Title: Mutual Fund Performance and Manager Style. J.L. Davis, FAJ, Jan/Feb 01, 19-27
1Mutual Fund Performance and Manager Style. J.L.
Davis, FAJ, Jan/Feb 01, 19-27
- Various studies examined the evidence of
persistence in mutual fund performance. - General consensus is that a few fund managers do
tend to consistently appear near the top of the
return rankings.
2Mutual Fund Performance and
- Strong evidence that some fund managers
consistently appear near the bottom of the
ranking. - The implication for investors Small likelihood
of consistently earning abnormal returns by
seeking individual fund managers.
3Mutual Fund Performance and
- Objective of this Study Examine the relationship
between equity fund performance and manager
style. - 1. Examine whether any investment style reliably
delivers abnormal performance - 2.Evidence of performance per- sistence based
on style.
4Mutual Fund Performance and
- Data and Methodology
- Data
- US Mutual Fund Database from CRSP.
- Time period 1965-1998.
- Data is free of Survivorship bias.
- Explain the problem with survivorship bias.
5Mutual Fund Performance and
- Selection Criteria for funds to be included in
the data set - 1. If a funds stated objective was growth,
growth and income, maximum capital gains,
small-cap growth, or aggressive growth - 2. Objective not listed but policy statement
indicated that they primarily invested in common
stocks.
6Mutual Fund Performance and
- Sample consisted of 4,686 funds covering 26,564
fund-years from 1962-98, i.e.,in 26,564 time the
fund was classified as an equity fund and had at
least one valid monthly return. - The median equity weight for the fund year 93
percent.
7Mutual Fund Performance and
- Style Identification
- Fama-French three factor model as presented
below is used to infer funds investment style. - RitRftaibi(Rm,tRf,t)si SMBt
- hiHMLmei,t
- Where,
- Rit the percentage return to fund i,
8Mutual Fund Performance and
- Rft the U.S. T-bill rate
- Rmt the return on the CRSP value weighted
index, - SMBt small cap return - large cap return,
- HMLt value return growth return.
- et the error term.
9Mutual Fund Performance and
- Factor loading (sign of coefficients)
- 1. Small company stocks positive si
- 2. Large company stocks negative si
- 3. Value factor positive hi
- 4. Growth factor negative hi
- 5. Intercept (?) measures
10Mutual Fund Performance and
- performance relative to three
- factor model.
- Funds were identified as small cap or large cap
on the basis of SMB slopes and as value or growth
on the basis of HML slope.
11Mutual Fund Performance and
- Portfolio Formation Funds were placed in style
portfolios at the beginning of each year from
1965. - Returns for the previous 36 months were used to
estimate performation slopes on the HML and SMB
factors. - Based on the slopes, funds were allocated into
portfolios, and returns
12Mutual Fund Performance and
- were calculated for each month of 1965.
- The process was repeated for each year.
- Univariate SMB and HML sorts to form decile
portfolios and bivariate sorts to form portfolios
based on intersection of the HML and SMB ranking.
13Mutual Fund Performance and
- Funds were divided into thirds (low, medium and
high) on the basis of SMB and HML rankings. - This 3x3 partition produced nine portfolios -
high SMB/Low HML (small/growth) portfolio.
14Mutual Fund Performance and
- Test of performance persistence used bivariate
sorts on HML and ? and SMB and ?. For example,
(Low HML/Low ?) implies growth emphasis and
performed poorly compared with the three-factor
benchmark.
15Mutual Fund Performance and
- Tests of Abnormal Returns
- 1. Davis, Fama and French - three factor model
(excess market returns, the size and value-growth
factors) have explanatory power - 2. If premiums associated with size and value
can be earned by passive strategy by buying
diversified portfolios with desired level of
risk,
16Mutual Fund Performance and
- therefore, an active manager should be able to
outperform such passive strategy. - Results- Style Based Portfolio
- Table 1- Three Factor Results, HML Sorts.
- Panel A Sorted by HML slope
- Panel B Regression Results
- R2 and t values indicate in favor
17Mutual Fund Performance and
- of three factor model.
- - HML coefficient growth
- HML coefficient value
- Deciles 1-6 growth based on HML
- Deciles 7-10 value based on HML
- Deciles 1-4 positive ?
- Deciles 5-10 negative ?
- Decile 10 (?-.2) and significant, i.e.,
underperformance
18Mutual Fund Performance and
- Summary Value fund did not performed better than
growth fund. - Table 2 - Three Factor Results, SMB Sorts.
- R2 close to 1.
- ? negative and insignificant.
- Three factor model is appropriate
19Mutual Fund Performance and
- Table 3- Independent HML and SMB sorts portfolios
- Panel A
- Low- Low corresponds to large-growth portfolio
- High-High corresponds to small-value portfolio
- No style portfolio show reliable abnormal
profit,a tendency for value
20Mutual Fund Performance and
- funds to under perform growth fund is clear when
SMB sensitivity is held constant. - Conclusion from Tables 1-3 Value funds performed
poorly over the past 30 year period. -
21Mutual Fund Performance and
- Table 4 Independent sorts on HML and ?.
- Panel A Average for each portfolio
- Panel B Regression coefficients
- Panel C Regression results one year after
formation - Panel A The spread in ? value between low and
high ? portfolio is about 1 percentage point.
22Mutual Fund Performance and
- Panel B For high HML portfolio, all three ?
values are negative, i.e, value funds have not
done well. - Low HML/high ? has a .14 ? value, i.e., some
growth mangers have been able to maintain good
performance over short period. Advantage
disappears in Panel C.
23Mutual Fund Performance and
- Table 5 Independent sorts on SMB and ?.
- Panel A The spread in ? value between low and
high ? portfolio is about 150 bps. - Panel B The spread falls to 25 bps.
- Panel C Spread is less than 10 bps.
- Conclusion Some evidence of persistence, but
dies quickly.
24Mutual Fund Performance and
- Conclusion
- No investment style generated abnormal returns
over 1965-98. - Small evidence of persistence among best
performing funds more evidence of performance
persistence among poor-performing funds - Funds did not capture value premium.