Plight of the fortune tellers
Many people see the repackaging of risking investments into new financial instruments as a large part of the problem but Rebonato disagrees - he sees such instruments as a good way of managing risk rather than a problem. Instead he sees the poblem as the belief in risk estimates with little basis in reality. The 99.9th percentile represents a risk which happens only once in a thousand years - how can we have a realistic estimate of such risks with only a few year's data. The distribution of such long tail risks can be very different from the Gaussian distribution which is often assumed by default. Also the use of frequentist statistical methods gives a false assurance about the estimates - Rebonato recommends Bayesian statistics for a more realistic estimate of risk.
The book makes important points, but after the first few chapters I felt it was just saying the same thing in several different ways. I would have liked more discussion of the ideas presented. For instance should repackaging risk get off the hook so easily or does managing risk mean hiding risk? The book seemed to me to be aimed more at the financial practitioner than at the reader interested in what went wrong with the economy - indeed as the first edition of the book was published 5 years ago, it's a pity that many more financial practioners didn't read it then.