UNLOCKING EFFICIENCY VIA MICRO ATTRIBUTION

Turning Average Funds Into World Class Teams

GOALS > STRATEGY > FUNDS > TEAM > PERFORMANCE FACTORS

Our “Team of Funds” approach is an innovation that provides truly customized solutions rooted in each client’s financial goals and risk tolerance. By selecting funds using “alpha diversification,” we maximize the value of risk management, producing higher active portfolio returns with only a fraction of the active risk of the individual funds held in the portfolio. This increases the likelihood of delivering strong alpha, even in downside situations. While this claim can rarely be made for individual funds, we show that this is a reasonable expectation for our teams.

By applying our risk attribution methodology, we show how alpha diversification produces efficiency through the interactions of the active processes at work across the funds in the portfolio. As we noted in our earlier papers, portfolios become efficient when their components contribute equally to return and risk. This depends on maximizing the value of these active interactions, which is a key feature of our team approach, along with our customization of the fund teams within each client’s asset strategy and alpha goal.

WHEN THE WHOLE IS MUCH GREATER THAN THE SUM OF ITS PARTS

Our portfolios produced total returns that exceeded their benchmarks by their target alpha, delivering these higher returns with slightly less volatility than their benchmarks. We had concluded that the active process was a source of total return efficiency. This perspective took the innovative approach of considering active efficiency in the context of the portfolio’s total risk. Our goal is now to identify and demonstrate the sources of that active efficiency.

Active efficiency is often evaluated using the Information Ratio (“IR”) which measures excess return relative to the volatility of that excess return. We substitute “alpha” for excess return, because we take into consideration the amount of volatility in each fund. We do not want to reward those funds that earn a small extra return at the cost of substantially more volatility risk; neither do we wish to punish those funds that underperform by a modest amount while reducing risk significantly. To correct these errors in the traditional approach (which ignores volatility risk) we compare each fund’s return to its “required rate of return” (given its volatility.) Our Information Ratio measures active performance relative to each fund’s “risk-adjusted benchmark” and then considering its true alpha relative to its active volatility.

We measured the Information Ratios for the funds used in our set of 24 portfolios. (These were created using 4 asset allocations with 6 alpha targets for each strategy.) We found that the average fund’s Information Ratio was 0.31; this is a typical value found in many funds. Quite frankly, this value does not provide statistical confidence in terms of delivering excess return. We could not conclude that these funds – in isolation – are likely to deliver alpha.

Enter alpha diversification and the team approach to portfolio construction…

The average information ratio across our set of portfolios was 1.23; this essentially quadrupled the information ratio of the portfolios’ component funds. This “Information Ratio Multiplier” is a new standard in quantifying the benefit of effective portfolio construction and evaluation.

HOW ALPHA DIVERSIFICATION DRIVES HIGHER INFORMATION RATIOS

Diversification is a bedrock principle of investing and risk management, and it is the cornerstone of the mean-variance optimization processes that are foundational to the asset allocation process. While individual asset classes are volatile, when combined in a portfolio, some of their individual volatility is reduced. Why? The key is finding assets where the correlations between all pairings tend to be low. We observe that in addition to finding low-correlation assets, increasing the number of assets in a portfolio, and removing concentrations also work to lower the overall volatility of the portfolio.

The key driver in lowering volatility risk is correlation, and this can be applied to active returns in the context of assembling a team of funds. Because alpha is an idiosyncratic effect (rather than a systematic market effect) the correlation of alpha between funds is quite low. This means that a smaller fraction of the tracking error of the funds remains in the portfolio. Our team approach maximizes alpha diversification and removes a significant portion of the funds’ individual tracking error.

On average, the tracking error of the funds used in our portfolio was about 375 bps, while the average tracking error of our portfolios was only about 90 bps, indicating a reduction of active risk of more than 75 percent. We had noted in our earlier paper that the correlation of alpha to portfolio total return was negative, which caused the portfolio’s tracking error to reduce total return volatility. This was a double benefit from our active process.

These findings demonstrate the superiority of the team approach, as compared with the traditional approach to fund selection and evaluation – where assets are viewed in isolation, relative to their benchmarks. That myopic view ignores the substantial opportunity to reduce active risk using diversification where it works best – around active returns.

ALPHA TARGET + INFORMATION RATIO: PATH TO HIGH-CONFIDENCE ALPHA

We introduce the concept of “Performance Factor Attribution” to identify the sources of a high-confidence minimum alpha. This analysis brings together alpha target and Information Ratio as the drivers of a higher minimum expected alpha.

This chart shows the alpha at the lower 95% confidence band (using the standard calculation of subtracting 1.96 standard errors from the expected alpha.) Our alpha diversification process minimizes the tracking error that drives the standard error. This results in less of a “haircut” from the expected alpha earned by the team of funds.

We see a clear relationship between the target alpha and the Information Ratio. We also see improving efficiency as we move to higher alpha targets. Combining higher alpha targets with increasingly-high Information Ratios provides two substantial benefits: higher minimum alpha and the delivery of a higher percentage of the expected alpha. The proportion of expected alpha delivered at 95% confidence is only about 20% at the lowest alpha target. This improves dramatically at higher alpha targets, with the minimum alpha reaching above 60% of expected alpha.

We can observe the powerful effects from alpha diversification, by comparing the 95% confidence minimum alpha of the portfolios with the average minimum alpha for their funds. Only 7 of the 38 funds used by our portfolios delivered positive alpha at a 95% confidence level, and so none of the averages of the individual funds delivered high-confidence alpha.

What turned these “losing” funds into “winning” teams? It was our team approach to portfolio construction!

EVALUATING ALPHA EFFICIENCY USING MICRO ATTRIBUTION

Our efficient portfolios are the result of equal contributions to active return and risk from the assets held in these portfolios.

Each portfolio’s funds were selected to deliver the allocations in each of the market segments within the four asset allocations. In aggregate, these funds also had to deliver the target alpha with a minimum of tracking error. This was equivalent to optimizing to the highest Information Ratio at each target alpha within each strategy.

We calculated the percentage contribution to alpha and to tracking error for each fund held in the portfolio. This allows us to evaluate each fund’s contributions, and then summarize these results to create a hierarchy of performance: from fund to style segments, to minor and major asset segments, and finally to asset classes. This also helps us to evaluate the sources of active diversification, which is driven by the interaction of the excess returns between all the funds. This provides a practical view of the complementary nature of the funds, and the optimization process which incorporates several aspects of risk:

• market exposure within an asset segment (i.e. “benchmark misfit risk”)
• periodic (and complementary) underperformance vs outperformance
• individual tracking error
• active diversification (i.e., low correlation) across all fund pairings

This chart illustrates four strategies, each aiming at an alpha target of 100 basis points. These portfolios hold about 23 funds across 11 style segments. Over 75% of these funds appear to be reasonably efficient, as they remain within 5% of the line of perfect efficiency. This appears to be true across all four asset strategies. We also see that that the stronger inefficiencies (10% - 15%) are offset with a set of “super-efficiencies.”

IN-DEPTH EVALUATION OF A SINGLE PORTFOLIO

We chose the 60-40 strategy with an alpha target of 100 bps as our representative portfolio. Our initial view focuses on the funds in the context of their contributions to active efficiency. The critical takeaway is that this enables us to immediately identify the few “outliers” that are substantial contributors or detractors of efficiency.

From the standpoint of the overall portfolio manager (“OCIO”) or the investment committee or the client, this analysis provides critical insights that become part of the investment decision process. In this case, we reviewed almost two dozen funds and quickly found three that should be given additional scrutiny.

This chart allows us to easily evaluate relative performance by viewing each fund’s position relative to the “Efficiency Line.” (We calculate the efficiency of each fund as its percent contribution to active return minus its contribution to active risk.)

The chart of fund efficiency provides insight into the active diversification and the complementary performance across all the funds. Every portfolio has its “winners and losers” and this exhibit identifies any substantial “weak links,” while also showing the pairings of funds of inefficient and “super-efficient” funds, which helps to smooth out both active losses while also lowering active return volatility. It is immediately clear that one of the large growth funds and one of the foreign value funds are the most significant detractors of efficiency, followed by one of the high yield funds. We also see how segments of US equity and the high-quality bond funds make up for these and other inefficiencies in the portfolio.

We must consider that even these “underperforming” funds do contribute to the overall interaction between the set of funds that make up this portfolio. From a practical perspective, these are the best funds that could be selected from the set of funds available on the company’s platform. These results may also encourage an effort to add funds to the platform in areas where the current offerings may not be optimal. These insights are all contributions to the investment process.

We find it helpful to view the efficiency metric across all funds. This is calculated by subtracting percent contribution to active risk from percent contribution to active return. This summarizes the individual characteristics presented on the prior decomposition of performance contributions.

Next, we present an attribution analysis of each fund’s contribution to portfolio efficiency. This shows the breakout of efficiency into its components: contribution to alpha vs contribution to tracking error. It is essential to moving beyond a naïve basis points approach and on to understanding the proportionality of these contributions to the overall portfolio.

SETTING CONTEXT: EVALUATING THE PERFORMANCE HIERARCHY

The portfolio’s equity was an inefficient contributor to active return, and this was offset by the super-efficiency of its bonds. This might be a surprising result, but our process explains that the equity alpha was more highly correlated to the portfolio’s total alpha, and therefore it contributes more of its individual active risk to the portfolio. One reason for this higher equity alpha correlation is simply that equity makes up most of the portfolio’s excess return.

We decomposed the efficiency of the major asset classes into their two components. This confirms the portfolio’s inefficiency that primarily comes from foreign developed equity, which is offset primarily by high quality bonds.

DRILLING DOWN INTO EFFICIENCY CONTRIBUTIONS

Our next step is to identify the drivers of efficiency within the major segments. This provided several insights:

• Emerging markets equity was quite efficient;
• Large Cap inefficiency was more than offset by the efficiencies in Mid Cap and Small Cap equity.

These offsetting effects confirm that our team approach is working! We are finding and including a set of funds that complement each other’s performance characteristics, as they contribute to delivering the asset allocation while providing a target alpha that helps the client meet their return goals and minimizing both active risk and total return volatility.

That is a “win!”

The deepest and most detailed structural analysis is at the “style” level. Our methodology allows each client to customize the asset groupings to best reflect their active decision process. This final chart confirms our performance evaluation: our three “efficiency detractors” were primarily offset by high quality bonds. The rest of the portfolio was reasonably efficient. Most importantly, “active diversification” from our team approach has been confirmed by these objective results. And best of all, these exhibits are compelling and easy-to-understand.

STEPPING BACKWARD: TRADITIONAL PERFORMANCE ATTRIBUTION

The current “state of the art” in performance attribution evaluates only the excess returns of asset segments in the portfolio, relative to their corresponding benchmark returns. This naïve approach ignores the volatility of both total return and active return, leaving the client with an often-incorrect measure of added return, with no information on the whether the active return was worth the active risk taken. This provides no insight into the sources of efficiency in the active process.

The only information traditional performance attribution provides is an accounting of the sources of excess return, as shown below. In our example, we see how this approach sets a common trap for investors who lack risk attribution: keeping your losers while firing your winners.

Sadly, ignoring active risk resulted in presenting the least-efficient segment in the portfolio (foreign equity) as one of its strong contributors to success. This is the opposite conclusion we saw from our risk-aware approach, where performance attribution is provided for both active return and active risk. As we have noted: “Risk Matters.”

SUMMARY: ALPHA DIVERSIFICATION AND THE TEAM APPROACH

Our portfolio construction methodology selects funds within the context of each client’s asset allocation and target active return. This ties the fund selection process to each client’s goals and risk tolerance, making this a unique approach and a genuine innovation within the investment process.

The drivers of success are rooted in the concept of efficiency, where every asset delivers an equal proportion of active return and risk. We identify active efficiency through a robust risk attribution process, which maximizes “alpha diversification” across all funds in the portfolio. The true strength of this process is rooted in our “team approach” to portfolio construction.

This stands in stark contrast to the traditional approach to performance attribution, which is an example of “not seeing the forest for the trees.” The typical performance attribution analysis provides a painstaking (and sometimes painful) level of detail in identifying sources of excess return, in the context of an arbitrary, one-size-fits-all set of asset groupings that often have little relationship to the investment manager’s decision process. Worse yet, this approach focuses solely on individual products, and is applied almost exclusively to equity. The attribution analysis for multi-asset, client portfolios (AKA “asset owner portfolios”) is rarely seen. When performance attribution analysis is provided, it is a one-sided accounting of excess return, with no insight into active risk. This pseudo-analysis often provides incorrect and sometimes opposite answers to the question of what really drives portfolio success.

What is next in our insights into portfolio construction?

Our analysis has been focused on a 100% active portfolio; it did not consider the role that passive investments could play in our “team of funds.” We have found that passive funds often play a constructive role in delivering efficient excess returns.

This may be a surprising insight. In our next paper we will begin with our fund platform, enhanced with passive products. This will give passive investments a chance to “earn their way” onto the fund team by proving their ability to make our existing solutions even more efficient.

Would it surprise you to find that we could deliver the same efficiency with a portfolio that is partly passive? Stay tuned.

.

.

.

Written in Partnership with Stephen Campisi