Customer Reviews:
Showing reviews 16-20 of 32
 By far, the most important risk management book I've read July 1, 2009
Jim Galt (St. Louis MO)
3 out of 5 found this review helpful
Doug Hubbard is a skeptic, a scientifically-minded consultant, and from what I can see an extremely pragmatic philosopher. He tells us why all of the "structured" or "formal" methods we seem to like in risk management, are no good at all. We are just foolng ourselves. We feel better about our decisions even when it can be shown that if we track a lot of our decisions they didn't get any better.
Fortunately, Hubbard also tells us about some methods that do stand up to the test of scientific rigor. It has been proven in a large number of experiments that some kinds of analysis methods are better than our gut feel.
This is more of an applied book than Nassim Taleb's well-known books and it covers a lot more ground than just financial markets (financial markets are discussed but are just part of the bigger story). Hubbard gives specific recommendations about what works. And he covers so much research about what works and what doesn't that regardless of how much experience you have in risk management, you are certain to learn something new.
I haven't yet read his book about measurement, but if its anything like this one, I know I will like that one, too.
You can't be a risk manager without this July 1, 2009
Dan Just
6 out of 8 found this review helpful
I've had the title of Sr. Risk Manager for the past 4 years and I find it incredible that I was trying to do my job without knowing the contents of this book. In my opinion, anybody who hasn't read this book can't really claim to be managing risk. The author debunks most of what my company was trying to do for years. The author supports some of what I've promoted (like Monte Carlo simulations) but now I finally have the strong case to make for these methods.
My takeaway: stop using what we know doesn't work, start using or learning about what we know does work, and measure the performance of risk management scientifically. Most important of all, quit fooling yourself! Most methods make us feel better without actually improving over the long run our estimates and decisions.
I am making this required reading for everyone on my team.
The one risk management book worth getting June 29, 2009
Mike Morgan (New York, NY)
8 out of 10 found this review helpful
This is a fantastic resource that dissects the problems of risk management like none other I've seen. There are several books out now about the failures of finance in these times but Hubbard sees a broader set of problems that apply to risk management in any field including natural disasters, public health, security, and more.
He starts out with the idea that the biggest single risk in most organizations is that the risk management methods themselves are flawed. Hubbard describes how to answer the question "Does this really work?" when evaluating methods that have been taught as best practices. Looking at the self-evaluations companies provide, it appears they all think they are doing well, but then we learn otherwise.
The second section of the book is the "Why It's broken" section and it starts with an interesting classification of methods based on what he calls the "Four Horseman" of risk management: the actuaries, the war quants, the economists, and the management consultants. All approached the problem in different ways but some were influenced by others. The stand out of this group are the management consultants who like to develop methods in less than scientific ways and seem to be the least aware of other methods. Yet, it is the management consultants that have the most influence on Chief Risk Officers.
Hubbard then systematically debunks several risk management methods that I have personally seen employed. I can vouch for Hubbard's observation that these methods seem to improve confidence in decisions without actually making the track records of decision makers any better. He criticizes well-known methods in finance but also much softer methods. I consider myself a quant and I have always thought, as Hubbard does, that soft scoring methods are pure voodoo. But he also makes a strong case for the flaws in even the most quantitative methods. While Hubbard is a booster for Monte Carlo simulations, he points out some flaws in the typical use of them. In all, I don't think he leaves any method out of the evaluation.
In the last section, Hubbard deals with the "How to fix it" side of the issue. These is where the real value of the book kicks in for me. He makes three recommendations:
1) Speak the language of probability. Hubbard denounces the use of subjective labels and scoring methods but explains that people can learn how to state their uncertainties with odds. He calls this a "calibrated probability" and it becomes a main pillar of much of what he talks about next.
2) Build models. Even for a habitual modeler like me, he gives plenty of useful tidbits. He is an empiricist and a skeptic. Models must be built and doubted at the same time. We use models not because the are perfect, but because they outperform the models between our ears (our common sense).
3) Build collaborations, institutions and incentives for risk management. This is especially helpful in finance today. Much of the problems of the market could have been avoided if Hubbard's ideas on incentives and collaborative model-sharing were applied.
All in all, this is perhaps the best indictment of the failures of the market precisely because it is not only about the market. It is about how we think about risk and the errors we continue to make when considering it.
 Required reading on risk management June 25, 2009
A. W. Sparks
6 out of 13 found this review helpful
As with his previous excellent book on measuring intangibles in business, Douglas Hubbard deals with the subject of Risk Management in concrete, practical terms. He manages to make a potentially dull subject absolutely rivetting.
I could not put it down - though had to take breathers to digest the impact of his messages and examples.
His questions and skeptical/empirical approach to the subject are simply brilliant to follow - great learning for me.
I can't quite give the book 5 stars due to (IMHO) the editing (NOT Mr Hubbards fault)
- some rather curious choices of text for highlighting in the frequent sidebars
- some simple editorial spelling/grammar misses (rather than technical errors)- luckily Hubbard provides a remedy for this via his website/fora see below
What Hubbard repeats with this book is a close integration with a matching website with very helpful downloads and discussion forums. Both of these are also frequently mentioned in the book and are well worth visiting. Unfortunately some elements of the website & fora are not as robust as one would hope and this does do a disservice to Mr Hubbard's simply excellent work.
 Right about F. Knight ,but way off the mark on Keynes,uncertainty,and risk June 24, 2009
Michael Emmett Brady (Bellflower, California ,United States)
6 out of 19 found this review helpful
The author of this book is simply ignorant of the applied work done by Keynes in the A Treatise on Probability(TP,1921).He is correct that F Knight supplied the reader of his 1921 book on uncertainty with no applied apparatus to analyze uncertainty with (p.63).His claim that Keynes provided no such apparatus means that he never read the TP.Keynes's technical contributions consist of his interval estimate approach to probability in chapters 15 and 17 of the TP.Keynes presents his own modified version of the original Boolean calculus.This was a momumental intellectual accomplishment unmatched by any economist or mathematician in the 20th century with the exception of Theodore Hailperin's 1986 and 1996 books on his extention of Boole's system into modern day " probabilistic satisfiability logics ".Keynes then constructed his work on analogy and induction on Boole's framework.Keynes presents a full blown analysis of sub and super additive,nonlinear decision weights in chapter 26 with his analysis of the conventional coefficient of risk and weight,c.Keynes also presented the first advanced safety first,risk minimization analysis(Tversky and Kahneman,ignorant of Keynes's contributions, renamed this "loss aversion").Keynes's analysis is superior to the original 1951 analysis of Roy.It is contained on the same page as his c coefficient.One also needs to understand the mathematical analysis provided by Keynes on pp.353-358,especially the end of section 13 on p.355.Keynes's injunction to minimize Risk = R= qpA=qE,where pA is the mathematical expectation ,E,or expected monetary value,easily explains the revised choices made by Savage and Paul Samuelson to the Allais paradox in the early 1950's.The Tversky -Kahneman Prospect Theory,of either 1979 or the Cumulative version of 1993,is basically an extension of the analysis provided by Keynes in sections 6,7, and 8 of chapter 26 of his A Treatise on Probability (TP;1921.These sections also appear in the earlier 1907-1908 Fellowship dissertations Keynes did at Cambridge under Bertrand Russell and Alfred North Whitehead).Keynes gets right to the heart of the matter : " The last difficulty concerns the question whether,the former difficulties being waived, the ' mathematical expectation' of different courses of action accurately measures what our preferences ought to be- whether ,that is to say,the undesirability of a given course of action increases in direct proportion to any increase in the uncertainty of attaining its object,or whether some allowance ought to be made for 'risk',its undesirability increasing more in proportion to its uncertainty " (Keynes assumed that a reader could then simply extend the analysis to the other case,which is " decreasing less in proportion to its uncertainty ".See TP,p.313 or p.345 of the CWJMK edition,Volume 8).
Keynes's technical analysis is presented on p.315 and ft.2 on p.315 of the TP.The heart of the T-K Prospect theory is that decision makers use decision weights that are non additive or super additive(sub-proportional or super-proportional), as opposed to the additive probability concept that assumes linearity.Keynes called his decision weight a "conventional coefficient of risk and weight,c ".Keynes presented it as c = 2pw/[(1+q)(1+w),where p is the probability of success ,q is the probability of failure,p+q =1,and w represents the weight of the evidence,w, defined on the interval [0,1].w measures the completeness of the relevant evidence upon which the probability estimates for p and q are based.Keynes defined w in the first paragraph of chapter 6 on p. 71 of the TP.The conventional coefficient of risk and weight is easily rewritten as c = p[1/(1+q)][2w/(1+w)].c consists of the usual linear ,additive p multiplied by two weights-the first weight,[1/(1+q)],deals with the problem of non linear , non-additive risk,while the second weight,[2w/(1+w)],deals with the uncertainty or ambiguity of the evidence w.w is practically the same as D.Ellsberg's rho variable used to deal with ambiguity.K-T assume that there is no uncertainty or ambiguity.Set w = 1 and you obtain a modified Keynesian decision weight,p[1/(1+q)].The other case is obtained simply by using p[1+q].It is a simple case of arithmetic to obtain the same solutions provided by the majority of the K-T experimental subjects in the following categories of decision problem -(a) certainty effects ,(b) reflection effects.
,(c)translation effects,(d)Allais paradox effects and (e) preference reversal effects .The crossover points,relative to the p-axis and the weighting function axis, pi =f(p),where the T-K weighting function,pi, is a function of p, are obtained easily by taking linear combinations of p,p[(1/(1+q)],and p[(1+q)].p[(1/(1+q)] generates a convex curvature.p[(1+q)] generates a concave curvature.Linear combinations of these two different curvatures(first convex then becoming concave or first concave then becoming convex) result in S-shaped curves that cross over the 45 degree line specifying where pi(p) = p.For example,a{p[(1/(1+q)]}+ (1-a){p[(1+q)]},where a and (1-a) sum to one,generates one of many possible different such S-shapes.Three dimensional graphics are easily obtained by using the Mathematica program.
The other part of the K-T theory,the value function,is not theoretically original either.The value function is used to deal with the fact that a large majority of experimental subjects felt that losses of ,say, $500,had a greater negative impact than positive gains of $500.Adam Smith was the first to point this out in his 1759 The Theory of Moral Sentiments.
In summary,there is little that is theoretically new,original,innovative,novel,or creative in the K-T Prospect Theory worked out by Tversky and Kahneman in 1979.Keynes and Smith had already provided the theoretical breakthroughs in 1759 and 1921(1907,1908),respectively.What is interesting is the gross ignorance of decision theorists, in general, when it comes to Keynes's work in the area of decisiion theory.Taleb is an exception who recognizes that it is all in the TP.
The author is also in error in his belief that the way to deal with the colossal risk management failures in financial markets over the last 3 years is to apply Monte Carlo techniques to probability distributions.Mandelbrot has demonstrated that the probability distributions in all financial markets are all Cauchy.The mean and variance of the Cauchy is infinite.You can not calculate the mean and variance.One must use Mandelbrot's rescaled range technique combined with estimating the Hurst ,H ,parameter .Mandelbrot's point is that this is all that one can do-protect yourself from losses by minimizing them.We are right back to Keynes's Minimize R criterion.This is what the banking system should be based on.An ounce of regulation ,as a preventative,is worth a pound of Free Market cure.
The author needs to revise this book throughly by integrating Keynes's insights, as well as properly accounting for Mandelbrot's insights.Mandelbrot's approach,as is the case with Keynes's,will not make you a lot of money.However,it will prevent you from suffering losses.This is what Adam Smith described as the decision making calculus of the " sober " people in The Wealth of Nations as opposed to the projectors and imprudent risk takers who currently run our commercial banks.The Investment banks of Wall Street are no longer with us as a result of their belief in the Efficient Market Hypothesis and its unsupported claims that all financial markets have time series data showing that the outcomes are normally distributed.
Showing reviews 16-20 of 32
|
