This study examines mutual fund performance around fund manager
replacement and the timing of the decision to replace mutual fund
managers. Fund manager replacement timing is explored to test whether
quick actions by boards of directors mitigate the negative fund
performance characteristics usually associated with manager replacement.
The study includes data from 507 instances of replacement of an
individual fund manager or the entire management team. While results
match previous findings that returns improve and standard deviation
falls following a manager change, several important new findings are
also presented. In using a unique control sample of funds matched on
prior period performance, net assets, and investment objective, it is
shown that poorly performing managers who retain their positions
actually improve fund performance to as great a degree as the new
managers hired after a termination. Both groups experience improved
returns and lower standard deviation of monthly return after the
replacement date.
Further, evidence indicates that for boards which decide to replace
poorly performing managers, stronger boards are more likely to complete
the replacement early in a period of underperformance. That is, boards
which have a larger percentage of independent directors and which are
smaller tend to be associated with early terminations.
INTRODUCTION
Mutual funds experiencing poor performance often replace the fund
manager in an effort to improve returns. Managers who are
underperforming are therefore motivated to increase performance in order
to retain their position. While this may initially appear to align
managers' goals with those of shareholders, a deeper examination is
required. Fund managers who are underperforming may feel a need to focus
exclusively on short-term results, and therefore increase risk in a
gamble to boost returns (Chevalier and Ellison (1997), Brown, Harlow,
and Starks (1996)). If the gamble pays off, the manager may keep his
position, while if the gamble fails, he only loses a position which he
was destined to lose anyway. An underperforming manager who remains in
place for several years may compound the severity of the problem,
resulting in a trend of increasing risk, lower performance, and lower
asset flows.
Prior research has shown that performance and flows generally
increase and risk decreases following a manager replacement (Chevalier
and Ellison (1999), Khorana (1996) Khorana (2001)). The improvement is
usually credited to the new manager. However, in this paper I compare
funds which have changed managers to a matched set of funds which have
not replaced managers in order to determine whether poorly performing
funds which retain their managers also experience improved performance.
This study is unique in determining a control group of funds with
similar performance rather than comparing the "change" funds
to a much broader control group based solely on investment objective.
Therefore, the first question addressed in this paper is whether credit
should be given the new manager or if there is a general mean reversion
that occurs in underperforming funds over time.
Results indicate that after the manger replacement date,
performance, asset flows, and risk are very similar for the funds which
replaced their managers and the control group. Excess objective returns
approach 0 for both groups in the post-replacement period, a significant
increase from negative excess objective returns prior to the manager
change. Flows, which appear to follow lagged return, fall during the
pre-replacement period, then also increase after the replacement date.
Secondly, since prior studies have shown that a manager replacement
is a positive event for a struggling fund, the timing of the decision to
replace mutual fund managers is examined. I test whether more timely
reactions by boards of directors mitigate the negative fund performance
characteristics usually associated with manager replacement. I find no
significant reaction of flows to an early manager change. Fund returns
appear to be the more likely driver of flows, and flows improve with
performance, regardless of the timing of the manager change. Finally, I
test whether an early replacement decision is associated with stronger
board governance characteristics. If directors assume that changing a
manager can boost performance, then stronger boards may be more likely
to replace a manager early. I find that boards of directors with a
greater percentage of independent members, fewer members, and an older
chair are associated with early manger changes.
LITERATURE REVIEW AND HYPOTHESIS DEVELOPMENT
The relation of mutual fund flows to performance has been the focus
of several studies. Chevalier and Ellison (1997) find that
underperforming fund managers attempt to gamble by increasing risk in an
effort to boost returns. A successful gamble may result in the manager
retaining his position, while a loss may simply result in the manager
losing the job he was likely to lose anyway. Further, Sirri and Tufano
(1998) report that the sensitivity of flows to performance is high
during superior performance, but low during under performance. Consumers
base mutual fund investment decisions at least in part on past
performance, putting more money into funds that perform very well in the
prior period, but they do not divest of poor performers. If outflows are
similar for under-performers regardless of the severity of poor returns,
there appears to be little incentive for managers to avoid steeper
losses in already underperforming funds. A poorly performing manager may
increase risk in order to move up in rankings at the risk of damaging
return even more.
Similar results are found by Brown, Harlow, and Starks (1996) who
describe the mutual fund market as a tournament in which all funds
compete for new assets based on relative performance. They report
mid-year underperformers tend to increase fund volatility in the latter
part of an annual assessment period to a greater extent than mid-year
winners.
Given the reasons to replace the manager of a poorly performing
fund, I next examine fund performance around a manager change. Chevalier
and Ellison (1999) find that firing a manager who has performed poorly
may reduce outflows by 45% compared to a control group. Khorana (1996)
provides additional evidence of improvement after manager replacement.
He finds that departing managers tend to have higher fund turnover and
higher expenses. Growth is also much slower for funds before a manager
replacement compared to a control group, and risk increases as
termination approaches. This market reaction of investors provides
support for the replacement of poorly performing managers as soon as
they are identified.
In a later study, Khorana (2001) finds that after a manager
replacement, underperforming funds experience significant improvement in
returns relative to past performance, and that changes are preceded by
decreasing net flows. Underperforming fund risk is higher
pre-replacement, then declines. In a study of Australian equity mergers
from 1994 to 2000, Gallagher, Nadarajay, and Pinnuck (2006) find that
after replacement, poorly performing funds improve performance, but not
because of better stock selection. Performance is enhanced primarily
through decreased momentum investment strategies and decreased portfolio
concentration.
If performance is affected by the replacement of the fund manager,
then the timing of the replacement may also be important. While the
timing of mutual fund manager replacement has not been previously
examined, Ertugrul and Krishnan (2007) study the timing of CEO
dismissal, and find that late CEO terminations result in worse ex-post
performance, more bankruptcies, and more de-listings than do early
replacements. Stock market reactions are negative for early replacement,
while there is no reaction to late, suggesting that the market has
already adjusted for poor performance of the CEO in office longer. They
also find that more effective boards replace lagging CEOs earlier,
before stock performance suffers. Less effective boards rely on a
history of poor performance rather than internal indications of
potential declines in management ability.
Mutual fund governing bodies are similar to corporate boards of
directors. Mutual fund governance consists of a board of directors
(trustees) who are elected by shareholders (fund owners) and have
specific responsibilities, including employing the fund advisor. The
fund advisor controls the management of the fund and hires the fund
manager, who is responsible for investment decisions.
There have been few studies into the effectiveness of mutual fund
boards. Khorana, Tufano, and Wedge (2007) find that fund mergers are
more likely when the board is composed of a higher percentage of
independent directors, and mergers are less likely when relatively
higher paid boards govern the merger target. Tufano and Sevick (1997)
find that shareholder fees are lower when boards are smaller and have a
greater fraction of independent directors. They also find some evidence
of better paid directors setting higher fund fees.
I first examine performance around a fund manager change. Manager
replacements are classified as either forced or voluntary, where forced
replacements are those instances in which a fund manager leaves a fund
and is not reported as a manager of a fund with greater assets within
two years. Further, since a main reason for increased performance after
a manager change may be prior-period underperformance, I also test
post-turnover performance to a matched sample of funds without a manager
change. The control sample is matched based on investment objective,
excess objective returns prior to the replacement date, and net assets.
H1: Following a forced turnover a fund will demonstrate: 1)
increased performance, 2) reduced risk, and 3) increased net flows.
I next examine the timing of the replacement decision. If fund
directors can determine at an early stage that performance is
deteriorating or that other factors indicate a fund manager needs to be
replaced, they will attempt to replace the manager as early as the
decision can be correctly made. The sample of forced turnovers is
divided into two additional subsets: those which replace early and those
which replace late. I calculate the mean number of months during the two
years prior to the manager change that the sample funds posted negative
excess objective returns. Those replacements which are preceded by more
than the mean are classified as late. Early replacements include those
with the same or a fewer number of months of negative return.
H2: Post replacement performance, risk, and net flows are different
for late and early replacements.
Finally, it has been shown in corporate finance that certain board
characteristics are associated with "better" boards. I test
this for mutual fund boards, and expect that boards which are smaller,
more independent, and have an independent chair will be associated with
a greater likelihood of early replacement.
H3: The probability of an early termination increases with stronger
board governance characteristics.
An overview of the relation of the hypotheses is shown in figure 1.
[FIGURE 1 OMITTED]
SAMPLE AND METHODOLOGY
To construct the sample, I examine the CRSP mutual fund database to
identify all changes of a sole manager or the replacement of an entire
management team for domestic equity funds for the period of January 2002
through December 2005. To be included in the sample, a fund must have at
least two years of pre- and post-turnover performance data and only one
manager change during the sample period. The sample includes 507 funds
meeting the selection criteria. The replacement date (t = 0) is set to
the six month period during which CRSP reports the change in manager.
I compare this "change sample" to a matched control
group. Fund matches are determined by using logistic regression, run
separately for each period and for each investment category, to estimate
the probability of manager replacement. The independent variables are
return and net flow during the pre-replacement period. The match is the
fund with an identical two-decimal probability of replacement with the
asset size nearest the target fund. If a twodecimal match is not found,
then the next closest probability is used. Each fund in the matched
sample is used only once. For the control funds, the replacement date is
set to the replacement date of their matched funds. Three areas of fund
performance are examined: return, net flows, and risk. Risk is defined
as the standard deviation of monthly returns for 12 month periods
surrounding replacement. Return is measured using fund excess objective
return. Objective return is determined by first dividing all domestic
equity funds in CRSP into the seven CRSP classifications of large
growth, aggressive growth, income and growth, growth and income, mid
cap, small cap, and S&P index. The mean return for each category is
calculated for each month. That mean is subtracted from the actual
return for each fund within the category to result in the excess
objective return, as shown below.
[EXCESS OBJECTIVE RETURN.sub.it] = ([R.sub.i,t])--([R.sub.o,t])
Where [R.sub.i,t] is the return of fund i for period t, and
[R.sub.o,t] is the mean return of all funds with the same investment
category for period t.
Net flow is defined as follows:
[PERCENT NETFLOW.sub.i,t] = [[ASSETS.sub.i,t]--[ASSETS.sub.i,t-1] *
(1 + [R.sub.i,t])]/[ASSETS.sub.i,t-i] Where [ASSETS.sub.i,t] is total
assets in fund i at the end of period t, and [R.sub.i,t] is the return
of fund i during period t. PERCENT NETFLOW measures the growth in assets
over and above the change in value of fund assets at the beginning of
the period.
To test hypothesis 2, the sample is divided into early and late
terminations as described earlier. I use the same methodology as in
hypothesis 1 above, but here the sample is divided into the groups of
early vs. late replacement rather than the sample vs. the control group.
The reason for this is to examine the economic impact to a fund, if any,
of allowing poorly performing fund managers to remain in charge longer.
To test hypothesis 3, I examine the board characteristics of the
sample funds. While the board does not have direct control over the fund
manager replacement decision (the fund advisor technically makes that
decision), it is assumed that this is an important enough matter that
the board would exude influence. Also, the culture or environment
established by the board may dictate the actions taken in employing the
manager. Data concerning the boards of directors are collected from the
SEC for all the funds in the sample. This is published in forms 485APOA
and 485BPOS, and is available on the SEC website at www.sec.gov.
I estimate the probability of early replacement using logistic
regression and board governance characteristics in the model below:
P (early replacement) = f {board governance characteristics} Where
board governance characteristics include:
* Board size: the number of directors on the board of the fund
* Compensation: the annual compensation earned by a board member
from the fund
* Independent board: the percentage of the board composed of
unaffiliated directors
* Independent chair: a dummy variable equal to one if the chair is
unaffiliated with the fund
* Retirement: dummy variable equal to one if the board members
receive pension benefits after retirement
* Director Age: the average age of fund board members
* Chair Age: age of the Board Chairman
RESULTS
Table 1 reports the excess objective return for each 6 month period
surrounding the manager change, as well as the return for each full 24
month period before and after replacement. Both the control group and
the sample exhibit poor performance before the manager change, as
expected. Both groups also exhibit performance not significantly
different from the mean of the fund objective after the replacement
date, except for a return of -0.63% for the change group in the 6 to 12
moth period after the change. While one may expect a new manager to be
able to improve performance, it appears that those managers who remained
at their funds also improved performance--at least up to the mean for
their fund objective. The only significant difference in performance
between the funds which changed managers and those which didn't was
in months 6 to 12 post-replacement, in which the funds with a change
actually performed worse.
Panel B shows that the late change group is associated with lower
performance before the change date. This is expected since by
definition, late changes have more months of negative performance.
Interestingly, the change group underperforms the control group for the
first 12 months after the replacement, with manager change funds
averaging excess objective returns lower than the control group by 1.02
and 1.11 percent each of the first two 6month periods. The only
significant negative period post replacement date for the control group
is for the 6 month period from 18 to 24 months after the change date.
Performance for both groups is neither significantly different from 0,
nor different from each other for the full 24 months post-replacement
date period, indicating again that performance is no better for those
funds which changed managers, even for these late changes which are
preceded by longer periods or poor performance.
Panel C of Table 1 examines changes between early and late manager
replacements. Performance prior to the manager change is not reported
since it is much worse for late changes due to the method of
classification. Those funds with an early manager change outperform
those with a late change by 99 basis points for the first 6 months after
the change date. Both groups are negative from months 6 to 12.
Performance is similar for the next 12 months, then those funds which
replaced the manager late outperform by 50 basis points, but the
difference is only marginally significant. Also, performance over the
full 24-month period is not significantly different for the two groups.
This suggests that the timing of a manager change does not significantly
impact post-replacement returns. The results reported in Table 1 are
also shown graphically in Figure 2, which more clearly demonstrates the
convergence to 0 of the excess objective return of both the replacement
and control groups for all three categories of comparisons. Taken
together, these results suggest there is not a significantly greater
increase in fund performance for funds which replace a manager either
early or late.
[FIGURE 2 OMITTED]
Table 2 Panel A shows that flows are not significantly different
for the replacement group and the matched sample before replacement, and
were generally falling and approaching 0 net flows by the time of
replacement. Flows continue to fall after the replacement date for both
groups, with the only significant difference in period 2 in which flows
were 2.43 percentage points lower for those funds with a replacement.
Flows for both groups are flat in the 12 month period surrounding
replacements, so there appears to be no significant signal (either
positive or negative) associated with a manager change.
Panel B reports the results for late changes. Flows prior to the
change date are negative and declining for both groups, with no
significant differences. After the change date, flows for periods 2 and
3 are significantly lower for the change group vs. the no change group
by 3.08 and 3.69 percentage points, respectively. It appears that a
manger change is a negative event for a fund, even when compared to
funds with equally poor performance. A new manager does not appear to be
a positive signal for a poorly performing fund. Flows are actually
higher for poorly performing funds which retain their managers.
Panel C compares early and late manager changes. Flows are higher
for the early changes for periods 2 and 3, which suggests that an early
change can better stop the trend of decreasing cash flows. Also, since
late replacement funds demonstrate lower flows prior to replacement as
well, an early replacement may mitigate some of the negative flows
associated with retaining a poorly performing manager longer.
The results reported in Table 2 are also shown graphically in
Figure 3. As with returns, the flows for both groups are similar except
for the early vs. late changes. It appears that the replacement of a
manager who has a longer history of poor performance may actually cease
the decline in flows during the six months after replacement, but only
temporarily as the trend of decreasing flows then continues for the next
12 month period.
[FIGURE 3 OMITTED]
Table 3 reports changes in standard deviation. For the forced
replacements in Panel A, the standard deviation of monthly returns is
similar for both the sample and the control group. Each group also
experiences lower standard deviation across time. This supports
Chevalier and Ellison (1997) in that standard deviation declines, but I
also find a decline for those funds which did not change their manager.
Panel B shows very similar results for late replacements.
Panel C shows that standard deviation is slightly greater for the
late changes for the period of return ending two years after the
replacement. This may be the result of greater changes to the holding of
the portfolio of late replacements. If a fund has performed poorly for a
long period, more significant changes in holdings may be needed.
The results from the left half of Table 3 are also shown
graphically in Figure 4, which more clearly shows the decreasing
standard deviation around the replacement, as well as the very similar
pattern for both the change and control groups.
[FIGURE 4 OMITTED]
While I have shown little incentive for replacing a poorly
performing fund manager, prior research as well as intuition suggests
that when portfolio performance is below average, the manager may be the
problem. In this section, I examine mutual fund board characteristics,
and test whether there is a connection between governance and the timing
of the decision to replace a mutual fund manager.
I use only the sample of funds which experience a forced
replacement, and compare the early changes to the late. Table 4 reports
the results from the logistic regression used to test for relations
between common board characteristics and the probability of a
replacement being an early replacement. Chair age, the percentage of
board members which are independent, and board size are all significant
predictors of whether a change in fund manager will be an early change.
Coefficients for chair age and percentage independent are positive and
significant, indicating that the probably of a change being an early
change are higher when the chair is older and the board is more
independent. The negative coefficient for board size implies that early
changes are more likely for smaller boards. Taken together, these
results suggest that the board characteristics generally regarded as
indicating stronger boards are associated with early replacement.
ROBUSTNESS CHECKS
A potential issue with the data is that the manager changes from
CRSP are grouped over a six month window. That is, a manager change in
January is reported with the same change date as one in June. To test
whether this impacts the results, I collect manager tenure information
from Morningstar, then manually calculate the change month. Of the 507
funds from CRSP, 340 were successfully matched against the Morningstar
database.
Results related to returns and flows are reported in Tables 10 and
11, and are very similar to the patterns revealed using the full CRSP
data. The returns and flows of the change sample and control group again
appear to converge after the change date. This again implies no
significant advantage to performance or asset flows in replacing a
manager.
CONCLUSIONS
While results match previous findings that returns and flows
improve and standard deviation is reduced following a manager change,
several important new findings are also presented. Most past studies
compare funds with a manager change to all other funds with the same
investment category, without matching on performance. In using a unique
control sample of funds matched on prior period performance, net assets,
and investment objective, I have shown that poorly performing managers
who retain their positions actually improve fund performance just as
well as those funds in which the manager is replaced. Both groups
experience improved returns and lower standard deviation of monthly
return.
Regardless of the lack of evidence supporting improved performance
due to manager replacement, I also find evidence that stronger boards
are more likely to replace a manager early in a period of
underperformance. Boards which have a larger percentage of independent
directors and which are smaller tend to be associated with early
replacements.
REFERENCES
Brown, K. W., V. Harlow, & L. Starks, (1996), Of tournament and
temptations: An analysis of managerial incentives in the mutual fund
industry. Journal of Finance 51, 85-110. Chevalier, J. & G. Ellison,
(1997), Risk taking by mutual funds as a response to incentives. Journal
of Political Economy 105, 1167-1203.
Chevalier, J. & G. Ellison, (1999) Career concerns of mutual
fund managers. Quarterly Journal of Economics 114, 389432.
Ertugrul, M. & K. Krishnan, (2007), CEO dismissal timing and
costs of delayed action: Do some boards act too late? working paper.
Gallagher, D., P. Nadarajay, & M. Pinnuck, (2006), Top
management turnover: An examination of portfolio holding and fund
performance. Australian Journal of Management 31, 265-292.
Khorana, A. (1996), Top management turnover, An empirical
investigation into mutual fund managers. Journal of Financial Economics
40, 403-427.
Khorana, A., (2001), Performance changes following top management
turnover: Evidence from open-end mutual funds. Journal of Financial and
Quantitative Analysis 36, 371-393.
Khorana, A., P. Tufano, & Lei Wedge, (2007), Board structure,
mergers, and shareholder wealth: A study of the mutual fund industry.
Journal of Financial Economics 85, 571-598.
Sirri, E. & P. Tufano (1998), Costly search of mutual fund
flows. Journal of Finance 53, 1589-1622.
Tufano, P., & M. Sevick (1997), Board structure and fee setting
in the U.S. mutual fund industry. Journal of Financial Economics, 46,
321-355.
Steve A. Nenninger, Sam Houston State University
Table 1: Mean excess objective return before and after replacement
This table reports the mean excess objective return for six-month
periods surrounding manager replacement for the sample, the control
group, and reports differences between the groups.
Panel A: Forced Change, n=388
Period -4 to -1 -4 -3 -2
Six months Full 24 -18 -12 -6
ending prior
no change -0.0267 -0.0078 -0.0108 -0.0121
0.019 0.145 0.002 0.000
change -0.0549 -0.0211 -0.0229 -0.0166
<.0001 <.0001 <.0001 <.0001
diff 0.0282 0.0133 0.0121 0.0045
0.071 0.059 0.025 0.327
Period -1 1 2
Six months 0 6 12
ending
no change -0.0129 -0.0007 0.0013
<.0001 0.769 0.607
change -0.0107 -0.0012 -0.0063
0.016 0.518 0.010
diff -0.0020 0.0005 0.0076
0.691 0.857 0.031
Period 3 4 1 to 4
Six months 18 24 Full 24
ending post
no change -0.0015 -0.0016 -0.0002
0.456 0.392 0.964
change -0.0003 0.0003 -0.0053
0.916 0.854 0.271
diff -0.0010 -0.0020 0.0051
0.684 0.447 0.492
Panel B: Late Change, n = 169
Period -4 to -1 -4 -3 -2
Six months Full 24 -18 -12 -6
ending prior
no change -0.1175 -0.0378 -0.0317 -0.0417
<.0001 <.0001 <.0001 <.0001
change -0.1793 -0.0655 -0.0613 -0.0442
<.0001 <.0001 <.0001 <.0001
diff 0.0618 0.0276 0.0296 0.0025
0.002 0.012 0.000 0.747
Period -1 1 2
Six months 0 6 12
ending
no change -0.0322 0.0034 0.0059
<.0001 0.335 0.126
change -0.0385 -0.0068 -0.0052
<.0001 0.009 0.074
diff 0.0063 0.0102 0.0111
0.445 0.020 0.022
Period 3 4 1 to 4
Six months 18 24 Full 24
ending post
no change -0.0049 -0.0086 -0.0025
0.148 0.002 0.772
change 0.0012 0.0034 -0.0072
0.655 0.190 0.204
diff -0.0060 -0.0120 0.0047
0.159 0.002 0.649
Panel C: Early vs. Late Change, n = 220/169, changed managers only
Period -4 to -1 -4 -3 -2
Six months Full 24 -18 -12 -6
ending prior
early
late
Diff
Period -1 1 2
Six months 0 6 12
ending
early 0.0031 -0.0072
0.229 0.052
late -0.0068 -0.0052
0.009 0.074
Diff 0.0099 -0.0020
0.007 0.665
Period 3 4 1 to 4
Six months 18 24 Full 24
ending post
early -0.0014 -0.0021 -0.0039
0.707 0.325 0.599
late 0.0012 0.0034 -0.0072
0.655 0.190 0.204
Diff -0.0030 -0.0050 0.0033
0.570 0.098 0.719
Table 2: Asset flow return before and after replacement
This table reports the mean asset flows for six-month periods
surrounding manager replacement for the sample, the control group,
and reports differences between the groups.
Panel A: Forced Change, n=388
Period -4 to -1 -4 -3 -2
Six months Full 24 -18 -12 -6
ending prior
no change 0.2987 0.0565 0.0491 0.0019
0.001 <.0001 0.001 0.846
change 0.1509 0.0359 0.0244 -0.0087
0.011 0.018 0.060 0.421
diff 0.1478 0.0206 0.0246 0.0106
0.155 0.303 0.203 0.467
Period -1 1 2
Six months 0 6 12
ending
no change -0.0162 -0.0014 -0.0239
0.168 0.938 0.019
change -0.0107 0.0069 -0.0481
0.658 0.875 <.0001
diff -0.0050 -0.0080 0.0243
0.838 0.861 0.060
Period 3 4 1 to 4
Six months 18 24 Full 24
ending post
no change -0.0464 -0.0112 0.0259
<.0001 0.461 0.789
change -0.0469 -0.0252 -0.0781
<.0001 0.078 0.261
diff 0.0005 0.0139 0.1040
0.966 0.504 0.382
Panel B: Late Change, n = 169
Period -4 to -1 -4 -3 -2
Six months Full 24 -18 -12 -6
ending prior
no change -0.0065 0.0523 -0.0019 -0.0486
0.883 0.006 0.899 <.0001
change -0.0661 0.0194 -0.0235 -0.0511
0.157 0.433 0.086 <.0001
diff 0.0597 0.0329 0.0216 0.0025
0.351 0.289 0.28 0.871
Period -1 1 2
Six months 0 6 12
ending
no change -0.0442 -0.0100 -0.0360
0.021 0.781 0.012
change -0.0445 0.0093 -0.0668
0.049 0.916 <.0001
diff 0.0003 -0.0190 0.0308
0.991 0.839 0.096
Period 3 4 1 to 4
Six months 18 24 Full 24
ending post
no change -0.0464 -0.0004 0.0791
0.002 0.990 0.708
change -0.0833 -0.0380 -0.1466
<.0001 0.090 0.275
diff 0.0369 0.0376 0.2257
0.049 0.308 0.367
Panel C: Early vs. Late Change, n = 220/169, changed managers only
Period -4 to -1 -4 -3 -2
Six months Full 24 -18 -12 -6
ending prior
early
late
Diff
Period -1 1 2
Six months 0 6 12
ending
early 0.0050 -0.0338
0.897 0.002
late 0.0093 -0.0668
0.916 <.0001
Diff -0.0040 0.0330
0.962 0.040
Period 3 4 1 to 4
Six months 18 24 Full 24
ending post
early -0.0189 -0.0154 -0.0255
0.096 0.409 0.703
late -0.0833 -0.0380 -0.1466
<.0001 0.090 0.275
Diff 0.0644 0.0226 0.1211
<.0001 0.435 0.419
Table 3: Changes among periods in standard deviation of monthly
return before and after replacement
This table reports the mean standard deviation of monthly return for
twelve-month periods surrounding manager replacement for the sample,
the control group, and reports differences between the groups. Also
reported are the differences in mean standard deviation between
consecutive six-month periods after manager replacement, as well as
differences in periods surrounding replacement.
Panel A: Forced Replacement, n = 388
Annual Standard Deviation of Monthly Return
Periods -4,-3 -2,-1 1,2 3,4
12 months ending -12 0 12 24
no change 0.0631 0.0516 0.037 0.0306
<.0001 <.0001 <.0001 <.0001
change 0.0636 0.054 0.0362 0.0297
<.0001 <.0001 <.0001 <.0001
diff -0.0005 -0.002 0.0008 0.0009
0.823 0.298 0.575 0.297
Panel B: Late Replacement , n = 169
Periods -4,-3 -2,-1 1,2 3,4
12 months ending -12 0 12 24
No change 0.0709 0.0581 0.0398 0.0331
<.0001 <.0001 <.0001 <.0001
Change 0.0725 0.0595 0.0365 0.0316
<.0001 <.0001 <.0001 <.0001
Diff -0.002 -0.001 0.0033 0.0014
0.732 0.732 0.199 0.332
Panel C: Early vs. Late Change, n = 220/169, changed managers only
Periods -4,-3 -2,-1 1,2 3,4
12 months ending -12 0 12 24
early 0.0359 0.0283
<.0001 <.0001
late 0.0365 0.0316
<.0001 <.0001
Diff -0.0006 -0.003
0.769 0.003
Table 4: Probability of a change being an early change
This table reports the results of a logistic regression testing the
effect of board governance variables on the probability of a manager
change being an early change.
P (early replacement) = f {board governance characteristics}
Odds Ratio Coefficient P
Point
Estimate
Chair age 1.024 0.0237 0.052
Percentage independent 1.02 0.0199 0.08
Size of board 0.855 -0.1565 0
Average age 0.992 -0.0081 0.802
Total Compensation 0.999 -0.0006 0.732
Independent chair dummy 0.974 -0.0268 0.918
N 446
Table 5: Mean excess objective return before and after replacement--
Morningstar Change Dates
This table reports the mean excess objective return for six-month
periods surrounding manager replacement for the sample, the control
group, and reports differences between the groups. The change date
used for this table is collected from Morningstar.
Panel A: Forced Change, n=340
Period -4 -3 -2 -1
Six months ending -18 -12 -6 0
no change -0.0131 -0.0152 -0.0051 -0.0176
0.071 0.005 0.897 <.0001
change -0.0222 -0.0148 -0.0113 -0.0164
<.0001 0.001 0 <.0001
diff 0.0091 -0.0004 0.0118 -0.001
0.315 0.954 0.015 0.828
Period 1 2 3 4
Six months ending 6 12 18 24
no change 0.0014 0.0001 -0.0029 -0.0011
0.648 0.982 0.244 0.643
change 0.003 -0.0043 0.0013 -0.0021
0.27 0.024 0.468 0.229
diff -0.002 0.0044 -0.004 0.001
0.705 0.166 0.171 0.723
Panel B: Late Change, n = 154
Period -4 -3 -2 -1
Six months ending -18 -12 -6 0
no change -0.0404 -0.0395 -0.0123 -0.0373
0.001 <.0001 0.282 <.0001
change -0.0611 -0.0488 -0.03 -0.0366
<.0001 <.0001 <.0001 <.0001
diff 0.0207 0.0093 0.0245 -0.0007
0.167 0.332 0 0.93
Period 1 2 3 4
Six months ending 6 12 18 24
no change 0.0014 0.01 -0.0068 -0.0082
0.728 0.006 0.058 0.006
change -0.0069 -0.0079 -0.0005 -0.0076
0.039 0.001 0.822 0.004
diff 0.0083 0.0178 -0.006 -0.0006
0.111 <.0001 0.138 0.881
Table 6: Asset flow return before and after replacement--using
Morningstar change dates
This table reports the mean net asset flows for six-month periods
surrounding manager replacement for the sample, the control group,
and reports differences between the groups. The change date used for
this table is collected from Morningstar.
Panel A: Forced Change, n=340
Period -4 -3 -2 -1
Six months ending -18 -12 -6 0
no change 0.1279 0.0566 0.0426 -0.1121
<.0001 0 0.028 0.296
change 0.0774 0.0401 0.0491 -0.0111
0.002 0.004 0.085 0.5
diff 0.0505 0.0165 -0.007 0.087
0.157 0.42 0.85 0.352
Panel B: Late Change, n = 154
Period -4 -3 -2 -1
Six months ending -18 -12 -6 0
no change 0.2663 0.0588 0.0108 -0.0344
0.107 0.08 0.68 0.2
change 0.0315 -0.0199 -0.0379 -0.0582
0.18 0.151 0.001 <.0001
diff 0.2348 0.0787 0.0487 0.0238
0.155 0.03 0.088 0.399
Period 1 2 3 4
Six months ending 6 12 18 24
no change -0.0545 -0.0365 -0.0443 -0.0219
<.0001 0.002 <.0001 0.3
change -0.0748 -0.0477 -0.0629 -0.0491
<.0001 0 <.0001 <.0001
diff 0.0561 0.0112 0.0186 0.0272
0.242 0.509 0.187 0.252
Panel B: Late Change, n = 154
Period 1 2 3 4
Six months ending 6 12 18 24
no change -0.0754 -0.045 -0.0465 -0.0052
<.0001 0.007 0.002 0.878
change -0.0865 -0.0678 -0.0904 -0.066
<.0001 <.0001 <.0001 0
diff 0.0111 0.0228 0.0439 0.0608
0.57 0.274 0.021 0.113