiple people are to blame. In this case, the overwhelming share of the blame goes to the bankers. I spent years participating in academic research in my first chosen field. I can tell you that most papers that get published at conferences and in journals are either very unimportant or are crap. Anyone who relies on them in the real world without examining the ideas carefully is foolish.

  10. While you're talking about academic papers, let me just add that a lot are incomprehensible too. Doing research for a geology paper in university, I had to have one hand on my geological dictionary at all times.

    Maybe what George Bernard Shaw wrote is true: "All professions are conspiracies against the laity. "

  11. Yes, I agree that the separation of banking functions into different entities is interesting, and the logistics may prove a little challenging. However, it used to be that they were largely separate. If I recall correctly, Glass-Stegal kept a corporation from running too many different divisions, but that was struck down due to lobbying by the big banks (I want to say Citigroup, but I can't say for sure).

    Having an investment bank attached to a brokerage caused huge problems in the late 90s, of course, with analysts giving favourable ratings to investment banking customers for continued business or access to privileged information. Buy ratings were given to stocks that were called pieces of crap in internal emails.

    I do like the idea of basic banking functions done under a utility company model. Shouldn't we have access to our chequing accounts just like we expect to access our light switches? Of course, both Canada and the US have deposit insurance, so probably the idea of bank runs or failures are highly unlikely.

    Even if a huge bank goes bankrupt, customers deposits can be transfered to a stronger bank, so I'm not sure it's necessary to formally separate the banking functions.

    Also, upon reflection, even the failure of a major investment bank causes huge systemic problems, even if they aren't deposit taking institutions. Letting them go bankrupt is not generally a risk the US government seems willing to take, Lehman Brothers excepted.