The potential benefits of diversification across asset classes to achieve an investment objective are well known. Constructing a portfolio diversified by market capitalization, geography and style can mitigate overall portfolio risk. But does the adage "Don't put all your eggs in one basket" also apply within a style category? In other words, does it make sense to combine multiple managers within, say, a portfolio's large cap value allocation? The answer, in a word, is "yes." Let's explore why, and how to put the concepts to practical use.

Some Types Of Portfolio Risk
There are many different types of risk associated with investment portfolios, including shortfall risk (the risk that a portfolio won't achieve a desired objective); systematic risk (or market risk that is not diversifiable); unsystematic risk (risk which is diversifiable, for example through increasing the number of portfolio holdings); standard deviation (which measures the variability of a portfolio's returns); and benchmark risk (often expressed through tracking error, which measures how closely a portfolio tracks its benchmark). For purposes of this article, we'll focus on benchmark risk, and ways to enhance the risk/return profile of a style category in relation to its benchmark.

Active Value And Sources Of Benchmark Risk
Active investment managers are continuously taking calculated risks in an effort to provide value over and above their portfolio's benchmark. Over time, good active managers must be able to demonstrate that they can in fact outperform the benchmarks while incorporating risk-control processes. If they can't, investors will choose another manager or a passive investment strategy.

There are of course several ways to deliver benchmark-beating returns, or active value, including (1) increasing the portfolio's allocation to cash when the market appears overvalued (market timing); (2) tilting the portfolio's allocations toward some factor (e.g., momentum) or style (e.g., growth/value, large cap/small cap) that is exhibiting strong performance characteristics; (3) overweighting sectors that are expected to perform best, and conversely, underweighting those with less positive outlooks (sector timing); and (4) selecting the individual securities with the best prospects (security selection).



Market Timing
Of these active bets a manager makes, market timing has proven to be the most difficult to consistently generate excess returns. A market timing decision is actually comprised of two decisions that both need to be correct in order to generate active value: the decision as to when to become conservative and allocate to cash, and a second decision about when to become aggressive again and put the assets back to work. Market timing is a macro-level tactical decision that many managers find difficult because they are so focused on the fundamentals of finding good securities, and less on overall market dynamics. It is for this reason that many financial advisors and pension consultants will give a manager little or no credit for any market timing ability the manager has previously demonstrated.

The other three means by which a manager can generate active value are closely related to the manager's investment strategy and process for identifying the best securities.

Factor Exposures
Maintaining risk factor exposures that are significantly out of line from the benchmark can likewise hurt performance. Risk factors are characteristics that have been statistically determined to drive performance of stocks, such as style (e.g., growth and value), market capitalization (e.g., large cap and small cap), relative strength and volatility. Every stock has a "risk factor fingerprint," meaning that each one has some exposure to each of the risk factors. Widely used risk models such as those developed by Barra (now owned by MSCI Inc.) and Northfield Information Services typically list 11-13 common risk factors.  As the risk factors for each of a portfolio's holdings are combined, the portfolio itself will exhibit certain risk characteristics. Generally, any variations between a portfolio's risk characteristics and those of the benchmark will result from the manager's security selection process, but sometimes a manager will try to overweight factors that are poised for a "run." For instance, as the economy begins to accelerate after a period of slow growth, cyclical companies tend to be strong relative performers, and a manager might overweight the stocks of those companies, increasing the portfolio's exposure to the value factor.

Sector Allocation
In terms of sector allocation, managers employing strong benchmark risk control processes will generally maintain sector weights that are relatively close to the benchmark's sector weights. There will usually be slight variances, but veering too far from the benchmark weights is a bet that can result in underperformance, even if the manager was able to select the right securities. For example, the sector with the largest representation in the Russell 1000 Value Index, which measures the performance of a group of large cap value stocks, is Financial Services, which recently accounted for more than 27% of the index's weight.  A manager who is unconvinced that the sector has seen its worst days may hold only a 17% allocation, meaning the portfolio is 10% underweight. However, if the Financial Services sector performs well on a relative basis, the portfolio's performance suffers. Even if the manager picks some of the underweighted sector's strong performers, the sector bet may be a drag on performance.

Security Selection
Security selection is perhaps the most important way active value can be generated: managers who can consistently demonstrate an ability to add value through picking the best relative performers at the individual security level are extremely valuable in constructing a multi-manager portfolio. The process for selecting individual securities is the essence of a manager's investment strategy; it ultimately defines the manager's long-term success.  Many investment consultants, including PMC, assign the greatest weight to the active value resulting from a manager's security selection abilities, because through combining managers we can neutralize the other risk factors and isolate active value from security selection. We'll discuss the perceived benefits of isolating security selection through combining managers below.

Neutralizing Risk Exposures, Isolating Security Selection
As far as investment consultants are concerned, in an ideal world investment managers would not try to time the market, they would maintain factor and sector exposures that are very close to the underlying benchmark, and they would generate 300 basis points of active value year in and year out solely through their security selection process. But unfortunately, that's not a realistic scenario. Managers with excellent security selection abilities but managing a more concentrated portfolio with, say, 40-50 holdings, may necessarily have factor and sector exposures that are at least somewhat different than the benchmark. Still other managers are good stock pickers, but have lax benchmark risk controls.

Table 1 below shows how combining managers in a portfolio can result in neutralized factor exposures.  According to the Northfield risk model, there are 11 fundamental factors which are the sources of a portfolio's active return. These include such metrics as Price/Earnings, Price/Book and Dividend Yield; EPS Growth Rate; and Market Cap, to name a few. The active exposures in the table are presented relative to the underlying benchmark, in this case the Russell 1000 Value Index. A widely used rule of thumb is that relative exposures of greater than +/- 0.20 are generally considered meaningful. For example, Table 1 shows Manager A having a +0.20 relative exposure to Price/Earnings, meaning that compared to the benchmark, Manager A owns higher P/E stocks. Manager B, on the other hand, owns stocks with a lower P/E on average than the benchmark, as exhibited by its -0.20 exposure. Manager C's weighted average P/E is relatively neutral to the benchmark.

One of the benefits that accrue from combining managers also comes to light in Table 1. Even though the managers individually exhibit meaningful exposures to certain factors, when blended together, the exposures can offset each other. For the Blend A-B-C, a portfolio equally weighted between the three managers, there are no meaningful exposures, meaning that risk factors should not be a key determinant of whether the blended portfolio outperforms or underperforms the benchmark. 


In a similar way, sector exposures can also be effectively neutralized by combining managers of the same asset class in an equal-weighted portfolio. Table 2 displays the average sector over- and underweights relative to the benchmark for a recent five-year period, with relative weights of +/- 5% highlighted. As is evident, Manager A tended to have meaningful sector exposures during this period, while Managers B and C managed sector weights somewhat closer to the benchmark. But each of the individual managers' relative sector exposures can be mitigated in a portfolio context. For example, Manager A's large underweight to Financials is offset by the neutral-to-slightly-overweight positions of Managers B and C, so that the Blend A-B-C portfolio has only a slight underweight to that particular sector. Across the board, the blend's sector weights are in line with the benchmark, meaning the portfolio's relative performance should not be driven by sector bets.


A "Free Lunch"?
The benefit of offsetting risk exposures by combining managers is perhaps most evident when assessing a portfolio's risk-adjusted returns.  Table 3 below shows portfolio characteristics for Managers A, B and C individually, as well as for Blend A-B-C and the benchmark.

During the five-year period ended 9/30/10, each of the managers outperformed the benchmark Russell 1000 Value Index, and Managers B and C did so with less volatility than the benchmark. Manager A delivered the highest return, but was more volatile than the benchmark and the other managers, and based on its tracking error of 9.0%, it is apparent that it was less concerned about tracking the benchmark, a characteristic confirmed by the factor and sector exposures in Tables 1 and 2. Managers B and C more closely adhered to the benchmark, but did not generate the same level of active returns as Manager A.

Information ratio is a widely used measure of how well a manager performs on a risk-adjusted basis, and is calculated by dividing the active return by the tracking error. The higher the ratio, the greater the active return a manager achieves for each unit of risk undertaken. All three of the managers individually generated impressive information ratios ranging from 0.37 to 0.64 during the period.


Blend A-B-C demonstrates excellent combined statistics, with solid active return and mitigated volatility. But the real power of diversification is apparent in the information ratio: Blend A-B-C's information ratio was actually higher than any of its constituent managers, meaning the blend generated a higher risk-adjusted return than the individual managers. In essence, diversification across managers within an asset class can result in a "free lunch."

Putting Concept Into Action
There are several ways to take advantage of these principles when constructing client portfolios on the Envestnet platform. First, Envestnet's Unified Managed Account (UMA) offers advisors the flexibility to combine multiple mutual funds or separate account managers within a given asset class, depending on the entitlements your firm has selected. For advisors who would prefer to not spend time selecting and blending managers on their own, the PMC Manager Blends can offer attractive alternatives constructed and managed by the PMC team. Finally, for more customized solutions, a PMC Investment Consultant can help you construct an optimal solution tailored to your client's specific needs.

Brandon Thomas is chief investment officer of Envestnet, one of the largest providers of wealth management solutions in the industry.