In his book, Proofiness, Charles Seife quotes a 2007 survey conducted by the Centers for Disease Control that found an American male, on average, has sex with seven women in his lifetime while an American female has sex with four men in her lifetime. Seife claims this finding is ridiculous, given there are roughly the same number of heterosexual males and females. He concludes, “people are habitual liars.”
Amazingly, Seife was able to come to this conclusion without much apparent exposure to the investment management industry.
I would estimate, in the 20 years, over 90% of the investment managers I have met have had above-median investment performance. This is also an amazing statistic. Perhaps this is related to the sex survey—this is another male-dominated sample group boasting their exploits.
Are they really outperforming?
In his 1991 article titled The Arithmetic of Active Management, William Sharpe points out the absurdity of expecting most active managers to beat the market. In essence, all investors as a group are the market; so it is ridiculous to expect more than half the investors, on a weighted average, to beat the market before fees. After fees, most should be expected to underperform the market return.
But investment managers can still appear to be outperforming. Statistics is the last refuge of an underperforming manager. If performance is bad in one period, simply shift attention to a different period. If there are no good periods, emphasize risk-adjusted returns. And most importantly, always have enough different investment products that at least one of them is bound to be outperforming at any given time.
The CFA Institute has spent a considerable amount of time and effort devising a set of performance standards to fight against cherry-picking and other investment marketing abuses. The Institute’s Global Investment Performance Standards (GIPS) provide rules that ensure that the performance numbers are as accurate and comparable as possible, with adequate disclosure to reveal what underlies the numbers. If performance numbers are to be relied upon, GIPS should be used.
How reliable are the numbers?
But should performance numbers be relied upon? There is ample evidence demonstrating that a manager’s past performance provides very little information concerning how the manager will perform in future. Most performance numbers have absolutely no statistical validity. The GIPS standards may provide more believable numbers, but not necessarily more useful ones.
However, pension committees continue to rely heavily on investment returns in their hiring and firing decisions. Past performance is not predictive of future performance, but it is a strong predictor of which investment manager is likely to get hired.
The myth that investment is really about a brilliant manager being able to outsmart the markets will continue to dominate much investment thinking. A committee member once asked me, “If it weren’t true, why would everyone believe it?” The short answer is that there are hundreds of thousands, if not millions, of people in the global investment industry whose livelihood depends on investors continuing to believe that such brilliance exists and is available for hire. So the support for this view will continue to dominate investment information.
Once it is believed that we need to seek out money manager brilliance, what can we rely on except performance numbers? Just a bunch of qualitative factors that are much less believable because they are imprecise. To quote again from Proofiness, if McCarthy had simply said there were some communists in the state department he would not have been taken seriously, but because he claimed there were 205 communists in the state department, he was believed.
Using numbers often implies greater certainty than actually exists. If one investment manager has a five-year return of 12.5% per annum and the other investment manager has a five-year return of 11.9% per annum, the implication is that the first investment manager is better. This is nonsense, both because the numbers have no predictive value and because the 0.6% difference is statistically insignificant.
Using numbers often implies greater certainty than actually exists.
The claim of precision that is implied by our use of numbers permeates the pension investment business, in investment performance analyses, in actuarial valuations, in asset/liability modelling. Far too often point estimates are given when broad ranges would be more realistic.
At a pension meeting I attended years ago, a consultant presented an investment policy with an expected real return calculated to three decimal places (something like 5.387%). When asked whether the number wasn’t overly precise, the consultant answered without hesitation, “I don’t believe in rounding.” At the time I just thought this was eccentric. Now I realize it was dangerously misleading, implying a level of precision that should not have been promised.
It is hard not to give in to “proofiness.” There is such comfort in the precision of numbers. The appearance of greater precision achieves greater credibility. But to manage pension investments responsibly, we should focus primarily on the imprecision of the numbers.