"Why do people fail to [obtain] the feedback to detect this [regression to the mean]? First, people fall prey to the 'fundamental attribution error' and attribute too much of a manager's performance to skill rather than luck. Second, people miss the importance of control conditions (i.e., a situation in which we had not fired the manager)-even if the data [for] such control conditions exists."
Post-termination returns remain practically invisible to committee leaders who fixate on skill.
A singular preoccupation with presently retained managers dovetails with regret theory. As mentioned above, committee members avoid personal angst associated with poor decisions; an investment process that saddles asset managers with blame also allows committee members to avoid personal and professional scrutiny related to a flawed investment process.
Dunning also mentions committee failure to control for this seemingly irrelevant data. Control conditions help to distinguish random data from causally related data. Typically, investment committees will switch managers and compare the cumulative returns from both managers to a benchmark. Dunning suggests controlling also for "a situation in which we had not fired the [active] manager." The returns of the terminated managers would then be weighed against both the benchmark and the returns of those managers who were retained.
The third trait, overconfidence, like fundamental attribution error may render post-termination returns irrelevant in the minds of committee leaders. Overconfidence describes a mindset that exaggerates personal ability and encourages risk-taking; when a committee member receives feedback on past judgments, it counteracts this overconfidence. Meteorologists, for example, show little overconfidence because they are publicly predicting the unpredictable. Instead, each day's weather provides them with feedback on their accuracy.
Dr. Erik Angner of the University of Alabama at Birmingham and author of "Economists as Experts," writes as follows:
"The psychological literature indicates that overconfidence is endemic, [unless] feedback is regular, prompt and unambiguous. If the latter condition doesn't [exist] for investment managers, you would expect them to suffer some degree of overconfidence."
Unfortunately, as Arnold Wood reports, investment committees function with anemic levels of feedback on their past decisions.
Dr. Anand Goel of DePaul University has studied CEO decisions and overconfidence. CEOs often feel ambivalent about the added expense and time of accumulating information in preparation for an imminent decision. If hasty, CEOs may not obtain sufficient data, while too much caution wastes resources and money. Overconfident CEOs introduce an extra element of risk: These men or women are more likely to cut short the gathering of information and injure their businesses through ill-advised decisions.
Goel suggests that committee leaders operate with similar limitations. Investment committees, Goel suggests, should "evaluate the effectiveness of their own decisions by observing the performance of retained versus fired managers (and controlling for changes in environment that may have affected the performance of fired managers)." This self-evaluation may validate past decisions or lead to more patient and informed decisions.
Status Quo
Again, a systematic analysis of a manager's post-termination returns rarely occurs in the retirement industry. A retirement plan consultant interviewed for this article remarked, "That's a great idea, but I don't know anyone who does it." Similarly, Russell Olson, a former director of pension investments for Eastman Kodak, does not mention the practice in his Handbook for Investment Committee Members.
Jason Zweig, author of Your Money and Your Brain, relates the following anecdote:
"I once gave a speech at a meeting of retirement plan consultants and was heckled for pointing out that they do a lousy job at manager selection. 'Prove it!' called one of the hecklers. I responded with a question: 'How many of you track the performance of the managers you fire AFTER you fire them? Let's see a show of hands.' There were roughly 100 people in the room, [but] not a single person raised a hand. In other words, they all believed their sell discipline worked, but not a single one of them had ever gathered a shred of empirical data to test whether it was actually true. It was as if Neil Armstrong had come back from landing on the moon and insisted that it was made of green cheese because the lunar soil had felt sort of squishy under his feet."