measuring success in fixed income - health | aon · types of investments throughout the modern...

14
Consulting | Investment Consulting Measuring Success in Fixed Income Avoiding the Unwanted Risks and Costs when Investors confuse Alpha and Beta November 2012 Hewitt EnnisKnupp, An Aon Company Copyright Aon plc 2012

Upload: nguyenthien

Post on 05-Sep-2018

227 views

Category:

Documents


0 download

TRANSCRIPT

Consulting | Investment Consulting

Measuring Success in Fixed Income Avoiding the Unwanted Risks and Costs when Investors confuse Alpha and Beta

November 2012

Hewitt EnnisKnupp, An Aon Company Copyright Aon plc 2012

1

This is the first part of a multi-part series of white papers on fixed income that Hewitt EnnisKnupp is developing. In this piece, we address what many investment managers are doing in their portfolios, how to assess whether they are successful, and how to adjust their mandates to improve their ability to meet the plan’s needs. In part 2, we will address constructing customized fixed income strategies within the context of the investor’s objectives and the rest of the portfolio. In the final installment, we will address implementation within the framework of the recommendations in the first two parts.

Executive Summary Benchmarks serve several critical functions in portfolio management. Setting benchmarks poorly or

improperly interpreting performance results can lead to costly investment errors, such as unwanted risk exposures or bad decisions about hiring and firing managers. As a result, it is critical that investors take care in utilizing benchmarks.

Not all active risk (the difference between portfolio returns and the benchmark) is true alpha (skill-based excess returns). We found a significant amount of style bias in fixed income as well as unintended exposures with bottom-up managers. These make it difficult to evaluate performance against standard benchmarks or peer groups.

Customizing manager benchmarks and investment guidelines can improve the likelihood that the portfolio will meet the investor’s objectives as well as better employ the strengths of each manager.

When adjusting manager benchmarks and guidelines, it is important to have an open dialogue with the managers to ensure that reasonable expectations are set and all parties understand the investor’s goals.

Introduction Investors often measure investment success using benchmarks, which have been used for nearly all types of investments throughout the modern investing era. Despite their prevalence, benchmarks are often misused, manipulated and misinterpreted, which causes unintended costs and risks to investors. This white paper takes a close look at many of these issues for fixed income and describes ways to improve how benchmarks are designed and used to more effectively measure and manage portfolio exposures.

Consulting | Investment Consulting 3

Purposes of Benchmarks A well-defined set of benchmarks has several key functions, including:

Setting the strategic direction for the investment program

Giving guidance to investment managers on the types of investments to hold

Measuring the performance of investment managers

These purposes are clearly important, so benchmarks should not be taken lightly. For active managers, not all active risk (the difference between portfolio returns and the benchmark) is true alpha (skill-based excess returns). There are three main explanations for why active managers deviate from benchmarks:

Pursuing alpha. Managers take calculated, tactical positions to produce greater returns than the benchmark.

Style-based structural mismatch. Some managers systematically over- or under-weight certain sectors (usually those with spread) year after year as a style bias.

Unintended mismatch. Fixed income managers with a bottom-up approach often pick the bonds they like most within the allowed guidelines, and the resulting total portfolio may be different from the benchmark in unintended ways.

These three reasons for deviating from benchmarks are very different, though they are interrelated. The last two are often incorrectly interpreted as pursuing alpha, but it is important to understand the differences. For example, style-based structural mismatches often over-weight areas in which the managers believe their alpha-generating capabilities are strongest or the market returns are greatest – maybe a manager systematically over-weights mortgages because it has strong capabilities within that sector. Alternatively, a manager could create a structural mismatch by systematically over-weighting spread sectors year after year. This manager would probably perform better than the benchmark over a market cycle, but the strategy employed does not require skill; it can be pursued at a lower level of fees than most active managers charge. As another example, a bottom-up manager might unintentionally hold a portfolio with a different duration or sector allocation than the benchmark simply because these are the bonds it favors, though the manager isn’t actually trying to express an explicit view on a sector or macro factor. These outcomes are not skill-based or worthy of active management fees. While there are good managers who use such strategies, it is not because of this that we believe they are good managers.

For the most part, pursuing alpha is the main reason to hire active managers. This includes being able to avoid downgrades and defaults, rotate between sectors at the right times, and position the portfolio ahead of macroeconomic movements. True alpha represents skillful decisions, rather than systematic style bias or unintended benchmark mismatches.

Consulting | Investment Consulting 4

Effects of Misinterpreting Manager Performance There are significant costs and risks of incorrectly assessing manager performance, which is one reason it is so critical to properly understand the reasons that performance differs from the benchmark. The interpretation of manager performance can influence hiring and firing decisions, ultimately driving both transaction costs and future performance. The round-trip transaction costs of changing fixed income managers are typically around 0.54-0.86% for Barclay’s Aggregate Mandates, 0.24-0.36% for Barclay’s Long Government, and 1.40-1.72% for Barclay’s Long Credit1.

These transaction costs may be a worthwhile price to pay to exchange a bad manager for a good one; however, there is evidence that managers recently hired tend to underperform those recently terminated, suggesting that the managers being terminated are not less skilled than those hired. Researchers have found that plan sponsors tended to terminate managers after underperformance, but these managers often subsequently outperformed after being fired. Further, plan sponsors tended to hire managers after periods of outperformance, but these managers tended not to outperform after being hired.2 That is, even before considering transaction costs, plan sponsors that terminated managers and then hired new ones typically would have been better off not having made the change. One possible explanation is that managers are hired and fired based on performance against their benchmarks, which may be influenced by investment style (beta performance) more than skill (alpha expectations).

In addition to causing bad hiring and firing decisions, if the investor is not aware of structural biases, it could have more risks than it intended; many investors suffered in 2008 because of massive structural over-weights to spread sectors.

This highlights how critically important it is for investors to understand how their managers performed and why. To get a sharper lens on this, we analyzed historical returns from six major fixed income managers3 to examine the results for several types of benchmark mismatch.4

1 Based on Hewitt EnnisKnupp’s experience for transactions during 2011. Includes explicit transaction costs, bid-ask spreads, and market impacts of trading. 2 Amit Goyal and Sunil Wahal, “The Selection and Termination of Investment Management Firms by Plan Sponsors.” Journal of Finance. August 2008. 3 Although this paper does not explicitly name the managers, they are all relatively large in the fixed income area and have several different portfolio management styles. The analysis is based on ten years of historical monthly data when available, with adjustments as needed when less information was available. 4 The analysis in this white paper focused on the results for mandates against the Barclays Capital Long Government/Credit Index. We selected this index because it only had two major sectors, government and credit, so an analysis more clearly illuminates the implied positions of the managers. We also reviewed results for the Barclays Aggregate Index and found the conclusions to be consistent.

Consulting | Investment Consulting 5

Performance Mismatch 1 – Over-weight to Credit Spreads Many fixed income benchmarks are composites of government bonds and securities from spread sectors, such as corporate bonds. Ideally, a manager would over-weight the spread sectors when they outperform governments and under-weight them when they underperform. But the spread sectors are generally viewed to have a higher expected return—that is, they are expected to outperform over a full market cycle. As a result, some fixed income managers over-weight spread sectors more often than not.

Investment managers who rotate in and out of spread sectors at the right time are skilled, but over-weighting certain sectors all the time is a strategy that can be implemented passively by using a custom benchmark. We wanted to identify whether performance was due to skill, so we compared rolling 12-month excess returns for each of the six managers to a passive strategy that simply over-weights credit by the same amount at all times (“over-weight credit”).5 Exhibit 1 shows this passive portfolio compared to managers 3, 4, 5 and 6; it is visually clear that the excess returns for these managers move up and down with a strategy that passively over-weights credit. Exhibit 2 shows the same comparison against managers 1 and 2; the picture is very different, as there doesn’t seem to be as discernable of a pattern. That is, managers 1 and 2 seem to be actively managing the exposure to credit spreads, but managers 3, 4, 5 and 6 seem to have a style-based mismatch that simply over-weights credit all the time.

5 The passive portfolio perpetually over-weights credit by 20%. The benchmark generally contains about 40-55% credit over the analysis, so this portfolio varies within 60-75% credit.

6

While some investors may want a structural allocation with more credit than what is available in standard off-the-shelf fixed income benchmarks, this can be achieved by adjusting the benchmark without the need for managers to systematically over-weight credit in every economic environment.

Performance Mismatch 2 – Over-weight to Securitized Sectors Although the Barclay’s Capital Long Government/Credit Index does not include securitized sectors such as mortgage backed securities, some managers include them in their portfolios. Managers can generate alpha by (1) rotating in and out of the securitized sectors and (2) selecting the best securities from these sectors. For each of the six managers, Exhibit 3 shows the allocation to securitized sectors.6

We can see that all of the managers are rotating in and out of this sector to some extent, as all of the lines seem to be moving up and down. Manager 2, however, is an outlier in that its allocation never drops below 23%. Every other manager in this analysis has an allocation of less than 5% at some point during the ten-year period in the analysis, and most of the other managers had allocations below 10% for the majority of the period. It appears that manager 2 always over-weights securitized sectors—a style-based mismatch against its benchmark.

6 Some managers were unable to provide the sector allocation back to 1/1/2002, which is why there are gaps in the data for the early years of this exhibit.

7

Manager 2 may believe it has particularly strong capabilities in selecting the best securitized assets, so it meaningfully over-weights the sector every time. Given how much manager 2 emphasizes securitized sectors, most of its clients are probably aware of this style and appreciate it. While this may be an appropriate strategy for some investors, it is important for their clients to realize that much of this manager’s tracking error relative to its benchmark will be because of structural over- weight to securitized sectors. So its performance relative to the benchmark is highly dependant on factors unrelated to the manager’s investment skill—instead, its performance is highly dependant on whether securitized sectors are in or out of favor.

This makes it misleading to compare the manager to this benchmark, and may cause problems for the investor. The investor will not have a good metric for evaluating the results of the manager’s decisions. Possibly worse than having no metric, the benchmark investors are using is a misleading metric and could influence the investors to make bad decisions regarding retaining or terminating the manager. For this manager, investors may be better served by using a customized benchmark of government, credit and securitized sectors that reflects the manager’s style.

Nevertheless, we believe that managers with strengths in specific styles of investing can be extremely good additions to portfolios. They can contribute unique skills and diversification to the portfolio. Customized benchmarks can be more effective at getting the appropriate exposures than using an off-the-shelf benchmark and asking the managers to systematically deviate from it. In the case of manager 2, which has strength in securitized sectors, it could be given a benchmark that emphasizes these sectors. This would allow the manager’s alpha to be measured against a more appropriate hurdle.

8

Performance Mismatch 3 – Duration For each of the six managers, Exhibit 4 shows how the portfolio duration compares to the benchmark duration.7 Most of the managers have durations that seem to remain relatively close to the benchmark, sometimes above and sometimes below. Manager 2 seems to be systematically below the benchmark, probably because the securitized sector has a shorter duration than the benchmark

Duration mismatch can be a source of excess return for managers. Managers can generate alpha by (1) accurately predicting whether rates will rise or fall and structuring their portfolio accordingly and (2) taking advantage of the slope in the yield curve to benefit from its term structure.

Though many managers exhibit similar patterns to these six, we found that some produced markedly different results. Exhibit 5 shows a similar analysis for two additional managers who were not included in the original analysis because they did not provide as long of a performance history. Manager 7 had a fairly steady duration, always about a year lower than the benchmark, while Manager 8 had wild swings in its duration, literally “off the chart” at some points. Why are they different from other managers?

7 Some managers were unable to provide the duration back to 1/1/2002, which is why there are gaps in the data for the early years of this exhibit.

9

Hewitt EnnisKnupp’s Global Investment Management (GIM) group has been meeting with and rating both of these managers for several years, and we believe we have a solid understanding of their investment processes. Both of these managers are bottom-up bond pickers. They select the bonds they like most from within the benchmark (and sometimes outside the benchmark), with a lesser focus on constructing a benchmark-aware portfolio. So the portfolios may inadvertently have a duration posture that is very different from the benchmark, even though the manager is not intending to express a view on interest rate changes.

For both of these managers—especially Manager 8—a large portion of the tracking error versus the benchmark could be due to unintended duration positions. One option these managers have is to use overlays to reduce the size of unintended duration positions. While there are advantages to bottom-up portfolio construction and there are several bottom-up managers we rate highly, it also requires careful consideration when evaluating performance.8

8 We also note that one of the managers in the main 6 we analyzed is viewed as a bottom-up bond-picker, but it did not appear to be to taking unintended interest rate positions of the same magnitude. We suspect that it may have better risk controls than managers 7 and 8.

Although this white paper focuses on fixed income, many of the same principles can be applied to other asset classes. For example, in the public equity markets, there are some managers with style biases toward growth or value. By giving them a broad benchmark that is inconsistent with their style, investors often find themselves praising the managers for “good” performance or explaining away “bad” performance (or terminating the manager), depending more on what part of the growth/value cycle we are in than how much value was added due to the manager’s skill. A custom benchmark would facilitate more meaningful performance evaluations.

10

Performance Mismatch 4 – Security Selection Many fixed income managers seek alpha through security selection. Security selection can be used to outperform the benchmark even if the portfolio matches the benchmark’s average credit quality, sector allocation, and duration. Pure security selection is almost exclusively an alpha pursuit, as it does not change any of the beta characteristics of the portfolio. While managers often over-weight sectors in which their skill in security selection is greatest, it’s important to differentiate performance due to security selection from that due to sector allocation. If the investor’s goal is to maximize the opportunity for alpha from security selection, then it should rework the sector allocation in the benchmark to help the manager implement this goal.

Ways Forward: Possible Plans for Plan Sponsors Benchmarks themselves don’t directly change portfolio performance, but they provide information that influences decisions about the portfolio. Investors may want to consider the following actions to reduce the likelihood that benchmarks will be misused:

Customize benchmarks There are some situations in which investors would benefit from using customized benchmarks relative to a standard index, including:

When a manager is more skilled in some aspects of portfolio management than others. For example, a manager whose strongest skill is security selection may have challenges exercising that skill with a benchmark containing a lot of government bonds. Consider separating out the government bond component of the allocation and having it indexed, giving the active manager a benchmark exclusively based on its strength.

When there are structural exposures in a benchmark that are not optimal for an investor. For example, an investor might want a strategic allocation weighted differently than the benchmark allocations to government, credit, and securitized sectors. In the second part of this white paper series, we will discuss how customized benchmarks can be used to help the investor achieve the desired risk exposures as well as provide a more legitimate basis for performance evaluation.

Customized benchmarks do make it more difficult to do peer group benchmarking of managers. However, given so many managers with styles biases, it has always been challenging to use peer group benchmarking to develop actionable conclusions. We believe it would be more effective to give managers better benchmarks and measure their performance against those benchmarks, rather than using standard benchmarks for managers with different styles and have more peer group benchmarking data.

There are some disadvantages to customized benchmarks, such as reduced ability to use pooled funds and potentially marginal increases in cost due to reduced efficiencies for the managers. Nevertheless, in many cases, the advantages could make customizing the benchmarks worthwhile.

11

Customize manager guidelines This can make sure each manager is able to implement a mandate consistent with its styles and skills as well as prevent the investor from having unintended risk exposures. For example, a manager that is viewed as skilled at rotating between sectors might be given guidelines with a lot of latitude to rotate among sectors, whereas a manager that is not strong in that area might be required to stay close to the sector allocation in the benchmark.

As a more detailed example, earlier in the paper, we observed that some managers appear to over-weight credit year after year. While this implies that these managers are not true sector rotators, they might have done this because they were good at security selection within the credit market. For them, custom guidelines might assign a benchmark that is over-weight credit (if this is where their security selection skills are), but prevent them from tilting even further into credit. With these customized guidelines, the investor would have greater knowledge and control over what the manager is doing without preventing the manager from actively managing the portfolio in the areas it is skilled. Further, the manager’s performance relative to the benchmark is much more meaningful. The fixed income portfolio can be rounded out with other managers with different strengths to make sure the total portfolio has an appropriate weight to credit. In this way, customizing guidelines to the manager’s skills and the investor needs can help insure that managers are able to use their full skill set to generate excess performance without creating unwanted exposures.

Consider performance thoughtfully Bad performance doesn’t necessarily mean that a manager should be terminated, and good performance doesn’t necessarily mean that a manager should be retained. Understand what decisions drove past performance and how that should influence views on expected future performance.

Alpha Expectations For investors who take these actions, one final step is for them to adjust their expectations of manager alpha. For some managers, a significant amount of their long-term performance relative to their benchmarks has been due to over-weighting spread sectors. By adjusting their benchmarks and guidelines, investors are essentially reclassifying some of that return from “alpha outperformance” to “strategic investment policy return.” Even if the investments in the portfolio have not changed, the hurdle for some managers would be raised to a more appropriate level. This will help investors better understand the true value they are getting from their actively managed fixed income.

Manager Reactions We expect managers to have mixed reactions about customized benchmarks and guidelines. Some may have previously found comfort in the ease of beating a benchmark by investing in less government bonds than is in the benchmark. Others may have appreciated the opacity of being able to attribute underperformance to their style, rather than active decisions they made. However, some managers may appreciate benchmarks that are customized to their styles, which allows their true performance to be seen more clearly and reduces the need to explain underperformance when their style is out of favor.

12

In order to make these changes successfully, investors should make sure the benchmarks reflect the manager’s strengths and style and the guidelines reflect the opportunity set the manager should be allowed to exploit. Further, the investor should work with the manager to develop expectations for tracking error against the benchmark so there is mutual agreement of the level of active share.

Conclusion Benchmarks are a necessary and critical piece of most investment programs, and the nuances of selecting them and interpreting performance should not be lost. There can be many challenges inherent in the use benchmarks, but we’ve found that the biggest problems often occur when investors misuse them: picking the wrong benchmarks, interpreting them poorly, and hiring and firing managers for the wrong reasons.

We recommend investors consider customizing benchmarks for their plan needs and manager styles and customizing manager guidelines to give active managers the appropriate opportunity set for their styles and skills. Above all, investors should always be thoughtful and ask “why” as they evaluate performance.

13

Contact Information Eric Friedman, Chicago, IL Hewitt EnnisKnupp +1.312.715.2973 [email protected] Francois Otieno, Chicago, IL Hewitt EnnisKnupp +1.312.715.3344 [email protected] John Geissinger, Norwalk, CT Hewitt EnnisKnupp +1.203.523.8898 [email protected]

14

About Hewitt EnnisKnupp Hewitt EnnisKnupp, Inc., an Aon plc company (NYSE: AON), provides investment consulting services to over 450 clients in North America with total client assets of over $2 trillion. More than 270 investment consulting professionals in the U.S. advise institutional investors such as corporations, public organizations, union associations, health systems, endowment, and foundations with investments ranging from $3 million to $700 billion. For more information, please visit www.hewittennisknupp.com.

About Aon Hewitt Aon Hewitt is the global leader in human resource solutions. The company partners with organizations to solve their most complex benefits, talent and related financial challenges, and improve business performance. Aon Hewitt designs, implements, communicates, and administers a wide range of human capital, retirement, investment management, health care, compensation, and talent management strategies. With more than 29,000 professionals in 90 countries, Aon Hewitt makes the world a better place to work for clients and their employees. For more information on Aon Hewitt, please visit www.aonhewitt.com.