Does Active Fixed Income Management Add Value.pdf

This article reviews the monthly performance of 43 Canadian fixed income funds over a ten-year period to June 30, 2009. The conclusions are that, after fees, a significant majority of actively managed fixed income funds failed to outperform their benchmark index; the outperformance achieved by a small minority of funds was attributable to chance; and past performance did not predict future performance. Before fees, the long-term returns for fixed income index funds were roughly equal to the returns of actively managed fixed income funds. After fees, however, index funds outperformed actively managed funds since index fund fees were 60% lower.

by Hubert Lum ©2010 International Foundation of Employee Benefit Plans

Introduction This article addresses three questions: First, does active fixed income management add value in excess of a benchmark index return? Second, is outperformance due to luck or skill? Third, does information on past performance help to predict future performance? The mostly calamitous investment results of 2008 heightened interest in these questions, which are the subject of ongoing debate. As shown in Figure 1, the single positive return result of 2008 was Canadian fixed income,

represented by the DEX Universe Index. Before the next market upheaval, pension funds may want to revisit their policy asset mixes and determine whether there is an adequate policy allocation to fixed income. Part of this determination should include a review of the style of fixed income management—Should it be active or passive?

Description of Data Sources Morneau Sobeco’s Pooled Fund Survey and eVestment Alliance, a subsidiary of Morningstar, supplied data for 85 and 166 Canadian fixed income funds, respectively. The two databases were screened for fixed income funds possessing the following characteristics: a consecutive tenyear, monthly performance record to June 30, 2009; the DEX Universe Bond Index (DEX Index) as the fund’s relevant benchmark; and a pooled fund structure. The screening produced 38 active funds with an average market value of $780 million and five passive funds with an average market value of $2.5 billion.

Most Active Managers Fail to Beat the Benchmark Active fixed income managers try to add value by exercising judgment on changes in one or more of the following areas: the direction of interest rates, the shape of the yield

Canadian Benefits & Compensation Digest  •  October 2010



FEATURE:  Does Active Fixed Income Management Add Value?

Does Active Fixed Income Management Add Value?

Feature Article Figure 1 Returns of Major Asset Classes in 2008 10.0

6.4%

% Return

0.0 -10.0 -20.0 -30.0

FEATURE:  Does Active Fixed Income Management Add Value?

-40.0

-29.2% -33.0% S&P/TSX

-22.6%

-25.9%

MSCI EAFE $C MSCI World $C

S&P 500 $C

DEX Universe

TABLE I Annualized Returns Before Fees for Periods Ending June 30, 2009

3 Years

5 Years

10 Years

Median of actively managed funds

6.15%

5.77%

6.33%

DEX Index

6.19%

5.90%

6.30%

% of actively managed funds underperforming the DEX

53.20%

58.80%

44.70%

% of actively managed funds outperforming the DEX

46.80%

41.20%

55.30%

curve, the yield spread between market sectors and the valuation of specific bond securities. The analysis considered returns before and after fees. The fees applied to a fund’s return were derived from the fee schedule specific to an individual fund for an investment at the $50 million level. Before fees, the finding was that a slim majority of managers outperformed the DEX Index in the ten-year period and underperformed the DEX Index in the three-year and five-year periods. As shown in Table I, there was no pronounced pattern. However, after fees, the majority of actively managed funds underperformed the DEX Index in each of the threeyear, five-year and ten-year periods ended June 30, 2009. The percentage of funds that failed to beat the DEX Index were 80.7%, 87.1% and 79.4%, respectively. These results are shown in Table II.

Excess Return Generated by Most Managers Can Be Attributed to Chance While a majority of actively managed funds failed to beat the benchmark, after fees, a minority did beat the benchmark. In the above example in the ten-year period, 20.6% of active funds outperformed the DEX Index. Here,

J

the question is: Was the outperformance due to skill or luck? The distinction is important since, if the positive outcomes were the result of luck, one would not expect these funds to outperform in the future. One way to measure the distinction between luck and skill is to equate the lucky result with random variation. The sum of this random variation will be zero. At different levels of probability, one can measure the amount by which an active manager beats his or her index and assess if that excess amount is statistically different from zero. At a 90% confidence level, the annualized return in excess of the DEX Index return that qualifies as statistically distinct from random fluctuation is 0.55%. In other words, there is only a 10% probability that an annualized return of benchmark return plus 0.55% is due to chance. For the tenyear period to June 30, 2009, this equates to a return of 6.85% (0.55% plus 6.30%). An analysis of the returns of the 20.6% of funds that outperformed the DEX over ten years showed the number of funds in the group that beat the DEX Index by 0.55% was zero. As shown in Figure 2, the highest return achieved was 6.77%. At lower confidence levels, 85% and 80%, the results were similar. The number of funds that achieved returns statistically distinct from random fluctuation were zero and one, respectively.

Canadian Benefits & Compensation Digest  •  October 2010

TABLE II Annualized Returns After Fees for Periods Ending June 30, 2009

3 Years

5 Years

10 Years

Median of actively managed funds

5.92%

5.54%

6.10%

DEX Index

6.19%

5.90%

6.30%

% of actively managed funds underperforming the DEX

80.70%

87.10%

79.40%

% of actively managed funds outperforming the DEX

19.30%

12.90%

20.60%

Figure 2 Annualized Returns After Fees for Periods Ending June 30, 2009

Return (%)

7.0

5.0

3.0

1.0 Maximum 25th Percentile Median 75th Percentile Minimum DEX Universe Index

3 Years 7.56 6.15 5.92 5.30 3.71 6.19

The Challenge in Finding the Manager Who Can Beat His Peers Most pension funds have at least two return objectives for their asset classes—one that refers to returns above a benchmark, such as the DEX Index, and a second objective that refers to above-median performance in a universe of similar funds. The earlier portions of this article examined performance against the DEX Index. The examination now turns to the second objective. There are many ways to measure persistently good or

10 Years 6.77 6.24 6.10 5.85 5.42 6.30

5 Years 6.76 5.78 5.54 5.15 4.17 5.90

bad performance. The measures have themselves been evaluated in scholarly literature. One simple, robust test is the chi-square test, described in the following sections. Consider the returns for a group of funds for a specific period—say, one quarter—and divide the returns into above median and below median. The above-median funds are identified as “winner” ( W ), and the below-median funds are labeled as “loser” (L). Divide the group in the same way into a W group and an L group in the next quarter. Continue in this manner for 40 quarters (ten years) after the initial quarter.

Canadian Benefits & Compensation Digest  •  October 2010

K

FEATURE:  Does Active Fixed Income Management Add Value?

9.0

Feature Article Figure 3 Chi-Square Test for Performance Persistence Chi-Square Value: 0.95 Probability Value 33.0% Change From Period 1 to Period 2 Period 2 Winner Loser

Period 1 Winner WW WL Expected Results if Random

FEATURE:  Does Active Fixed Income Management Add Value?

Period 2 Winner Loser

Loser LW LL Period 1

Winner 380 380 Expected Results % if Random

Period 2 Winner Loser

Loser 380 380 Period 1

Winner 50.0% 50.0% Actual Results

Period 2 Winner Loser

Loser 50.0% 50.0% Period 1

Winner 399 361 Actual Results %

Period 2 Winner Loser

Loser 361 399 Period 1

Winner 52.5% 47.5%

The change in the fund’s status from one quarter to the next quarter over the ten-year period is one of the following four combinations: WW, WL, LL and LW. One sums each of the four combinations for all funds and all periods. The sums observed are then compared to the sums expected if the process is random. The hope is that one or more funds can consistently outperform the peer group. This outcome would produce a disproportionately high number of WWs. The information can then be used to try to predict future above-median managers. Figure 3 shows that the pattern of change from quarter to quarter over ten years is not significantly different from a random pattern. If the pattern is random, the expectation is to observe 380 as the total in the category WW, which is simply one-quarter of all observations. The actual total observed in the category WW was 399. The lack of a pattern of persistent outperformance within the peer group is confirmed by a weak chi-square result at 0.95 at a 67% confidence level. The chi-square values at a 90% and 95% confidence level are 2.71 and 3.84, respectively. The conclusion is that one period’s result is independent of the prior period’s result. Outperformance in the

L

Loser 47.5% 52.5%

past is not helpful in determining outperformance in the future.

Risk Reduction Is Negligible Many plan sponsors employ active fixed income management for a purpose other than to generate extra return. They use active management to lower portfolio risk relative to a passive approach. Consequently, it is useful to examine the contribution of active fixed income management to the reduction of portfolio risk. Consider a portfolio that is split 60% equity and 40% fixed income where the equity component is represented by the S&P/TSX Index and the fixed income by the DEX Index. Assume that the long-term historic correlation between the two asset classes of 0.4 is applicable in the current environment. With these assumptions, one can measure the portfolio volatility in terms of the annualized standard deviation of monthly returns for the ten-year period to June 30, 2009. A portfolio comprising 100% S&P/TSX had a volatility of 15.31%. A portfolio made up of 60% S&P/TSX and 40% DEX

Canadian Benefits & Compensation Digest  •  October 2010

Figure 4

Annualized Return

Annualized Return Before Fees for Periods Ending June 30, 2009 7.0%

6.30% 6.29% 6.33%

6.19% 6.11% 6.15%

5.90% 5.86% 5.77%

6.0% 5.0% 4.0%

5 Years

DEX Universe Index

10 Years

Passive Median

Active Median

Figure 5

Annualized Return

Annualized Return After Fees for Periods Ending June 30, 2009 7.0% 6.19% 6.0%

6.01% 5.92%

5.90% 5.75%

6.30% 6.18% 6.10% 5.54%

5.0% 4.0%

3 Years DEX Universe Index

Universe Index had a volatility of 9.86%. Bonds reduced the volatility by 5.45%. The reduction in volatility was significant. What happens to volatility when active fixed income management replaces passive fixed income management? When combining the portfolio of 60% S&P/TSX with the median volatility of the above sample of 38 active fixed income managers, the blended portfolio volatility declined to 9.85%. The additional reduction in portfolio volatility, attributable to active management, at 0.01% (9.86% minus 9.85%), was negligible.

Passive Management Outperforms Active Management Unlike active managers who exercise judgment to try to generate excess returns or lower risk, passive fixed income managers use quantitative models. These models allow the passive manager to mimic the risk and return of an index without purchasing the entire index, which in the case of the DEX Index comprises over 1,000 medium-term fixed income issues.

5 Years

10 Years

Passive Median

Active Median

All passive managers in this study use a similar technique (called stratified sampling) to produce a representative sample of the key characteristics of the sectors (federal, provincial, municipal and corporate) and maturities of the DEX Index. A comparison of the returns of the passively managed sample of funds with the actively managed sample of funds produced the following findings. Before fees, the median active and the median passive fund returns were similar over the three-year, five-year and ten-year periods. Further, one group did not consistently outperform the other group over all three time periods measured. Finally, returns were roughly equivalent to index returns. This is shown in Figure 4. After fees, the median passive fund return exceeded the median active fund return in all periods measured. Although neither the active median nor passive median matched the index return, the passive median was consistently closer. This is shown in Figure 5. Insofar as the passive group, at five funds, was small, the number of passive funds exceeding the active median was identified. In the three-year period and ten-year

Canadian Benefits & Compensation Digest  •  October 2010

M

FEATURE:  Does Active Fixed Income Management Add Value?

3 Years

FEATURE:  Does Active Fixed Income Management Add Value?

Feature Article period, all five passive funds exceeded the active median. In the five-year period, four of the five funds exceeded the active median. The superior performance of the passive funds is attributable to their low fees. While the before-fee returns were similar for active and passive funds, the after-fee returns favoured passive funds since passive funds cost 60% less than active funds. At a $50 million investment level, the median cost of passive funds was 0.10% while the median cost of active funds was 0.25%. While the previous discussion has focused on the upside potential, one could also compare the downside risk of active and passive management. As a simple gauge of downside risk, consider the spread between the index return and the worst return of a fund in either the active or passive fund group. The analysis shows that over the three-year period, the worst passive fund underperformed the DEX Index by 0.47% per year while the worst active fund underperformed the DEX Index by 2.48% per year. Over a ten-year period, the worst passive fund underperformed the DEX index by 0.18% per year while the worst active fund underperformed the DEX Index by 0.88% per year. Not surprisingly, the opportunity for significant underperformance was greater with active management.

Caveats The findings are accompanied by several caveats. While many major active and passive fixed income managers are included in the examination, the samples are small. Further, the returns are subject to survivorship, backfill and self-reporting bias. Survivorship is an issue since the samples do not account for managers who may have stopped reporting to the database due to weak returns. Also, the samples do not account for managers who, after accumulating a strong return record, may have joined the database and may have added or backfilled strong historic returns to the database. Self-reporting refers to the fact that returns are reported by the investment managers, not by an independent third party. Hopefully, these biases are similar in the active and passive groups. The examination was restricted to fixed income funds managed against the DEX Index since this is a common benchmark for pension funds. There may be greater opportunities for generating excess returns in strategies that include high-yield bonds, distressed bonds or foreign bonds. There may also be greater chances to add value in periods with higher interest rate volatility. Other caveats include the following: a relatively short, ten-year time horizon; the potential for end-point bias; and

N

the validity of simplifying assumptions such as bell-shaped returns and stable asset class correlations. Finally, on a practical basis, it is also acknowledged that passive management may be unsuitable for small pension funds, since passive management can have high minimum account sizes.

Conclusions The conclusions concerning active fixed income management are: � After fees, most active fixed income funds failed to outperform their benchmark index. � Outperformance was attributable to chance. � Past performance did not predict future performance. � Active fixed income management did not reduce portfolio risk relative to the benchmark index. The conclusions concerning passive fixed income management are: � On an after-fees basis, passive fixed income management outperformed active fixed income management; long-term median returns for passive fixed income managers were roughly equal to the median returns of active fixed income managers but the fees for passive managers were 60% lower. � Passive management meant a reduced risk of significant underperformance. Plan sponsors should adopt a “show me” attitude. They should insist that fixed income managers present a compelling case for active management. In the absence of such evidence, a pension plan’s default policy position should be to obtain low-cost, indexlike returns from passive management for what is normally the largest asset component of its fund. • Author’s note: The views expressed in this article are the author’s alone and do not necessarily reflect the views of Morneau Sobeco. Hubert Lum, M.B.A., CFA, is a senior asset management consultant at Morneau Sobeco in Toronto, Ontario. He has provided consulting services to some of the world’s largest pension funds. His research experience includes investment and governance research collaborations with scholars from the Tuck School of Business, Dartmouth College, the Rotman School of Management, the University of Toronto, the World Bank and Maastricht University. Lum’s work and collaborations have been published in Canada, the United States, Europe and Australia.

Canadian Benefits & Compensation Digest  •  October 2010

Copyright of Canadian Benefits & Compensation Digest is the property of International Foundation of Employee Benefits and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.