New research says stock selection is what matters

CAMBRIDGE, Mass. - Two researchers are challenging the conventional wisdom that asset allocation is a far more important investment decision than stock selection.

"Contrary to the widely held view, it turns out that choosing stocks within the equity component of a portfolio is substantially more important than choosing a portfolio's exposure among stocks, bonds and cash," wrote Mark Kritzman, managing partner of Windham Capital Management Boston LLC, and Sebastien Page. Mr. Page is senior associate at State Street Associates, a research venture of State Street Global Advisors, Windham and FDO Partners LLC. Both Windham and State Street Associates are based in Cambridge, Mass.

Normally, the news that stock-picking beats asset allocation by a mile might change the way thousands of investment professionals would go about making investment decisions.

The problem, according to some finance academics, is that Messrs. Kritzman and Page unintentionally fixed the race.

A flaw in their methodology "gives you asset allocators who aren't doing any asset allocation," said John Y. Campbell, a partner with Arrowstreet Capital LP and the Otto Eckstein Professor of Applied Economics at Harvard University, both in Cambridge.

"I think they've handcuffed the asset allocator," echoed Roger Ibbotson, chairman of Ibbotson Associates Inc., Chicago, and a professor at the Yale School of Management, New Haven, Conn. "You just can't make any conclusions because (the study is) so dependent upon how they did their methodology."

If the critics are right, the results of the research might just be a flash in the pan.

But Mr. Kritzman, research director of the Association for Investment Management and Research's research arm, is a highly regarded quantitative manager.

Plus, the research is being published in some of the most prominent U.K. and U.S. academic journals.

The first of two papers, which looks at the tradeoffs in importance between asset allocation and security selection, appeared last year in the British Journal of Asset Management. The second paper, which expands the work to include country allocation, global and country sector selections, is slated for publication in the summer issue of the Journal of Portfolio Management.

Ever since Gary P. Brinson, L. Randolph Hood and Gilbert L. Beebower published their landmark 1986 study showing that asset allocation accounts for 93.6% of variability of total returns, industry experts have widely touted the importance of selecting the right asset classes over security selection. An updated study put the figure at 91.5%.

In recent years, however, other experts have argued that those studies have been misinterpreted to suggest asset allocation explains more than 90% of a fund's returns.

A 1997 article by William Jahnke, founder of Comprehensive Wealth Management LLC, Orinda, Calif., and an early pioneer in Wells Fargo Bank's trust department, argued that consultants and financial advisers were misusing the study to overstate the importance of asset allocation - and hence to exaggerate their own role.

What the Brinson-Hood-Beebower study shows is how much of the volatility of returns comes from asset allocation, said Mr. Ibbotson. But that's "not an answer to what anybody ever asks," he said.

"What most people ask is what percent of my return comes from asset allocation," he added. "The answer to that question is about 100% because, on average, stock selection doesn't add any extra return."

The more interesting question is how much asset allocation explains differences in returns among funds, Mr. Ibbotson said. In a study published in the Financial Analysts Journal in 2000, Mr. Ibbotson and Paul D. Kaplan, now director of research at Morningstar Inc., Chicago, found that policy differences accounted for 40% of the variability in returns for 94 balanced mutual funds and 35% of the variability in returns for 58 pension funds.

Messrs. Kritzman and Page took an entirely different approach from that used in the previous studies.

Instead of examining actual portfolios, they looked at the variation in returns across hypothetical portfolios. "Which choices should investors emphasize if they are skillful, or equivalently, which choices should they avoid if they lack skill?" they asked in the second paper.

They did this using a "bootstrap" methodology. From a universe of 1,512 global securities, they randomly picked 100 stocks, calculating the annual return for each one. They repeated the procedure 10,000 times, creating 10,000 portfolios for each year. For each portfolio, they also weighted each stock return by the stock's relative market capitalization, and held the overall asset mix at 60% stocks, 30% bonds and 10% cash.

Then they repeated these steps for each year from 1987 through 2001.

The researchers applied similar bootstraps to calculating variability in returns for global asset allocation, country allocation and global sector allocation.

The upshot: Security selection provided more dispersion of returns than the other categories. In order of importance, the other important factors were country sector allocation, country selection, global sector allocation and - ranked dead last - global asset allocation.

In a swipe at another of Mr. Brinson's most cherished theses, Messrs. Kritzman and Page wrote that they are unconvinced that global sector allocation has become more important than country allocation. When "new economy" stocks - defined as information technology, telecommunications and media - are taken out of the universe, the dispersion of results between country and global sector allocation shrinks drastically, they found.

Mr. Brinson, who is now president of GP Brinson Investments, Chicago, said the authors are "wrong." Mr. Brinson said they confused currency effects with country effects. "That's a big mistake."

Using an options-pricing model, the researchers found that a top-quartile stock picker is nearly four times as valuable as a top-quartile asset allocator.

In practice, however, stock pickers demonstrate a far more narrow dispersion of returns than Messrs. Kritzman and Page's hypothetical model shows, Mr. Kritzman said during a presentation at an AIMR meeting.

This leads them to conclude that money managers pick stocks from a much smaller subset of the stock universe because they are worried about business risk.

Investors, Mr. Kritzman said, "focus very intensely on relative performance," and their risk expectations are based on fear of losing money at the end of an investment horizon rather than throughout the period.

However, that doesn't mean the authors think asset allocation is a waste of time. To the contrary: Their findings show that stock pickers have to pass a much tougher hurdle to add value.

While an asset allocator "should expect to outperform the average asset allocation portfolio by 1.15 (percentage points) per year in order to be 75% confident of adding 100 basis points to average performance," a stock picker should expect to outperform the average portfolio by 1.71 percentage points, they wrote.

The bottom line: If pension funds cannot pick managers with sufficient skill, "the inescapable conclusion of our analysis is that most investors should focus on asset allocation, because it is the least demanding investment choice," they wrote.

There are two key problems with the papers, explained Mr. Campbell, the Arrowstreet partner and Harvard professor.

For one thing, "people are not trying to make random bets; they are trying to make systematic bets." Secondly, the researchers have constrained the asset allocator by limiting the change in asset mix to one percentage point from the 60% stocks, 30% bond and 10% cash base portfolio. Repeating the experiment with a mix of 60% stocks and 40% bonds, Mr. Campbell found that 95% of the time, the asset allocator will have an equity allocation between 57% and 63%.

"(Mr.) Kritzman has created a robot asset allocator and is simulating the performance of that robot. What is obscured is this robot is very close to a fixed-weight index," Mr. Campbell said. "I would argue that (Mr.) Kritzman has assumed his conclusion."

Rob Arnott, chairman of First Quadrant LP, Pasadena, Calif., an asset-allocation firm, agreed that tactical asset allocators make much bigger bets in practice. He said if the asset allocator "is allowed to make bets twice as large, the dispersion of returns would be twice as wide."

In response, Mr. Kritzman acknowledged the asset allocator was greatly constrained, but he doubted whether allowing the asset mix changes of five or 10 percentage points "would make that big a difference."