Has Bill Gross — the great “bond king” who investment researcher Morningstar in 2010 named fixed-income manager of the decade — lost his touch?
It certainly seems to many as though he has. Since leaving PIMCO in October 2014, the bond fund that he has managed for Janus Capital Management — the Janus Henderson Global Unconstrained Bond Fund JUCAX, +0.23% has produced a loss in total return terms of 0.5% annualized, more than two percentage points below a benchmark bond-market index fund. Just this month, even Gross’s boss at Janus went on CNBC to say that Gross has “been wrong and wrong badly.”
I wouldn’t be too quick to write Gross’s obituary, however. If “being wrong” were grounds for concluding that someone has lost his touch, then every adviser — including such luminaries as Warren Buffett, the CEO of Berkshire Hathaway and widely thought to be the most successful investor alive today — would come up short.
Is there a more rigorous and objective standard we can use to determine if an adviser has lost his touch? Fortunately, there is. This column has applicability not just for investors in Gross’ fund but for all of us, since at some point or another every adviser will at least appear to have lost his touch.
Let’s start by reviewing Gross’ performance. From the end of 1987 through September 2014, the PIMCO Total Return Fund PTTRX, +0.00% that he was managing surpassed an unmanaged bond index by 1.14 percentage points per year, annualized. From October of that year to the end of this past July, in contrast, the Janus fund he managed has lagged the index by 2.1 percentage points, annualized. (See accompanying chart.)
In the normally staid world of fixed income investing, in which managers typically beat or lag an index over the long-term by narrow margins, such a large swing certainly seems significant.
But is it really? A big part of the challenge is properly defining the issue in statistical terms. It makes no sense, for example, to ask whether a positive alpha of 1.14 percentage points is different than a negative alpha of 2.1 percentage points. Of course it is. But the proper question is whether the difference is statistically significant — so big that’s unlikely to have been caused by luck alone.
Gross’s losing stretch isn’t as unusual as it appears.
One good first question to ask, when trying to determine if Gross has lost his touch, is whether he previously ever produced negative alpha over a period as long as he has been recently with Janus. The answer is “yes.” So from this perspective, Gross’s losing stretch isn’t as unusual as it appears.
As an aside, I note how rare this simple question gets asked by investors. They so impulsively want to get rid of their adviser for bad performance, they don’t stop to check whether the crime for which they are convicting their adviser was ever committed previously. Another recent example is value investing, which historically has far outperformed growth. But it has largely lagged in recent years, leading many to declare that the approach is “dead” — even though value lagged growth for at least as long a period in the 1990s.
Another, slightly more involved, question to ask about Gross: Has there been a significant reduction in the percentage of months in which he produces positive alpha? In the 46 months since Gross began managing the Janus fund, for example, he has generated positive alpha in 23 of them, or 50% of the time. In the 321 prior months while managing the PIMCO Total Return Fund, in contrast, he generated positive alphas in 215 months, or 67% of the time.
How significant is this reduction in the frequency of positive alpha months from 67% to 50%? This is just the kind of question that statisticians have ready tests to answer. They imagine a situation in which they are asked to pick 46 ping-pong balls at random from a bag of 321 ping-pong balls, 215 of which have a “+” on them (representing the positive alpha months) and 106 with a “-” on them. What are your chances of picking 23 balls with a “-” on them?
The answer? The odds are fairly high — much higher than the 5% threshold below which statisticians often are comfortable concluding that a pattern is genuine.
To be sure, the real world is more complicated than this simple thought experiment, since it’s important not only whether the balls have a “+” or a “-” on them but also the magnitude of their alphas. But when applying more sophisticated tests to Gross’s record since 1987 that take into account both the direction and magnitude of the alphas, the answer is the same: The frequency of Gross’s negative monthly alphas while at Janus is not so high as to cause statisticians to sit up and take notice.
The general investment implication, for all advisers and not just for Gross: It is extremely difficult to conclude that an investment manager has become significantly worse than he was in a prior period — that he has lost ability. That’s because luck and randomness play such an outsized role in performance over the short- and even intermediate terms. Performance must deteriorate by a huge amount, and stay awful for an extended period of time, before a rigorous statistician can be confident that what’s going on is more than just a period of bad luck.
The corollary, of course, is that you need to pick investment professionals with extreme care. You must be willing to stick things out, through thick and thin, since it will take a serious transgression — more than we’re seeing with Gross’s recent performance, for example — to justify leaving. Keep this in mind the next time you’re eager to fire a fund manager because of disappointing performance.
None of this is to say that Gross’s bond fund won’t perform dismally in coming months and years. It’s just that we can’t conclude, from his past performance alone, that it will.