From the February 2007 issue of Investment Advisor • Subscribe!

February 1, 2007

Investing With Style

The authors present their vision of "adult" portfolio management

In a series of articles appearing in Investment Advisor (September 2005, January 2006, and February 2006), we criticized the market-cap/value-growth characteristic grid used for categorizing and evaluating equity managers. Among other things, we argued that keeping equity managers in a box hurts performance and provides few, if any, risk reduction benefits.

The information we and others present confirms a performance deficit that could be as high as 300 basis points, with risk reduction benefits one-seventh of that provided by traditional diversification alternatives. We present statistics that boxes (e.g., small-cap value) are not really asset classes and, in addition, selecting a fund based on its box designation provides no risk reduction benefits when compared to randomly selecting funds without regard to their box designation.

In a June 2006 Investment Advisor article, John Rekenthaler, VP of research and new products at Morningstar, challenged some of our tests, but agreed with our conclusions regarding the box problem. There is thus support for our proposition that boxes are a problem when it comes to categorizing managers.

What's an Advisor to Do?

The challenge for advisors is to find a box alternative that provides a way to categorize and evaluate equity managers that does not trip over itself with respect to investment performance. In this article, we present just such an alternative which harkens back to the original meaning of equity style. Style, we argue, is the way a manager goes about analyzing, buying, and selling stocks. Style is not what the manager ends up holding, but rather how the manager goes about making investment decisions. Thinking of equity style in this way, and categorizing and evaluating managers accordingly, solves many problems in the current system.

As a starting point, we believe that equity managers ought to tell investors how they go about making investment decisions and then stick to that style. So the first step in the new process is to ask managers how they manage money and, in turn, use the answer as the basis for categorizing and evaluating the manager.

We refer to this system as "adult" portfolio management, as compared to the current "playpen" portfolio management, in which a manager receives a box designation from an outside service and is then expected to stay in that box when making investment decisions.

The focus on box consistency is an unhealthy consequence of the current system. Successful managers cannot be both box and style consistent. They must choose one over the other. We believe, and our evidence backs us up, that style consistency is more important than box consistency.

A true style-based system, with a dozen or so well-defined styles and many associated elements, has a number of advantages when compared to the current boxed system. Most importantly, a manager, pursuing a true equity style rather than staying in a box, has the opportunity to deliver improved performance. This is because managers can focus on the most highly-rated stocks rather than being forced to limit themselves to less desirable picks within their box. As we show in our September 2005 IA article, forcing a style manager into a box seriously degrades the quality of her stock selections and, consequently, leads to performance deterioration.

Some have expressed concern about allowing managers such freedom and wonder whether this will lead to a manager simply doing whatever they wish without regard to the stated style. We have a similar concern, so for a style based system to have integrity, there must be a matching style audit that holds the manager accountable and thus assures the investor that their money is being managed as advertised.

Performance Evaluation

One of the greatest challenges facing both investors and managers is how to evaluate investment performance. Despite hundreds of indexes available for benchmarking performance, it is often the case that neither the manager nor the investor is happy with the resulting benchmark. We believe this is the result of a fundamental flaw in the current system.

Let's say that we live in a world in which people are categorized by physical characteristics, with hundreds of measurements provided for each person, such as height, weight, hair color, and so forth. In such a world, a company might request people within a particular height/weight range to apply for a job opening. After going through the selection process, a person is hired for the job. The performance benchmark for that person is based on the average for all people falling within the height/weight range requested.

It is not long before those within the company begin to feel something has gone wrong. The new employee might be performing well with respect to the benchmark averages, but is not accomplishing what is needed. This represents a particularly perplexing situation since, on the one hand, the benchmark comparison is flashing good performance, while, on the other, the results within the company are unacceptable.

Of course, the problem in this hypothetical example is that the initial categorization system, while very objective, fails to capture what is important for predicting employee performance. This is further compounded by creating a performance benchmark based upon largely irrelevant physical characteristics.

The obvious solution here is to categorize and evaluate employees based on those elements that are believed to predict job performance, such as education, aptitude, punctuality, motivation, communication skills, and others. While some of these are difficult to measure objectively, they are much better predictors of performance and represent a sound basis for constructing a job performance benchmark.

Equity manager performance is measured relative to a characteristic-based benchmark, such as small-cap value. Unless the manager's stated style is focused on picking the best small-cap value stocks, for example, then using a characteristic-based index as a performance benchmark makes no sense. The characteristic-based index does not capture the essence of the investment process, much as height generally tells us little about employee performance.

Instead, why not create the investment benchmark based on the performance of those managers pursuing the same style? We believe this represents a true peer comparison, much as one would compare the performance of a quarterback to other quarterbacks. Now ranking within the peer group makes sense, since managers pursuing the same style are being compared. If a manager is in the top quartile, this means that at least 75% of all managers pursuing the same style are performing worse.

Complementary Styles

It is widely believed that hiring an active equity manager necessitates investing in several active managers to achieve the desired level of portfolio risk. In the current system, boxes are used for this purpose, but boxes are not asset classes in the traditional sense. Instead we believe that the U.S. equity market is a single asset class in which managers pursue different equity styles.

In such a market, then, the goal is to create a portfolio comprising managers pursuing complementary styles; that is, styles that lead to the best overall risk/return portfolio performance. Our preliminary results reveal that style indexes are less correlated than are box indexes, and so diversifying across complementary styles produces greater risk reduction than does diversifying across boxes. Combining this with the already mentioned improved performance of style consistent managers suggests that complementary style portfolios produce superior risk/return performance.

Return Attribution

The typical return attribution approach is to determine what portion of the under- or over- performance is due to the portfolio's characteristics/sectors/industries weightings relative to a benchmark. On the contrary, a style-based system is focused on the manager's style and elements relative to the style benchmark. This provides the direct link between a manager's investment decision process and the resulting portfolio performance. Accordingly, future research along these lines by us and others will point to what is and is not important in the investment process and can be used to understand and adjust the investment process accordingly.

Some will argue that besides style attribution, returns should be adjusted for both characteristic (i.e., P/E ratio, market-cap) differences as well as sector/industry differences. Indeed, the P/E and small-cap return effects are well established in the research literature. We contend, however, that characteristic and sector/industry tilts are a residual of the investment process and are thus inseparable from the manager's style. We argue that return attribution should be based on style alone and ignore characteristic and sector/industry tilts. This is a contentious area, so we anticipate our theory may generate considerable controversy.

If an investor believes that a long-term characteristic or sector/industry tilt (e.g., tilt towards small-cap value stocks) is desirable, then that should be accomplished at the portfolio level. This can be done by combining complementary styles that, on average, are tilted as desired or by investing in an appropriate index fund or ETF. If the tilt is to be varied over time, however, we feel such changes should be left to the manager as part of the investment process. We not do believe that changing tilts over time can be successfully implemented at the overlay level.

0207STYLE.gif

Reducing System Inefficiency

Returning to our example of the company that hires based on the physical characteristics of its applicants, both the company and the applicants experience considerable frustration because the resulting candidate pool, while characteristically consistent, has a wide range of skills and abilities with respect to job requirements. Furthermore, applicants may chafe at being categorized based on physical characteristics when the strengths they bring to the job have little to do with such measures.

This leads to a great deal of inefficiency since the company must laboriously rework the applicant pool based on what really matters, while applicants are constantly describing their job skills to potential employers.

We contend that a style-based system eliminates system inefficiencies, allowing advisers and managers to devote more time to what really matters to investors. And, since the description of style emanates from the manager, the current plethora of fund categorization systems, one for each information service, simplifies into a single system used by everyone. We believe such standardization will lead to significant industry-wide efficiency gains.

Reprints Discuss this story
This is where the comments go.