Last summer, Baruch, Zubin, and I got into a discussion about the oft-cited statistic that 75% of fund managers underperform their benchmark. Is it true? Baruch concluded that no one really knows where it came from: "it seems the 75% rule will remain unattributable," he said.
But now there's an empirical study out:
Standard & Poor's Index Services has relaunched its Spiva scorecard, which compares the performance of US mutual funds and benchmark indices. Using data corrected for survivorship bias, the scorecard shows the benchmarks outperformed the managers in at least 70 per cent of cases in almost all categories.
"This is true even in relatively inefficient segments of the market such as small capitalisation stocks and emerging markets," said Srikant Dash, head of global research and design at S&P Index Services.
Here, for instance, is the US scorecard for mid-2008:
Over five years ending June 2008, S&P 500 outperformed 68.6% of actively managed large cap funds, S&P MidCap 400 outperformed 75.9% of mid cap funds and S&P SmallCap 600 outperformed 77.8% of small cap funds.
What's more, the fund performance figures do take into account annual fees, but they don't take into account any up-front "loads" -- the fee paid by investors to get into the fund in the first place, which can be as high as 5%.
Interestingly, in a case of creative destruction, the new, improved Spiva rose from the ashes of the dismembered old Spiva:
SPIVA has been a popular keeper of statistics on the active versus passive debate for more than five years. Till first quarter of 2007, it was based upon the S&P Mutual Fund database, a continuous, consistent, survivorship-bias free database. In 2007, that database lost much of its continuity and consistency following its sale and restructuring. Therefore, we had to seek alternative data sources to which we could apply the SPIVA methodology.
The new database, put together with combining data from the Center for Research in Security Prices with Lipper fund data, includes more than 3,500 fund portfolios. Which I think makes it the last word on this discussion, at least for the foreseeable future.