Harnessing The Power Of Counter-Intuition
Navel-gazing can pay
By Daniel M. Ryan
If Think Twice: Harnessing The Power Of Counter-Intuition can be summed up in a single sentence, it would be called an ideal book for navel-gazers. Its author, Michael J. Mauboussin, is Chief Investment Strategist of value-investing firm Legg Mason. His background does show, as value investors tend to be critical thinkers and methodical.
Methodicality is the practice that Mauboussin recommends, in several places in the book. More than once, he suggests using a checklist so as to not forget procedures that could cause trouble if omitted. He elaborates on the theme that, in a probabilistic system with a wide variety of outcomes like the stock market or business in general, success is a combination of skill and luck. The only way to determine skill level is to looks at the methods used; measuring by outcome confuses the luck component with the skill component. There's only one exception to the rule, at least in baseball: skilled players tend to have longer and more frequent winning streaks. That's a result of better skill improving the probabilities of a successful performance.
Being methodical, though, is more of a challenge than it looks. One of the lessons in the book is that treating a complex system as if it were a tightly-coupled machine, or treating a correlation as if it were a causal relationship, or treating a circumstance as if it were a stable attribute, is an easy way to go wrong. The first part ties into the wisdom of crowds part of the book, and his discussion of complex systems as possessing emergent systems. One of the chapters is entitled "More Is Different," and it is. Take rock music, for example. It isn't just the music of the Beatles that had a different character than its members when the band is no more. Sports, too, have examples of that sort. A star lifted from a dynasty doesn't make for a new one, regardless of his or her star quality. Wayne Gretzky fans saw it when he went from the Oilers to the Kings.
Although he does discuss business in general, many of Mauboussin's examples come from the stock market. His discussion of regression to the mean focuses on the surplus or deficit of firms' return on capital net of their weighted average cost of capital. He marshals a graph showing that investment-manager firms terminated for nonperformance tend to outperform the new hires subsequent to the termination. The "stars" he discusses are hot analysts, hired away for a bigger pay packet. He notes that, like many mergers, such hires end up being a bad deal for the acquirers as the star power evaporates.
Mauboussin does recommend thinking of teams rather than individuals. Drawing on behavioral finance, as he does throughout the book, he discusses the "fundamental attribution error" of ascribing individual actions to the individual alone, rather than taking into account the social context.
This point, although valid, does lead him into an apparent inconsistency that he doesn't seem to be aware of. Before discussing social contextualization, he summarizes two iterations of a famous psychological experiment testing the effect of social pressure on an individual to come up with a wrong answer. The experiment consists of an individual invited to join a group assessment of a perceptual puzzle that the average person can figure out easily. The other members are actually confederates of the experimenter, and their job is to deliberately give a wrong answer. What's being tested is degradation of performance in the face of group encouragement to do so. The experiments found that more than a few people will go with the group, even though the answer is wrong, to the point where they fool themselves into thinking the wrong answer is right. This experiment does demonstrate the power of social pressure, 'tis true. But there's a bit of an oddity in seeing it followed by counsel to pay more attention to social contexts. A skeptical reader is left wondering, "what if the people who pay more attention to social are the ones who crack under pressure to be wrong? Doesn't that mean that social contextualization makes one more vulnerable to making bad decisions?"
Mauboussin is better at reconciling the paradox of the wisdom of crowds coexisting with the madness of crowds. He says that, in normal conditions, diversity of opinion means that individual mistakes tend to mostly cancel out. The overall accuracy of the crowd, found in (say) prediction markets, only comes into play when the members have an incentive to be right, their decision models are heterogeneous, and their different opinions are aggregated easily. The mania phase of the crowd kicks in when diversity of opinion drains away, and the crowd becomes a homogenous mass like a tightly-coupled machine. Once that state is reached, only a small amount of panic can cascade into an all-out panic – just as a tightly-coupled machine can fail by losing only one small part. In markets, loss of diversity of opinion heralds a phase transition: a panic, if the homogenous crowd is bullish. As with all phase transitions in complex systems, we can't predict the tipping point.
Some people won't like this book, as it really requires a fund of experience to draw on before the points stick in. Although Mauboussin is good enough to provide action lists at the end of each chapter, and does provide a rich fund of anecdotes, people who require examples instead of anecdotes will be a little put off. Others will devour it and read it several times in order to cultivate the ability to think critically. Hayekians will appreciate the point that spontaneous orders, and unintended bad consequences of meddling with complex systems, apply far more generally than to government intervention and market economies. Hayek himself, had he been alive, would be amazed at how widespread the applicability of his two principles are.
One point, I appreciated: his explanation of why the subprime and derivative crisis got so bad. There's been a lot of moralizing about the subject, and too little forensic analysis. On pp. 110-11, he explains that the CDOs that caused so much trouble appeared safer than they were because of a widely-used risk function developed by statistician David Li; it purported to show the correlation between the default of one loan asset and another. The trouble with this function, a kind of Gaussian copula function, was that it assumed the correlations were stable. As we now know, they weren't. So, what appeared to be an adequately diversified portfolio of loan assets was in fact dangerously vulnerable to a systemic string of defaults which the equation didn't acknowledge, let alone capture. We know that now too.
This fact makes it more difficult to blame "the free market," "greedy banksters" or "government" for the disaster. Its root was in a flawed equation, which bamboozled regulators as well as lenders. This is the kind of information necessary to prevent a repeat, as blamesmanship does little to point to any needed change in course. What about the next one that comes along, which impresses regulators as well as lenders as being innovatory? Just like the last one?
I found this book a good read. It can be gotten though in a day, and I found that it contained enough thought-provoking material to make it worth more than one read-through. For some, it will serve as a handbook. Myself, I can say that Mauboussin has done a good job in showing how navel-gazing can pay off.
Daniel M. Ryan is currently watching The Gold Bubble.
Buy Think Twice at Amazon.com for only $19.77 (34% off)
Get weekly updates about new issues of ESR!