The Making of Stansky's Monster

by

Ross M. Miller
Miller Risk Advisors
www.millerrisk.com
February 26, 2007

"Stansky's Monster" is my second paper based on a simple idea: Most portfolios can be statistically decomposed into two parts—one part that is easy to reproduce with "off-the-shelf" parts and a second part that makes the investment distinctive. For mutual funds and similar investment portfolios, the easy-to-reproduce part is equivalent to an index fund (or the right combination of them) and the leftovers are equivalent to a market-neutral hedge fund. ("Stansky's Monster" is currently available at http://ssrn.com/abstract=964824 and the button to download it is at the bottom of that page.)

 I was not the first person to see this relationship, but it does appear that I was the first to work out the underlying math in a way that I could get some useful numbers out of it. I derived a simple formula that converts the published R-squared of any mutual fund into the share of its assets that are being "actively" managed. For funds that closely track their benchmark index, the active share of the fund will be quite low. For a hardcore shadow indexer with an R-squared of 99%, is only really purchasing a bit over 9% active management. Like any financial theory, there are some assumptions that underlie the analysis, but the real test of the theory is that the results it generates seem reasonable.

The first thing that I noticed was that if one looked at the implied cost of active management, the typical active mutual fund was much more expensive than it appeared and many hedge funds, especially those with little or no index component in them, where cheaper than they appeared. With some simple algebra, it was possible to reverse engineer what the active part of a mutual fund cost by netting out the cost of the passive part. The output of this research was an article entitled "Measuring the True Cost of Active Management by Mutual Funds," which has led a fruitful existence as a working paper and will appear any moment now in the First Quarter 2007 issue of the Journal of Investment Management. (An electronic reprint of that article is available here.)

It was not surprising to find that shadow index funds are really quite expensive, charge someone even 50 basis points for a fund that contains 9% active management and hedge funds, and implicit cost of active management runs 5% and up. Viewed this way, the portable alpha strategies used by many sophisticated institutional investors make sense not only from a performance perspective, but also from a cost perspective.

While I was doing this research, I noticed something else. There were a handful of shadow index funds that not only were costing their investors a lot for what they were delivering, they were consistently losing more money than could easily be accounted for in management fees and related expenses. And it was not just obscure funds that behaved this way. Fidelity Magellan Fund had an R-squared that frequently exceeded 99% between 2002 and 2004 and yet managed to trail the S&P 500 by nearly 3% per year over that timeframe. By my analysis, Magellan is actively managing roughly 9% of the money it receives. Of that 9% at the very most 1% per year is being consumed by costs that are either stated or hidden. (Magellan's expense ratio averaged around 70 basis points during that period and its turnover averaged 13%, so even though it may have overpaid its brokers in exchange for receiving freebies like invitations to dwarf-tossing parties, there is no way its legitimate expenses could have exceeded 1% in total.) This means that Magellan was losing a hefty chunk of the funds (22% or more each year) that it was implicitly actively managing even if one considers its expenses to be money down the rat hole.

The problem is that the typical way that people look at mutual funds, which is encouraged by outfits like Morningstar and Lipper, helps funds like Magellan hide what is really going on from its investors. This is easily illustrated by two graphs. First, consider how Magellan did relative to its benchmark, the S&P 500, between 2002 and 2004:

The nearly 9% shortfall is not good, but it is not that awful either, at least not until we see what Magellan was doing with the funds that it was actively managing. In that case, based on a formal analysis that gives Magellan every benefit of the doubt, we get this picture that shows the active part of Magellan dropping by over 50% while the worst market-neutral hedge tracked by the University of Massachusetts falls by only 12%:

"Stansky's Monster" presents my dogged effort to figure out how Magellan's manager back then, Robert Stansky, could lose so much money. (It also provides an explanation of how one gets from the first graph to the second.)

Normally, data snooping (formerly known as "data mining" until a positive spin was placed on that term) is a bad thing in finance because it undermines the statistical significance of the reported results. In the case of "Stansky's Monster" it only reinforces the significance of the results because what I find is that there is no legitimate (or even semi-legitimate) analysis of Magellan's behavior that can explain where all the money went no matter how deeply I dig.

In Fidelity's official filings with the SEC, Robert Stansky claims that a combination of his industry and major stock selections accounts for his shortfall. For the fiscal year ending March 31, 2004, I find that they explain essentially none of it. Indeed, anyone with a Wal-Mart calculator and a passion for digging up numbers on the Internet could spot the holes in Stansky's alibi. Things work out a little better the following fiscal year that ends on March 31, 2005: Horrific stock selections—Stansky had an amazing ability to invest in companies that would become embroiled in scandal—do account for some, but not a majority, of Magellan's risk-adjusted shortfall relative to the S&P 500, but the majority of it remains missing in action.

Overall, the missing money comes to well in excess of $2 billion at a time when the fund had about $60 billion in assets under management. Unlike fictional mystery stories, there is no good answer to where to the money went for now. For one thing, I do not have access to Magellan's trading records—I have to get by with whatever I can squeeze out of the net asset values that Magellan publishes each day and the portfolios that it reports to the SEC, which presumably contain some amount of window dressing. Furthermore, if Fidelity "manipulated" Magellan's assets, those manipulations would likely inject statistical noise into their net asset values and make the job of squeezing information out of those numbers more difficult.

The explanation that I provide is that in the process of doing the shadow indexing, Robert Stansky turned Magellan into a monster that ultimately did in its creator by costing him his job.
If you want to find out the gory details, you will have to read the paper for yourself. I get the feeling, however, that new theories will be forthcoming in the days ahead. The full story of Stansky's Monster has yet to be told.