Bear Stearns News and the Role of Economic Models and Predicitions

Pretty dramatic news tonight that JP Morgan has acquired Bear Stearns. After Bear Stearns was trading in the 90s just a few months ago, this sale represents a purchase by JPM at a $2/share price — wow !

[ Begin Rant ]

Right after I read that, I saw an article in the FT, that Alan Greenspan thinks our financial models were partly at fault for this housing / loan crisis, and that better statistical models & better data in the future will help improve the stability of our markets (although he acknowledges that we’ll “never have a perfect model of risk”):

The most credible explanation of why risk management based on state-of-the-art statistical models can perform so poorly is that the underlying data used to estimate a model’s structure are drawn generally from both periods of euphoria and periods of fear, that is, from regimes with importantly different dynamics.

If we could adequately model each phase of the cycle separately and divine the signals that tell us when the shift in regimes is about to occur, risk management systems would be improved significantly

Greenspan is obviously a guru and a huge figure of our time — Bob Rubin’s book that I read a few years ago, In an Uncertain World: Tough Choices from Wall Street to Washington, had some great insights into how Greenspan operates.

But this general idea that better models are the solution strikes me as counter intuitive. There is a whole debate in certain quarters about economics being a hard or soft science — that is, is it closer to psychology and philosophy (soft – about trends / ideas / different for each person or situation – no less important a science, but not about injecting repeatable test-driven methodologies) or closer to math and physics (hard – repeatable tests, about math & data, etc). I tend to think it’s a softer science for a few reasons:

1) With nearly every economic model there is an underlying assumption that people will act in an economically rational way. That they will maximize their economic situation. We simply know this to not be the case. People overspend, overreach, and are easily manipulated into doing things that are absolutely not economically rational. Having a killer sound system in a beat-up broken down car is just one example 🙂 Investing money in the markets while carrying high interest credit card debt is another common and simple example of not being economically rational.

2) If somehow you could build a model that took in every data point, you would need infinite data points. You would need to model in the probability of a new york sports team winning and how that will impact $$ spent in bars in NYC, and will therefor impact real estate in the east village. Just seems out of our current reach — the perfect model needs to encompass the entire real world.

3) Economic models on future events have been mostly wrong, and continue to be wrong – regardless of source – government & private sector. Stock pickers tend to be wrong over the long run, as index funds have been shown to do much better. Basically the finance experts who have tons of resources, sophisticated models, and 100% focus have a pretty bad hitting average. A friend of mine works in energy and told me that all the economic and financial prediction models that are put out are basically useless to his trading strategy & that I should look up up oil price estimates from the major banks in 2006. A quick google search and I found this from a top 3 bank:

We have increased our projection for prices of crude oil (WTI) for 2006 and 2007 from $50.00/bbl to $57.50/bbl and $45.00/bbl to $55.00/bbl, respectively.

Not even close — we are well over $110 today for WTI

I don’t necessarily have the answers on how to do it better — but it’s worth looking at the track record of the experts and perhaps not being as surprised when they get everything wrong and their models fail to predict the outcomes that we see.

But to me the inspiring part of this situation is that the US is the best positioned economy to deal with any challenges that lie ahead. In the US companies fail fast (the whole JP Morgan deal took 3 days basically over a weekend ). By failing fast and dealing with it, reallocation of capital can happen efficiently and expeditiously. Other economies are saddled with slow processes that keep failing companies on life support for years and actually suck energy out of the economy and prevent smart people from going to work for companies that should be thriving.

[ End Rant ]

Then again — I could be totally wrong 🙂 I’m also influenced right now by a book I just started reading the other day, The Black Swan, that takes a very similar line towards economic models and predictions. I’ll write up a few thoughts on that book soon.

2 thoughts on “Bear Stearns News and the Role of Economic Models and Predicitions

  1. I agree – whenever the financial markets begin to breed a “get rich quick” movement that the common person can understand and execute, we end up with a surge that has to correct itself. We went through the surge of betting on internet stocks and now we have the housing rush – both correcting strongly. So a metric that measures the “loophole” in the system that allows this surge to start and the amount of people exploiting it at any given time – may be a tool for predictive defense. Americans are suckers for the false confidence that *paper equity* breeds and they will rush in wherever possible. Maybe call it the “paper equity metric” that would track this kind of issue. I think no matter what we learn here, when the American consumer rushes in for a quick fix, American companies will always follow – that’s the American way right? The trick is how to head it off before it becomes out of control.

  2. I agree with your premise here. It’s a given that people in general like to think that there is logic and rules (science) to systems that do not have it. The financial markets have been prone to this dreaming for decades. There is always a belief in the next magic equation that will make billions. The reality is the markets are driven by people, making decisions, and not always logically.

    I would challenge your point #2 though. I don’t believe that an accurate model must cover every data point in a system. I’m not nearly good enough at math to dive into this for real, but I would expect some mathematical proof that this is not the case exists. It likely involves integrals. 🙂

    BTW — If you like Black Swan, read his earlier book Fooled by Randomness. I liked that one so much I read it twice.

Leave a Reply

Please log in using one of these methods to post your comment: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s