Alice in Factorland. Rob Arnott Founder and CEO Research Affiliates, LLC

Transcription

1 Alice in Factorland Rob Arnott Founder and CEO Research Affiliates, LLC

2 Our Adventure in Factorland» Factor timing is difficult but possible» Relative valuation of the strategy or the factor (i.e., relative to its own history) is a powerful predictor of future return» Most investors already practice a form of market timing unfortunately in the wrong direction by chasing past performance which can erode the benefits of factor investing» They fund the success of contrarian investors» Emphasizing factors or strategies that are trading cheap relative to their own historical norms and deemphasizing the more expensive factors or strategies can improve performance 2

3 Continue Our Journey in Factorland» Do mutual funds capture their factor returns?» Simulated factor returns are believed to describe investor opportunities» Instead, we observe the incredible shrinking factor returns in live assets» Smart beta is becoming synonymous with factor investing» In our view, this is not correct: Smart beta originally covered strategies that break the link between price and portfolio weight» Factor tilts don t do this! Factor tilts are not true smart beta» Manager evaluation: factor tilts can help predict fund returns» Past performance is worse than useless» Funds with factor tilts that are trading cheap will tend to outperform» Troubles with momentum» Harvesting momentum premia appears to be mission impossible» Can we save momentum? Maybe. 3

4 PART I. Our Factor Timing Research 4

5 Trend Chasing Everywhere Survivorship Bias Practitioners look for best historical performance. Academics look for best historical performance. Asset Owners look for best historical performance.» Problem: Not all factors are robust.» Selection bias and data mining are mistaken for persistent alpha 1» Rising valuations are mistaken for persistent alpha 2 Harvey, Liu, Zhu (2015); Beck, Hsu, Kalesnik, Kostka (2016). Fama, French (2002); Arnott, Bernstein (2002); Campbell, Shiller (1988); Cochrane (2008). 5

6 Alpha Decomposition Portfolio Alpha Return Due to Change + in Relative Valuation Valuation- Adjusted Alpha Revaluation Alpha Structural Alpha» Alpha due to change in relative valuation» is mean reverting and averaging roughly zero in the long run» contributes significantly to strategy performance in the short run» Short run can mean decades!» Alpha adjusted for change in relative valuation is a good measure of unconditional expected return of a strategy 6

7 Valuation Cycle for Value Factor 4.00 Value vs. Growth, United States (July 1968 December 2016) D Relative Performance, Value vs. Growth 1.00 A A B B C D E E Relative Valuation,* Value vs. Growth 0.50 Nifty Fifty Biotech Bubble Tech Bubble Global Financial Crisis C 0.13 Value Performance Relative Valuation *Based on a blend of four valuation metrics: Price/Book, Price/5yrSales, Price/5yrEarnings, Price/5yrDividends. Source: Research Affiliates, LLC, using data from CRSP and Compustat. 7

8 Factor Valuations Are Predictive of Future Returns: Example: The Value Factor 30% Value vs. Growth (July 1968 December 2016) 15% Subsequent Five-Year Return 0% -15% Relative Valuation (Aggregate) US Developed EM Median Valuation Source: Research Affiliates, LLC, using data from CRSP, Compustat, Worldscope, and Datastream. 8

9 Subsequent 5-Yr Return Subsequent 5-Yr Return Subsequent 5-Yr Return Subsequent 5-Yr Return Factor Valuations Are Predictive of Future Returns: 15% 10% 5% 0% -5% -10% -15% Gross Profitability Factor Correlation: t-stat: -2.06** Relative Valuation (aggregate) 25% 20% 15% 10% 5% 0% -5% -10% -15% Size Factor Relative Valuation (aggregate) Correlation: t-stat: -7.53*** 25% 20% 15% 10% Momentum Factor Correlation: t-stat: -1.79* 20% 10% Low Beta Factor Correlation: t-stat: % 0% 0% -5% -10% -15% -10% -20% Relative Valuation (aggregate) Relative Valuation (aggregate) Long-Term Forecast Near-Term Forecast Source: Research Affiliates, LLC, using data from CRSP and Compustat. As of December Two-Tail statistical significance: * = 10% threshold; ** = 5% threshold; *** = 1% threshold. 9

10 Relative Valuation of Factors What We Saw in June 2016 U.S. (1967 Mar 2016) Aggregate Dev ex U.S. (1983 Mar 2016) Aggregate Emerging Markets (1996 Mar 2016) Aggregate 8 4 Legend Factor is Expensive Current Valuation 2 1 Median Valuation Factor is Cheap Gross Profitability Momentum Low Beta Illiquidity Small Cap Investments Value (B/P) Value (Blend) Source: Research Affiliates, LLC, using data from CRSP and Compustat, 1967 Mar The chart was originally published in To Win With Smart Beta Ask if the Price is Right, June 2016, Arnott, Beck, and Kalesnik. 10

11 Emerging Market Developed Market United States What Happened Afterwards in 2016 Value Won, Quality & Momentum Lost, and Low-Vol Cratered Everywhere, Jul Dec 2016 Region Index Performance Relative to the Benchmark Jul-Dec 2016 Prior 3 Years Prior 5 Years Jul-Dec 2016 Absolute Index Performance Prior 3 Years Prior 5 Years S&P % 39.7% 76.8% FTSE RAFI US % -5.3% -2.8% 11.7% 34.4% 74.0% Russell 1000 Value 2.6% -7.0% -5.6% 10.4% 32.6% 71.2% S&P 500 Low Volatility -9.6% 9.4% 21.6% -1.7% 49.1% 98.4% MSCI USA Quality -0.8% 0.8% 1.6% 7.0% 40.5% 78.5% S&P 500 Momentum -4.2% -3.0% -7.9% 3.6% 36.6% 69.0% MSCI World 6.8% 22.3% 37.8% FTSE RAFI Developed 4.1% -5.3% -8.4% 10.9% 17.0% 29.5% MSCI World Value 3.9% -6.5% -7.2% 10.7% 15.9% 30.7% S&P Developed Low Volatility -8.7% 9.2% 18.1% -1.9% 31.5% 56.0% MSCI World Quality -3.3% 10.8% 20.4% 3.5% 33.2% 58.3% S&P Momentum Developed -5.7% -0.5% 8.5% 1.1% 21.9% 46.3% MSCI Emerging Markets 4.5% -4.6% -17.5% FTSE RAFI Emerging 10.6% -3.1% -7.7% 15.1% -7.7% -25.3% MSCI Emerging Markets Value 2.5% -4.9% -7.2% 7.0% -9.6% -24.8% S&P Emerging Markets Low Volatility -7.0% -6.2% 10.7% -2.5% -10.8% -6.8% MSCI Emerging Markets Quality -4.1% 7.5% 11.5% 0.4% 2.9% -6.0% S&P Emerging Markets Momentum -3.7% 12.7% 26.2% 0.8% 8.1% 8.7% Source: Research Affiliates, LLC, using Bloomberg data. 11

12 Most Academics Are Trend Chasers! Return Degradation Before and After Factor Publication United States (Jan 1967 Aug 2016) Annualized Results Value (Blend) Value (B/P) Momentum Size Illiquidity Low Beta Profitability Investment Average Year Published Before Publication 9.8% 9.1% 5.4% 7.0% 2.5% 7.4% 1.2% 3.5% 5.8% After Publication 2.3% 1.4% 3.7% 0.8% 5.0% 2.1% 5.0% -1.0% 2.4% Difference -7.5% -7.8% -1.8% -6.2% 2.5% -5.4% 3.8% -4.5% -3.3%» After-Publication Alpha is Not Large!» 2.4% is for long-short portfolio 1.2% per side» That s before trading costs, implementation shortfall, and fees» Residual alpha for end customers could easily be zero! Source: Research Affiliates, LLC, using CRSP/Compustat and Worldscope/Datastream data. 12

13 Most Product Providers Are Trend Chasers! Return Degradation Before and After Smart Beta Index Launch United States (Jan 1967 Aug 2016) Annualized Results Fundamental Index Equal Weight Low-Vol Index FTSE RAFI Low Vol Quality Index Dividend Index Risk Efficient Maximum Diversification Average Year Launched Nov-05 Jan-03 Feb-11 Apr-13 Dec-12 Nov-03 Jan-10 Nov-11 Before Launch 2.0% 1.3% 1.2% 2.2% 0.4% 2.9% 2.7% 1.6% 1.8% After Launch 0.5% 2.3% 2.1% 0.1% 0.1% 1.3% 0.9% 4.1% 1.4% Difference -1.5% 1.0% 0.9% -2.1% -0.4% -1.6% -1.9% 2.5% -0.4%» Here, at least, there s some hope» 1.4% after launch is not bad, not far below prior simulated results» Again, this is before trading costs, implementation shortfall, and fees» But, many of these have low turnover, and most have delivered live results ahead of benchmark since launch, net of all fees and costs Source: Research Affiliates, LLC, using CRSP/Compustat and Worldscope/Datastream data. 13

14 Most Investors Are Trend Chasers! 8.81% 8.38% 9.36% 8.05% 8.23% 9.78% 8.66% S&P 500 Index 8.97% 6.87% 6.76% 5.22% All Funds Growth Funds Value Funds Small-Cap Funds Large-Cap Funds Dollar-Weighted Return Buy-&-Hold Return S&P 500 Index Source: Hsu, Myers, and Whitby, Timing Poorly: A Guide to Generating Poor Returns While Investing in Successful Strategies, Journal of Portfolio Management (Winter 2016). 14

15 Trend Chasing Is Costly Performance Characteristics of Trend Chasing and Contrarian Allocations, United States (Jan 1977 Aug 2016) Smart Beta Strategies Trend Chasing and Contrarian Strategies Factors Trend Chasing and Contrarian Strategies Value Add (Ann.) Information Ratio Average Alpha (Ann.) Sharpe Ratio 6.1% % 1.2% % 2.4% 1.2% 0.14 Equally Weighted Smart Beta Allocation Three Best Performing Smart Beta Strategies (1,3,5,10 yr Performance) Three Cheapest Smart Beta Strategies Equally Weighted Factor Allocation Three Best Performing Factors (1,3,5,10 yr Performance) Three Cheapest Factors Source: Research Affiliates, LLC, using CRSP/Compustat and Worldscope/Datastream data. 15

16 Timing with Valuations Is Not Always a Value Tilt Allocation of Strategies and Factors Used in the Most and Least Expensive Series, Relative to Own History, United States (Jan 1977 Aug 2016) Presence of Smart Betas Presence of Factors Dividend Index Fundamental Index FTSE RAFI Low Vol Low-Vol Index Value (Agg) Value (B/M) Investments Low Beta Risk Efficient Size Equal Weight Maximum Diversification Illiquidity Momentum Quality Index Profitability Portfolio of Least Expensive Portfolio of Most Expensive Note: Cheapest and most expensive valuations are computed as a blended average of Five-Year Average Earnings-to-Price, Five-Year Average Sales-to-Price, Five-Year Average Dividends-to-Price, and most recent Book-to-Price ratios. For each factor/valuation metric we compare their current valuation relative to long run average valuation. Source: Research Affiliates, LLC, using CRSP/Compustat and Worldscope/Datastream data. 16

17 Expected 5-Year Returns (Ann.) Relative Valuation Research Affiliates Smart Beta Interactive Site 12% Real Long-Term Expected Return, Net of Transaction Costs, US Smart Beta Strategies (as of 12/31/2016) 4 Relative Valuation, Select Smart Beta Strategies (as of 12/31/2016) EM, RAFI Fundamental Index 8% 2 1 4% 0% Quality Income US, RAFI Fundamental Index Value Momentum Low Volatility Small Cap % 0% 5% 10% 15% 20% 25% Volatility US EM Source: Research Affiliates, LLC, using CRSP/Compustat and Worldscope/Datastream data. 17

18 Expected 5-Year Returns (Ann.) Relative Valuation Research Affiliates Smart Beta Interactive Site 12% Real Long-Term Expected Return, US Factors (as of 12/31/2016) 4 Relative Valuation, Select Factors (as of 12/31/2016) 8% 2 4% Profitability EM Value (Aggregate) EM Value (P/B) Momentum US Value (Aggregate) % Investment Illiquidity Size US Value (P/B) Low Beta % 0% 5% 10% 15% 20% 25% Volatility US EM Source: Research Affiliates, LLC, using CRSP/Compustat and Worldscope/Datastream data. 18

19 Expected 5-Year Returns (Ann.) Relative Valuation Research Affiliates Smart Beta Interactive Site Developed 12% Real Long-Term Expected Return, Developed Factors (as of 12/31/2016) 4 Relative Valuation, Select Factors (as of 12/31/2016) 8% 2 4% 0% Investment Illiquidity Profitability Size Momentum Value (Aggregate) Value (P/B) Low Beta -4% 0% 5% 10% 15% 20% 25% Volatility Source: Research Affiliates, LLC, using Worldscope/Datastream data. 19

20 Expected 5-Year Returns (Ann.) Relative Valuation Research Affiliates Smart Beta Interactive Site Developed 8% Real Long-Term Expected Return, Net of Transaction Costs, Developed Smart Beta Strategies (as of 12/31/2016) 4 Relative Valuation, Select Smart Beta Strategies (as of 12/31/2016) RAFI Fundamental Index 2 4% Income Value 1 Quality Momentum 0% Low Volatility 0.5 Small Cap % 0% 5% 10% 15% 20% 25% Volatility Source: Research Affiliates, LLC, using Worldscope/Datastream data. 20

21 PART II a. The Incredible Shrinking Factor Returns 21

22 Our Methodology» Our database» All funds from Morningstar Direct survivorship bias-free fund universe» US open-ended long-only active equity funds, with at least two year return history from January 1990 to December 2016» Share-class inclusion: A-share, No-load-share, and Institutional-share» We can reverse-engineer the factor returns earned by managers» We first measure mutual fund s average factor loadings over time by regressing fund returns against conventional constructed factor returns» We then run a cross-sectional regression, of fund returns on fund factor loadings, to estimate monthly factor returns, as realized by live funds» The slippage in factor returns» While the conventional factor returns and the factor returns realized by the managers show ~0.9 correlation, there is often a huge shortfall 22

23 Value Factor Return Captured by Managers Momentum Factor Return Captured by Managers Market Factor Return Captured by Managers Size Factor Return Captured by Managers Long Short Factor Returns vs. Realized Factor Returns Captured by Managers 12% 4% Market Correlation = 0.92 Slope = % 4% 0% Size Correlation = 0.96 Slope = % -4% -12% 8% 4% -12% -8% -4% 0% 4% 8% 12% Value Observed Market (Mkt - RFR) Factor Return Correlation = 0.89 Slope = % 16% 8% -8% -4% 0% 4% 8% Momentum Correlation = 0.90 Slope = 0.98 Observed Size (SMB) Factor Return 0% 0% -4% -8% -8% -8% -4% 0% 4% 8% Observed Value (HML) Factor Return -16% -16% -8% 0% 8% 16% Observed Momentum (UMD) Factor Return Source: Research Affiliates, LLC, using data from CRSP, Compustat, Morningstar Direct, and the website of Kenneth French. Jannuary 1991 December Monthly Correlations are shown in all four graphs. 23

24 Growth of $100 (log scale) Growth of $100 (log scale) Factor Slippage Momentum Factor $400 Cumulative Returns $120 Cumulative Difference, Realized Versus Theoretical Shortfall (per ann):- 5.2% t-stat = $247 $200 $60 $100 $111 $30 $ $ Return Captured by Manager Theoretical L/S Factor Return Source: Research Affiliates, LLC, using data from CRSP, Compustat, Morningstar Direct, and the website of Kenneth French. Jannuary 1991 December Note: Cumulative returns are compounded using arithmetic returns. 24

25 Growth of $100 (log scale) Growth of $100 (log scale) Factor Slippage Market Factor Market-Risk Free $400 Cumulative Returns $200 Cumulative Difference, Realized Versus Theoretical $314 $200 $205 $100 $100 $50 Shortfall (per ann): -4.2% t-stat = $ $ Return Captured by Manager Theoretical L/S Factor Return Source: Research Affiliates, LLC, using data from CRSP, Compustat, Morningstar Direct, and the website of Kenneth French. Jannuary 1991 December Note: Cumulative returns are compounded using arithmetic returns. 25

26 Growth of $100 (log scale) Growth of $100 Factor Slippage Value Factor $256 Cumulative Returns $125 Cumulative Difference, Realized Versus Theoretical $194 $128 $157 $100 $75 $64 $50 $32 $25 Shortfall (per ann): -1.4% t-stat = $ $ Return Captured by Manager Theoretical L/S Factor Return Source: Research Affiliates, LLC, using data from CRSP, Compustat, Morningstar Direct, and the website of Kenneth French. Jannuary 1991 December Note: Cumulative returns are compounded using arithmetic returns. 26

27 Growth of $100 (log scale) Growth of $100 Factor Slippage Size Factor Small Cap Large Cap $200 Cumulative Returns $185 $125 Cumulative Difference, Realized Versus Theoretical $167 $100 $100 Excess (per ann): 0.7% t-stat = 1.13 $ $ Return Captured by Manager Theoretical L/S Factor Return Source: Research Affiliates, LLC, using data from CRSP, Compustat, Morningstar Direct, and the website of Kenneth French. Jannuary 1991 December Note: Cumulative returns are compounded using arithmetic returns. 27

28 Do Mutual Funds Capture Their Factor Returns? No! Annualized Factor Returns, US Equity Funds (Jan 1991 Dec 2016) 8.2% 5.7% 4.1% 2.6% 3.3% 3.6% 2.2% 0.4% Mkt (Mkt-Rf) Size Value Momentum Theoretical L/S Portfolio Realized Return in Live Portfolio» The returns realized by mutual fund managers can fall short due to:» Transaction costs, fees, bid ask spreads, and trades that get away» Hard to replicate features of theoretical factor portfolios (usually constructed ex post, bringing in data-mining and selection bias) Source: Research Affiliates, LLC, using data from CRSP, Compustat, Morningstar Direct, and the website of Kenneth French. 28

29 Similar Slippage in International Equity Funds Annualized Factor Returns, International Equity Funds (Jan 1991 Dec 2016) 6.3% 6.6% 4.9% 1.6% 1.6% 2.3% 2.1% Mkt (Mkt-Rf) Size Value Momentum -0.6% Theoretical L/S Portfolio Realized Return in Live Portfolio Source: Research Affiliates, LLC, using data from Worldscope, Datastream, and Morningstar Direct. 29

30 PART II b. Why Factor Tilts Are Not Smart Smart Beta 30

31 Factor Tilts = Smart Beta?» The definition of smart beta has been vastly extended» Now, almost anything formulaic, other than a full-market cap-weighted index, seems to qualify for the smart beta label» Factor tilt and multi-factor strategies are sold as smart beta even though most of these strategies begin with, and anchor on, cap weighting» Let s replicate a few generation one smart beta strategies with factors:*» Fundamental Index» Equally Weighted Index» Minimum Volatility Index» How well do the factor replicated smart beta strategies fare?» much is lost in translation. *These theoretical factors are Market, SMB, HML, MOM, and BAB. 31

32 Early Criticism of the Fundamental Index It Is Repackaged Value Return Performance and Factor Loadings for Fundamental Index, Equal Weight, and Minimum Variance Strategies (January 1974 June 2016) Panel A: Fama French Three Factor Model Plus Momentum Alpha (Ann.) Market Value Size Momentum R-Sq. Fundamental Index Equal Weight Low Volatility US 0.97% *** 0.98 *** 0.35 *** *** *** 97.92% (2.74) (147.47) (33.46) (-8.07) (-10.23) 0.67% 1.03 *** 0.18 *** 0.24 *** *** 97.23% (1.48) (120.96) (13.78) (19.59) (-2.73) 1.59% ** 0.83 *** 0.16 *** *** 0.05 *** 87.38% (2.1) (58.05) (6.97) (-7.61) (3.1) Panel B: Fama French Three Factor Model Plus Momentum and BAB Factor Alpha (Ann.) Market Value Size Momentum BAB R-Sq. Fundamental Index Equal Weight Minimum Variance 0.67% * 0.98 *** 0.33 *** *** *** 0.04 *** 98.01% (1.9) (150.06) (29.2) (-8.31) (-11.47) (4.83) 0.38% 1.02 *** 0.16 *** 0.24 *** *** 0.04 *** 97.30% (0.83) (121.87) (11.25) (19.77) (-3.8) (3.57) 0.51% 0.82 *** 0.08 *** *** *** 88.94% (0.71) (61.28) (3.31) (-8.24) (0.26) (8.45)» What about those alphas? *** Significance at the 1% level, **Significance at the 5% level, * Significance at the 10% level. Source: Research Affiliates, LLC, based on data from CRSP and Compustat. 32

33 Factor Replications Miss the Mark Return Performance of Various Strategies (January 1974 June 2016) Relative to Cap US 1000 Investment Allocation Returns Volatility Sharpe Ratio Value Add Tracking Error Information Ratio Fundamental Index original 12.9% 15.3% % 4.3% 0.42 Full Long/Short Factor Replicated 12.0% 15.2% % 3.4% 0.28 Factor Replicated Long Only 12.0% 15.7% % 3.2% 0.29 Equal Weight US 1000 original 13.1% 16.9% % 4.8% 0.41 Full Long/Short Factor Replicated 12.8% 17.4% % 5.3% 0.32 Factor Replicated Long Only 12.6% 16.8% % 5.1% 0.30 Minimum Variance original 12.4% 13.3% % 5.7% 0.23 Full Long/Short Factor Replicated 12.2% 13.7% % 3.0% 0.38 Factor Replicated Long Only 12.3% 14.4% % 2.6% 0.46 Cap Weight US % 15.4% 0.41 *These theoretical factors are Market, SMB, HML, MOM, and BAB. Source: Research Affiliates, LLC, based on data from CRSP and Compustat. 33

34 When It Comes to Implementation They Miss the Mark Big Time! Portfolio Trading Costs, Capacity, Turnover, and Leverage (January 1974 June 2016) Annual Return Ann. Return Net of Costs Long Leg Turnover Short Leg Turnover Trading Cost (bps) Capacity ($Bn) Average Long Leg Leverage Average Short Leg Leverage Fundamental Index original 12.9% 12.9% 11% % 0% Factor Replication 12.0% 11.7% 57% 15% % 13% Factor Replication Long Only 12.0% 11.8% 50% % 0% Equal Weight original 13.1% 13.0% 18% % 0% Factor Replication 12.8% 12.6% 36% 2% % 1% Factor Replication Long Only 12.6% 12.4% 65% % 0% Minimum Variance original 12.4% 12.2% 25% % 0% Factor Replication 12.2% 11.9% 37% 14% % 11% Factor Replication Long Only 12.3% 12.1% 75% % 0% Source: Research Affiliates, LLC, based on data from CRSP and Compustat. 34

35 Why Factor Tilts Are Not True Smart Beta» Fundamental Index, as one example, is not a repackaged value» Smart beta strategies have factor tilts, but they are much more than this» Smart beta strategies deliver alpha net of their Fama French four- or five factor regressions; factor-replicated strategies do not» Factor loadings show us some of the systematic drivers of return» But, simple factor tilt strategies based on theoretical factors are not the best way to capture return premiums» Replication portfolios have lower performance, higher turnover, and smaller capacity» Construction details matter!» Efficient implementation taking into account transaction costs can help better capture the premia 35

36 PART II c. Manager Evaluation 36

37 The Big Picture» Trend chasing is tempting and damaging» Investors tend to focus overwhelmingly on past performance in making investment decisions, especially for fund selection» But, past performance is worse than useless» To win, ask if the price is right» Valuation is a powerful tool in gauging future performance» Fees and turnover can eat away take-home returns» Implication for manager evaluation» We find reversal, not persistence, in mutual fund manager performance» We think there are reliable quantitative metrics that can help» Fund style return expectation*» Fees and turnover * The first metric is drawn from the methodology of Arnott, Beck, Kalesnik (2016a, b, and c). 37

38 Mutual Fund Data Sample» Fund inclusion criteria» We used Morningstar Direct survivorship bias-free fund universe» US open-ended long-only active equity funds» We included A-share, No-load-share, and Institutional-share» For funds with multiple subclasses (share-classes), we select the subclass with the oldest inception date» Funds with at least two-year return history from January 1990 to December 2016» Our sample data» A total of 3,331 funds, classified into 9 Morningstar categories:» Large Blend/Growth/Value» Mid-Cap Blend/Growth/Value» Small Blend/Growth/Value» Fund returns at monthly frequency and are net-of-fees 38

39 Does Past Skill Predict Future Skill?» The belief that manager outperformance* is persistent is false» Manager performance is mean-reverting. Among many possible predictors for return,* one of the most powerful and reliable predictor is the simple past return with the wrong sign!» Pooled correlation between past three-year simple return and subsequent three-year simple return is -26%, with an adjusted t-stat of -6.75» Positive reasons to hire a manager (i.e., good things, like past alpha) are never even half as powerful *We looked at an array of return measurement which included: simple return, excess return relative to benchmark or peer group, CAPM risk-adjusted return, and Fama French alpha, net of factor returns. 39

40 A Naïve Contrarian Strategy Can Work Average Mutual Fund Subsequent Three-Year Performance, Sorted by Prior Three-Year Returns, US Long-Only Equity Funds (Jan 1990 Dec 2016) 11.7% 11.0% Subsequent 3-Year Average Annualized Return 10.6% 10.3% 10.2% 10.1% 9.9% 10.0% 9.8% 9.7% Average 10.3% = Lowest Decile Decile Portfolios, Based On Prior 3-Year Returns = Top Decile This result is arguably created by our industry s favorite decision rule: Three bad years and out! Source: Research Affiliates, LLC, based on data from Morningstar Direct. 40

41 Factor Tilts Can Help Predict Future Relative Performance» Managers with exposure to out-of-favor factors outperform the market» We can measure fund factor loadings» We can estimate factor expected returns based on valuations*» Therefore, we can predict fund returns» Pooled correlation between fund style expectation (i.e., fund return expectation based on manager factor tilt) and subsequent one-year excess return relative to market is 29%, with an adjusted t-stat of 10.30» This is also the strongest relationship we found * The factor return forecasts (valuation-ratio based model plus structural alpha) are done following the methodology introduced in Arnott, Beck, Kalesnik, West (2016), and Arnott, Beck, Kalesnik (2016a and b). 41

42 Fees and Turnover» Managers with higher fees tend to underperform their peers» In the median-run (future three-year) and long-run (future five-year and above), 10 bps more expense delivers more than 10 bps lower relative net-of-fee performance» The relationship between turnover and future performance is inconclusive» Managers with higher turnover seem to underperform their peers» But, high turnover seems to be associated with higher return relative to market 42

43 PART II d. Momentum: Crowded Space and the Problem of Stale Momentum Crashes 43

44 Troubles with Momentum: Factor Premia Hard to Capture and Funds Underperform Average Annualized Relative Performance Across Funds (Jan 1991 Dec 2016*) Keyword No. of Funds Average Value-add Relative to Market Average FF4 Alpha RAFI** Funds 9 2.0% 0.8% Small % -1.0% Multi-Factor*** % -1.0% Value % -0.3% Opportunity % -0.7% Growth % -1.0% Research % -1.1% Advantage % -1.0% Dynamic % -2.1% Fundamental % 0.2% Dividend % 0.0% Contrarian % -0.2% Volatility % -2.0% Large % -1.1% Momentum % -2.5% Income % -0.4% Quality % -0.3% Source: Research Affiliates, LLC, based on data from Morningstar Direct. *Data sample includes all US equity mutual funds, including non-survivors, with "A" or No-load or institutional share-classes. The oldest share class is kept for funds with multiple share classes. The reported performance stats are averaged within each month, and compounded over the full span. ** Funds with "RAFI" or "RAE" or "Fundammental Index" in the name ***Funds (ETFs and Open-ended Funds) labeled as "multi-factor" in their "Strategic Beta Attribute" by Morningstar Direct, with at least 12 month return history. 44

45 Troubles with Momentum: Factor Premia Hard to Capture and Funds Underperform Average Annualized Relative Performance by Style, Jan 1991 Dec 2016* Panel A. Average Value-Add Relative to Market Sorting Variable Top Decile Funds Bottom Decile Funds Top Decile minus Bottom Decile t-stats (Top - Bottom) Market Beta 1.3% -0.8% 2.0% 0.72 Size Beta 2.3% -1.0% 3.3% 1.21 Value Beta 1.4% 0.5% 0.9% 0.34 Momentum Beta 1.2% 0.1% 1.1% 0.53 Panel B. Average FF4 Alpha Sorting Variable Top Decile Funds Bottom Decile Funds Top Decile minus Bottom Decile t-stats (Top - Bottom) Market Beta -1.9% 0.5% -2.4% Size Beta -1.2% -0.4% -0.8% Value Beta -0.2% -1.0% 0.8% 0.65 Momentum Beta -1.7% -0.4% -1.3% Source: Research Affiliates, LLC, based on data from Morningstar. *Data sample includes all US equity mutual funds, including non-survivors, with "A" or No-load or institutional share-classes. The oldest share class is kept for funds with multiple share classes. Decile funds are sorted based on fund full-sample FF4 factor loading estimates. 45

46 Does Momentum Trade Get More Crowded in Bubble Stocks? 8% Cumulative Returns Over Subsequent Months, United States (Jan 1963 Dec 2016) 4% Cumulative Returns 0% -4% -8% Months Standard Momentum Stale Momentum Source: Research Affiliates, LLC, based on data from Compustat and CRSP. 46

47 Stale Momentum Poor Performance and Crashes Dollar Growth of Standard and Stale Momentum, United States (Jan 1963 Dec 2016) Dollar Growth (Log) Year Standard Momentum Stale Momentum Source: Research Affiliates, LLC, based on data from Compustat and CRSP. 47

48 Saving Momentum» Harvesting momentum premium seems to be mission impossible» First, there may be no momentum effect left; it s basically flat-lined since 1999» And that s for long/short paper portfolio investors, before fees!» And before trading costs!» Is momentum trade crowded?» What s the Achilles Heel for momentum? Well, there are two:» How about shunning stale momentum?» Once momentum has been running for two years or more, stocks are very expensive, running out of gas.» It s dreadful!! Shun stale momentum.» How about momentum conscious trading of other strategies?» If a strategy wants to buy a stock in free fall, why not wait a month?» If a strategy wants to sell a stock that s soaring, why not wait a month?» Trading costs? Zero-to-negative! 48

49 Thank You research-affiliates 49

50 Important Information By accepting this document you agree to keep its contents confidential. You also agree not to disclose the contents of this document to third parties (including potential coinvestors) without the prior written permission of Research Affiliates, LLC (including its related entities). The material contained in this document is for information purposes only. This material is not intended as an offer or solicitation for the purchase or sale of any security or financial instrument, nor is it advice or a recommendation to enter into any transaction. Any offer to sell or a solicitation of an offer to buy or sell shall be made solely by the legally permitted entity registered or licensed with the applicable regulator(s) and only to qualified investors through a regulatory registration statement or subject to an applicable exemption and disclosure document such as a private placement memorandum for pooled investment vehicles, or investment management agreement for separately managed accounts. This information is intended to supplement information contained in the respective disclosure documents. The information contained herein should not be construed as financial or investment advice on any subject matter. Research Affiliates, LLC and its related entities do not warrant the accuracy of the information provided herein, either expressed or implied, for any particular purpose. The index data published herein may be simulated and in such instances, no allowance has been made for trading costs, management fees, or other costs, are not indicative of any specific investment, are unmanaged and cannot be invested in directly. Past performance, including simulated performance, is no guarantee of future performance and actual investment results may differ. Any information and data pertaining to an index contained in this document relate only to the index itself and not to any asset management product based on the index. With the exception of the data on Research Affiliates Fundamental Index, all other information and data are based on information and data from third party sources. Investors should be aware of the risks associated with data sources and quantitative processes used in our investment management process. Errors may exist in data acquired from third party vendors, the construction of model portfolios, and in coding related to the index and portfolio construction process. While Research Affiliates takes steps to identify data and process errors so as to minimize the potential impact of such errors on index and portfolio performance, it cannot guarantee that such errors will not occur. Certain indices referenced herein may be calculated by S&P Dow Jones Indices LLC or its affiliates. S&P is a registered trademark of Standard & Poor s Financial Services LLC ( S&P ) and Dow Jones is a registered trademark of Dow Jones Trademark Holdings LLC ( Dow Jones ) and have been licensed for use by S&P Dow Jones Indices LLC and its affiliates. Investment products based on any such indices are not sponsored, endorsed, sold or promoted by S&P Dow Jones Indices LLC, Dow Jones, S&P or their respective affiliates and none of S&P Dow Jones Indices LLC, Dow Jones, S&P or their respective affiliates makes any representation regarding the advisability of investing in such product(s). S&P Dow Jones Indices LLC, its affiliates, sources and distribution agents, and each of their respective officers, directors, employees, agents, representatives and licensors (collectively, the Index Calculation Agent ) shall not be liable to Research Affiliates, LLC, any customer or any third party for any loss or damage, direct, indirect or consequential, arising from (i) any inaccuracy or incompleteness in, or delays, interruptions, errors or omissions in the delivery of any index or any data related thereto (the Index Data ) or (ii) any decision made or action taken by Research Affiliates, LLC, any customer or third party in reliance upon the Index Data. The Index Calculation Agent does not make any warranties, express or implied, to Research Affiliates, LLC, any of its customers or anyone else regarding the Index Data, including, without limitation, any warranties with respect to the timeliness, sequence, accuracy, completeness, currentness, merchantability, quality or fitness for a particular purpose or any warranties as to the results to be obtained by Research Affiliates, LLC, any of its customers or other person in connection with the use of the Index Data. The Index Calculation Agent shall not be liable to Research Affiliates, LLC, its customers or other third parties for loss of business revenues, lost profits or any indirect, consequential, special or similar damages whatsoever, whether in contract, tort or otherwise, even if advised of the possibility of such damages. Frank Russell Company is the source and owner of the Russell index data contained or reflected in this material and copyrights related thereto. Frank Russell Company and Research Affiliates, LLC have entered into a strategic alliance with respect to the Russell RAFI Indexes. The Russell RAFI Indexes are calculated by Russell in conjunction with Research Affiliates LLC. All intellectual property rights in the Russell RAFI Indexes (the Index ) vest in Russell and RA. Neither Russell nor RA or their licensors accept any liability for any errors or omissions in the Index and / or Index ratings or underlying data. No further distribution of Russell Data is permitted without Russell s express written consent. Russell is a trademark of Frank Russell Company. Research Affiliates, LLC is the owner of the trademarks, service marks, patents and copyrights related to the Fundamental Index and the Fundamental Index methodology. The presentation may contain confidential information and unauthorized use, disclosure, copying, dissemination, or redistribution is strictly prohibited. This is a presentation of Research Affiliates, LLC. Frank Russell Company is not responsible for the formatting or configuration of this material or for any inaccuracy in Research Affiliates presentation thereof. 50

51 Important Information The FTSE Research Affiliates Fundamental Indexes are calculated by FTSE International Limited ( FTSE ) in conjunction with Research Affiliates, LLC, ( RA ). All rights and interests in the FTSE Research Affiliates Fundamental Indexes vest in FTSE. All rights in and to the RA Fundamental Index weighting methodology used in the calculation of the FTSE Research Affiliates Fundamental Indexes vest in RA. All rights in the FTSE indexes and / or FTSE ratings (together the FTSE Data ) vest in FTSE and/or its licensors. Except to the extent disallowed by applicable law, neither FTSE nor RA nor their licensors shall be liable (including in negligence) for any loss arising out of use of the FTSE Research Affiliates Fundamental Indexes, the FTSE Data or underlying data by any person. FTSE is a trademark of the London Stock Exchange Plc and is used by FTSE under license. Research Affiliates and Fundamental Index are trademarks of RA. FTSE is not an investment adviser and makes no representation regarding the advisability of investing in any security. Inclusion of a security in an index is not a recommendation to buy, sell or hold such security. No further distribution of FTSE Data is permitted without FTSE s express written consent. The RAFI Bonds US Investment Grade Index and RAFI Bonds US High Yield Index is calculated by ALM Research Solutions, LLC. in conjunction with Research Affiliates, LLC. All rights and interests in the RAFI Bonds US Investment Grade Index and the RAFI Bonds US High Yield Index vest in Research Affiliates, LLC. All rights in and to the Research Affiliates, LLC Fundamental Index concept used in the calculation of the RAFI Bonds US Investment Grade Index and the RAFI Bonds US High Yield Index vest in Research Affiliates, LLC. The RAFI Bonds US Investment Grade Index and the RAFI Bonds US High Yield Index are not sponsored, endorsed, sold or promoted by ALM Research Solutions, LLC., and ALM Research Solutions, LLC. makes no representation regarding the advisability of investing in such product(s). Neither ALM Research Solutions, LLC. nor Research Affiliates, LLC make any warranties, express or implied, to any of their customers nor anyone else regarding the accuracy or completeness of any data related to the RAFI Bonds US Investment Grade Index or the RAFI Bonds US High Yield Index. All information is provided for information purposes only. Neither ALM Research Solutions, LLC. nor Research Affiliates, LLC accept any liability for any errors or any loss arising from the use of information in this publication. Copyright MSCI. All Rights Reserved. Without prior written permission of MSCI, this information and any other MSCI intellectual property may only be used for your internal use, may not be reproduced or redisseminated in any form and may not be used to create any financial instruments or products or any indices. This information is provided on an as is basis, and the user of this information assumes the entire risk of any use made of this information. Neither MSCI nor any third party involved in or related to the computing or compiling of the data makes any express or implied warranties, representations or guarantees concerning the MSCI index-related data, and in no event will MSCI or any third party have any liability for any direct, indirect, special, punitive, consequential or any other damages (including lost profits) relating to any use of this information. The trademarks Fundamental Index, RAFI, Research Affiliates Equity, RAE, and the Research Affiliates trademark and corporate name and all related logos are the exclusive intellectual property of Research Affiliates, LLC and in some cases are registered trademarks in the U.S. and other countries. Various features of the Fundamental Index methodology, including an accounting databased non-capitalization data processing system and method for creating and weighting an index of securities, are protected by various patents, and patent-pending intellectual property of Research Affiliates, LLC. (See all applicable US Patents, Patent Publications, Patent Pending intellectual property and protected trademarks located at which are fully incorporated herein.) Any use of these trademarks, logos, patented or patent pending methodologies without the prior written permission of Research Affiliates, LLC, is expressly prohibited. Research Affiliates, LLC, reserves the right to take any and all necessary action to preserve all of its rights, title, and interest in and to these marks, patents or pending patents. Neither Research Affiliates, LLC nor its related entities provide investment advice outside of the United States, Canada and Ireland or offer or sell any securities, commodities or derivative instruments or products. Any such business may only be conducted through registered or licensed entities and individuals permitted to do so within the respective jurisdiction and only in conjunction with the legally required disclosure documents and subject to the all legally required regulatory filings. Research Affiliates, LLC is an investment adviser registered under the Investment Advisors Act of 1940 with the U.S. Securities and Exchange Commission (SEC). Our registration as an Investment Adviser does not imply a certain level of skill or training. Research Affiliates, LLC, its affiliates, agents, and each of their respective officers, directors, employees, agents, representatives and licensors (collectively RA ) shall not be liable to any third party for any loss or damage, direct, indirect or consequential, arising from or related to (i) any inaccuracy or incompleteness in, errors or omissions in the enclosed information (the Presentation Material ) or (ii) any decision made or action taken by any third party in reliance upon the Presentation Material. RA does not make any warranties, express or implied, to anyone regarding the Presentation Material, including, without limitation, any warranties with respect to the timeliness, sequence, accuracy, completeness, currentness, merchantability, quality or fitness for a particular purpose or any warranties as to the results to be obtained by any third-party in connection with the use of the Presentation Material. RA shall not be liable to anyone for loss of business revenues, lost profits or any indirect, consequential, special or similar damages whatsoever, whether in contract, tort or otherwise, even if advised of the possibility of such damages Research Affiliates, LLC. All rights reserved. Duplication or dissemination prohibited without prior written permission. 51