A search for intelligent life in the active equity management universe

Transcription

1 INVESTMENT INSIGHTS A search for intelligent life in the active equity management universe FOR INSTITUTIONAL/WHOLESALE OR PROFESSIONAL CLIENT USE ONLY NOT FOR RETAIL DISTRIBUTION

2 MICHAEL CEMBALEST Chairman of Market and Investment Strategy J.P. Morgan Asset Management Michael Cembalest is Chairman of Market and Investment Strategy for J.P. Morgan Asset Management, a global leader in investment management and private banking with $1.5 trillion of client assets under management worldwide (as of September 3, 213). He is responsible for leading the strategic market and investment insights across the firm s Institutional, Funds and Private Banking businesses. Mr. Cembalest is also a member of the J.P. Morgan Asset Management Investment Committee and a member of the Investment Committee for the J.P. Morgan Retirement Plan for the firm s 26, employees. Mr. Cembalest was most recently Chief Investment Officer for the firm s Global Private Bank, a role he held for eight years. He was previously head of a fixed income division of Investment Management, with responsibility for high grade, high yield, emerging markets and municipal bonds. Before joining Asset Management, Mr. Cembalest served as head strategist for Emerging Markets Fixed Income at J.P. Morgan Securities. Mr. Cembalest joined J.P. Morgan in 1987 as a member of the firm s Corporate Finance division. Mr. Cembalest earned an M.A. from the Columbia School of International and Public Affairs in 1986 and a B.A. from Tufts University in 1984.

3 FOREWORD Our Investment Insights series focuses on the building blocks of managing money. Some of our reports deal with the way we construct portfolios while others address opportunities we see in the various countries and markets that we invest in. From time to time, we also look at the asset management industry and the value it can provide for institutional clients. In the attached report, we focus on the question of active equity management, which represents the cornerstone of many diversified portfolios. What we find confirms the notion that despite its ups and downs during different parts of the business cycle, active equity management has the potential to add value to client portfolios over long periods of time. This has been particularly true for equity markets outside the United States, which now comprise more than 5% of managed institutional equity assets. We hope that the research and insights presented here will help you to anticipate and realize the investment opportunities that lie ahead.

4 ABOUT J.P. MORGAN GLOBAL INSTITUTIONAL ASSET MANAGEMENT J.P. Morgan Global Institutional Asset Management is a global leader in investment management, dedicated to creating a strategic advantage for institutions by connecting clients with J.P. Morgan professionals globally. With more than 75 investors on the ground in more than 35 countries, the firm seeks to deliver first-class investment results to some of the world s most sophisticated organizations, including corporate pension plans, endowments, foundations, insurance companies, sovereign wealth funds and government-affiliated institutions. J.P. Morgan Global Institutional is distinguished by its capital markets knowledge, global investment expertise and the long-term, proactive partnerships it establishes with clients. Our innovative strategies span equity, fixed income, real estate, private equity, hedge funds, infrastructure and asset allocation. J.P. Morgan Global Institutional is part of J.P. Morgan Asset Management, which has assets under supervision of $2.2 trillion and assets under management of $1.5 trillion (as of September 3, 213).

5 TABLE OF CONTENTS 1 Executive summary 3 Data universe and specifications 5 The frequency of manager outperformance: A current assessment 8 Manager outperformance: Trends over time 11 The impact of rising correlation on manager performance 14 Manager outperformance trends and market direction 16 On outperformance trends and manager size 18 The performance lifecycle of successful active managers 2 The question of manager consistency 22 Measuring outperformance trends on a risk-adjusted basis 23 The ongoing academic debate 24 Appendix and sources

6 E X E C U T I V E S U M M A R Y

7 The prospects for finding outperforming equity managers The inception of open-ended equity investment funds and stock market benchmarks dates back to the 192s. Ever since that time, portfolio managers of the former have been trying to beat the latter. Index funds and analytical services such as Lipper were launched in the 197s, creating alternatives to actively managed products and systematic means of comparing active and passive investments. Exchange-traded funds (ETFs) followed in the 199s, deepening the set of equity investment alternatives. What works best? A lot has been said and written about active management, efficient and rational market hypotheses, random walk theories and similar subjects. Rather than debate the hypothetical, our goal in this paper is straightforward: to assess the prospects for institutional investors looking for equity managers that outperform. We took a deep dive into the last 2 years of history using a database of managers focused on institutional investors state and corporate pension plans, endowments & foundations, sovereign wealth funds, insurance companies, etc. The results can be summarized as follows: As Exhibit 1 illustrates, over the last five years, the chance of picking outperforming managers has been well over 5% in many investment styles, including: all U.S. value and U.S. small cap strategies, U.S. equity income and the entire range of non-u.s. strategies in both developed and developing markets. The challenges have been in U.S. large cap core, large cap growth and mid cap growth. On a trailing seven-year basis, results for U.S. large cap core and mid cap growth improve markedly, while most large cap growth managers still struggle. Manager outperformance trends have been generally positive over the last 5 and 7 years, except for large cap growth EXHIBIT 1: PERCENTAGE OF ACTIVE EQUITY MANAGERS OUTPERFORMING THEIR RESPECTIVE ETFs LC Core LC Value LC Growth 5-year basis MC Core MC Value 7-year basis MC Growth SC Core Source: J.P. Morgan, evestment; data as of September 213. SC Value SC Growth Multi Core Multi Value Multi Growth Equity Income EAFE Emerging Markets Global All Europe Japan Asia Country ex-japan ex-u.s. J.P. MORGAN ASSET MANAGEMENT 1

8 EXECUTIVE SUMMARY Outperformance trends computed on a risk-adjusted basis tell a similar story to the one shown in Exhibit 1. Trailing five- and seven-year periods include the financial crisis and global recession, a time during which pairwise stock correlations rose and created greater challenges for some (but not all) investment styles. If historical patterns repeat themselves, a decline in correlations in the future can be expected to coincide with improving manager performance. We have finally begun to see evidence of declining correlations in large cap and small cap stocks. We find evidence that for some investment styles, active manager performance has been strongest when overall market returns are weak or rising in single digits. In these style categories, manager outperformance trends decline as equity market returns rise sharply (e.g., during low quality equity rallies). We find no evidence that manager outperformance is systematically influenced by the size of a manager s fund. In the lifecycle of managers that outperform over five-year periods, their year-by-year performance can vary. Most will underperform in two or even three of the five individual years. Manager outperformance trends are volatile and cyclical, and of course, the universe of equity managers changes over time. Through a period which encompassed two greater than 4% declines in equity markets, pairwise correlations rising to the highest levels since the 193s, and changes in rules governing corporate disclosure to investors (Reg-FD), our analysis shows that manager outperformance trends have been generally positive. Should financial conditions continue to normalize, we would expect outperformance trends to improve in some of the investment styles most affected by these changes. One thing to keep in mind about the institutional portfolio manager universe is that outperformance is not a zero-sum game. For example, in the U.S. we estimate that actively managed institutional assets are only 21% of the entire stock market, with another 12% actively managed on behalf of individuals. The remainder is held directly by institutions, hedge funds and individuals in passive strategies and as individual stocks (Exhibit 2). In other words, the theoretical notion that active equity management is a zero-sum game where excess returns vs. the market must average out to zero does not apply in reality. Since external, active institutional managers account for only a portion of actively managed equity assets, the sum of their excess returns vs. the market can be greater (or less) than zero EXHIBIT 2: ACTIVELY MANAGED EQUITIES COMPARED TO THE OVERALL EQUITY MARKET, $ TRILLION Actively managed institutional: separate accounts, commingled funds, and mutual fund share classes $4.7 $7.3 $22. Actively managed institutional plus actively managed retail Total market cap of all U.S. publicly traded stocks Source: J.P. Morgan, evestment, Federal Reserve, Investment Company Institute (ICI), Lipper; data as of June 213. The analysis presented in this paper, by its nature, depends on the performance of active equity managers over the time period analyzed. As a result, we intend to update this assessment in the years ahead. 2 A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

9 Data universe and specifications Where does the data come from? evestment offers the deepest dataset for undertaking the kind of institutional investor-oriented analysis presented in this report. Its database primarily focuses on portfolio managers running separate accounts and commingled funds, and is widely relied upon by institutional investors around the world and their consultants during manager searches. There is also data available from Lipper on the performance of institutional share classes of mutual funds. However, evestment s universe is much larger, with $7.9 trillion in institutional long-only equity assets tracked across all styles, compared to approximately $2.4 trillion in institutional equity share class assets tracked by Lipper. As a result, we use the evestment database of longonly active equity managers, and equal-weight each one (rather than weighting each manager by asset size, which is constantly shifting). Our goal is to assess the frequency of manager outperformance, rather than how many dollars of assets eventually find their way to outperforming managers. The evestment database includes the past performance of managers that no longer report, and is not just confined to surviving managers. What about fees? While evestment has what we consider to be the richest dataset, its performance information by manager is gross of fees paid by investors. To approximate the investor experience, we include fee estimates for each equity investment style. Manager fees often vary based on the size of an investor s commitment to the strategy. Fortunately, evestment also compiles a comprehensive set of fee data by manager, style, size of investor commitment and account type (commingled fund or separate account). We use this data in deriving our fee estimates. J.P. MORGAN ASSET MANAGEMENT 3

10 DATA UNIVERSE AND SPECIFICATIONS Scope Exhibit 3 lists the categories of equity investment styles we analyze in this paper, the current assets under management (AUM) tracked by evestment and the respective number of manager fund vehicles for each category as of June 213. The total current AUM of strategies we analyze is over $7 trillion. The Appendix maps each investment style to its respective benchmark. EXHIBIT 3: EQUITY INVESTMENT STYLES INCLUDED IN THE ANALYSIS Investment style Assets under management ($ millions) Total number of manager fund vehicles Large Cap Core 59, Large Cap Value 1,4, Large Cap Growth 744, Mid Cap Core 51, Mid Cap Value 175,94 16 Mid Cap Growth 173, Small Cap Core 113, Small Cap Value 25,39 2 Small Cap Growth 137, Multi Cap Core 7, Multi Cap Value 13, Multi Cap Growth 15,64 87 Equity Income 196,161 6 EAFE 675, Emerging Markets 656, Global 1,198, All Country ex-u.s. 41, Europe 28, Japan 199, Asia ex-japan 289, There are other categories we did not analyze in this paper since they either overlap with the existing categories, have too few managers on which to base statistical analysis or are deemed to be too small in terms of AUM. The following are examples of excluded strategies: $189 billion in 24 U.S. smid cap funds. These managers fall in between small and mid cap, categories which we already analyze. $38 billion in 29 international large cap growth funds $12 billion in 61 U.S. micro cap funds Country-specific categories in the UK and Canada Source: J.P. Morgan, evestment; data as of June A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

11 The frequency of manager outperformance: A current assessment The question we address in this paper is the likelihood of finding outperforming managers: Is it something that can be done with reasonable due diligence and frequency, or is it like finding a needle in a haystack? Exhibit 4 suggests that it s more the former than the latter. Many investment styles show well over 5% of managers outperforming over the last five years, including all U.S. value and U.S. small cap strategies, as well as U.S. equity income and the entire collection of non-u.s. strategies in both developed and developing markets. The challenges have been in U.S. large cap core, large cap growth and mid cap growth. For U.S. large cap core and mid cap growth, using a sevenyear instead of a five-year analysis period increases the likelihood of finding an outperforming manager to well over 5%. All things considered, the challenges over the last few years have been greatest for U.S. large cap growth managers (see box on page 7 for more details). As seen in Exhibit 4, there are two means of computing outperformance frequency relative to stated benchmarks and relative to ETFs. The first looks at how often managers beat their stated benchmarks and the second looks at how often managers beat an investible exchange-traded fund. From a practical perspective, the second method is more meaningful since that s what an investor s passive alternative often is (given that benchmarks are not investible). In U.S. markets the difference between the two approaches is usually not large; using ETFs as a comparison point usually increases manager In many investment styles, well over 5% of managers outperformed over the last five years EXHIBIT 4: PERCENTAGE OF ACTIVE EQUITY VEHICLES THAT OUTPERFORMED, BY INVESTMENT STYLE Annual Outperformance vs. stated benchmark, trailing 5 years (%) Outperformance vs. ETF... trailing 5 years (%) Source: J.P. Morgan, evestment; data as of September 213. Based on five-year trailing performance through September 213 and fee levels for segregated accounts of $25 million. trailing 7 years (%) Large Cap Core Large Cap Value Large Cap Growth Mid Cap Core Mid Cap Value Mid Cap Growth Small Cap Core Small Cap Value Small Cap Growth 6 59 Multi Cap Core Multi Cap Value Multi Cap Growth 4 45 Equity Income EAFE 68 8 Emerging Markets Global All Country ex-u.s Europe Japan Asia ex-japan 5 68 J.P. MORGAN ASSET MANAGEMENT 5

12 THE FREQUENCY OF MANAGER OUTPERFORMANCE: A CURRENT ASSESSMENT outperformance by less than 5%. In non-u.s. markets, the manager outperformance differences in Exhibit 4 are larger, averaging around 15%. This results from many non-u.s. ETFs substantially underperforming stated benchmarks by more than 1% per year over the last five years. The performance differences between benchmarks and ETFs derive from ETF management fees and each ETF s tracking error (see the Appendix for more details). In Exhibit 4, we used fee levels assuming a $25 million separate account 1 (e.g., the lower end of the institutional asset commitment scale and the higher end of the fee scale). Do the chances of finding outperforming managers improve if we assume larger commitment sizes and lower fees? In Exhibit 5, we show the original results compared with the frequency of outperformance assuming a $1 million separate account and lower fees. The differences in outperformance trends are pretty small, on the order of 1% 3%. Conclusions from a current assessment With the exception of U.S. large cap growth managers, outperformance trends have generally been positive over the last five and seven years. These results are not highly sensitive to whether the fee level for a $25 million vs. $1 million separate account is assumed. In U.S. markets, the results are also not highly sensitive to whether stated benchmarks or ETFs are used. In non-u.s. markets, benchmarks have often considerably overstated the achievable returns for many passive investors, resulting in larger outperformance measures when using ETFs as a comparison point. We end this current assessment with a note on the magnitude vs. the frequency of manager outperformance (Exhibits 6A 6B, next page). In the following section, we take a look at results across a broader period of time, before the global recession. Lower fee schedules associated with higher commitments result in only modest increases in manager outperformance EXHIBIT 5: FREQUENCY OF MANAGER OUTPERFORMANCE VS. RESPECTIVE ETF, BY STYLE AND FEE SCHEDULE Investment style $25 million segregated account fee (bps) Outperformance vs. ETF, trailing 5 years (%) $1 million segregated account fee (bps) Outperformance vs. ETF, trailing 5 years (%) Source: J.P. Morgan, evestment; data as of September 213. Note: Analysis based on five-year trailing performance through September 213. Outperformance difference (%) Large Cap Core Large Cap Value Large Cap Growth Mid Cap Core Mid Cap Value Mid Cap Growth Small Cap Core Small Cap Value Small Cap Growth Multi Cap Core Multi Cap Value Multi Cap Growth Equity Income EAFE Emerging Markets Global All Country ex-u.s Europe Japan Asia ex-japan evestment tracks fees for both separate accounts and commingled funds. However, the differences between them are very small, generally less than five basis points. 6 A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

13 THE FREQUENCY OF MANAGER OUTPERFORMANCE: A CURRENT ASSESSMENT Note on the magnitude of manager outperformance While the primary goal of our paper is to assess the frequency of manager outperformance, we also looked at the magnitude of manager outperformance. As shown in Exhibits 6A 6B, in all circumstances where the frequency of outperformance over the ETF was over 5%, the equally weighted excess returns were positive as well. Furthermore, the degree of outperformance is roughly proportional to the frequency of manager outperformance. The frequency and the degree of manager outperformance have been roughly proportional EXHIBIT 6A: RELATIONSHIP BETWEEN OUTPERFORMANCE FREQUENCY AND NET EXCESS RETURNS, TRAILING 5 YEARS Frequency of outperformance (%) Each circle represents a different investment style Equally weighted net excess return over ETF by style (%) Source: J.P. Morgan, evestment; data as of September 213. Note: Based on five-year trailing performance through September 213 and fees for $25 million segregated accounts. Within investment styles, outperformance by more than 5% of managers coincided with positive excess returns EXHIBIT 6B: FREQUENCY OF MANAGER OUTPERFORMANCE VS. NET EXCESS RETURNS OVER ETF Investment style Frequency of outperformance vs. ETF, trailing 5 years (%) Equally weighted net excess return over the ETF, trailing 5 years (%) Large Cap Core Large Cap Value Large Cap Growth Mid Cap Core Mid Cap Value Mid Cap Growth Small Cap Core Small Cap Value Small Cap Growth Multi Cap Core 45.9 Multi Cap Value Multi Cap Growth Equity Income EAFE Emerging Markets Global All Country ex-u.s Europe Japan Asia ex-japan THE RECENT OUTPERFORMANCE CHALLENGES FOR LARGE CAP GROWTH MANAGERS FRANK RUSSELL RESEARCH LOOKED AT THIS QUESTION IN JUNE 213, AND CAME TO THE FOLLOWING FOUR CONCLUSIONS: 1. While mega cap stocks outperformed, many growth managers viewed these issues as fully priced and did not own them. 2. Growth managers had allocations to international stocks, which underperformed U.S. counterparts. 3. Many growth managers employed an earnings momentum strategy. Given the unevenness of the global recovery, there have not been as many secular earnings growth winners as in prior cycles. 4. Growth managers were often underweight consumer staples. Many of these stocks pay high dividends, and became bond-like proxies by the summer of 213. Empirical Research Partners found that the highest quintile of dividend-paying stocks exhibited 8% correlation with Treasury bonds in 213, by far the highest levels in almost 1 years. Bond-proxy behavior often works against fundamental stock-picking approaches. J.P. MORGAN ASSET MANAGEMENT 7

14 Manager outperformance: Trends over time The manager outperformance trends shown in the prior section are a snapshot in time, using trailing five- and seven-year periods ending in September 213. There were some unique aspects to the last five years, notably the worst global financial crisis in 7 years, which in turn led to many stocks being indiscriminately sold (and then bought) with less consideration to fundamentals than usual. In this section we look at manager outperformance trends over time, including periods prior to the financial crisis. Exhibits 7A 7H (on the following two pages) show the percentage of managers outperforming over rolling five-year periods compared to their respective investible ETFs. We selected a start date of 1996 for the rolling performance analysis, mostly to ensure a statistically significant number of funds in the sample set (see Appendix for more details on sample sizes). 2 In addition, we restricted the rolling analysis to periods for which there are at least 6 managers. 2 The other reason we chose 1996 is that going too far back before 2 would mean including a period before Reg-FD, an SEC rule change related to the timing and transparency of corporate disclosures. 8 A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

15 MANAGER OUTPERFORMANCE: TRENDS OVER TIME The clearest observations from the analysis of U.S. large, mid and small cap core, value and growth strategies are highlighted in Exhibits 7A 7D. Large cap and mid cap growth manager outperformance is still substantially below pre-crisis levels EXHIBIT 7A: LARGE CAP CORE AND GROWTH % OF FUNDS OUTPERFORMING ETF, 5-YEAR BASIS SINCE Large Cap Growth Large Cap Core 5% line EXHIBIT 7B: MID CAP CORE AND GROWTH % OF FUNDS OUTPERFORMING ETF, 5-YEAR BASIS SINCE Mid Cap Growth Mid Cap Core Small cap core and growth outperformance has improved recently EXHIBIT 7C: SMALL CAP CORE AND GROWTH % OF FUNDS OUTPERFORMING ETF, 5-YEAR BASIS SINCE Small Cap Growth Small Cap Core Source: J.P. Morgan, evestment; data as of September 213. Note: Rolling five-year analysis is restricted to periods in which available data includes at least 6 managers. Value manager performance has changed little from pre-crisis levels, with the majority of managers outperforming EXHIBIT 7D: LARGE, MID AND SMALL CAP VALUE % OF FUNDS OUTPERFORMING ETF, 5-YEAR BASIS SINCE Small Cap Value Large Cap Value Mid Cap Value J.P. MORGAN ASSET MANAGEMENT 9

16 MANAGER OUTPERFORMANCE: TRENDS OVER TIME Exhibits 7E 7H capture manager outperformance trends for U.S. multi cap core, growth and value managers as well as for non-u.s. strategies. As tracked by evestment, the non-u.s. strategies below have current AUM of more than $3 trillion, in aggregate. Multi cap core, growth and value managers show outperformance characteristics similar to their large and small cap counterparts EXHIBIT 7E: MULTI CAP CORE, GROWTH AND VALUE % OF FUNDS OUTPERFORMING ETF, 5-YEAR BASIS SINCE % line Multi Cap Growth Multi Cap Core Multi Cap Value Outperformance trends in these non-u.s. strategies (Exhibits 7F 7H) have been quite positive EXHIBIT 7F: EMERGING MARKETS AND EAFE % OF FUNDS OUTPERFORMING ETF, 5-YEAR BASIS SINCE EAFE Emerging Markets EXHIBIT 7G: GLOBAL EQUITY AND ALL-COUNTRY EX-U.S. % OF FUNDS OUTPERFORMING ETF, 5-YEAR BASIS SINCE Global Equity All Country ex-u.s Source: J.P. Morgan, evestment; data as of September 213. Note: Rolling five-year analysis is restricted to periods in which available data includes at least 6 managers. EXHIBIT 7H: EUROPE, JAPAN AND ASIA EX-JAPAN % OF FUNDS OUTPERFORMING ETF, 5-YEAR BASIS SINCE Europe Japan Asia ex-japan Conclusions from the trend assessment Looking at the U.S. core and growth charts, we infer that if some of the factors affecting equity prices and investor sentiment were to normalize, there would be plenty of room for manager outperformance improvement. That is a topic we address in more detail in the next section. 1 A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

17 The impact of rising correlation on manager performance What are the prospects for some of the weaker manager outperformance trends reverting back to pre-crisis levels? A decline in the correlation of stocks with each other would probably help. We can track pairwise correlation of U.S. stocks back to 1926, looking at a universe of stocks across capitalization levels. Correlations began to spike in 28, with three factors often held responsible for many stocks moving in unison: macroeconomic factors increasingly affecting investor buy/sell decisions; the increase in exchange-traded funds and S&P futures contracts (which are, in effect, simultaneous buy/ sell orders on all constituent stocks), and the rise of highfrequency trading strategies such as index arbitrage. While changes related to rising ETFs/futures volumes and high-frequency traders may be a factor, we have always believed that eventually, there would be a cyclical decline in macroeconomic factors driving pairwise correlations. As seen in Exhibits 8A 8B, we are finally seeing evidence of such a downshift in correlations among U.S. large cap, European and U.S. small cap stocks. Whether this trend will persist is unclear, but a break below post-crisis levels is a welcome sign. A downshift in pairwise correlations is finally materializing among U.S. large cap, U.S. small cap and European stocks EXHIBIT 8A: AVERAGE CAP-WEIGHTED RETURN CORRELATIONS EXHIBIT 8B: AVERAGE CAP-WEIGHTED RETURN CORRELATION AMONG U.S. LARGE CAP AND EUROPEAN STOCKS AMONG SMALL CAP STOCKS 7 U.S. Large Cap Europe year moving average Source: Empirical Research Partners; data as of September 213. Source: Empirical Research Partners; data as of September 213. J.P. MORGAN ASSET MANAGEMENT 11

18 THE IMPACT OF RISING CORRELATION ON MANAGER PERFORMANCE Another take on the correlation issue can be inferred from option contracts trading on the Chicago Board Options Exchange. Using options on the S&P 5 and on individual constituent stocks, the implied expected pairwise return correlation on the largest 5 stocks can be derived. As shown in Exhibit 9A, such implied correlations rose during the global recession, and have since been gradually declining, along with realized correlations. Large U.S. stocks have seen a gradual decline in implied and realized pairwise correlations EXHIBIT 9A: IMPLIED AND REALIZED PAIRWISE CORRELATIONS FOR THE 5 LARGEST STOCKS IN THE S&P Implied Realized To complete our thoughts on correlations, Exhibits 1A 1F (next page) confirm the notion that rising correlations have coincided with weaker manager outperformance trends in the U.S. In the charts, the pairwise correlations are on the x-axis, and manager outperfor mance trends are on the y-axis. We use shorter, non-overlapping one-year time periods to better assess the impact of correlation changes on manager outperformance trends. The recent decline in correlations (see page 11, Exhibits 8A 8B) is a sign that may bode well for many active investment styles if the relationships indicated by the following exhibits persist. Source: Bloomberg, Empirical Research Partners; data as of October 213. Exhibit 9B looks at pairwise correlations within each sector. They are the highest in utilities, where most stocks trade like bonds these days, given Fed policy, and the lowest in technology, where winner-take-all products often predominate and create greater differentiation across companies. Bond-like stocks and energy stocks are currently exhibiting high pairwise correlations EXHIBIT 9B: S&P 5 INTRA-SECTOR STOCK RETURN CORRELATIONS FOR 12 MONTHS ENDING OCTOBER 213 Utilities Financials Energy Telecom Services Industrials Consumer Staples Materials Consumer Discretionary Health Care Info Tech Source: J.P. Morgan Asset Management; data as of October A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

19 THE IMPACT OF RISING CORRELATION ON MANAGER PERFORMANCE Rising correlations have coincided with weaker manager outperformance trends in the U.S. markets EXHIBIT 1A: LARGE CAP VALUE EXHIBIT 1B: MID CAP CORE % OF ACTIVE MANAGERS OUTPERFORMING ETF, 1-YEAR BASIS % OF ACTIVE MANAGERS OUTPERFORMING ETF, 1-YEAR BASIS U.S. large cap stocks pairwise return correlation (%) U.S. large cap stocks pairwise return correlation (%) EXHIBIT 1C: MID CAP VALUE % OF ACTIVE MANAGERS OUTPERFORMING ETF, 1-YEAR BASIS 1 8 EXHIBIT 1D: MID CAP GROWTH % OF ACTIVE MANAGERS OUTPERFORMING ETF, 1-YEAR BASIS U.S. large cap stocks pairwise return correlation (%) U.S. large cap stocks pairwise return correlation (%) EXHIBIT 1E: SMALL CAP GROWTH % OF ACTIVE MANAGERS OUTPERFORMING ETF, 1-YEAR BASIS 1 8 EXHIBIT 1F: MULTI CAP VALUE % OF ACTIVE MANAGERS OUTPERFORMING ETF, 1-YEAR BASIS U.S. small cap stocks pairwise return correlation (%) U.S. large cap stocks pairwise return correlation (%) Source: J.P. Morgan, evestment, Empirical Research Partners; annual data from June 1997 to June 213. Note: Analysis is restricted to periods in which available data includes at least 6 managers. J.P. MORGAN ASSET MANAGEMENT 13

20 Manager outperformance trends and market direction We also examined manager outperformance trends based upon whether respective equity market returns for that style were rising or falling. For many investment styles, particularly the non-u.s. strategies, there was no clear trend. But for some, including those shown in Exhibits 11A 11D (next page), active managers showed considerably stronger performance when markets were falling or rising in single digits. When equity markets rose more than 1% per year over a five-year period, these investment categories demonstrated falling levels of manager outperformance. Manager cash balances (often on the order of 3%-5%) held in rising markets explain part of the story here, but we also believe that low quality stock rallies play a role as well. What do we mean by low quality? Here s one example: Since 1956, S&P has ranked stocks in the S&P 5 based on the growth and stability of their earnings and dividends. The stocks in the bottom half of the ranking have been combined into an index, shown, along with the S&P 5, in Exhibit 12 (next page). During periods when lower quality stocks outperform, fundamentally based active managers whose stock-picking relies on valuation and earnings quality may trail their benchmarks. What Exhibits 11A 11D may also imply is that in the selected investment styles, many managers have a beta to the equity market of less than 1.. In other words, their portfolios have less exposure to the markets than their benchmarks, such that when market returns are very high, even fully-invested portfolios will trail the market. Conversely, many such portfolios would have a greater tendency to outperform in weaker market conditions. 14 A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

21 MANAGER OUTPERFORMANCE TRENDS AND MARKET DIRECTION Active manager outperformance for some investment styles has been higher during declining or single digit rising equity markets EXHIBIT 11A: LARGE CAP CORE EXHIBIT 11B: LARGE CAP VALUE % OF ACTIVE MANAGERS OUTPERFORMING ETF, ROLLING 5-YEAR BASIS % OF ACTIVE MANAGERS OUTPERFORMING ETF, ROLLING 5-YEAR BASIS Annualized ETF returns (%) Annualized ETF returns (%) EXHIBIT 11C: MID CAP VALUE % OF ACTIVE MANAGERS OUTPERFORMING ETF, ROLLING 5-YEAR BASIS EXHIBIT 11D: ALL COUNTRY EX-U.S. % OF ACTIVE MANAGERS OUTPERFORMING ETF, ROLLING 5-YEAR BASIS Annualized ETF returns (%) Annualized ETF returns (%) Source: J.P. Morgan, evestment, Empirical Research Partners; five-year rolling monthly observations, September 21 to September 213. Note: Rolling five-year analysis is restricted to periods in which available data includes at least 6 managers. When low quality stocks rise, fundamental managers may underperform EXHIBIT 12: S&P 5 VS. S&P 5 LOW QUALITY RANK INDEX TOTAL RETURN INDICES (SEPTEMBER 3, 28 = 1) S&P 5 S&P 5 Low Quality Index Index Oct-6 Oct-7 Oct-8 Oct-9 Oct-1 Oct-11 Oct-12 Oct-13 Source: J.P. Morgan, Bloomberg; data as of October 213. J.P. MORGAN ASSET MANAGEMENT 15

22 On outperformance trends and manager size Can funds become too big to generate outperformance? One way to address this question systematically is by looking at whether larger funds tend to outperform corresponding ETFs more or less frequently than smaller funds. Using the same evestment database, we divided each investment style universe by size into four quartiles, with Quartile 1 capturing the smallest funds, and Quartile 4 capturing the largest funds. If smaller size were an advantage, we would presumably see consistently lower outperformance percentages in Quartile 4 compared to Quartile 1. Exhibit 13 shows the difference between the outperformance trend in Quartile 4 less the same trend in Quartile 1. In other words, if the number is positive, larger funds, in aggregate, have outperformed more frequently and vice versa if the number is negative. It is important to note that there are computational complexities associated with this kind of analysis, since the concept of size is not constant over a five-year investment horizon. In contrast, if we were to analyze performance trends based on whether a manager is fundamental or quantitative, such factors are static throughout the five-year analysis period. However, assets under management are not static, and are always changing. Outperformance does not appear to be systematically influenced by fund size EXHIBIT 13: OUTPERFORMANCE FREQUENCY OF LARGEST QUARTILE FUNDS MINUS OUTPERFORMANCE FREQUENCY OF SMALLEST QUARTILE FUNDS, 5 YEARS ENDING JUNE 213 Largest funds outperformed more often % LC Core Smallest funds outperformed more often LC Value LC Growth MC Value MC Growth SC Core SC Value SC Growth Multi Core Multi Value Multi Growth EAFE Emerging Markets Global All Country ex-u.s. Europe Japan Asia ex-japan Source: J.P. Morgan, evestment; data as of June 213. Minimum: 6 managers per observation period. 16 A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

23 ON OUTPERFORMANCE TRENDS AND MANAGER SIZE To mitigate the impact of changing AUM, we used average assets under management in each five-year period to categorize each fund s size. However, such an approach does not eliminate the imprecision involved. As shown in Exhibit 13, we did not find systematic, consistent biases in favor of or against smaller managers (+/- 5% 7% should be considered noise from a computational perspective). Larger funds in the small cap core space did better, as did the smaller EAFE and Asia ex-japan funds, but there was no consistent pattern. We also employed a rank-order correlation analysis to examine the relationship between manager fund size and the magnitude of excess return. As with the prior approach, we found no systematic bias demonstrating a clear pattern: EAFE and large cap growth categories showed a modest negative correlation between fund size and return. There was also one large negative correlation observation for Japan, although the data set is more limited due to gaps in the evestment AUM data reported by Japanese managers. For the rest of the strategies, rank order correlations generally ranged from -1% to +1%. We do not see these levels as being statistically meaningful. An analytical approach which uses shorter discrete periods of less than five years for purposes of defining large and small funds would not be as affected by intermittent funds flows. However, using shorter periods would be less meaningful in terms of assessing performance, since many active management strategies are designed to work over a business cycle rather than over a one-year or two-year period. From research we have performed on the subject, using performance assessment periods that are too short can lead to suboptimal results in terms of manager selection. This is a topic we address in the next section. J.P. MORGAN ASSET MANAGEMENT 17

24 The performance lifecycle of successful active managers The prior sections deal with the likelihood of managers outperforming their investible benchmarks over five-year time periods. We consider five-year periods long enough to assess a manager s value-added. If that s the case, what should investors expect regarding shorter-term, interim performance of successful managers? This is a question that investment committees deal with when looking at high-frequency measures of outperformance and underperformance vs. market benchmarks. The purpose of this section is to provide insight about performance lifecycles, since short-term outcomes often do not have a lot of bearing on longer-term performance. Many investment approaches employed by active managers take time to work. Markets do not immediately reward or penalize companies for their results there can be periods of time when investor risk appetite, monetary policy, industry dynamics and other exogenous factors get in the way. That s why most long-term investors analyze their managers over a multi-year cycle rather than over shorter time periods. To get a sense for the performance lifecycle, we examined the individual year-by-year performance of managers that outperformed over a five-year period ( successful managers ). For example, as seen in Exhibit 14A, for the five-year period ending June 213, 42% of successful large cap core managers outperformed in three out of those five years, and 33% outperformed in only two out of five years. In other words, more than threefourths of all successful managers had two or more individual years of underperformance. Exhibits 14B 14C (next page) show the distribution of performance for successful managers in the U.S. large cap value and growth categories, with the number of individual years of outperformance on the x-axis. Even successful* managers underperform their respective ETFs in some years EXHIBIT 14A: LARGE CAP CORE, % OF ACTIVE MANAGERS BY YEARS OF OUTPERFORMANCE OVER A 5-YEAR PERIOD % 42% Number of individual years of outperformance Source: J.P. Morgan, evestment; data as of June 213. *Successful managers are defined as those outperforming their respective ETFs over a given five-year period, regardless of performance in any individual year. 18 A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

25 THE PERFORMANCE LIFECYCLE OF SUCCESSFUL ACTIVE MANAGERS EXHIBIT 14B: LARGE CAP VALUE, % OF ACTIVE MANAGERS BY YEARS OF OUTPERFORMANCE OVER A 5-YEAR PERIOD EXHIBIT 14C: LARGE CAP GROWTH, % OF ACTIVE MANAGERS BY YEARS OF OUTPERFORMANCE OVER A 5-YEAR PERIOD Number of individual years of outperformance Number of individual years of outperformance Source: J.P. Morgan, evestment; data as of June 213. Source: J.P. Morgan, evestment; data as of June 213. Exhibit 15 takes a look across all styles, and shows the percentage of successful managers that underperformed in at least two years out of five. The pattern is consistent across almost every equity investment style: Good managers have extended periods of underperformance. Even a number in the 3% 4% range can be considered high in this context, since it implies that a very large subset of the successful managers underperformed in two or three years out of five. One possible reason for this trend is suggested by an academic study of over 1 million individual active equity management transactions from 1999 to The study found that the vast majority of excess return was earned on positions held for between two and three years. If these are the holding periods that correspond to outperformance on individual positions, it follows that on a portfolio basis, outperformance would be best evaluated over a longer time frame. This observation has important implications for how investors evaluate shorter-term manager performance outcomes. Performance lifecycle of successful managers shows substantial periods of interim underperformance EXHIBIT 15: PERCENTAGE OF MANAGERS WHO UNDERPERFORMED IN AT LEAST 2 OUT OF 5 INDIVIDUAL YEARS, BUT OUTPERFORMED ON A 5-YEAR BASIS LC Core LC Value LC Growth MC Core MC Value MC Growth Source: J.P. Morgan, evestment; data as of June 213. SC Core SC Value SC Growth Multi Core Multi Value Multi Growth EAFE Emerging Markets Global All Country ex-u.s. Europe Japan Asia ex-japan 3 Chakrabarty (St. Louis), Moulton (Cornell) and Trzcinka (Indiana), Institutional Holding Periods, (April 213). J.P. MORGAN ASSET MANAGEMENT 19

26 The question of manager consistency In the prior section, we reviewed the interim performance lifecycles of managers that are successful over five-year periods. In this section, we consider the question of manager consistency. In other words, does ex-ante excess return performance over one five-year period tell us anything about the likelihood of ex-post outperformance in the next five-year period? One can imagine two extremes regarding successful managers: The group of successful managers remains the same over time or is constantly changing, with low instances of outperformance over consecutive five-year periods. What we find, unsurprisingly, is that the answer lies in the middle. Before computing anything, we have to further refine the question we re asking. It makes the most sense to analyze the issue of manager consistency during a period in which manager outperformance trends are roughly constant. Otherwise, we would simply be capturing the overall trend of manager outperformance rising or falling. As a result, we confine our analysis here to consecutive, non-overlapping fiveyear periods in which overall manager outperformance trends are stable (to be exact, when they are within +/- 1% of each other). Having done so, we can assess the degree of churn that is, the degree to which the pool of outperforming managers changes from one period to the next. Exhibit 16 (next page) shows the results. In the five-year period ending 213, manager consistency ratios 4 are in the range of 35% 45%. Before 213, we found manager consistency ratios in the 5% 6% range. These levels indicate modest consistency at an industry level, and do not point to either high, reliable levels of consistent outperformance by successful managers, or the opposite, the notion that manager outperformance is a constantly mean-reverting trend. In this regard, public equity manager outperformance trends are similar to their private equity counterparts: A study from the Universities of Virginia, Chicago and Oxford in showed that since 2, private equity managers in the top two quartiles had a 5% chance of being above median when measuring the quartile performance of their subsequent funds. Before the year 2, more private equity manager consistency was seen, but it has since declined. 6 4 Manager consistency ratio is the percentage of managers outperforming in one five-year period who go on to outperform in the next five-year period. 5 Harris (UVA Darden), Jenkinson (Oxford), Kaplan (Chicago Booth) and Stucke (Oxford), Has persistence persisted in private equity? Evidence from buyout and venture capital funds, (April 213). 6 See our Eye on the Market special issue on private equity, Private Investigations, July 9, A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

27 THE QUESTION OF MANAGER CONSISTENCY On an industry basis, analysis suggests modest levels of repeat outperformance by managers from period to period EXHIBIT 16: PERCENTAGE OF MANAGERS OUTPERFORMING IN PERIOD 1 THAT ALSO OUTPERFORM IN PERIOD 2 Source: J.P. Morgan, evestment; data as of June year period ending... Investment style Period 1 Period 2 Manager consistency ratio (%) Large Cap Value 6/3/28 6/3/ Mid Cap Value 6/3/28 6/3/ Small Cap Core 6/3/28 6/3/213 4 Small Cap Value 6/3/28 6/3/ Small Cap Growth 6/3/28 6/3/ Multi Cap Value 6/3/28 6/3/ EAFE 6/3/28 6/3/ Global 6/3/28 6/3/ Asia ex-japan 6/3/28 6/3/ Large Cap Core 6/3/26 6/3/ Mid Cap Growth 6/3/26 6/3/ Multi Cap Value 6/3/26 6/3/ EAFE 6/3/26 6/3/ All Country ex-u.s. 6/3/26 6/3/ Europe 6/3/26 6/3/ Asia ex-japan 6/3/26 6/3/211 6 Large Cap Core 6/3/24 6/3/29 63 Large Cap Value 6/3/24 6/3/29 52 Small Cap Value 6/3/24 6/3/29 61 EAFE 6/3/24 6/3/29 59 Europe 6/3/24 6/3/29 61 Small Cap Value 6/3/22 6/3/27 62 Note: The table above shows manager consistency ratios for investment styles that meet certain tests. First, there must be at least 6 managers in both five-year periods. Second, the difference in industry-level manager outperformance between Period 1 and Period 2 for a given investment style must be within a +/- 1% range, since we are primarily attempting to measure the degree of manager rotation within a stable outperformance trend. J.P. MORGAN ASSET MANAGEMENT 21

28 Measuring outperformance trends on a risk-adjusted basis We address one last question about manager outperformance trends: What about the risk that managers took in the process? There are a lot of ways to address this question: portfolio position or sector concentration, tracking error versus benchmark, out-of-index exposures, etc. We chose a method which looks at risk on an ex-post basis: a comparison of the manager s volatility compared to the volatility of the respective ETF. Exhibit 17 contains the outperformance trends first shown in Exhibit 4 (page 5), measuring the percentage of managers outperforming their respective ETFs. Exhibit 17 also shows the percentage of managers whose ratios of return to risk (e.g., annualized five-year net return divided by the annualized volatility of monthly returns) are higher than that of the ETF. As shown, for most investment styles, the risk-adjusted outperformance measures are higher than when measured on a nominal basis. Equity income in particular shows a large jump, which is intuitive given a typical focus on safer, dividend-paying stocks. There are a lot of ways to measure risk, and this is simply one approach. However, the results suggest that the generally positive manager outperformance trends described in this paper were not simply a by-product of managers with higher risk exposures than the market. Measuring returns on a risk-adjusted basis tells a similar story EXHIBIT 17: FREQUENCY OF MANAGER OUTPERFORMANCE VS. ETF ON A NOMINAL AND RISK-ADJUSTED BASIS, TRAILING 5 YEARS Investment style Nominal outperformance (%) Risk-adjusted outperformance (%) Difference (%) Large Cap Core Large Cap Value Large Cap Growth Mid Cap Core Mid Cap Value Mid Cap Growth Small Cap Core Small Cap Value Small Cap Growth Multi Cap Core Multi Cap Value Multi Cap Growth Equity Income EAFE Emerging Markets Global All Country ex-u.s Europe Japan Asia ex-japan Source: J.P. Morgan, evestment; data as of September 213. Based on five years trailing performance and fees for $25 million segregated accounts. 22 A SEARCH FOR INTELLIGENT LIFE IN THE ACTIVE EQUITY MANAGEMENT UNIVERSE

29 The ongoing academic debate This paper takes an empirical rather than a theoretical approach to assessing manager alpha. Even so, we briefly review below competing theories on whether active management should be able to work in the first place. In the 197s, some academics embraced the concept of an efficient market hypothesis (EMH). 7 Simply put, this concept implies that markets and investors are rational and optimize all decisions, and that all available information is already reflected in the price of any security. As such, the ability of an active manager to generate consistent outperformance based on anything other than luck would be a statistical anomaly. Even before the efficient market hypothesis, many academics described the random walk of financial securities a concept which has been interpreted to mean that fundamental approaches to predicting securities price movements would be doomed to failure. However, others disagree with the concept and implications of the efficient markets hypothesis. To illustrate the point, a recent Nobel Prize was awarded jointly to one of the most well-known proponents of the efficient markets hypothesis, and also to a research peer whose career has been focused on research arguing against it. The anti-emh arguments are as follows, all of which are components of Behavioral Finance Theory: Excess volatility: the S&P 5 is effectively a forecast of future dividend payments, discounted back to today. Importantly, the volatility of actual dividends is much lower than the volatility of the S&P 5 itself. As a result, markets (and individual stock prices) generate excess volatility, which is by its existence not efficient, and which investors can capture. 8 Patterns and tendencies: If markets were truly efficient, market researchers would not find consistent patterns and seasonal trends in stock prices. Yet they do exist: the January effect, stock price movements around earnings releases, the size effect benefitting smaller stocks, etc. The momentum effect: There are academic studies that demonstrate that buying and selling stocks based on price momentum can generate excess returns above the market. An efficient market would not yield opportunities such as this. I am not sure that there are substantial benefits to immersing oneself in epistemological debates such as these when contemplating the prospects for active equity management. It can be interesting to read the theoretical literature, but in the end, industry data and investor experience will inform us as to how active equity management is faring. That is why we intend to update this paper over time to evaluate whether substantial segments of the active management community are continuing to deliver excess net-of-fee returns to their investors. 7 Elroy Dimson and Massoud Mussavian, Market efficiency, (London Business School, 2). 8 Robert Shiller, From efficient markets theory to behavioral finance, Journal of Economic Perspectives (Winter 213). J.P. MORGAN ASSET MANAGEMENT 23