SEI Investments describes the benefits of multi-manager style structured approach

What causes a portfolio to perform poorly relative to a benchmark? Investment style exposures explain most of the variations in equity portfolio returns. Investment styles have a tendency to outperform and underperform one another in no predictable pattern. For example, in the US equity market in 1991, large capitalisation growth stocks outperformed large capitalisation value stocks by more than 16%. In contrast, in 1993, large cap growth stocks underperformed large cap value stocks by 15%.

Many analysts, portfolio managers, and investment advisers believe they can predict which style will be in favour during a particular time period, but studies have shown time and time again that market timing and style timing are not consistent methods of generating excess return.

Equity styles tend to be uniquely volatile due to the fact that they are less diversified and have higher industry concentrations than the market as a whole. For example, equity value indices in the US tend to be dominated by utility, energy and financial stocks, whereas growth indices tend to be mainly technology, retail and health care stocks. Industry concentration will also vary by capitalisation, as is demonstrated by the fact that consumer staples account for approximately 20% of the large cap growth index, but less than 1% of the small gap growth index. Equity styles also differ significantly from one another in average weighted market capitalisation, price to book ratio, and dividend yield. The US large cap growth style has a weighted average market capitalisation of almost $40bn and a price to book ratio of approximately eight. The small cap value index has a weighted average market cap of less than $600m and a price-to-book ratio of approximately two.

Lastly, the correlation of returns between style indices is relatively low.For example, the correlation between the large cap growth and small cap value styles was less than 0.8 over the period 1980 to 1996. These style characteristics are unique to the equity market and can add an additional element of risk to a portfolio.

Style volatility can negatively impact an investor's return. For example, from 1989 to 1991growth outperformed value in the US equity market. As a result, in 1991many investors replaced their US value managers with US growth managers, only to see value outperform growth by a considerable amount in the following years. This movement in and out of various styles kept most investors from reaping the benefits of either one of the styles being in favour! The same pitfall affected managers. In 1990 and 1991, many value managers began 'drifting', or tilting their portfolios towards growth as growth outperformed value. As a result, in 1992 and 1993 they significantly underperformed their value benchmarks.

Due to the volatility with which various styles move in and out of favour, and the large differences in return, multiple manager portfolios should be diversified across all four US equity styles. To give one or more managers a broad US equity mandate is not optimal. If the manager favours a particular style, or drifts in search of the best performing style at the time, the portfolio will be exposed to a higher level of risk, which may not be rewarded by incremental return.

Additionally, managers who try to time styles or markets are more likely to underperform a benchmark. Specialists who adhere to a distinct investment discipline have an opportunity to hone their approach and to anticipate favourable changes within their area of the market. This focus provides the opportunity for greater consistency and predictability of results. Many managers, who roam about the market in search of 'good' opportunities, will end up missing most. All of this activity contributes to unneeded volatility in the portfolio.

A portfolio that is not style neutral is not only exposed to this style risk but also runs a considerable risk of being inconsistent with the asset allocation strategy. If the asset allocation strategy an investor had elected is not maintained, the risk and return balance will be compromised. This will most likely result in an unpleasant surprise for an investor over the long term. A style neutral US equity portfolio will have style characteristics that are significantly similar in its allocation to the Wilshire 5000. Style neutrality reduces the tracking error of a portfolio relative to the broad market benchmark.

Within each equity style, there should be appropriate allocations to each sub style. The allocations to individual managers within a particular style minimise deviations across various factors relative to the appropriate style index. These factors include such characteristics as size, book-to-market ratio, momentum and yield. Industry weightings are also considered. Specialist managers, because they are experts in one area, have an increased ability to consistently add alpha in one particular sub style.

The question then becomes, how do we determine allocations within a style for each money manager?

The relationship between alpha, tracking error and the information ratio can help to guide us in the construction of the portfolio. Alpha is the portfolio's return above the benchmark return. Tracking error is the variability of the difference between the returns and the benchmark returns. The more tightly controlled the returns are relative to an index, the smaller the tracking error. For example, the tracking error of an index fund should be fairly close to zero, whereas the tracking error of the average actively managed US equity fund is about four. Generally, the higher the alpha, the higher the tracking error and the greater the probability of underperformance over short time intervals.

Therefore, throughout the portfolio construction process, there needs to be a balance between alpha and tracking error. This balance will provide a reasonable expected excess return while reducing the risk of underperforming the benchmark. The optimal relationship between tracking error and alpha is best illustrated through the construction of an efficient frontier. For each level of alpha we minimise tracking error, or conversely, for each level of tracking error, we maximise alpha.

Each sub-style allocation should support the risk controlled portfolio construction model outlined above. A carefully structured portfolio combining higher alpha, higher tracking error managers with highly differentiated investment processes can result in a portfolio with a lower tracking error. For example, the relationship between alpha and tracking error for two equity value managers is shown in Figure 3. Manager A is a contrarian value manager, with extremely low price-to-book exposures, while Manager B is a relative yield manager, with significant high yield exposures. The combination of these managers in a large cap value fund, reduces tracking error because the correlation of their residual returns is very close to zero. This will provide a reasonable excess return while minimising the likelihood of under-performing our benchmark over reasonable periods of time. Thus, the contribution to tracking error is an important consideration in the manager selection process.

The relationship between alpha and tracking error is expressed through another important concept of risk-controlled portfolio construction called the information ratio. Information ratio is the ratio of a portfolio's alpha to its tracking error, or the ratio of excess return relative to a benchmark divided by the dispersion of returns relative to a benchmark. In a sense, information ratio is a risk-adjusted alpha. The larger is the information ratio, the more confident we are that a portfolio will outperform a benchmark in any particular period.

Once the portfolio design has been determined, the actual selection of managers begins. SEI employs a three-step approach to its manag er selection process. Faced with a universe of thousands of US equity managers, SEI needed to develop an efficient way to identify style-specific managers. This is achieved by using a returns-based analysis. A regression analysis done on the portfolio manager's returns versus a benchmark's returns is used to determine which style bias the manager's portfolio exhibited at a given point in time. Looking at this information for at least 16 quarters, we can track the degree of 'style drift' that has occurred within a particular portfolio. This will help us to assess if that manager can be considered style specific. The returns based analysis can also help us attribute manager skill relative to a benchmark. For example, by looking at the regression analysis, we can see if a manager's out-performance is due either to style timing or due to stock selection abilities in a within a given discipline. This returns based analysis allows us to eliminate a significant number of managers early on in the selection process.

The second step of SEI's process is a fundamental or 'holdings based' analysis. Examining monthly portfolio holdings for each money manager helps us to further dissect the attributes of that manager's return. The analysis measures contributions to return from such factors as size, valuation, and momentum and enables the analyst to isolate out the security selection skills of the manager. Above all, we are looking for those managers whose return can be attributed to pure stock selection abilities.

The final step is a qualitative analysis and leads us to the actual selection of a manager. The qualitative assessment involves multiple conference calls and on-site visits with a prospective manager. Due to the significant amount of research that is done up-front prior to meeting with the manager in-person, the on-site visit is primarily intended to confirm the findings of the research teams.

Once selected, it is critical that a rigorous monitoring practice is implemented. Because performance is a lagging indicator, trends that may cause problems within a management firm need to be detected prior to the returns being affected. At SEI, a manager's trading activity is monitored daily for adherence to the mandated style. In addition to monitoring, weekly, monthly and quarterly performance review meetings are held with each manager.

Overall, a style-structured portfolio consists of multiple styles and multiple managers within each style. Each manager is style specific, and plays a particular, well-defined role, within the total portfolio. Each manager is expected to add value through stock selection skills, and not style drift. The total portfolio is constructed to be style neutral, and to have exposures similar to the broad market index. Such a portfolio will support the asset allocation policy of each investor, while allowing each manager to contribute excess returns within a risk controlled framework.

Joseph Ujobai is SEI's market manager for Europe, based in London

Topics