Survivor Bias' Skews Morningstar Data
April 3, 2006
Investors who pick actively managed funds simply because data shows they beat the index might want to look a little more closely, according to a report released by a coalition of eight independent financial advisory firms known as the Zero Alpha Group.
"The bottom line here is that we are very concerned that the mutual fund industry systematically and significantly overstates fund performance in a way that makes actively managed the mutual funds occasionally look competitive with indexes when that is not the case," said Brent Brodeski, a co-author of the report, and managing director of Rockford, Ill.-based Savant Capital Management.
The report singled out "survivor bias" as the source of the misconception, and placed the bulk of the blame with Morningstar, the Chicago-based company from which the group drew its data.
"Survivor bias' occurs when the funds with the worst performance are made to disappear from the database while strong performers move forward," the report notes. Fund companies can weed out fund weaklings either by merging them with better-performing products or dissolving them altogether.
The result is that when survivors are examined as a group, their average performance is skewed upward to the tune of 1.6% per year, according to Savant's review of Morningstar data.
"When properly measured, active fund managers fail in almost every category," said Glenn G. Kautt, president, chairman and chief investment officer of The Monitor Group of McLean, Va., another member of Zero Alpha. In fact, after accounting for failed funds, the study found that actively managed mutual funds actually lagged their corresponding indices in all nine Morningstar Principia "style boxes" between 1995 and 2004. In the 42 more narrow Morningstar fund categories, survivor bias inflated actively managed fund returns in all but one category.
"It appeared as if there was a glimmer of hope that fund managers were providing value, when in fact, they're not," Brodeski said.
Criticism of survivor bias is nearly as old as the debate between passive and active management of funds, the report acknowledged. The study cited Burton Malkiel, a researcher at Princeton University who analyzed Lipper data between 1982 and 1991 and found that when adjusted for survivor bias, funds ranked by the New York-based company trailed the Standard & Poor's 500 Index for that period by 1.83%.
Mark Carhart, head of Goldman Sachs' quantitative analysis team, found in 1997 that accounting for survivor bias dropped investor returns, on average 2%, the study noted.
"This is not a Morningstar issue," Brodeski said. "It really is an industry issue."
Lipper and Morningstar each acknowledge the presence of a survivor bias in their products. "Lipper admits that this is a shortcoming but considers it a minor issue," the report said. "Morningstar also conducted a study of survivor bias using their database. They continue telling the investment community that it does not change the average U.S. stock performance at all. We disagree with their conclusion."
Morningstar Managing Director Don Phillips disagreed with that characterization. He said that Zero Alpha took these comments out of context and that they pertained only to a short period of time in 1997. And for that specific period, survivor bias did not, in fact, significantly affect performance, he maintained.
In fact, survivor bias is so clearly an issue when comparing fund categories that Morningstar CEO Joe Mansueto even made it a company objective to develop figures free of survivor bias by the end of 2006, Phillips said.
But the data the Zero Alpha Group zeroed in on wasn't meant to be used for that type of comparison. It was meant for investors who want to compare individual funds to their peers within a category, Phillips said.
"If you're looking at a fund you're considering picking today, you will ask, How has this done relative to the other funds I can buy today?'" Phillips said. Extinct funds speak for themselves.
"Survivor-bias-free data is like motherhood and apple pie. It all sounds good," he said. "But there's a dark side to what they're advocating." If failed funds were included in the category averages, individual survivors would break into the top half simply by staying open, he said. Suddenly, individual funds that were once average-performing would appear to be superstars.
Not only is that misleading, it's also dangerous to the typical Morningstar customer, he said.
"I think this whole press conference was an attempt to get the names of nine small investment advisory firms in the paper," Phillips said. "More power to them, but it's just a little unfair to us."
Just as the Zero Alpha Group study criticized active managers for using Morningstar's "skewed" category data to push their products, it also argued that bias-free data offers irrefutable proof that passively managed index funds, the type they support, are superior.
The truth is probably somewhere in the middle, said Geoff Bobroff, principal of Bobroff Consulting in East Greenwich, R.I.
"There is survivor bias in almost every measurement," he added. Even in the S&P 500, companies merge, liquidate and are taken out of the index, he said. At the same time, investors and analysts, including Morningstar, have been critical of actively managed funds, which cost more and have provided lackluster returns in recent years.
Whether they're considering an actively managed or an index fund, Phillips said, investors should be shrewd and strike the right mix. "They're painting us into a corner saying we're apologists for active managers because of our methodology," he said. "That's a pretty hard stretch of the facts."
(c) 2006 Money Management Executive and SourceMedia, Inc. All Rights Reserved.