Greenspan said (on the topic of the present financial crisis):
"The whole intellectual edifice, however, collapsed in the summer of last year because the data inputted into the risk management models generally covered only the past two decades — a period of euphoria."
That's odd... the last two decades included the 1991 recession, which had a good-sized impact on housing prices, and the 2001 recession, which was shortly before the housing boom started. Maybe if the data had only covered the previous five years, say 2002-2006, I could understand "euphoria", but going back to 1988?
What would have happened if the data went all the way back to 1980?
As for priors, whose prior am I going to use? Greenspan's? Given his reputation, one could hardly have been faulted (much) for using Greenspan's prior in 2005 rather than, say, Roubini's, although of course it was a mixture of ideology and data from the euphoric times.
If cross-validation is going to give a bad result because the data covers an unrepresentative range of possibilities, the posterior is going to suffer too, as will bootstrapping, etc. Maybe the posterior will suffer less, but you need an informative prior for that, and where the "informative" part comes from matters, since it can't come from the data.
I'm not expert on this but ... if you base your risk models on the last 20 years you exclude the depression. Presumably that means setting the probability of a depression-like event at zero or close to zero. More generally if recessions are infrequent events you won't get an accurate estimate of frequency from a 20 year snapshot.
I suspect that one reason the 20 year period was used was precisely because it produced less risk. Given the choice between a 20 year and 100 year sample there is a clear incentive to choose the one that gives you or your bosses the bigger bonus.
That's a good news from a statistician point of view!
The model is not wrong, the data is!....
(of course I am kidding)
So we're blaming (a) data entry clerks???? (b) the high cost of data storage?? (c) sheer laziness or (d) failure to even look at the last 20 years very clearly.
I'm going with (d).
I'm sure others can note various unfortunate events in the last 20 years that indicate all was not positive: the failure of the giant hedge fund Long Term Capital Management, the Russian meltdown, the Thai meltdown, various problems in Argentina, the failure of a US municipal bond insurer ... these are just a few things one might mention.
In terms of major data series being available: sure, there is more data now, but DRI already had the major econometric data series in the US on-line in the late 1970's.
Perhaps relevant to John's question of whether an adequate prior would be assigned is this humorous juxtaposition of articles:
The August 2007 American Statistician (an ASA journal) has 3 articles on "The Black Swan". There's a heavy dose of "real statisticians wouldn't make errors like this", such as this quote: "the book is statistically reckless about many issues in our profession". Taleb has a reply.
IMMEDIATELY AFTER these articles is an article by Frey providing a method to use in case you want to be sure a probability is exactly zero or one. [The classic problem is how to assign a probability of 1 to "the sun will rise tomorrow"]
Of course, assigning a probability of 0 to events such as "Black Swans" that have never occurred but that can actually occur (or too low a probability) is exactly the problem Taleb was talking about.
this is a joke, right?
1. I saw a black swan in Australia.
2. The issue here is not the model. Any cursory reading of reports by OECD, BIS, IMF, would have alerted you to the risk way back (if not exact timing).
3. The problems are the incentives of politicians, money managers, etc. to keep playing musical chairs until the music runs out.
4. Priors, data who cares? No politician is going to take the punch bowl away while the party is raging.
Taleb thinks the financial sector is ripping everyone off during the period of expansions (saving the rich), and the monetary system (taxpayers to some extent, but everyone with holdings in money) pays up in a contraction - motivated by the political interest to inject liquidity into the system (saving the poor).
The way the society "pays up" is indirect - through inflation in the next expansion period.
All in all, this scheme is pretty effective at crushing the middle class with their savings.