‘Why can’t you just give me the number’ is an entertaining read that provides a great introduction to modern Monte Carlo-based uncertainty analysis. Leach used to be a geophysicist and ‘in-house evangelist’ for probabilistic methods with Texcao. The book is a continuation of Leach’s evangelization and contains a wealth of interesting stuff and opinions that pack almost every page.
Some purveyors of ‘risk management’ seem to imply that, so long as you use their technology, you can make risk ‘go away.’ None of this stuff for Leach. In the opening chapter, he explains that ‘uncertainty is the ultimate source of all value in business.’ The job of management thus becomes ‘the identification and understanding of key uncertainties in your business.’ (You may like to read about how Leach’s successors in Chevron are doing this on page 11.)
The book is packed full of intriguing insights and sideswipes at current practices. Leach is more than skeptical of the SEC’s requirement that companies report the sum of their ‘low’ reserve estimates. It may be ‘the law’ but it is statistical nonsense.
Leach offers clear definitions to the terminology. In no time you will be familiar with S Curves, Tornado, Decision Trees, Portfolio Management, Efficient Frontier analysis and more—no statistical stone is left unturned. If you were already familiar with such stuff, I guarantee that Leach will have you re-thinking your previous notions of some or other aspect of probabilistic analysis. There is insight on every page and potential controversy in every chapter.
Some of the most interesting content covers the impact of human psychology. For instance, consider the $125 jacket and $15 tie purchase problem. A deal that involves a drive across town offers a $5 discount on the $15 tie is taken up by 68% of the study group. When a similar group is offered a $5 discount on the $125 jacket, only 29% take up the offer! For Leach, this blatantly inconsistent**. Other psychological work demonstrates asymmetry in risk aversion. People are seemingly averse to ‘high potential, long shot projects that could be company makers.’ Another asymmetry is witnessed by the mild pleasure obtained from finding $20 contrasted with the foul mood into which the same person is thrown on discovering a similar loss!
I would be failing in my task as a reviewer if I didn’t point out a couple of weak points. The chapter headings are a bit too jokey for easy navigation in anything other than a full cover to cover read. The index is also a bit skimpy. Finally, I have to chip in with my own mild and idiosyncratic complaint that I have addressed at other writers in the risk management field. Leach preaches for intellectual rigor when treating our intuitive (and wrong) analysis of statistical paradoxes like the ‘Monte Hall’ problem or when trying to eliminate psychological bias. But the rigor is replaced by guesswork and coercion when trying to squeeze error estimates from an unforthcoming engineer.
Why not just buy it!
But if you are untroubled by such philosophical difficulties and want to consume the whole risk uncertainty enchilada at one sitting—‘Why can’t you just give me the number?’ is the book for you. I have seen nothing that comes close in terms of readability and overall usefulness. It’s a great title too.
* Probabilistic Publishing, 2006. ISBN 0964793857 www.decisions-books.com.
** I wondered though if it might also explain perhaps why at least 68% of the population are not running a company!
This article originally appeared in Oil IT Journal 2007 Issue # 2.
For more information or to comment on this topic email here.