Sorry for putting up a second plug in a few days, but it seems as if, after the usual delays, quite a few things of mine are coming out that might be of broader interest than most of my academic work. I’m a contributor to a new book, Uncertainty and Risk: Multidisciplinary Perspectives, edited by Gabrielle Bammer and Mike Smithson. It’s discussed in this piece on the ABC website, which talks about Rumsfeld and ‘unknown unknowns’, a topic I’ve talked about before (here at Crooked Timber and here on this blog).
There’s lots of interesting views of uncertainty, in all sorts of fields, from statistics to jazz. You can watch a slowTV video (parts 1 and 2) or hear a more complete podcast of the book launch, with a public lecture on uncertainty and intelligence (in the CIA sense) by Michael Wesley.
One thing that is, unfortunately, certain is that the price of the book will be far too high for all but the keenest readers, so you’ll probably have to wait for it to reach the library if you want to read it – there’s not even “Search Inside” on Amazon.
"Uncertainty and Risk: Multidisciplinary Perspectives (The Earthscan Risk in Society Series)" (Earthscan Publications Ltd.)
You can, however, get 15 per cent off the UK price (save ten quid!) with this flyer
updateHere’s an extended ‘teaser‘ (4.4 Mb) with TOC and one chapter. More to come at the website of the book.
5 thoughts on “New book on uncertainty”
John, the podcast link doesn’t work – is there an alternative?
Fixed now, thanks for the alert.
Looks fascinating. I presume there’s some material on the massive discrepancy between perceived and actual risks to life and limb from various sources?
For confidentiality reasons, I won’t use my usual name.
I spent a couple of years working in a professional capacity on the interpretation and implementation of Basel II Accord AMA for big bank types. Operational risk class, although barely represented in the BIS documents, required a fairly instense level of research. Perhaps the greatest thing to come out of the Australian banks implementing Basel II, is that they needed to look closely at how they monitored and controlled their exposure to operational risks; these tend to get to the heart of management and systems, as opposed to the more narrowly focussed financial risks dealt with in the previous Basle documents.
Because of the paucity of operational risk loss event data, and indeed the difficulties in distinguishing a loss event (eg duration, antecedent and precedent events, cascading effects, correlative factors, and so on), the outcomes of the models (whether simple or more involved Monte Carlo and variants) is useful in aiding detective work towards unearthing unintended exposures, rather than just giving a numerical assessment of value at risk (in my personal opinion only; it’s an arguable point).
For those not familiar with the characteristics of operational risk, they are basically that:
a) severity of losses may be quite heavy tailed,
b) frequency of loss events is often very low,
c) a + b implies hard to determine even remotely plausible probability distributions or parametrically robust distribution fitting.
d) reporting bias (eg less than million bucks not worth considering as material at group level) adds a truncated data component to any empirical data sets. On the flip side, under-reporting of major losses is not unknown either.
e) post-event organisational change modifies the likelihood of future occurrence and/or extent of impact.
f) Basel II Accord uses VaR (Value at Risk) as the central measure. Unfortunately, VaR has some less than nice mathematical properties (related to additivity). Perhaps something like expected shortfall would be a better measure.
g) Risk and uncertainty have technically distinct meanings but unfortunately this is not well appreciated.
Anyway, it is good to see that some decent literature is now becoming available in this rather interesting area. I’d be most interested if you (JQ) gave a future blog to this sort of topic.
I might well be interested. Can you post or email the table of contents, or better, get the publisher to put the front matter on Amazon?