Investment style performance has been the most divergent that we have ever observed. Any comparison of a manager's recent performance with the market's performance may have nothing to do with the manager's skill and everything to do with the manager's style. Performance evaluation should be about skill, not style.
Performance comparisons relative to a market index, e.g., the Standard & Poor's 500, are fraught with danger because of systematic risks inherent in a manager's portfolio. This is particularly true recently.
In place of market indexes, many fund sponsors during the 1990s adopted generic investment style indexes for performance comparisons, such as the S&P/Barra Value or the Russell 1000 Growth.
We've never supported the use of these style indexes as benchmarks to evaluate manager performance. The main reason is that the universe of stocks a manager researches and invests in can be materially different from the universe of stocks that comprise a style index. A manager's benchmark should comprise securities about which the manager has an informed opinion.
In a broad market in which performance of individual securities is relatively similar, the security composition of different benchmarks might not have a major impact on performance. However, in a narrow market like the one we've seen recently, in which a limited number of stocks account for an abnormal amount of the market's performance, security composition is a very important matter.
The return dispersion observed in the past two years was unusually large among the major styles.
From 1976 through 1999, the average annual difference between large value and large growth has been around 13 percentage points a year. In 1999, it was 55.75 points, the maximum annual difference ever and more than four times the average.
Likewise the average difference between small value and small growth has averaged around 15 percentage points. In 1999, the difference was 60.63 percentage points, the highest ever and also more than four times the average.
If a portfolio had a slight value orientation relative to a benchmark or market index, the resulting underperformance during 1999 would have been four times as great as expected based on historical experience.
These extreme results offer credence to the argument that managers be evaluated relative to their investment styles. Performance relative to "the market," or generic style indexes, is confused by style bias and true value-added skills are overshadowed.
Consider the performance of the following published large value index results:
R&T Large Value -11.63%
S&P/Barra Value 12.72
Russell 1000 Value 7.35
The differences in the performance arise because the processes of building these three indexes are very different. The R&T is built from a universe of 750 stocks. The S&P/Barra comes from the S&P 500. The Russell is built from the 1,000 largest stocks. Also, the R&T seeks to create pure value and growth indexes and does not include all securities in either a value or growth index. The S&P/Barra and Russell include all stocks, but S&P/Barra divides the stocks exclusively, while the same stock can be included in both the Russell value and growth indexes.
An investment manager's decision to use the most simple and easily understood index as a benchmark might be completely inappropriate and very painful if the manager does not have opinions about all of the stocks in the index. The manager may have underperformed either the S&P/Barra or the Russell index in the past two years, simply because his definition of value stocks is different from that of the companies that build the indexes. Underperformance relative to standard indexes says nothing about a manager's ability to select stocks within his realm of expertise.
To study this idea further, we looked at the breadth (or conversely, the narrowness) of the market during the past several years. We compared how much individual stock returns were contributing to overall market returns in recent years vs. history. The analysis is based on the R&T 2500, a market-cap-weight index of the largest 2,500 stocks traded by institutional domestic equity managers.
A key concept to help us understand the narrowness of the market is the performance impact of each stock in the market.
In 1999 the return of Microsoft Corp., the largest stock in the index, was 68.36%, while the R&T 2500 returned 20.18%. Microsoft's return had the largest positive impact on the R&T 2500. Without Microsoft, the R&T 2500 would have returned 18.86%. Eliminating the 10 top impact stocks from the R&T 2500 would have dropped its return to 11.19%. By eliminating the top 50 stocks with the greatest positive impact on the R&T 2500's performance, the return would have been a mere 1.3%. In 1998, a similar phenomenon occurred.
To put this in a historical perspective, we performed the same analysis for all years between 1976 and 1999. For each year, we removed each of the 50 largest contributing stocks from the index and calculated the performance of the index without them. As you can see from the accompanying chart, the slopes of the declines in returns were much more sudden and sharp in 1998 and 1999 than the slopes in the prior years.
In all years between 1976 and 1997, the impacts of individual stock were much smaller. The analysis illustrates the vast majority of the 20%-plus rates of return can be attributed to only a few key stocks in 1998 and 1999. These have been the most narrow markets in the 24 years we were able to analyze.
The upshot is: Every active manager has a universe, or list, of stocks he actively researches. The manager will hold either a positive or negative opinion about each stock in this list. This list of stocks should form the basis of a fair and appropriate benchmark to measure a manager's investment skill.
If stocks like Microsoft are being actively researched, then the manager should be held accountable. If not, then the manager shouldn't be. Popular style indexes contain numerous stocks not often part of a manager's research universe. It is very easy for fund sponsors to make an error in evaluating managers as a result of the recent domestic equity market environment. Good benchmarks allow fund sponsors to avoid the costs of firing managers unnecessarily and increase the probability of a successful investment program.
Thomas M. Richards is principal and Sandra K. Weiskirch is director-investment analytics at Richards & Tierney Inc., Chicago. This commentary is adapted from a lengthier analysis produced by the firm.