Limits of Statistics

[gothwalk]

Nassim Nicholas Taleb's The Fourth Quadrant: A Map of the Limits of Statistics is a fascinating read, particularly if you're into economics. I commend it particularly to giftederic, sares2000, olethros and nhw.

I reckon he's completely right, for what it's worth.

Comments

[giftederic.livejournal.com]

Intriguing stuff.

My god does that article need editing though...

[ulaire-daidoji.livejournal.com]

Yeah what we need is a degree that does maybe 2 weeks introductory in every subject. It'd be really nice to be 90% educated in everything. You'd never be an expert but you'd be able to confidently have a discussion about any topic :)

[sares2000.livejournal.com]

Bookmarked for weekend reading. Thank you for bringing this to my attention.

[giftederic.livejournal.com]

I wanted time to think about all this before I posted a comment on the content, but here goes, 1 nights sleep later...

The general thrust of his argument is that trying to predict the frequency of rare events is nigh impossible for lack of data. And that the more geared your positions are, the more the consequences of your risk are similarly unmeasurable. I won't argue with either of these points, because I think that they are obvious and true statements. Don't go borrowing money to buy assets for investment purposes with any kind of risk attached. Many of us already know this to be good advice.

The bits of this article that don't sit right with me are the sideswipes he takes at economics as a whole. Economics is a very young discipline, and modern markets, whilst they generate lots of data, haven't been doing so for very long.

As a predictive science you can compare it loosely to meterology. Both have given us new vocabularies to describe the world better. Both have given us broad brushstrokes of understanding about their respective areas. Both areas of study are riddled with countless variables and seeming chaos.

Meteorology has improved in leaps and bounds over the last few decades, by analysing data, creating new tools to measure future data and developing better models of reality. And their predictive skills are improving, though they are still far from perfect. Economics is alot like this too, only younger and having to deal with a constantly evolving system. Measuring weather and trying to predict future weather is at the heart of the global warming movement and daily life alike. Economics seeks to find the same value.

Yes, economists are often wrong, and I take alot of what economists say about any kind of near future with a bunch of salt. But, I think they are right to try.

I also think that he missed the point of Bernankes comments on the economy. It is important to understand that comments made by the Fed seek to create reality as much as describe it. What the fed says changes the way people behave, and they are aware of that power. They are charged with a broad mandate of protecting the financial system as a whole, and many Bernanke comments should be read in that light.

(the following is less relevant to the article, but is pertinent to the broader debate)
At the heart of all these problems is the fact that the approach to investment by people en mass is wrong. Capital does not make money "just because". It makes money if someone puts the time and effort into analysing how to use the capital most efficently, and then puts in the day to day grind to eek out that efficency. Too many people don't do any of this, leave themselves open to unwarranted risk as a result, and then expect the government to bail them out. How many people do you know with pension funds or investment accounts that have actually done anywhere near the work to justify the risk with their savings that they are taking? Nothing in life is free, especially return on capital.

This current credit crunch arises from what seems blindingly obvious were just bad bets by financial institutions (buying stupidly bad "sub prime" debt), and years of low interest rates that supported high leverage, and this has multiplied the consequences of the 'crunch' as a result. I am of the opinion that they should let the banks fail, and make society face the true consequences of its lazy approach to capital. The bad remuneration policies for managers mentioned in the article are a symptom of this too.

(I'm going to stop now, I'm rambling off point... :)