A Few Thoughts on "The Plight of the Fortune Tellers"

(A) I suppose that the first lesson from talking to people about Riccardo Rebonato's The Plight of the Fortune Tellers is how foolish we academics are in thinking that there are things that go without saying: a great deal of the very large value of the book comes in saying things that, when I would say them in academic seminars, would be followed by somebody saying "But of course that goes without saying." The biggest example is Rebonato's hammered-home point that the 99.9% value-at -risk for a weekly return distribution cannot be determined. You would need 200 years of data from an unchanging return distribution before you could expect to have ten observations above the 99.9th percentile.

We academics say: of course. You cannot estimate the tails: they are tails. What 99.9 VaR, 1000-year flood, and their cousins mean is that you (a) took the standard deviation in the normal-time non-tail data you observed, and (b) multiplied it by 3.3. To call it the "thousand observation" level does not mean it is a one-in-a-thousand chance because the central limit theorem does not help you out in the tails. But talking to non-academics who have read this book I have found that they feel betrayed by the idea that a breach of the 99.999 weakly VaR can take place much more often than once every 2000 years--it is, after all, only 5.1 times the normal weekly change as defined by the standard deviation.

So: things that are to without saying should instead be said: as often as possible.

---

(B) I do think Rebonato has a tendency to throw the baby out with the bathwater. The tools of modern finance are very useful in dealing with one kind of risk: that associated with frequently-repeated transactions in normal times. Then the central limit theorem is your best friend, and the tools are wonderful. But there are a number of ways things go wrong:

• When transactions are not frequently-repeated enough so that you do not learn that your model is wrong, and you then discover that you were trading against somebody who had a much better mode of the situation.
• When for exogenous reasons times stop being normal, and you wind up in the tails.
• When you and your friends and your peers through your own actions create abnormal times.

These are three very different cases that deserve very different handling, but Rebonato runs them together...

---

(C) The four-part classification of transactions:

• Sold lottery tickets
• Bought insurance
• Sold insurance
• Bought lottery tickets

is very useful, and I wish that he had expanded on it more.

---

(D) The Greenspan quote:

the management of systemic risk is properly the job of the central bank. Individual banks should not be required to hold capital against the possibility of overall financial breakdown. Indeed, central banks, by their existence, appropiately offer banks a form of catastrophe insurance against such events...

Is true but incomplete. Banks should carry out their operation anticipating rescue by the central bank from systemic risk. But banks need to take care that they do not by their actions greatly multiply the systemic risks that otherwise exist.

(E) All in all, a highly recommended book.

Comments

[RedCharlie]

Happy Republican Day!

You know, the day we turn the clock back, only to later wonder why the darkness comes so soon...

W extended DST so that Republican day now comes right before Election Day! It also nicely coincides with All Saints Sunday, when we celebrate the deaths, I mean memories, of all those pesky do-gooders who are no longer here to bother us.

[Bonapart O Cunasa]

A very good book - but it's Riccardo, not Ricardi

[MattF]

I've only just seen the cover of the book on Amazon (thanks to your post)-- sounds interesting. I've always thought that it's much more 'likely' that tails of distributions are the places where your model doesn't work than the places where very rare events occur.

[Jack]

When you say:

When you and your friends and your peers through your own actions create abnormal times.

are you assuming a corporate intelligence? It can be very hard for an individual to stop a stampede. I mean I've been trying to buy good tomatoes in my supermarket for years but those rubbery ones are so cheap!

[Oskar Shapley]

Also on the same topic , the Black Swan, by Taleb Nassim.

[riccardo rebonato]

From Riccardo Rebonato (the author of Plight of the Fortune Tellers).

Thank you for the thoughtful review. As for throwing away the baby with the bath water, I do agree that the tools of modern finance allow us to deal well with 'normal' risk (frequently-repeated transactions in normal times). This is indeed the point I try to make (obvioulsy, not clearly enough) in Chapter "What Type of Probability Matters in Risk Management". We violently agree on everything else - especially about positive feedback mechanisms, and the existence of well defined regimes ("normal" and "excited").