Active Share and Emerging Market Equity Funds

1. Active Share and Emerging Market Equity Funds Aron Gottesman And Matthew Morey

2. Motivation • Prevailing wisdom that emerging markets are less efficient than developed markets. • Much more political and economic risk but significant growth opportunities. • Higher volatility in these markets creates opportunities for active fund managers to outperform.

3. Mixed Past Evidence • Mixed past evidence as to whether active management outperforms passive in emerging market equity funds • Little or no evidence of outperformance: Gottesman and Morey (2007) and Basu and Huang-Jones (2015) • Active Management outperforms: Huij and Post (2011) and Lin (2013). • Eling and Faust (2010) find hedge funds outperform but mutual funds do not.

4. We revisit the Question using Active Share Developed first by Cremers and Petajisto 2009), it examines how active is the portfolio. Uses the fund’s holdings data to measure what percentage of the fund’s holdingsdiffer from the fund’s benchmark index

5. Active Share

6. Active Share (100 percent)

7. Active Share (other examples) 0 percent Active Share: Let us consider a fund with a $100 million portfolio benchmarked against the S&P 500. Imagine that the manager starts by investing $100 million in the index, thus having a pure index fund with 500 stocks. The active share here would be zero. A pure index fund. 50 percent active share: Assume the manager only likes half of the stocks, so he eliminates the other half from his portfolio, generating $50 million in cash, and then he invests that $50 million in those stocks he likes. This produces an Active Share of 50% (i.e. 50% overlap with the index). .

8. Active Share Another Example 90 Percent Active Share: If he invests in only 50 stocks out of 500, his Active Share will be 90% (i.e., 10% overlap with the index). So active share is giving one an idea of how different the portfolio is from the index. 100 percent is completely active and 0 is a completely passive.

9. Range of Activeness According to Petajisto (2013), funds that have an active share above 80 percent are truly active funds as they are holding a portfolio that is substantially different from the benchmark index. Funds with active shares of 60 percent or below are closet indexers as they are holding a portfolio that bears a substantial resemblance to an index fund. Funds with active shares of 20 percent or below are pure index funds.

12. Dynamics of Active Share • Some funds are consistent in their level of activeness. Keep their active share about the same over time. • Some funds drastically change their activeness. They time their activeness.

13. Down Markets and Active Share • The literature has found that during down markets there is more closet indexing. Underperforming in a down market is seen as worse than underperforming in an up market. • We should, rationally, see the opposite. We should see higher activeness in down, volatile markets when return dispersion is greater. This is when a truly skilled active manager can outperform.

14. Data and Methodology(Typical Approach) Typical papers using active share use a pooled approach. They take all the funds in a year, examine the fund characteristics (including active share) and then examine the 1-year performance metrics using the Carhart approach. They take the past 3 years returns and estimate a 4-index alpha. They then subtract the expected return from the realized fund return to estimate the fund abnormal return (alpha) in each year, which is measured as the sum of the intercept of the model and the residual as in Carhart (1997).

15. Problems with Typical Approach It give an extremely large sample that makes it too easy to find significance. Indeed, in one paper they have 346,711 observations! It also creates pooling problems (that are not discussed in the papers) due to serial correlation and unbalanced pools of data. It does not allow for examining within a fund, the time series of the active share as you are instead pooling individual years together. Uses a non-straightforward measure of performance that is difficult for practitioners to understand. Not typical for a retail investor at all.

16. We then use a more Straightforward Approach We take the approach of an actual investor at the end of 2008, who buys a fund and then holds it for six years. We use this period as it when Morningstar starts providing the data we need. Hence, we select all the funds that meet our criterion .We then follow all these funds for six years (2009-2014). If a fund drops outs before the end of the 2014, we still include the fund so as not have a survivorship bias issue.

17. Choice of Funds For a fund to be selected the fund has be: 1)On the Morningstar database as of end of 2008 (hence only U.S. based funds) 2) Be Diversified Emerging Market fund. 3) Have a self-stated prospectus benchmark of the MSCI emerging market index

18. Choice of Funds We choose these funds because: • The largest number of emerging market equity funds on the Morningstar database who shared the same benchmark. • There currently exist size, value and momentum indices for the MSCI emerging markets index. Allows us to calculate 4-index alpha. • Frazzini, Friedman and Pomorski (2016, FAJ)

19. Frazzini, Friedman and Pomorski They show that funds (U.S. based domestic equity funds) with high active share have benchmarks that consistently underperform the benchmarks of funds with low active share. Hence, funds with high active share are outperforming because their benchmarks perform poorly and funds with low active share are underperforming because their benchmark returns perform relatively well.

20. Frazzini, Friedman and Pomorski Indeed, according to Frazzini, Freidman and Pomorski, when one examines funds only within one benchmark, active share does not predict fund performance.

21. Frazzini, Friedman and Pomorski In light of this issue, we choose to examine only one style of fund (diversified emerging markets) with one prospectus benchmark (the MSCI emerging markets equity index). If we had chosen to include other styles of funds (such as Asia ex Japan, Latin America and China funds) we would have had to use other benchmarks and our results may have the problem described above, i.e., high active share funds that are doing well because they are being compared to benchmarks that underperform. Also there are just not that many other types of funds. Limited amount of U.S. based emerging market funds.

22. Data and Methodology So in sum, we then take all the diversified emerging market funds with a stated benchmark index of the MSCI emerging market index as of January 1, 2009. We then follow these funds for six years. This produces a sample of 67 funds. (Relatively small sample is somewhat problematic)

23. Non-Surviving Funds If a fund’s quarterly active share data disappears for more than one quarter or its monthly returns or other fund characteristics (net assets, expense ratio, and turnover) discontinue due to a merger or liquidation we consider the fund to be a non-surviving fund. 53 of the 67 funds survive the entire 6-year horizon period. 14 funds do not survive the entire sample.

24. Non-Surviving Funds For the funds that drop out of the sample, we use the funds actually data before the fund drops out of the sample. After the fund drops out we assume the fund takes on the characteristics of the average surviving fund. Hence, after the fund has dropped out, the monthly returns, active share, net assets, expense ratio, and turnover of the fund are those of the average surviving fund.

25. Costs/Benefits of Our Approach Cost—much smaller sample, survivorship bias issues. Benefits— • Use standard measure of performance over a standard performance holding period. Similar to what actual investors might do. • No pooling issues. • Time series of active share within each fund.

26. Hence For each surviving fund—we have 24 quarters of active share data (2009-2014). For expense ratio, fund size, fund turnover we have 6 years of observations for each fund (at the beginning of each year)

27. Sample Period: 2009-2014

29. Regression

31. Without Controls

32. Only Surviving Funds

33. Only Surviving Funds (without Controls)

34. Consistency of Active Share • Some funds alter their active share over time. Again, in down markets, active share seems to fall. • We wanted to examine how important the consistency of active share was to performance. • To measure the consistency of active share we use the standard deviation of a fund’s active share over the 24 quarters (2009-2014). • Advantage of our data is that we can calculate a meaningful standard deviation of active share.

35. Correlation between Standard deviation of active share and other variables

37. No Controls

38. Only Surviving Funds

39. Only Surviving Funds (no controls)

40. Interaction Effects We examine the interaction between a fund’s average level of active share and the fund’s standard deviation of active share. To do this we create a dummy variable called (High Active Share Dummy) that takes a value of 1 if a fund was in the top quintile of funds in terms of average active share. Similarly, we create another dummy variable, (Low Active Share Dummy) that takes a 1 if a fund was in the bottom quintile of funds in terms of average active share. We then regress the standard deviation of active share on performance but use the above created dummies to examine the interaction between the level of average active share and the standard deviation in active share. price.

42. Interaction Results This result indicates that highly active funds that alter their level of active share over time will have significantly worse performance than highly active funds that do not alter their active share over time. Hence, it is the highly active funds that remain consistently active (have a low standard deviation of active share) who truly outperform other funds. On the other hand, we find that altering the activeness of the portfolio over time does not significantly impact the performance of low active share funds

43. Closet Indexing in Emerging Market Funds To measure the closet indexing effect in emerging market equity funds we first find for each separate quarter from 2009-2014 the number of funds whose prospectus benchmark is the MSCI emerging market equity index. Hence, this is a different sample than used in the previous analysis. We then calculate, for each separate quarter, the percentage of these funds that had an active share of 60 percent or below and thus are said to be closet indexing according to Cremers and Petajisto (2009).

45. Summing Up Results • More active emerging market equity funds outperform less active funds. • Funds that alter their active share over time have significantly worse performance. However, this effect seems to be mostly in funds that are highly active. Funds that are not as active do not seem to suffer as much in performance if they vary their level of activeness over time. • Expenses matter. In every regression we run, expenses are negatively and significantly related to performance. • Closet Indexing has fallen over time

46. Caveats • Small Sample but six years of time series data for most funds. Tries to be consistent with what actual investor might do and give a sample that allows one to measure, within a fund, the time series of activeness. • Survivorship issues, but we try to adjust for these issues.

47. Conclusion Truly active fund that are consistent in their activeness and have lower expense ratios seem to outperform in emerging markets.