|
|
|
Quarterly Review
Won't Get Fooled Again?
Fooled by Randomness: The Hidden Role of Chance in the Markets and in Life
by Nassim Nicholas Taleb (Texere, 2001, $27.95)
Reviewed by Arthur J. Schwartz
Nassim Taleb is a professional options trader who runs ahedge fund in Connecticut. The fund has done well by taking a wisely considered contrarian approach to a number of potential market bubbles. Aside from the relation between insurance and options, this book is of interest to actuaries because of Taleb's focus on the nature of risk and of how we perceive and respond to risk.
A major theme is Taleb's hypothesis that our desire to see patterns in data often leads us to see patterns that are not really there. Positive outcomes are a result of a combination of skill and luck, but given a consistent pattern over time, we attribute the positive outcomes to skill when they really may be due more to luck (randomness). Also, when we find a pattern that has stood the test of time, and that works well, we are less likely to recognize a regime change when the fundamental rules of the game have changed. Thus, we may not react quickly enough when it becomes important to respond to the new order. Understandably we blame bad data, or a host of other possibilities, rather than the possibility that the underlying rules of the game have changed and that we must now expect different results. This is often seen in the cyclical nature of insurance pricing and affordability. By the time we have assimilated the "new rules" of inflation, or a steep trend, for example, perhaps we should be predicting the "next rule"—of deflation!
Taleb discusses two knotty problems that have plagued science and philosophy over the centuries. One is the black swan problem. Specifically, based on my observations of many white swans in my life, I may hypothesize that no black swans exist. Yet, it takes the sighting of only one black swan to disprove that hypothesis. Likewise, in the insurance world, the possibilities of a large-scale terrorist attack on American soil were largely dismissed prior to the events of 9/11. Now we have models that we believe will help us estimate long-term loading for terrorist attacks.
Another knotty problem is the skewness issue. Specifically, it does not matter how rarely something happens if failure is too costly to bear. Applied to insurance, the possibility of my home burning down in any year may be minuscule, but it is a loss that very few people would willingly bear.
An interesting and unusual strategy discussed in the book is to consider alternative histories. These are simply the alternative events that could have taken place. This concept will be familiar to actuaries and claim adjusters, whose experience may eventually include all kinds of claims that were not originally considered. Sometimes these lead to restrictions of coverage, other times, they lead to new coverages. This discussion naturally segues into Monte Carlo engines. Constructing models and using Monte Carlo methods to consider the range of possible events is something actuaries using dynamic financial analysis should always consider when we "stress test" a hypothetical business plan.
Taleb includes an entertaining sketch of two traders, Carlos and John, and six lessons learned from their behavior. With some adaptations, the lessons can be applied to insurers. If an insurer experiences losses in a certain geographic area, line of business, or type of risk, it may make sense to view this in the light of a trader. Three of these lessons—having a game plan for handling losses ahead of time, not getting married to a particular type of market or risk exposure, and deciding when to use a "stop loss" (exit a given market or use reinsurance)—may have saved some insurers from suffering large losses or even from liquidation. When large stock losses occur, or a line simply is unprofitable, how many insurers perform a rigorous analysis of the cause, including the need for an improvement in underwriting? Sometimes managers are quoted as saying that they will a) tolerate losses in the hopes that the market will turn and "we're in this for the long term" or b) exit a line simply because of current bad results. Either course of action may be valid. However, a better decision would be based on a well-reasoned study of why events turned out very much against the game plan.
Of course, many managers who blindly keep writing the same type of business are bailed out when the market does turn. The profits can be enormous and they then can look like heroes. On the other hand, if month after month of losses continue, even if the analysis of an expected turn-around may be right, the firm's capital may not last that long!
Taleb continues with an illuminating discussion of the rare event. One key point he makes is that the probabilities of rare events, if computed at all, are often computed under the assumption that probability distributions are stationary. It's a credit to casualty actuaries that we often revisit distributions and test not only for changes in parameters over time but also for changes in the type of distribution fitted.
Taleb includes a refreshing discussion of philosopher Karl Popper's skepticism. Popper felt there were only two types of theories: theories known to be wrong and theories that have not yet been shown to be wrong! Popper even dared to imagine that a mere increase in data does not necessarily lead to an increase in useful information! I find Popper's extreme skepticism helpful to actuaries, if only because it leaves us less inclined to be satisfied with the status quo and appropriately determined to find better models.
Towards the end of the book, Taleb discusses how he approaches randomness. I enjoyed his discussion of how he begins staff meetings: "My lesson from [George] Soros is to start every meeting…by convincing everyone that we are a bunch of idiots who know nothing and are mistake prone, but [we] happen to…have the rare privilege of knowing it."
There are many other entertaining aspects of the book. I strongly encourage every actuary to read it. It will inspire critical thinking about our own and our employers' risk taking. I highly recommend this book not only for its discussions about risk (also called randomness) but also for its discussions on how we perceive and respond to risk.
