Investor Attrition and Fund Flows in Mutual Funds

Transcription

1 Investor Attrition and Fund Flows in Mutual Funds

2 Liquidity and Mutual Funds Open-end mutual funds provide liquidity to investors Investors can liquidate shares at any point Creates a large management problem for managers Predicting when investors will leave Reduces returns by as much as % because of forced trades, price impact, or cash that needs to be held Is there a way to easily identify funds with more stable set of investors? Could be valuable when thinking of acquiring another fund Could be valuable when trying to evaluate the value of a management fee Could be valuable to managers in deciding liquid assets Could be valuable to investors wanting to know if others leave

3 Performance-flow We know fund performance is one of the biggest factors influencing investor decisions to invest or not Would like to estimate the likeliness investors will leave (Rank below 50% funds expect to lose money)

4 Problems of relying on flows Often need long history of information of a fund and for a large crosssection of a funds to estimate flows Sensitivities of investors today may be different than several years in the past Does not really allow us to determine if one fund is more likely to have outflows than another aside from looking at performance No fund specific characteristics Consider how the fund does relative to others Is there an alternative way to measure flow-sensitivities which is fund specific?

5 Attrition Conditions under which investors arrive, leave, and stay reveal preferences Samuelson (1938, 1948) Take the example of a fund with two types of investors: (i) Performancesensitive (green) and (ii) service-sensitive (blue) Fund A Poor performance Fund A After poor performance, investors remaining are stable and insensitive to performance Also no new performance-sensitive investors would join

6 Attrition Want to find an easy measure to identify stable investors The case of Dreyfus Worldwide Fund Current Size Attrition = 1 Historical Max Size

7 Hypothesis 1 Objective for this paper is to determine if attrition is indeed a good predictor of flow-performance Relating attrition with flow-performance Attrition is easy to calculate and fund-specific No need to rely on long time-series H1: After a period of bad performance, performance-sensitive investors will leave as well as avoid investing in the fund. Fund attrition therefore associates with a lower flow-performance sensitivity of outflows, inflows, and net flows.

8 Homogenous beliefs and attrition Attrition may not be a particularly good predictor of performance sensitivity if it arises for some non-performance related reason Maybe customer has liquidity shock Fund A Fund A Fund A experiences attrition, unrelated to performance Remaining investors will still leave if performance sours

9 Attrition Conditions under which investors arrive, leave, and stay reveal preferences Samuelson (1938, 1948) Take the example of a fund with two types of investors: (i) Performancesensitive (green) and (ii) service-sensitive (blue) Fund A Poor service Fund A After experiencing poor service, investors who care about service will leave but remaining investors remain sensitive to performance Attrition has limited effect on performance-sensitivity

10 Hypothesis 2 and 3 Are there conditions when attrition may be a better predictor of how investors will react? H2: Attrition will have limited effect on the flow-performance sensitivity of outflows, inflows, or net flows if all investors have homogenous performance sensitivities. H3: Attrition will have no effect on the flow-performance sensitivity of outflows, inflows, or net flows if investors are leaving or avoiding a fund for non-performance reasons.

11 Attrition Attrition may be influenced by the conditions under which people arrive Take two funds which advertise and attract investors based on different qualities: (i) Performance vs (ii) Service Inflows to these funds attract different types of investors High Performer Super Service Period of bad performance for the High Performer fund will imply attrition has a much larger effect on flow-performance We can only associate with inflows but linking conditions of inflow with outflow could be very powerful

12 Data CRSP Attrition Quintile Attrition Size (mil) Age (years) Expenses (%) Ln Accum Return Flow (%) Mean % %

13 Return and Attrition Poor performance has two effects on attrition: Decreases the asset value which makes the fund shrink Increases the number of shareholders who want to leave the fund which decreases the number of shares Need to focus on the latter effect Remove direct effect of log accumulated returns on asset size Accumulated return is the return calculated from historical maximum

14 Predicting attrition Explain significant portion of attrition Use these as controls for expected attrition and the excess attrition is what we relate to fund flows Attrition Dummy =1 if attrition above median and 0 if below

15 H1: Attrition and Flows Want to determine if ATTRITION serves as a good proxy for flow-performance sensitivities by fund High attrition funds should be left with investors who are less sensitive to performance Low attrition funds should be left investors who are relatively more sensitive to performance Does attrition tell us anything about the investors who enter a fund in addition to those who have left? High attrition may mean investors who enter the fund are coming in for reasons aside from performance

16 H1: Attrition and Flows Predict flows as function of attrition and returns Age often associated with more stable assets but attrition seems to replace this High attrition funds have lower flowperformance sensitivity Observe both for inflows and outflows

17 H1: Attrition, Flows, and Returns Theses tests are unconditional but what happens if condition on poor performance? What if we condition on the fund actually experiencing bad returns Predictions should get stronger after bad performance Revealed preferences for performance more observable after bad performance

18 H1: Attrition, Flows, and Returns Run same test conditioning on fund with positive or negative cumulative returns Only after bad performance do we observe high attrition being good predictor of low performance sensitivity Consistent with revealed preference

19 H1: Attrition, Flows, and Returns Above- and below-median attrition Non-parametric estimation Observe the flatter flowperformance relation for high attrition Explain convexity in flows?

20 H2: Institutional vs. Retail If investors are homogeneous then attrition is not going to do the same job of separating investor types Institutional investors are more homogeneous in terms of pricesensitivities Expect attrition to have a better prediction of investor responsiveness amongst retail investors where greater diversity and heterogeneity

21 H2: Institutional vs. Retail

22 H3: Attrition and Fund Type For certain funds, relative performance is a much less important criteria for investment so attract investors who are less sensitive to this fund characteristic Index funds Investors don t care about relative performance but rather interested in Funds in categories with few competitors Investors looks less to relative performance since fewer competitors Expect attrition for these funds to poorly project fund-flow performance

23 H3: Index vs. Active

24 H3: Competitive vs. Non-Competitive

25 Summary Thoughts on Attrition Attrition is a good proxy for investor sensitivity to performance and easy to calculate by fund Help predict likeliness of investors leaving at the fund level Our focus is on leaving as it relates to poor performance One would expect stable investors in a fund With more attrition With poor performance and more attrition With more attrition and retail investors With more attrition and focus on active management With more attrition and with many competitors

26 Future Thinking For fund companies which have much more information on the investor, could improve on the attrition measure If knew an investor had arrived after a period of good performance, predict this investor is likely to leave with bad performance Tying the conditions of when an investor purchased shares in a fund could be helpful in predicting their preferences and liquidity demands Could measure the time between inflows and outflows and relate to attrition Length of time investors stay in a fund change with attrition?

27 Future Thinking Suppose want to purchase a fund or were a manager of a fund, how much would you pay? Typically pay some fraction of assets 2-3% Does the amount you pay change with the estimated stability of the asset base

28 Questions? Thank you