Active vs. Passive Management: How to Separate SAMs from IAMs

Transcription

1 Active vs. Passive Management: How to Separate SAMs from IAMs Russ Wermers Bank of America Professor of Finance Director, Center for Financial Policy University of Maryland

2 Agenda 1. Does active management add value? 2. Can we identify superior active managers (SAMs) on an ex ante basis? 3. What are the best quantitative tools for assessing active managers? 2

3 Summary of Findings: 1. The average active manager does not add value, when strictly measured using returns vs. benchmark 2. Many active managers do add value and it may be possible to identify them in advance 3. (A) Portfolio-holdings based, and (B) Macroeconomic returns based are two recent quantitative advances of importance to find superior managers 3

4 Does active management add value? 4

5 The Efficient Market Hypothesis Market prices reflect all (public) information (?) Average active manager cannot add value A tautology? (The Arithmetic of Active Management, Sharpe) Market = Σ (All Active Managers) + Σ (All Index Funds) + Σ (All Individual Investors) Active management is a zero sum game? Depends on costs of indexing Active managers charge fees and incur expenses Return Σ (Active Managers) = Overall market return minus fees and expenses, unless they provide liquidity to index funds or overweight securities/sectors in aggregate 5

6 EMH Empirical Results Numerous studies support the EMH Early studies Jensen (1968), no CAPM alpha in average mutual fund returns, net of fees and expenses More recent studies Barra, Scaillet and Wermers (2010) study mutual funds and adjust for 4-factor model Goyal and Wahal (2010) study institutional separate accounts and also adjust for risk After tax comparisons are even worse 6

7 Recent Tests of the EMH Four factor alphas, in percent per year, for Equity Mutual Funds and Equity Institutional Separate Accounts, net of fees and expenses 0.6 Annualized Four Factor Alpha (net of fees and expenses) 0 Based on Mutual Fund Data (from Jan 1975 Dec 2006) Based on Institutional Separate Account Data (from ) -0.6 All Funds All Growth Funds Aggressive Growth Growth and Income Equally- Weighted Net Value- Weighted Net Source: The mutual fund data is from Barras, Scaillet and Wermers (2010), while the institutional separate account data is from Busse, Goyal and Wahal (2010) 7

8 Why Does Active Management Persist? Are investors irrational? If so, how can the market be efficient? Empirical evidence of biases, but may cancel out Grossman and Stiglitz (1980) Information is costly Markets must be mostly but not completely efficient to encourage investors to gather and analyze information An efficient degree of inefficiency ( informationally efficient markets ) 8

9 A Zero-Sum Game? Contest between active managers may be a zero-sum game (but, not a certainty at this point) Active managers are essential to efficient capital markets and will always exist: Improves market efficiency and capital allocation Leads to greater economic efficiency and growth Producing greater wealth for society as a whole And, do not hold market portfolio in aggregate! 9

10 Can we identify superior active managers (SAMs) on an ex ante basis? 10

11 Playing the Zero-Sum Game Active management is a SAM-IAM contest ( Superior Active Managers and Inferior Active Managers ) SAMs exploit IAMs (and other investors) by Better information or analysis (skill) Supplying liquidity and immediacy But Skill and liquidity needs can vary over time and across stocks Today s SAMs can be tomorrow s IAMS So who is whom? Σαµ Ι αµ 11

12 Finding Tomorrow s SAMs Today Academic studies focus on four basic approaches, looking at 1) Past performance (properly adjusted) 2) Macro economic relationships 3) Fund/manager characteristics 4) Fund holdings Investors should consider all of these when selecting funds 12

13 Past Performance If skill persists, alpha should persist Offsetting effects Poor performers get replaced Strong performers may grow too big or raise fees Changing opportunity set (e.g., internet bubble) Hence, lack of persistence does not necessarily imply lack of skill Evidence indicates modest persistence in properlyadjusted returns 13

14 Proof of Persistence: Recent Techniques Show that Skill is Persistent! Theme: Must use proper methods. Harlow and Brown (2006) Style-adjusted returns improve odds of finding SAMs from 45% to 60% That is, controlling for value/growth and small cap/large cap tilts allows a more precise capture of stock-selection skills Pastor and Stambaugh (2002) Better results after adjusting for sector biases For example, evaluating technology funds is better conducted by adding a technology index to the model; for financials, a financial industry index Kosowski, Timmerman, Wermers and White (2006) Better results adjusting for non-normality (fat tails, skewness) Dynamic strategies of funds create non-normally distributed returns Even relatively static strategies can hold portfolios with non-normally dist d returns Poses a challenge to standard modeling, which usually assumes normality Bootstrapping which is a simulation procedure that takes random samples from real fund return data helps to correct this 14

15 Persistence Persistence in Past Performance Decile by Prior Alpha Kosowski, Timmerman, Wermers and White Four Factor Alpha Model* Harlow and Brown Three Factor Alpha Model** * Harlow and Brown use a three factor alpha methodology rebalanced quarterly using the time period ** Kosowski, Timmerman, Wermers and White use a four factor alpha methodology using a three-year ranking period, with a bootstrapping technique to model non-normality, rebalanced annually using the time period

16 Past Performance: Conclusions Need a sophisticated performance attribution system to distinguish luck from skill Even then, persistence is modest When changing managers, make sure expected gains exceed switching costs Goyal and Wahal (2008) find that, for institutions, hired managers do not outperform fired managers, on average Past performance is less useful after a manager change Baks (2003) finds 10% 50% of persistence is from the manager, rest from other factors (firm, research team, etc.) 16

17 Macro Economic Correlations Moskowitz (2000) and Kosowski (2006) Active managers do better in periods of greater uncertainty (recessions, volatile markets) But it is hard to accurately forecast these periods in advance Alpha performance in recession and expansion periods, Kosowski (2006) All Funds All Growth Funds Aggressive Growth Annualized Four Factor Alpha Annualized Four Factor Alpha Annualized Four Factor Alpha Full Sample ( ) Expansion -2 Full Sample ( ) Expansion -2 Full Sample ( ) Expansion Growth Growth and Income Balanced Income Annualized Four Factor Alpha Full Sample ( ) Expansion Annualized Four Factor Alpha Full Sample ( ) Expansion Annualized Four Factor Alpha Full Sample ( ) Expansion 17

18 Macro Economic Forecasts Avramov and Wermers (2006) Relate future U.S. equity mutual fund returns to current macro conditions Short rates; dividend yield; default spread; term structure Style-adjusted (4-factor) alphas of 600 bps/yr when able to rotate to expected top-performing funds Banegas, Gillen, Timmerman and Wermers (2009) Use additional predictor variables and find similar results for European equity mutual funds Avramov, Kosowski, and Teo (2007) Include VIX and get even better results (1200 bps of alpha) for hedge funds 18

19 Some Cautions These strategies have high turnover 200% 300% annually Not practical when manager transition costs are high E.g., early withdrawal penalties, long lock-up periods, high trading costs Possible solution: Diversify and rebalance at the margin May conflict with a return-persistence strategy Buy recent losers when macro conditions change 19

20 Characteristics: Fund Manager Experienced managers of large funds outperform; not so for small funds Ding and Wermers (2009) Social connections lead to better performance Cohen, Frazzini and Malloy (2008) Graduates of better colleges Chevalier and Ellison (1999): Undergraduate degree Gottesman and Morey (2006): MBA degree CFAs manage risk better, but don t outperform Dincer, Gregory-Allen and Shawky (2010) Hedge fund managers who invest in their own funds De Souza and Gokcan (2003) 20

21 Characteristics: Fund Management Company Larger companies with more research resources Chen, Hong, Huang and Kubik (2004) Busse, Goyal and Wahal (2010) More independent directors Ding and Wermers (2009) Flatter organizational structure Massa and Zhang (2009) 21

22 Characteristics: Fund Itself Lower expense ratios outperform in every time period and fund category (but this doesn t mean low ER is the only factor that matters!) Kinnel, Morningstar (2010) Less cash drag = better performance Edelen (1999) indicates need to examine flow volatility Industry/sector specialist funds outperform Kasperczyk, Sialm and Zheng (2005) More style drift = better performance Wermers (2002) 22

23 Characteristics: Hedge Funds High water marks, higher incentive fees and longer lock-up periods Liang (1999) Goldilocks funds: not too large or too small Getmansky (2005): concave relationship of performance and fund size Conflicting evidence on fund age Howell (2001): Young funds better De Souza and Gokcan (2003): Seasoned funds better 23

24 Fund Holdings Smaller/positive return gap = better performance Kasperczyk, Sialm and Zheng (2008) Risk shifting = worse performance Huang, Sialm and Zheng (2010) Contrarian managers outperform herding managers Wei, Wermers and Yao (2009) 24

25 Active Share Cremers and Petajisto (2009) Higher active share = better performance CP claim this is due to higher conviction Higher tracking error better performance Qualifications Don t control for benchmark capitalization Small-cap portfolios tend to have higher active shares Better performance of SC funds consistent with costly information thesis of Grossman and Stiglitz Overconfidence literature warns against excessive conviction 25

26 Recent Research Provides Further Valuable Insights on Active Management 1. Low-turnover mutual funds outperform (on average) highturnover funds, all else equal A. Cremers and Pareek (2016) find that high Active Share is especially predictive when coupled with low turnover (Q5 Duration is the lowest turnover; Q5 Active Share is the highest Active Share): 26

27 Recent Research Provides Further Valuable Insights on Active Management 2. Competition reduces the probability of achieving active alpha A. Hoberg, Kumar, and Prabhala (2016) measure competition as the number of rival funds that are close (in Euclidian distance) to a subject fund in the style dimensions of (1) size (2) book-to-market and (3) momentum. B. Funds fare better, in the future, when they have fewer competitors that are close to their style space High past alpha funds in style areas with low competition generate an alpha of 4.5%/year High past alpha funds in style areas with high competition generate an alpha of only 1%/year 27

28 Recent Evidence of Active vs. Passive We can draw some evidence from Paul (2009), who shows that the advantage of active management is highly reliant on the dispersion of returns of stocks In other words, periods of high correlations among stocks are not conducive to active manager outperformance; the recent drop in such correlations should prove advantageous to active management 28

29 Conclusions Average active manager does not outperform But active management is essential for market efficiency SAMs should continue to earn an economic rent Academic evidence suggests it may be possible to identify SAMs in advance Multiple methods seem promising: (1) past returns properly adjusted for risk, (2) holdings-based, (3) macroeconomic returns-based, and (4) manager/fund/company characteristics Comprehensive approach works best Portfolio holdings-based measures and macroeconomic returns-based measures are components most underused by investors; however, even past returns are often misused! 29