- 1. Performance Persistence in Institutional Investment
ManagementJerey A. Busse Amit Goyal Sunil Wahal March
2006AbstractUsing new, survivorship-bias free data, we examine the
persistence in perfor- mance of 6,260 institutional portfolios
managed by 1,475 investment managers between 1991 and 2004.
Persistence in domestic equity portfolios is signicant and
economically large up to one year after portfolio formation. Unlike
retail mu- tual funds, this persistence is entirely in winner
portfolios. Similar patterns are evident in international equity
portfolios, but persistence in xed income portfo- lios lasts up to
three years. Better performing portfolios are more likely to oer
incentive fees and most-favored-nation clauses but also charge
higher fees. Fee size is insucient to eliminate excess returns. Top
performers draw an inux of assets from plan sponsors, and in the
year following such inows, alphas sharply decline. Overall, the
results are consistent with the supply-side quantity-based
equilibrating process modeled by Berk and Green (2004). Busse is
from the Goizueta Business School, Emory University, email: Je
[email protected]; Goyal is from the Goizueta Business School,
Emory University, email: Amit [email protected]; and Wahal is
from the WP Carey School of Business, Arizona State University,
email: [email protected]. We are indebted to Margaret Tobiasen at
Informa Investment Solutions, and to Jim Minnick and Frithjof van
Zyp at eVestment Alliance for graciously providing data. We thank
Byoung-Hyoun Hwang, George Benston, Narasimhan Jegadeesh, and
seminar particpants at Emory University, UCLA and the University of
Oregon for helpful suggestions.
2. 1 IntroductionPerformance persistence in delegated investment
management represents a signicant challenge to ecient markets.
Academic opinion on whether persistence exists is Bayesian we
update our priors based on the most recent evidence incorporating
either new data or improved measurement technology. Although
Jensens (1968) original examination of mutual funds concludes that
funds do not produce abnormal performance, later stud- ies provide
compelling evidence that relative performance persists over both
short and long horizons (see, for example, Grinblatt and Titman
(1992), Elton, Gruber, Das, and Hlavka (1993), Hendricks, Patel,
and Zeckhauser (1993), Goetzmann and Ibbotson (1994), Brown and
Goetzmann (1995), Elton, Gruber, and Blake (1996), and Wermers
(1999)). More recently, Carhart (1997) shows that accounting for
momentum in indi- vidual stock returns eliminates almost all
evidence of persistence among mutual funds.1 However, Bollen and
Busse (2005) nd that performance persists over short quarterly
holding periods, even after controlling for momentum. The attention
given to persistence in retail mutual fund performance is entirely
war- ranted. The data are good, and this form of delegated asset
management provides millions of investors access to readily-built
portfolios. As a result, at the end of 2004, there were 7,101
equity, bond, and hybrid mutual funds responsible for investing
$6.1 trillion in assets (Investment Company Institute (2004)).
However, there is an equally large arm of delegated investment
management that receives much less attention but is no less
important. At the end of 2004, over 50,000 plan sponsors (public
and private retirement plans, endowments, foundations, and
multi-employer unions) allocated over $6 trillion in assets to
about 1,500 institutional asset managers (Money Market Direc- tory,
2004). In this paper, we comprehensively examine the performance
persistence of portfolios managed by institutional investment
management rms.2Institutional asset management rms draw xed amounts
of capital (referred to as mandates) from public and private dened
benet retirement plans, endowments, foundations, unions, and
trusts. Although these mandates span a variety of asset classes,
including domestic equity, xed income, international equity, real
estate securities, and 1One exception is the continued strong
underperformance of the worst performing funds. Berk and Xu (2004)
show that this persistence occurs in both the current and prior
year. They argue that the unwillingness of investors to withdraw
capital from these funds causes this remaining pocket of
persistence.2Persistence in another largely unexplored form of
delegated asset management, hedge funds, is studied by Jagannathan
and Novikov (2005), among others. They estimate persistence in
hedge returns to be approximately 1 percent per year. 1 3.
alternative assets, we restrict our attention to the rst three. To
our knowledge, we are the rst to examine persistence in asset
classes beyond domestic equity. Invest- ment styles within these
asset classes run the gamut in terms of size and growth-value
gradations for equity portfolios, and credit and maturity
dimensions for xed income portfolios. Investment style identication
is important to plan sponsors because it al- lows plan sponsors to
make better-informed asset allocation decisions. Conveniently, it
also provides us with self-identied natural benchmarks to judge
returns. Our data are drawn from two independent sources: Informa
Investment Solutions (IIS) and eVestment Alliance. Both rms collect
self-reported returns and other infor- mation from investment
management rms and provide data, services, and consulting to plan
sponsors, investment consultants, and investment management rms.
IIS provides a longer time series of returns and has better
cross-sectional coverage; it provides quar- terly returns for 6,260
portfolios managed by 1,475 institutional asset management rms from
1979 to 2004. Since these data suer from survivorship bias prior to
1991, we de- rive our estimates from the post-1991 sample period.
In addition to returns, eVestment provides detailed information
about fee arrangements, schedules, and whether portfolios are
closed to new investors. We use these data to add texture to our
persistence analysis and sharpen inferences with respect to net of
fee performance. We investigate persistence over both short and
long horizons. We form deciles using one-year raw returns as well
as benchmark-adjusted returns, and then evaluate the future excess
performance of these deciles. Using raw returns to form deciles
mirrors the typical method employed in mutual fund studies, and
allows us to compare our results with those in such papers. Using
benchmark-adjusted returns is consistent with performance
evaluation methods used by plan sponsors, and also aords us the
ability to show benchmark-adjusted returns and information ratios
in the post portfolio formation periods. We estimate alphas using
both unconditional and conditional three- and four- factor models.
One quarter after deciles are formed based on raw returns, mean
decile alphas are positive, large, and statistically signcant in
the majority of winner portfolios. For in- stance, the one-quarter
Fama and French (1993) three-factor alpha from the extreme winner
decile is 2.88 percent per quarter with a t-statistic of 3.77.
Accounting for mo- mentum using the Carhart (1997) procedure
reduces the point estimate to 1.07 percent per quarter with a
t-statistic of 1.85. Using a factor model that includes an
aggregate bond index, the term spread, and the default spread,
unconditional one-quarter alphas for the winner xed income decile
vary from 1.03 percent to 1.53 percent depending2 4. on the
specication. Consistent with Christopherson, Ferson and Glassman
(1998), we nd that conditioning information materially inuences,
but does not reduce, estimates of alphas for xed income portfolios.
The alphas of the winner international equity decile vary from 2.04
percent to 2.67 percent per quarter. When we form deciles based on
benchmark-adjusted returns, the alphas shrink somewhat but remain
economically large and statistically signcant. Performance
continues to persist over longer horizons. For domestic equity
portfolios, persistence continues up to one year after portfolio
formation; the quarterly alpha of the extreme winner decile from a
three-factor model over a one-year evaluation period is 1.44
percent. Interestingly, after the rst year, domestic equity alphas
are statistically indistinguishable from zero. For xed income
portfolios, the rst year alpha for the extreme winner decile is
0.62 percent. In the second and third year thereafter, the alpha
shrinks marginally (to 0.60 and 0.51 percent respectively) and
retains statistical signicance. International equity portfolios
show a pattern similar to domestic equity. One year after decile
formation, the alpha for the extreme winner decile is 2.76 percent,
but is indistinguishable from zero in the second and third year.
Our persistence results are based on returns that are gross of fees
(but net of trading costs). Investment management fees could
potentially wipe out any gains to plan spon- sors from chasing
prior winners. To determine if that is the case, we use fee
schedules from our second data source (eVestment Alliance). These
schedules provide a pro forma fee that typically declines with the
size of the mandate supplied by the plan sponsor (as in a step
function). The marginal fees for each breakpoint vary across asset
classes, investment styles, and individual rms. We nd that annual
fees are a fraction of the alphas of the extreme winner portfolios
and insucient to eliminate excess returns. For example, using the
most aggressive fee schedule for domestic equity portfolios, a $10
million mandate in small cap growth portfolios has an annual fee of
1 percent. For these portfolios, the annual alpha is 5.76 percent
(a quarterly alpha of 1.44 multiplied by 4), which is an order of
magnitude higher than fees. Moreover, the schedules themselves
represent an upper limit on actual fees charged by investment
managers; individual fee arrangements between plan sponsors and
investment managers typically include rebates, some of which can be
quite large. Beyond the sheer magnitude of fees, there is other
interesting heterogeneity in fee arrangements. Two types of
arrangements are more common among better performing portfolios:
performance-based fees and most-favored nation (MFN) clauses. The
former typically involve low or no minimum fees, and link actual
fees to beating a prescribed3 5. benchmark. In domestic equity
portfolios, almost 55 percent of portfolios in the extreme winner
decile oer performance-based fees, compared with 47 percent for the
extreme loser decile. By comparison, the use of incentive fees is
less common in mutual funds. Elton, Gruber and Blake (2003) report
that only 108 funds oered incentive fees in 1999 (see also Golec
and Starks (2004)). MFN clauses require that the investment manager
match fees for comparable mandates and plan sponsors. For example,
if plan sponsor A negotiates a fee of x basis points, and plan
sponsor B subsequently negotiates a fee of x-5 basis points for a
similar sized mandate with similar restrictions, the investment
management rm is obliged to reduce As fees by 5 basis points.
Again, investment managers with portfolios in the extreme winner
decile are more likely to oer MFN provisions than in the extreme
loser decile (38 versus 33 percent).There is considerable economic
and practical signicance to persistence in institu- tional
portfolios. From a practical perspective, if performance persists
net of fees, then plan sponsors could benet from picking winners.
Interestingly, and in stark contrast to retail mutual funds,
predictability in institutional portfolios arises entirely from the
winner portfolios. Since loser portfolios cannot be shorted, the
upshot of this asymme- try is that picking winners is a feasible
strategy for plan sponsors. Consistent with this implication, Goyal
and Wahal (2005) show that plan sponsors hire investment managers
that have provided superior historical returns. The economic impact
of chasing past winners is that capital inows should follow
returns. Del Guercio and Tkac (2002) and Heisler, Knittel, Neumann,
and Stewart (2004) show that the ow-performance relation for
institutional asset managers is linear, unlike the convex relation
for mutual funds (Ippolito (1992), Sirri and Tufano (1998), and
others). If there are diseconomies of scale in investment
management, such inows should cause future alphas to deteriorate.We
investigate the relation between performance and ows across
portfolios in each of the three asset classes. For domestic equity
portfolios, capital ows one year after portfolio formation increase
monotonically from loser to winner deciles. The spread in ows
between the extreme winner and loser deciles is as much as 36
percent of total net assets. In dollar terms, the average portfolio
in the loser decile loses $27 million of assets while the average
portfolio in the winner decile gains $132 million. Monotonicity is
also evident for xed income and international equity portfolios.
Winner deciles have considerably higher inows than loser deciles;
the extreme winner-loser spread in ows is 16 percent for xed income
and 24 percent for international equity. These large capital inows
appear to have severe consequences for future perfor- mance, at
least for domestic and international equity portfolios. In domestic
equity4 6. portfolios, the winner decile that had a Fama-French
alpha of 1.44 percent per quarter in the rst post-ranking year
(with an accompanying inow of 31.80 percent of assets), has an
alpha of 0.28 percent (statistically insignicant) in the second
post-ranking year. In international equity portfolios, the alpha
for the extreme winner decile deteriorates from 2.76 percent per
quarter in the rst post-ranking year to 0.36 percent in the second
year and -0.35 percent in the third year. Persistence and ows in
xed income portfolios remain an enigma: alphas in winner deciles
remain high up to two years, and some- times three years, after
decile formation. Moreover, ows do not appear to diminish
signicantly over time. Even for equity portfolios, making the
causal leap that ows drive down alphas is a dicult, if not
impossible, task. To do so, one would need to generate an estimate
of capacity as well as diseconomies originating from investment
ideas and/or execution costs a task that is impossible without
proprietary data. We can, however, make some progress on the issue
at a more aggregate level. We do so by taking our extreme winner
portfolio at the end of the rst year and double sorting it with
respect to total assets and capital ows. Thus, we split the extreme
winner portfolio into four groups. If ows drive down alphas for
portfolios that are closer to capacity, then the subsequent decline
in alpha should be larger in the high total assets and high ow
intersection. The advantage of this approach is that it holds
pre-inow alpha approximately constant and allows us to focus on
alpha changes. We nd that the decline in alpha is indeed higher for
the high-assets/high-ow group than for the low-assets/low-ow group.
For example, in the domestic equity asset class using a four-factor
model, the decline in alpha is - 0.53 percent for the former group
and -0.21 percent for the latter. While this certainly does not
nail the causal issue, it does reinforce the idea that ows and
performance are endogenous. This notion is central to papers that
seek to understand the ow- performance relation (Lynch and Musto
(2003), Berk and Green (2004), and others). It is this very
endogeneity that, when combined with diseconomies of scale,
produces the quantity-based equilibrating mechanism that is at the
heart of Berk and Green (2004). Our results show that such
equilibration is not instantaneous some persistence is necessary to
draw the ows that subsequently extinguish persistence. In retail
mutual funds, redemptions and capital inows are rapid. In contrast,
in an institutional setting, capital ows are sticky. As a result,
the equilibration process appears to take one to two years.
Mechanisms exist to control asset ows that are endogenous to
investment man- agers. For instance, investment managers that
recognize potential diseconomies and 5 7. understand the
ow-performance relation could simply refuse capital inows by shut-
ting winner portfolios to new investors. In retail mutual funds,
Bris, Gulen, Kadiyala, and Rau (2005) report that over a 10 year
period, 143 equity funds that delivered posi- tive excess returns
subsequently closed to new investors. In our sample, 208 out of
2,881 domestic equity portfolios are closed to new investors, a
rate that is higher than that for retail mutual funds. More
interestingly, portfolios that are closed to new investors have an
average benchmark-adjusted quarterly return of 1.15 percent,
compared to 0.65 percent for all portfolios. Thus, it appears that
some investment management rms close portfolios to new investors
due to concerns about diseconomies.3 Another way to con- trol ows
is through fees. Although we can only observe fee schedules (not
actual fees), we can determine if portfolios in the winner decile
charge more than those in the loser decile. Our data provide some
weak evidence that this is indeed the case: for similar sized
mandates, marginal fees are higher for better performing
portfolios. To the extent that better-performing portfolios are
likely to oer smaller rebates than portfolios in the loser decile,
the spread in fees may in fact be larger than we can detect. Our
paper builds on the small but growing literature in institutional
asset manage- ment. The progenitors in this area are Lakonishok,
Shleifer, and Vishny (1992), who examine the performance of
equity-only portfolios managed by 341 investment manage- ment rms
between 1983 and 1989. Lakonishok et al. (1992) nd that performance
is poor on average, and although there is some evidence of
persistence, they conclude that survival bias and a short time
series prevent them from drawing a robust conclusion. Coggin,
Fabozzi, and Rahman (1993) also focus on equity portfolios and nd
that in- vestment managers have limited skill in selecting stocks.
Del Guercio and Tkac (2002) and Heisler, Knittel, Neumann, and
Stewart (2004) examine the relation between ow of funds and
performance and conclude that plan sponsors withdraw funds from
poorly performing investment managers. They also nd that ows are
positively related to Jensens alpha but negatively related to
tracking error. Goyal and Wahal (2005) exam- ine the selection and
termination of investment managers by plan sponsors and nd that
investment management rms are hired after superior performance and,
generally, but not exclusively, red after poor performance.
Post-hiring excess returns are zero, rather than positive, as one
would expect if performance persists. However, the dierence be-
tween their results and ours is largely due to horizon: hiring
decisions are conditioned on three-year returns, at which point
persistence has died out. Ferson and Khang (2002) use portfolio
weights to infer persistence, and Tonks (2005) examines the
performance of 3 For example, Aronson+Johson+Ortiz, a prominent
investment management rm, recently stopped accepting new capital
into its large cap value portfolio precisely because of such
worries. 6 8. UK pension fund managers between 1983 and 1997. Both
nd some evidence of excess performance. Perhaps the closest study
to ours is Christopherson, Ferson, and Glass- man (1998), who study
persistence among 185 equity-only investment managers between 1979
and 1990. Although their sample suers from survival bias, using a
conditional approach, they nd some evidence of persistence,
particularly among poorly performing investment managers.4 Our
paper proceeds as follows. Section 2 discusses our data and
methodology. Section 3 presents results. Section 4 concludes.2 Data
and Methodology2.1 DataAs mentioned earlier, we obtain data from
two independent data providers: Informa In- vestment Solutions
(IIS) and eVestment Alliance. Both rms provide data, services, and
consulting to plan sponsors, investment consultants, and investment
managers. Since there are dierences in composition of each of the
databases, we describe them in detail below, noting issues that are
particularly relevant for our results.IIS provides quarterly
returns of portfolios managed by investment management rms from
1979 to 2004. Panel A of Table 1 presents some basic descriptive
statistics of the IIS database. Prior to 1991, this database only
contains live portfolios. Subsequently, data gathering policies
were revised such that investment management rms that exit the
universe due to closures, mergers, and bankruptcies were retained
in the database. Thus, data over the 1979-1990 sample period suers
from survivorship bias, while the returns thereafter are free of
such problems. We therefore report separate statistics for the two
subperiods, 1979-1990 and 1991-2004. Not surprisingly, both the
total and average number of rms (and portfolios) per year are much
higher in the second part of the sample period. In general,
coverage of the database is fairly comprehensive; we cross-check
the number of rms with data contained in the Mercer Performance
Analytic database and nd that our database coverage is slightly
better. Our coverage also corresponds favorably to that found in
publications such as the Money Market 4 Using the structural break
in survivorship in our sample (1979-1990 and 1991-2004), we later
show that the magnitude of the survivorship can be quite large. For
example, the survivorship-biased alpha computed using Fama-French
procedures in the extreme loser portfolio is 0.82 percent per
quarter versus -1.19 percent in the non-survivorship-biased period,
a spread of 2 percent per quanter. 7 9. Directory of Investment
Advisors. As expected, the attrition rate of portfolios between
1979 and 1990 is zero. Carhart (1997) reports that one-third of all
retail mutual funds disappear over a 31 year period, which
corresponds to about 3 percent per year. Attrition in our
non-survivorship biased sample period (1991-2004) is slightly
higher and varies from 3.2 to 3.6 percent per year. Several
features of the data are important for understanding the results.
First, since investment management rms typically manage more than
one portfolio, the database contains returns for each portfolio.
For example, Aronson+Johnson+Ortiz, an invest- ment management rm
with over $22 billion in assets, manages 10 portfolios in a variety
of capitalizations and value strategies. The returns in our
database correspond to each of these 10 portfolios, and our unit of
analysis is each portfolio return. Second, the database contains
composite returns provided by the investment management rm. The
individual returns earned by each plan sponsor client (account) may
deviate from these composite returns for a variety of reasons. For
example, a public dened benet plan may ask an investment management
rm to eliminate sin stocks from its portfo- lio. Such restrictions
cause deviations from composite returns, but these deviations are
typically quite small. Therefore, composite returns are
representative of actual earned returns. Third, the returns are net
of trading costs but gross of investment management fees. The
database also contains two other critical pieces of information:
style assignments and assets at the end of the year. For domestic
and international equity portfolios, each portfolio is associated
with a primary style and a market capitalization. Twenty-nine
primary equity styles and four market capitalization categories
exist. The majority of the data reside in value, growth, and
core-diversied styles. The market capitalization categories include
micro ($7 billion). Geographic parameters are not available for
international equity portfolios (e.g. EAFE or EAFE excluding
Japan). Twenty- eight primary xed income styles exist, but again,
most of the data reside in just a few categories, including core,
maturity controlled, government, and high yield. Fixed income
maturity breakpoints are 1, 3, and 7 years. Unlike returns, total
assets in each portfolio are only recorded at the end of the year.
Moreover, asset information is only available for approximately 30
percent of investment management rms. The database contains both
active and passive portfolios, but since our interest is in the
performance persistence of active managers, we remove all passive
portfolios from the sample. We break statistics down by the three
major asset classes - domestic equity 8 10. (including all size and
value-growth intersections), xed income (domestic xed income
portfolios containing corporate and/or government debt securities),
and international equity (including global portfolios). This is in
contrast to Lakonishok et al. (1992), Del Guercio and Tkac (2002),
Heisler et al. (2004), and Christopherson et al. (1998), all of
whom focus exclusively on domestic equity portfolios. Our secondary
data source, eVestment Alliance, provides quarterly composite re-
turns, fee information, and an identier that tags portfolios that
are closed to new investors. Unlike IIS, the names of investment
management rms are not hidden. Panel B of Table 1 provides
descriptive statistics for the eVestment data. The time series cov-
erage is shorter, starting in 1991. The cross-sectional coverage is
also smaller than the IIS database. For example, the IIS database
covers 1,137 domestic equity investment management rms and 3,381
portfolios between 1991 and 2004. In contrast, eVestment provides
data on 805 rms and 2,682 portfolios. The attrition rate is
approximately 1 percent, substantially smaller than for the IIS
database.5 Because of these dierences, we generate estimates of
persistence from the IIS database, and use eVestment data to
provide a better understanding of fees and portfolio closures. 2.2
Methodological ApproachOur empirical approach to measuring
persistence follows the mutual fund literature with some minor
adjustments to accommodate certain facets of institutional
investment management. For instance, we follow Carhart (1997) and
form deciles based on raw returns during a ranking period and
examine returns over a subsequent evaluation period. However, we
also form deciles based on benchmark-adjusted returns for two
reasons. First, plan sponsors frequently focus on
benchmark-adjusted returns, at least in part because expected
returns from benchmarks are useful for thinking about broader asset
allocation decisions in the context of contributions and retirement
withdrawls. Second, sorting on raw returns could cause portfolios
that follow certain types of investment styles to systematically
fall into winner and loser deciles. For instance, small cap value
portfolios may fall into winner deciles in some periods, not
because these portfolios delivered abnormal returns, but because
this asset class generated large returns over that period. Although
alpha estimates from three- and four-factor models adjusts for
these characteristics in the post-ranking period, they do not
inuence decile formation if5 We note, however, that a direct
comparison of the attrition rates is not possible because the
cross- sectional coverage is also smaller; it could be that
attrition rates in the rms not sampled are higher. 9 11.
assignments are based on raw returns. By contrast, using
benchmark-adjusted returns to form deciles circumvents this
problem. Beginning at the end of 1979, we sort portfolios into
deciles based on the prior annual raw or benchmark-adjusted
return.6 We then compute the equal-weighted return for each decile
over the following one-quarter. As we expand our analysis to
examine persistence over longer horizons, we compute this return
over appropriate future intervals (for one- year results, we
compute the equally-weighted return over quarters 1 through 4, and
so forth). We then roll forward, producing a non-overlapping set of
post-ranking quarterly returns. Concatenating the evaluation period
quarterly returns results in a time series of post-ranking returns
for each portfolio; we generate estimates of abnormal performance
from these time-series. We assess post-ranking abnormal performance
by regressing the post-ranking returns on K factors as
follows:KUrp,t = p + p,k fk,t + p,t ,(1)k=1 where r is the return
on portfolio p, and fk is the k th factor return. For domestic
equity portfolios we use the Fama and French (1993) three-factor
model with market, size, and book-to-market factors. Since Carhart
(1997) shows that incorporating individual stock momentum
(Jegadeesh and Titman (1993)) removes most of the persistence
evident in mutual funds, we also estimate models that include a
momentum factor. We obtain these four factors from Ken Frenchs web
site. For xed income portfolios, we again follow Fama and French
(1993) and estimate a three-factor model with the Lehman Brothers
Aggregate Bond Index return, a Term Spread Return computed as the
dierence between the long-term government bond return and the
T-bill return, and a Default Spread Return computed as the dierence
between the corporate bond return and the long- term government
bond return. We obtain aggregate bond index returns from Mercer
Performance Analytics, available from 1981 onwards. We obtain the
default and term spread returns from Ibbotson Associates, available
from 1979 onwards. For international equity portfolios, we employ
an international version of the three-factor model. We obain the
international market return and book-to-market factor from Ken
French. We6 We sort based on prior one-year returns for two
reasons. First, this choice is consistent with that of the mutual
fund literature and allows us to directly compare estimates of
alpha. Second, longer ranking periods are more likely associated
with large dierences in fund size, and, given the evidence in Chen,
Hong, Huang, and Kubik (2004), we might not expect performance to
persist across large variations in fund size (due to diseconomies).
10 12. compute the international size factor as the dierence
between the S&P/Citigroup PMI World index return and the
S&P/Citigroup EMI World index return, both of which exclude the
United States (see http://www.globalindices.standardandpoors.com).
Christopherson et al. (1998) argue that unconditional performance
measures are inappropriate for two reasons. First, they note that
sophisticated plan sponsors pre- sumably condition their
expectations based on the state of the economy. Second, to the
extent that plan sponsors employ dynamic trading strategies that
react to changes in market conditions, unconditional performance
indicators may be biased. They advocate and show that conditional
performance measures can improve inferences. We follow their
prescription and estimate conditional models in addition to the
unconditional models described above. We estimate the conditional
models as:KLC0lrp,t = p + p,k + p,k Zl,t1 fk,t + p,t , (2)k=1l=1
where the Zs are L conditioning variables.We use four conditioning
variables in our analysis. We obtain the 3-month T-bill rate from
the economic research database at the Federal Reserve Bank at St.
Louis. We compute the default yield spread as the dierence between
BAA- and AAA- rated corporate bonds using the same database. We
obtain the dividend-price ratio, computed as the logarithm of the
12-month sum of dividends on the S&P 500 index divided by the
logarithm of the index level, from Standard & Poors. Finally,
we compute the term yield spread as the dierence between the long
term yield on government bonds and the T-bill yield, using data
from Ibbotson Associates.3 Persistence3.1Short-term
PersistenceTable 2 shows estimates of one-quarter evaluation period
alphas where raw returns deter- mine decile assignment. The table
reports results for both unconditional and conditional methods. The
t-statistic reported next to the alpha estimate for each decile is
based on the standard error of the regression coecient. At the
bottom of each column we report a Spearman correlation coecient. 11
13. Panel A of Table 2 shows results for domestic equity. Although
we do not report 22 each of the individual R from the regressions,
the average R is 0.91 (0.94) for the 2 unconditional three-factor
(four-factor) model. The corresponding average R for the2
conditional models are greater, 0.95 and 0.97 respectively. In
general, the R are large and comparable to those in prior studies
(Fama and French (1993) and Carhart (1997)).Three facts immediately
stand out. First, the eect of survivorship bias is quite large. For
instance, the alpha of the rst decile (extreme losers) using the
three-factor model over the survivor-biased 1979-1990 sample period
is 0.82 percent per quarter with a t-statistic of 2.11. The
corresponding alpha in the non-survivorship-biased period of
1991-2004 is -1.19 percent, a dierential of 2.01 percent. For the
four-factor model, the dierential is 1.74 percent per quarter.
Another way to see the eects of survivorship bias is to examine the
spread between the extreme winner and extreme loser decile. Based
on the unconditional three-factor model, this spread is 1.21
percent per quarter in the survivorship biased sample and 4.07
percent per quarter in the unbiased sample. Second, there is
virtually no persistence among the worst performers. Concentrating
on the survivor-bias free sample period, the alphas of loser
deciles are statistically in- distinguishable from zero. This is in
stark contrast to the evidence in mutual funds, where persistence
is strongest in the extreme loser decile. Third, persistence in
deciles 7 through 10 (and sometimes in decile 6) is statistically
signicant. The alphas across these deciles increase monotonically,
with alphas in the extreme winner deciles ranging from 1.07 percent
per quarter to 3.33 percent per quarter. Again, in contrast to the
evidence for retail mutual funds, the addition of momentum to the
factor model reduces but does not eliminate persistence; between
1991 and 2004, the unconditional (condi- tional) alpha drops from
2.88 percent (3.33 percent) per quarter to 1.07 percent (1.62
percent), but remains highly statistically signicant. Even by
conservative standards, the coecients are economically large,
generating annualized abnormal returns greater than 4 percent,
which compares favorably to retail mutual funds. Panel B of Table 2
provides estimates of one-quarter alphas for xed income port-2
folios. The average R of the unconditional regressions is
substantially lower than for domestic equity portfolios, 0.79, but
improves to 0.91 for the conditional regressions. 2 This increase
in R is perhaps not surprising since some of the conditioning
variables capture the eects of variations in yields in xed income
securities. Since conditioning information is especially relevant
for this asset class, we focus our attention on estimates generated
by conditional models. Because factor information is not available
prior to 1981, we present results for 1981-1990 and 1991-2004
subperiods. 12 14. Evidence of persistence in the loser deciles
does not exist in either subperiod. How- ever, the winner deciles
persist considerably. In conditional models estimated for the
1991-2004 subperiod, alphas are positive, increase monotonically,
and are statistically signicant for deciles 6 through 10. For the
extreme winner decile, the alpha from the unconditional
(conditional) model is 1.03 percent (1.53 percent) per quarter with
a t- statistic of 3.76 (4.68). These abnormal returns are
economically large, particularly for xed income portfolios. Panel C
of Table 2 shows one-quarter alphas for international equity
portfolios es- timated using three-factor models. Since we only
have factor data from 1989 onwards, and since very few
international equity portfolios exist prior to 1991, we calculate
al- phas for the non-survivor-biased sample from 1991 onwards.
Again, the performance of loser deciles does not persist. Winner
deciles, by contrast, continue to show large positive alphas. For
example, for deciles 9 and 10, unconditional and conditional alphas
range from a low of 2.04 to a high of 2.67 percent per quarter, and
three out of four are statistically signcant. In addition to
forming deciles based on raw returns, we also form deciles based on
benchmark-adjusted returns. Our returns data come with style
assignments and self designated benchmarks for domestic equity and
xed income portfolios. For domes- tic equity, benchmarks are based
on size and value-growth grids. For xed income, benchmarks are
based on credit and maturity terms. We use benchmark-adjusted an-
nual returns to form deciles and estimate alphas in the subsequent
quarter using the same methods as before. Unfortunately, benchmark
designations are not available for international equity portfolios.
In Table 3, we ignore the survivorship-biased sample period and
present results us- ing benchmark-adjusted returns for 1991-2004.
Compared to the earlier results, some changes are evident in the
results for domestic equity portfolios. First, the point esti-
mates of alphas for loser deciles are larger than those generated
by raw returns sorts, but remain statistically insigicant. Thus,
although portfolios in loser deciles do not perform as poorly, no
evidence of persistence or reversal continues to exist among
losers. Second, the alphas of portfolios in winner deciles are
smaller by approximately 0.5 percent per quarter. In the extreme
winner decile, the alpha from an unconditional three-factor model
drops from 2.88 percent per quarter (based on the raw return sort)
to 2.03 per- cent per quarter. In one case, the alpha loses
statistical signcance (the unconditional four-factor model based on
benchmark-adjusted returns). Despite these dierences, the general
pattern of results remains the same loser deciles do not persist
and extreme13 15. winner deciles do.For xed income portfolios,
ranking on benchmark-adjusted returns produces a sim- ilar set of
results, at least for conditional alpha estimates. Portfolios in
extreme loser deciles show no predictability. However, deciles 4
through 10 show positive and statisti- cally signicant alphas. For
deciles 4 through 9, alphas range from 0.10 to 0.31 percent, and
the alpha for the extreme winner decile is 0.80 percent per
quarter. 3.2 Longer-term PersistenceSignicant persistence in
performance one quarter after portfolio formation is certainly
economically meaningful. However, plan sponsors typically do not
deploy capital in a portfolio for one quarter, since the
transaction costs from exiting a portfolio after one quarter and
entering a new one are large and potentially prohibitive. Even if
plan spon- sors employ transition management rms that seek to
minimize such costs, the frictions are simply too large to justify
a performance-chasing strategy. In addition, adverse rep- utation
eects likely exist from trading in and out of institutional
portfolios. If excess returns remain high for a sucient period of
time after portfolio formation, however, then plan sponsors may be
able to exploit performance persistence. Accordingly, in this
section, we examine persistence in institutional portfolios over
longer horizons.Methodologically, we follow the same procedure as
before, with some minor adjust- ments. We roll forward annually and
calculate post-ranking quarterly returns for three years following
portfolio formation. Thus, we compute the alpha for the rst year
from the beginning of year 1 to the end of year 1, we compute the
alpha for the second year from the beginning of year 2 to the end
of year 2, and so forth. For domestic equity and xed income, we
assign deciles using benchmark-adjusted returns. Since benchmarks
are unavailable for international equity portfolios, we assign
their deciles using raw returns. Also, since conditioning
information clearly aects inference for xed income portfo- lios, we
estimate conditional xed income alphas; alphas for domestic and
international equity portfolios are unconditional. We do not show
results for two- and three-quarter evaluation periods so that we do
not overwhelm the reader with too many results.7 Instead, we simply
show alphas for each decile in years 1, 2, and 3 after portfolio
formation.7 The alphas for winner portfolios over two- and
three-quarter holding periods are generally large and statistically
signicant.14 16. Panel A of Table 4 shows alphas for domestic
equity. Using a three-factor model, the performance of winner
deciles persists. For example, deciles 9 and 10 have alphas of 0.56
(t-statistic=2.51) and 1.44 (t-statistic=3.34) percent
respectively. However, the four-factor model alphas for these
portfolios shrink to 0.21 and 0.47 percent respectively and lose
their statistical signicance. In the second year, alphas sharply
reverse: the loser decile alphas are actually positive (1.04 and
0.80 percent) and statistically signicant, whereas the winner
decile alphas are indistinguishable from zero using both three- and
four-factor models. What appears to be reversal in the loser decile
is most likely driven by greater attrition in that decile. The
attrition rate in the extreme loser decile is 7.6 percent in the
rst year, 5.3 percent in the second year, and 5.6 percent in the
third year. In comparison, corresponding attrition rates in the
extreme winner decile are 1.9 percent, 2.7 percent, and 2.8 percent
respecitvely. Since we do not assign a delisting return to the
portfolios that exit a decile, the returns of loser deciles are
positively biased. Three years after portfolio formation, almost
all alphas have deteriorated to zero. Alphas in xed income deciles
appear to follow a U-shape in the rst year (Panel B of of Table 4)
alphas are greatest in the extreme winner and loser deciles and
weaker in intermediate deciles. Again, reversal in the extreme
loser decile is most likely driven by higher attrition rates among
loser portfolios and the fact that we do not assign a delisting
return to exiting portfolios (the cumulative attrition rate over a
three year period is 11.9 percent for the extreme loser decile and
7.4 percent for the extreme winner decile). In the second year, the
alphas of the extreme deciles are dampened and by the third year,
the alpha of the extreme loser decile is indistinguishable from
zero. In all cases, the extreme winner decile has the highest alpha
and it is economically quite large. For instance, the alpha is 0.62
percent per quarter in the rst year and 0.58 percent per quarter in
the third year. In winner deciles of international equity
portfolios (Panel C of of Table 4), alphas are 1.98 and 2.76
percent per quarter for deciles 9 and 10 in the rst year. These
alphas decline sharply in the following years. The decile 10 alpha
drops from 2.76 percent in the rst year to 0.97 percent in the
second year, and 0.48 percent in the third year. Interestingly,
alphas for the loser deciles climb over time, although the
t-statistic of the extreme loser decile never achieves signicance.
In general, persistence appears to last up to one year in equity
portfolios (both domestic and international). The xed income
results, however, are puzzling to the extent that persistence
continues up to three years after decile formation. It is possible
that the composition of the extreme deciles is unusual and causes
this pattern. We15 17. examine whether any particular types of
investment styles load in these deciles. To do so, we calculate the
percentage of each deciles observations associated with each
investment style, and then subtract this percentage from the
unconditional mean across deciles. The nine styles are: money
market, municipal, core, short-term, intermediate-term, long- term,
mortgages, convertibles, and high-yield. This procedure produces a
deviation for each decile and style, which sums to zero within a
decile. We nd that deciles 1 and 10 have the largest loadings on
high yield portfolios.8 Since the default spread variable that we
use is the dierence between corporate and government bond returns,
it likely poorly adjusts for high yield securities. Consequently,
the persistence results that we observe are likely caused by
imprecise factor adjustments. To help with inference, in addition
to computing alphas based on factor models, we also compute
benchmark- adusted returns and information ratios for each decile
over the three-year period following decile assignment. Our hope is
that benchmarks more precisely account for high yield securities in
the high yield portfolios. Naturally, we sacrice the regression
approach and instead compute simple benchmark-adjusted returns and
information ratios: e rt = rp,t rb,t(3)ertIR = e , (4)(rt )where rp
is the return on portfolio p, rb is the return on the corresponding
benchmark, r e is the excess return, and IR is the information
ratio. Table 5 shows these returns and information ratios each year
for domestic equity (Panel A) and xed income portfolios (Panel B).
The domestic equity results mirror those from the regression
approach - persistence lasts up to one year in the winner deciles
and disappears thereafter. For xed income portfolios, persistence
only appears in the intermediate deciles during the rst year, which
represents an important departure from the alphas presented in
Table 4, and is consistent with our suspicion that the xed income
factor model inadequately handles high yield portfolios. Overall,
performance persists at least up to the end of the rst year after
portfolio formation for domestic and international equity
portfolios. These excess returns revert to zero thereafter.
Persistence in xed income lasts longer in some specications, and in
many ways remains enigmatic.8 Decile 1 (10) shows postive deviation
of 5.3 percent (18.3 percent) from the unconditional mean.
Additionally, the intermediate deciles all have negative loadings
on high yield investment styles. We also check loadings of various
investment styles in domestic equity and nd no substantial
deviations. 16 18. 3.3Persistence and FlowsThe evidence thus far
suggests that the performance of institutional portfolios persists
over short- to intermediate-term horizons. Excess returns are
generally positive up to one year after portfolio formation and
revert to zero thereafter. Although the magnitude of the excess
returns and the reversals varies across asset classes, the general
pattern appears widespread. Prior performance is an important
screen used by plan sponsors in selecting investment management rms
and allocating capital. Thus, new capital likely ows into extreme
winner portfolios and ows out of extreme loser portfolios. If there
are decreasing returns to scale in investment management, then
capital ows could account for the patterns in persistence that we
observe. We measure percentage asset ows, Cfp,t , for each
portfolio p during the year t as:Ap,t Ap,t1 (1 + rp,t) Cfp,t
=,(5)Ap,t1where Ap,t measures the dollar amount of assets in
portfolio p at the end of year t, and rp,t is the gross return on
portfolio p during the year (not quarter) t. Measurement of ows in
this manner (as fractional ows) is analagous to that typically
employed in the mutual fund literature. We truncate ows from below
by 1 so that small asset values in the denominator do not produce
large outlier ows that could distort our results. Two important
considerations exist in measuring institutional portfolio ows.
First, total assets for each portfolio are only available on an
annual basis, whereas returns are available quarterly. This
mismatch could potentially induce discreteness into the
ow-performance relation. In practice, however, it unlikely makes
much of a dierence, because investment management rms largely gain
or lose assets when they are hired or red by plan sponsors. Since
such selection and termination decisions are in themselves
discrete, no articial discreteness is induced in the ow data. It is
also worth noting that this capital ow process is quite dierent
from retail mutual funds, where purchases and redemptions take
place daily. Second, total asset data are available for a smaller
sample than that of returns. To ensure that we assign portfolios to
the correct decile, we assign deciles based on all portfolios that
report returns, and then calculate the mean ow for each decile
portfolio with available data. This procedure could in and of
itself cause selection bias, an issue that we address in section
3.4.Table 6 shows average capital ows into domestic equity, xed
income, and interna- tional equity portfolios in each decile 1, 2,
and 3 years following decile formation. In 17 19. domestic equity,
a monotonic relation exists between ows in year 1 and decile
ranking based on the prior year return. In year 1, decile 10
receives a ow of almost 32 percent of assets whereas decile 1 loses
almost 4 percent of assets, a spread of 36 percent. Since the
average portfolio in decile 1 (10) has total assets of $466 million
($420 million), this implies a loss (gain) of $27 million ($132
million). In the following year (year 2), the alphas revert. For
winner portfolios, the three- (four-) factor alpha declines from
1.44 percent (0.47) per quarter in year 1 to 0.09 percent (-0.22
percent) in year 2. Interest- ingly, the loser decile three-
(four-) factor alpha goes from 0.05 percent (0.40 percent) in year
1 to 1.04 percent (0.80 percent) in year 2. Portfolios in the
intermediate deciles follow similar patterns: deciles with the
highest capital ows have the lowest alphas during the next year. In
xed income, alphas persist over longer horizons. Capital ows are
smaller in percentange terms than domestic equity, but the spread
between winner and loser deciles remain. Perhaps even more
interesting is the fact that changes in ows over time are not as
dramatic. For instance, in the extreme winner decile, the change in
ow from year 1 to year 2 is only 4 percentage points, less than
half the 9 percentage points for domestic equity. The ow patterns
for international equity portfolios are more similar to those for
domestic equity. Alphas in the rst year are quite high (the extreme
winner decile alpha is 2.76 percent) and followed by extremely high
capital ows (29 percent). The year after these capital inows,
decile 10 alpha shrinks to 0.97 percent, following which capital
inows decline to 23 percent. Again, similar patterns exist in other
deciles. Despite the barrage of statistics above, the general
pattern that emerges for domestic and international equity
portfolios is straightforward. Performance lasts up to a year after
portfolio formation. Among winners, this persistence elicits capial
inows, and excess returns subequently revert. It is appropriate at
this point to examine our re- sults in the context of the
assumptions and implications of Berk and Green (2004). In Berk and
Greens model, performance does not persist, even though there is
dierential ability across fund managers. Investors rationally
respond to past performance. Capital ows into superior performers,
which in conjunction with assumed diseconomies of scale, causes
future excess returns to disappear. We observe persistence. We
cannot directly measure diseconomies of scale, although some
evidence indicates it exists (Perold and Salomon (1991)). We also
observe ows that appear to follow performance after which excess
returns disappear.18 20. The individual moving parts of the
evidence are consistent with ows aecting future persistence. To
establish a clear causal link, one would need to know the capacity
of an existing portfolio and then the impact of ows on future
returns. That is an impossible task without proprietary data. We
can, however, make a modest attempt in that direction. To do so, we
separate the extreme winner decile into four groups at the end of
the rst year based on total assets and the degree of capital ows.
Eectively, we do a double-sort on the winner portfolio using assets
and ows. Our hope is that total assets provide a crude measure of
capacity, and that if diseconomies of scale reduce future alphas,
then changes in alpha should be larger for big portfoliso that
experience larger inows. We nd that the decline in alpha is indeed
greater for the high-assets/highow group than for the
low-assets/low-ow group. In domestic equity, using a four-factor
model, the decline in alpha is -0.53 percent for the former and
-0.21 percent for the latter. This evidence certainly suggests a
link between ows and future performance. At a minimum, it provides
circumstantial support for the quantity-based equilibrating process
that is at the heart of Berk and Green (2004). The key dierence is
that in our setting it takes one year, and in some cases two, for
the equilibrating mechanism to take eect, most likely because
capital ows in the institutional arena are sticky (particularly
when compared to retail mutual funds). 3.4Fee ArrangementsWe base
our results thus far entirely on returns that are net of trading
costs but gross of fees. The possibility exists that investment
management fees eliminate the post- ranking period excess returns.
Cross-sectional variation in fees could also be so large that it
swamps alphas in winner portfolios; in other words, average fees
may not be high enough to eliminate excess returns across all
portfolios, but only in the extreme winner deciles. In this
section, we analyze fees charged by investment management rms for
institutional portfolios. As noted earlier, to study fees we employ
data from eVestment Alliance. This rm provides composite quarterly
returns and fee information. The proto-typical fee struc- ture is
such that management fees decline as a step function of the size of
the mandate assigned to the investment management rm by the plan
sponsor. Although variation undoubtedly exists in the breakpoints,
eVestment collects marginal fee schedules using standardized
breakpoints. Specically, each investment manager identies fees for
$10, $25, $50, $75, and $100 million mandates. These marginal fees
are based on fee 19 21. schedules; actual fees are individually
negotiated between investment managers and plan sponsors. Such
individual negotiations involve rebates to the marginal fees as
well as other structural fee arrangements (e.g. performance
linkages). To our knowledge, no available database details
individual fee arrangements. As a result, we regard our analy- sis
as exploratory in nature and designed only to address issues
pertaining to persistence. While our data are new and unique, they
are not rich enough to provide a comprehensive understanding of
actual fee arrangements in institutional investment management.
Table 7 provides descriptive information on fee arrangements. In
Panel A, we report the percentage of portfolios that oer
performance-based fee clauses in contracts across each asset class
and decile.9 Performance-based fees often have no minimum and link
actual fees to performance above a prescribed benchmark. For
domestic equity and xed income portfolios, a substantial variation
exists in the percentage of portfolios that oer performance-based
fees across the deciles. In domestic equity, for example, only 47
percent of the portfolios in the loser decile oer performance-based
fees while almost 55 percent in the winner decile oer such an
arrangement. The corresponding percentages for xed income are 38
percent for decile 1 and 46 percent for decile 10. Little variation
exists in international equity portfolios. Panel B of Table 7 shows
the percentage of portfolios that oer most-favored nation (MFN)
clauses by asset class and decile. MFN provisions typically state
that the invest- ment manager will charge the plan sponsor a fee
that is the lowest of that charged to similar mandates from other
comparable plan sponsors. If properly enforced, an MFN clause
benets incumbent plan sponsors in the sense that the investment
manager is required to match lower fees provided to a new plan
sponsors.10 Again, some evidence suggests that portfolios in winner
deciles have greater propensity to oer MFN clauses compared to
portfolios in loser deciles. The spread between extreme
winner-loser deciles for domestic equity, xed income, and
international equity are 5, 4, and 3 percentage points
respectively. Table 8 shows the distribution of annual fees across
deciles for domestic equity, xed income, and international equity
(Panels A, B and C respectively). For each decile, we show the
average marginal fee (in basis points) in each breakpoint described
above.9 The sum of the Yes and No columns does not add up to 100
percent because this information is missing for a small number of
portfolios.10 As with most such contracts and clauses, many of the
benets are dependent on the details of the contract and its
enforcement. For instance, the investment management rm and plan
sponsor might reasonably disagree on whether mandates from two-plan
sponsors are comparable because of size or specic portfolio
restrictions (e.g. no sin stocks or use of directed brokerage). 20
22. Since fees vary widely across investment styles within domestic
equity, we show four major intersections of the size and
growth-value grid: large cap growth, large cap value, small cap
growth, and small cap value. Similarly, for xed income, we only
show fees for four styles: municipal, high yield, intermediate
term, and mortage-backed securities. The results in Table 8 display
several important elements. First, the magnitude of the fees are
such that they are unlikely to inuence our inferences with regard
to persistence; the fees are simply not large enough to eliminate
the alphas. Take domestic equity, for example. The largest fee
reported in the table is 100 basis points for the extreme winner
decile corresponding to the smallest ($10 million) mandate in small
cap growth. The quarterly alpha one year after decile formation
using a three-factor model was 1.44 percent, implying an annual
alpha of almost 6 percent. Even if one were to use the four-factor
alpha of 0.47 percent (which is statistically insignicant),
removing maximum fees still leaves an annual alpha of 1 percent.
The results for xed income and international equity are
substantially stronger. The quarterly alpha one year after decile
formation for the extreme winner decile was 0.62 percent, implying
an annual alpha of 2.4 percent. The maximum annual fee (from the
high yield investment style) is 59 basis points. The corresponding
quarterly alpha for international equity was 2.76 percent (9.5
percent annually), where the maximum annual fee is 88 basis points.
We have been quite conservative in assessing the impact of fees on
two accounts. First, fees are based on reported schedules. Actual
fees involve signicant rebates and are likely to be substantially
lower. Second, we have deliberatedly cherry-picked the largest fees
in an asset class to shrink alphas. Despite this conservatism,
investment management fees are clearly not large enough to
eliminate the alphas discovered in the persistence regressions. One
might legitimately ask whether better-performing portfolios charge
more in fees than worse-performing portfolios. An examination of
the variation in fees across deciles in each of the panels shows
evidence that this is indeed the case. Reported fees are
consistently higher for portfolios in decile 10 than in decile 1;
the dierentials are modest in domestic equity and xed income, but
larger in international equity. To the extent that investment
managers are likely to oer larger rebates for worse performing
portfolios and less likely to oer large rebates for better
performing portfolios, the dierences may in fact be larger than can
be estimated using these data.21 23. 3.5Alternative Mechanisms to
Control FlowsPortfolio managers cognizant of diseconomies of scale
and wishing to preserve their reputation for providing consistent
excess gross returns may prefer to restrict asset ows rather than
suer declines in alpha. Several possible mechanisms could be used
to achieve this result. One obvious mechanism is price. Fee
increases could be used to control asset ows, preserving gross
performance and persistence. In retail mutual funds, fees and loads
vary widely, aecting redemptions and capital inows (see Nanda,
Narayanan, and Warther (2000) for a model in which heterogeneous
fees appear endogenously). To the extent that better-performing
investment managers charge higher fees, there is some
circumstantial evidence that fees are used to control ows. However,
as noted earlier, actual fee arrangements between institutional
investment management rms and plan sponsors are private. As a
result, we cannot detect fee increases or cleanly observe the use
of heterogenous negotiated fees to discourage asset inows from
particular types of (perhaps short-term) plan sponsors. Another,
more direct, way to control asset ows is to simply stop accepting
new money in winner portfolios. Bris, Gulen, Kadiyala, and Rau
(2005) nd that 143 re- tail equity mutual funds closed to new
investors over a 10-year period, and document that these funds
delivered positive excess returns prior to closing. We are
anecdotally aware of some investment advisors that have followed
this approach in the institutional marketplace. For example,
returning to our example of Aronson+Johnson+Ortiz, this investment
management rm closed its agship large cap value portfolio to new
institu- tional portfolios under the belief that it could not
continue to generate superior returns with a larger asset base.
Data from eVestment tags portfolios that have been closed to new
investors, and we can use these data to see if this mechanism is
employed by institutional investment man- agers. Out of 5,122
portfolios, eVestment reports that 277 are closed to new investors,
a rate that appears higher than that for mutual funds. The majority
of these closures (270) are in domestic and international equity
portfolios (208 and 62 respectively, out of 2,881 and 910
portfolios in each of these asset classes). Only 7 out of 1,331 xed
in- come portfolios are identied as closed. More interestingly, the
returns are substantially higher for closed portfolios than
portfolios that are open to new investors. The average quarterly
raw return for closed portfolios are 4.9 percent, 3.2 percent, and
3.8 percent respectively for domestic equity, xed income, and
international equity respectively. The corresponding returns for
portfolios open to new investors are lower in each case: 3.9,22 24.
1.9, and 3.0 percent respectively. These dierentials are also
evident in benchmark- adjusted returns, which can be computed for
domestic equity and xed income only. The average quarterly
benchmark-adjusted return in closed domestic equity portfolios is
1.15 percent, compared to 0.65 to open portfolios. The comparable
and corresponding returns for xed income portfolios are 1.05 and
0.13 percent respectively.11 3.6 Robustness and Other IssuesOur
results could be inuenced by backll bias, similar to that observed
in hedge fund databases (Liang (2000)). Specically, it could be
that only portfolios that have been successful over some period of
time enter the database. To determine how this bias may aect our
results, we follow Jagannathan and Novikov (2005), and eliminate
the rst 2 years of returns for each portfolio in our sample, and
then re-estimate our regressions. The results of this exercise are
not reported in the paper, but our basic estimates of persistence
are almost identical. Another possibility is that a selection bias
exists in the investment management rms that report total assets
under management per portfolio. Although ex ante the source of such
a bias is hard to identify, it is a possibility that could cloud
our inferences. In a perfect world, we would be able to observe
characteristics of rms that report assets and those that do not,
and then examine whether their ow-performance relations dier.
Unfortunately, this is not possible, particularly since we do not
have rm identities. We can, however, examine whether our
persistence results dier for the subsample of rms that include both
assets and returns data. We replicate our results for this
subsample and nd that the alpha estimates do not dier materially
from those reported in Table 4, and in some cases they are even
larger. Since the database used to generate estimates of fees diers
from the data on which we base persistence estimates, the two
databases may not be comparable. To examine this possibility, we
estimate one-year alphas (equivalent to those reported in Table 4).
The alpha estimates from eVestment data are very similar to those
estimated using the larger sample obtained from IIS. As shown
earlier, although our data is survivorship-bias free, greater
attrition exists 11 Ideally, we would like to know the date on
which a portfolio was closed to new investors so that we can
compute returns before and after closure. Unfortunately, our data
do not include closure dates. If investment management rms do in
fact avoid diseconomies by closing portfolios, then our computed
return dierentials are downward biased. 23 25. in the loser decile
portfolios than in the winner deciles. Since portfolios are not
assigned a delisting return, this implies that post-ranking returns
in loser deciles are biased upward. If we were to somehow assign a
delisting return, this might generate persistence (and negative
alphas) in the loser decile. We choose not to do so for two
reasons. First, there is no obvious choice of delisting return.
Second, and more importantly, since institutional portfolios cannot
be shorted, there is little practical advantage in doing so.
Finally, a reconciliation of our results with those reported in
Goyal and Wahal (2005) are in order. They report that one, two, and
three-year excess returns of investment management rms prior to
being hired by plan sponsors are signicantly positive. Post- hiring
excess returns are statistically indistinguishable from zero. On
the surface, this may seem at odds with our evidence of
persistence. However, the key lies in horizons. Hiring decisions
reported by Goyal and Wahal occur after three years of superior
per- formance. Our evidence of persistence is that after two years
(the portfolio formation year and the subsequent evaluation year),
reversals start to occur. Consequently, these two sets of results
suggest that deciles based on three-year return rankings (rather
than one-year return rankings) should not persist. To conrm this,
we re-estimate the results in Table 4 after forming deciles based
on three-year returns. This exercise (results not tabulated) shows
that performance does not persist when deciles are formed in this
way.4 ConclusionIn this paper, we examine the persistence in
performance of 6,260 portfolios managed by 1,475 investment
management rms between 1991 and 2004. These portfolios provide
exposure to domestic equity, xed income, and international equity
asset classes to public and private dened benet retirement plans,
endowments, foundations, multi- employer unions, and trusts. A
large number of active investment styles and strategies are
included; for equity strategies, all size and growth-value
gradations are represented, and all maturity and credit risk
dimensions are included in xed income portfolios. To our knowledge,
we are the rst to examine persistence beyond the traditionally
studied domestic equity funds.We form deciles using raw and
benchmark-adjusted returns over one year and ex- amine the
persistence in performance thereafter using a variety of three- and
four-factor models. The factor models do a good job of explaining
the return series, but we nd signicant excess returns in the
post-decile formation period. Unlike retail mutual funds,24 26.
however, these excess returns are concentrated entirely in winner
deciles. The magni- tudes of the alphas themselves are economically
large. Moreover, typical investment fees are not large enough to
eliminate net-of-fee excess returns.From a practical perspective,
our results suggest that the widespread practice of hiring
investment managers that have delivered superior returns is both
rational and potentially protable. Indeed, the organizational
structure of institutional investment management and, in
particular, the use of consultants to pick investment managers is
conducive to this eort. However, the persistence that is the source
of potential gains for plan sponsors is its very own death knell:
we nd that portfolios in the winner deciles draw an inux of capital
from plan sponsors, and in the year following this capital inow,
the excess returns disappear. Berk and Green (2004) argue that
diseconomies of scale in investment management, combined with
capital ows following superior performance, could generate this
result. Our results are consistent with their model and highlight
the transmission mechanism by which this may take place.
Particularly noteworthy is the fact that, within the institutional
money management arena, it takes time for asset ows to eliminate
persistence. 25 27. References Berk, Jonathan and Richard C. Green,
2004, Mutual fund ows and performance in rational markets, Journal
of Political Economy 112, 1269-1295.Berk, Jonathan and Jing Xu,
2004, Persistence and fund ows of the worst performing mutual
funds, Working paper, University of California at Berkeley.Brown,
Stephen and William Goetzmann, 1995, Performance persistence,
Journal of Finance 50, 679-698.Bollen, Nicolas and Jerey Busse,
2005, Short-term persistence in mutual fund perfor-mance, Review of
Financial Studies 18, 569-597.Bris, Arturo, Huseyin Gulen, Padma
Kadiyala, and P. Raghavendra Rau, 2005, Clos-ing time? The
signaling content of mutual fund closures, Working paper,
PurdueUniversity.Carhart, Mark M., 1997, On persistence in mutual
fund performance, Journal of Finance 52, 57-82.Carhart, Mark M.,
Jennifer N. Carpenter, Anthony W. Lynch, and David K. Musto, 2002,
Mutual fund survivorship, Review of Financial Studies 5,
1439-1463.Chen, Joseph, Harrison Hong, Ming Huang, and Jerey D.
Kubik, 2004, Does fund size erode mutual fund performance? The role
of liquidity and organization, American Economic Review 94,
1276-1302.Coggin, T, Frank Fabozzi, and Shaqur Rahman, 1993, The
investment performance of US equity pension fund managers: An
empirical investigation, Journal of Finance 48,
1039-1055.Christopherson, Jon A., Wayne E. Ferson, and Debra A.
Glassman, 1998, Condition- ing manager alphas on economic
information: Another look at the persistence of performance, Review
of Financial Studies 11, 111-142.Del Guercio, Diane and Paula Tkac,
2002, The determinants of the ow of funds of managed portfolios:
Mutual funds versus pension funds, Journal of Financial and
Quantitative Analysis 37, 523-557.Elton, Edwin J., Martin J.
Gruber, and Christopher R. Blake, 1996, The persistence of
risk-adjusted mutual fund performance, Journal of Business 69,
133157.Elton, Edwin J., Martin J. Gruber, and Christopher R. Blake,
2003, Incentive fees and mutual funds, Journal of Finance 58,
779-804.Elton, Edwin, M. Gruber, S. Das, and M. Hlavka, 1993,
Eciency with costly informa- tion: A reinterpretation of the
evidence for managed portfolios, Review of Financial Studies 6,
1-22.26 28. Fama, Eugene and Kenneth French, 1993, Common risk
factors in the returns on stocksand bonds, Journal of Financial
Economics 33, 3-56.Ferson, Wayne and Kenneth Khang, 2002,
Conditional performance measurement usingportfolio weights:
evidence for pension funds, Journal of Financial Economics
65,249-282.Goetzmann, William and Roger Ibbotson, 1994, Do winners
repeat? Patterns in mutual fund performance, Journal of Portfolio
Management 20, 9-18.Golec, Joseph and Laura Starks, 2004,
Performance fee contract change and mutual fund risk, Journal of
Financial Economics 73, 93-118.Goyal, Amit and Sunil Wahal, 2005,
The selection and termination of investment man- agement rms by
plan sponsors, Working paper, Arizona State University.Grin, John,
Spencer Martin, and Susan Ji, 2003, Momentum investing and business
cycle risk: Evidence from pole to pole, Journal of Finance 58,
2515-47.Grinblatt, Mark and Sheridan Titman, 1992, The persistence
of mutual fund perfor- mance, Journal of Finance 47,
1977-1984.Grinblatt, Mark, Sheridan Titman, and Russ Wermers, 1995
Momentum investing strategies, portfolio performance and herding: A
study of mutual fund behavior, American Economic Review 85,
1088-1105.Hendricks, D, J. Patel, and R. Zeckhauser, 1993, Hot
hands in mutual funds: Short-run persistence of performance,
1974-1988, Journal of Finance 48, 93-130.Heisler, Jerey,
Christopher R. Knittel, John J. Neumann, and Scott Stewart,
2004,Why do institutional plan sponsors re their investment
managers? Working paper,Boston University.Investment Company
Institute, 2004, Trends in mutual fund
investing,www.ici.org/stats/index.html.Ippolito, Richard, 1989,
Eciency with costly information: A study of mutual fund
performance, 1965-1984, Quarterly Journal of Economics 104,
1-23.Jagannathan, Ravi and Dmitry Novikov, 2005, Do hot hands
persist among hedge fund managers? An empirical evaluation, Working
paper, Northwestern University.Jegadeesh, Narasimhan and Sheridan
Titman, 1993, Returns to buying winners and selling losers:
Implications for stock market eciency, Journal of Finance 48, 65-
91.Jensen, Michael, 1968, The performance of mutual funds in the
period 1945-1964, Jour- nal of Finance 48, 389-416. 27 29.
Lakonishok, Josef, Andrei Shleifer, and Robert Vishny, 1992, The
structure and per- formance of the money management industry,
Brookings Papers: Microeconomics, 339-391.Liang, Bing, 2000, Hedge
funds: The living and the dead, Journal of Financial and
Quantitative Analysis 35, 309-326.Lynch, Anthony W. and David K.
Musto, 2003, How investors interpret past fund re- turns, Journal
of Finance 58, 2033-2058.Nanda, Vikram, M.P. Narayanan, and Vincent
A. Warther, 2000, Liquidity, investmentability, and mutual fund
structure, Journal of Financial Economics 57, 417-443.Perold, Andre
and Robert Salomon, Jr, 1991, The right amount of assets under man-
agement, Financial Analysts Journal 47, 31-39.Sirri, Erik and Peter
Tufano, 1998, Costly search and mutual fund ows, Journal of Finance
53, 1589-1622.Tonks, Ian, 2005, Performance persistence of
Pension-fund managers, Journal of Busi- ness 78, 1917-1942.Wermers,
Russ, 1999, Mutual Fund Herding and the Impact on Stock Prices,
Journalof Finance 54, 581-622. 28 30. Table 1: Descriptive
Statistics This table presents descriptive statistics on the sample
of institutional investment management rms and their portfolios.
Statistics are presented for the survivorship biased sample period
of 1979 to 1990 and survivorship bias free sample period of 1991 to
2004. Firm and portfolio size is in millions of dollars. The
attrition rate is calculated by summing the number of portfolios
that disappear from the database during a year and scaling by the
total number of portfolios at the beginning of the year.Domestic
Domestic InternationalEquity Fixed Income EquityPanel A: Data
Source: IIS Sample: 19791990 Total # Firms 670 397144 Total #
Portfolios 1123 742331 Avg. # Firms per year 359 204 57 Avg. #
Portfolios per year516 309108 Avg. Size of Firm 8231439 1188 Avg.
Size of Portfolio510 797553 Attrition Rate 0.000.00 0.00 Sample:
1991-2004 Total # Firms 1137602330 Total # Portfolios3381 1683 1196
Avg. # Firms per year 873 470236 Avg. # Portfolios per year2146
1192746 Avg. Size of Firm 2530 3764 3648 Avg. Size of
Portfolio9811427 1121 Attrition Rate 3.083.18 3.48Panel B: Data
Source: eVestment Alliance Sample: 1991-2004 Total # Firms 805 356
237 Total # Portfolios2682 1290 821 Avg. # Firms per year 630 292
169 Avg. # Portfolios per year1657914 491 Attrition Rate
1.011.021.21 29 31. Table 2: Post-Ranking One-quarter Alphas with
Deciles formed using Raw ReturnsThis table lists the post-ranking
alphas for deciles of funds sorted according to the raw return
during the ranking period of one year. The portfoliodeciles are
rebalanced at the end of every quarter and are held for one
post-ranking quarter. Unconditional alphas are calculated from the
factormodel K rp,t = U + pp,k fk,t + p,t ,k=1 while the conditional
alphas are calculated from the factor model K L0l rp,t = C + p p,k
+ p,k Zl,t1 fk,t + p,t , k=1l=1 where f are K factors and Zs are L
instruments. The list of instruments includes T-bill, dividend
price ratio, term spread and default spread. Thefactors for
domestic equity are the three Fama and French (1993) factors and
the Carhart (1997) momentum factor. The factors for domestic
xedincome are the Lehman Brothers Aggregate Bond Index returns,
Term Spread Return, and Default Spread Return. The factors for
internationalequity are the international versions of Fama and
French (1993) factors. We report alphas for the survivorship biased
sample period of 1979 to 1990 30and survivorship bias free sample
period of 1991 to 2004. All alphas are in percent per quarter and
t-statistics are reported in parentheses next toalphas. Decile 1
contains the worst performing portfolio and decile 10 contains the
best performing portfolios. Panel A: Domestic Equity Unconditional
Conditional 19791990 1991200419791990 19912004 DecileFF Carhart
FFCarhart FF Carhart FFCarhart1 0.82 (2.11) 1.59 (4.16) -1.19
(-1.64) -0.17 (-0.23)0.38 (0.86) 0.72 (1.56) -1.13 (-1.62) -0.37
(-0.52) 2 0.26 (1.01) 0.68 (2.55) -0.81 (-1.56) -0.13 (-0.25) -0.00
(-0.00)0.34 (0.92) -0.78 (-1.55) -0.08 (-0.16) 3 0.58 (2.77) 0.70
(2.89) -0.62 (-1.69) -0.12 (-0.34)0.40 (1.48) 0.49 (1.42) -0.54
(-1.47) -0.06 (-0.16) 4 0.64 (3.56) 0.61 (2.92) -0.35 (-1.29) -0.06
(-0.23)0.32 (1.53) 0.37 (1.47) -0.32 (-1.22)0.07 (0.25) 5 0.74
(3.85) 0.65 (2.89) -0.01 (-0.03)0.15 (0.76) 0.52 (2.66) 0.59
(2.53)0.03 (0.23) 0.25 (1.54) 6 0.70 (4.07) 0.50 (2.61)0.18 (1.30)
0.16 (1.08) 0.72 (3.33) 0.62 (2.55)0.22 (1.84) 0.28 (2.01) 7 1.10
(5.29) 0.74 (3.45)0.67 (3.47) 0.44 (2.22) 0.89 (3.81) 0.79
(3.78)0.74 (3.75) 0.57 (2.79) 8 1.09 (4.81) 0.74 (3.07)0.97 (3.52)
0.47 (1.86) 0.79 (3.57) 0.59 (2.64)1.05 (3.69) 0.70 (2.61) 9 1.37
(4.34) 0.69 (2.31)1.51 (3.70) 0.66 (1.89) 1.13 (2.93) 0.90
(2.56)1.65 (3.98) 0.91 (2.36) 102.03 (4.41) 1.11 (2.46)2.88 (3.77)
1.07 (1.85) 1.61 (2.85) 1.23 (2.18)3.33 (4.13) 1.62
(2.59)SpCorr0.79 (0.01) 0.13 (0.73)1.00 (0.00)1.00 (0.00)0.93
(0.00) 0.65 (0.05) 1.00 (0.00) 1.00 (0.00) 32. Panel B: Domestic
Fixed IncomeUnconditionalConditional Decile1981199019912004
19811990 199120041-0.53 (-1.77)0.27 (0.98)-0.08 (-0.31) -0.33
(-1.62) 2-0.06 (-0.35)0.14 (1.38) 0.18 (1.14)-0.13 (-1.34) 3-0.01
(-0.11)0.09 (1.83) 0.16 (1.31)-0.01 (-0.25) 4 0.10 (1.26) 0.06
(1.43) 0.22 (2.38) 0.01 (0.29) 5 0.11 (1.30) 0.04 (1.09) 0.10
(1.16) 0.03 (0.79) 6 0.32 (3.18) 0.05 (0.94) 0.24 (2.58) 0.13
(2.35) 7 0.17 (1.43) 0.03 (0.56) 0.11 (0.93) 0.17 (3.33) 8 0.37
(2.12) 0.09 (1.42) 0.24 (1.37) 0.31 (5.77) 9 0.51 (2.29) 0.23
(2.28) 0.13 (0.69) 0.52 (5.52) 100.70 (2.23) 1.03 (3.76) 0.38
(1.62) 1.53 (4.68)SpCorr0.99 (0.00) 0.02 (0.97)0.47 (0.18) 1.00
(0.00) Panel C: International Equity (19912004 only) Decile
Unconditional Conditional1-0.23 (-0.24) 0.02 (0.01) 2-0.32 (-0.43)
0.19 (0.20) 3-0.12 (-0.19) 0.23 (0.31) 4 0.46 (0.94)0.68 (1.28) 5
0.55 (1.26)0.77 (1.56) 6 0.98 (1.87)0.96 (1.84) 7 1.37 (2.40)1.10
(1.90) 8 1.60 (2.47)1.08 (1.68) 9 2.37 (3.29)2.05 (2.73) 102.67
(2.55)2.04 (1.70)SpCorr 0.99 (0.00) 0.98 (0.00)31 33. Table 3:
Post-Ranking One-quarter Alphas with Deciles formed using
Benchmark-Adjusted Returns This table lists the post-ranking alphas
for deciles of funds sorted according to the benchmark-adjusted
return during the ranking period of one year. The portfolio deciles
are rebalanced at the end of every quarter and are held for one
post-ranking quarter. Unconditional alphas are calculated from the
factor modelKrp,t = U +p p,k fk,t + p,t ,k=1while the conditional
alphas are calculated from the factor model KL rp,t = C + p0 p,k
+lp,k Zl,t1 fk,t +p,t , k=1l=1where f are K factors and Zs are L
instruments. The list of instruments includes T-bill, dividend
price ratio, term spread and default spread. The factors for
domestic equity are the three Fama and French (1993) factors and
the Carhart (1997) momentum factor. The factors for domestic xed
income are the Lehman Brothers Aggregate Bond Index returns, Term
Spread Return, and Default Spread Return. We report alphas only for
survivorship bias free sample period of 1991 to 2004. All alphas
are in percent per quarter and t-statistics are reported in
parentheses next to alphas. Decile 1 contains the worst performing
portfolio and decile 10 contains the best performing portfolios.
Domestic Equity Domestic Fixed IncomeUnconditional Conditional
Unconditional ConditionalDecile FFCarhartFFCarhart FF FF 1-0.08
(-0.21)0.62 (1.63) -0.05 (-0.17) 0.34 (1.00) 0.35 (2.20) 0.19
(1.18)2-0.08 (-0.25)0.43 (1.40)0.01 (0.05)0.50 (2.19) 0.19 (2.77)
0.08 (1.01)3-0.03 (-0.11)0.25 (0.95)0.05 (0.25)0.33 (1.77) 0.11
(2.66) 0.06 (1.24)4 0.05 (0.24) 0.25 (1.13)0.16 (1.11)0.42 (2.94)
0.11 (3.21) 0.10 (2.33)5 0.08 (0.45) 0.13 (0.69)0.16 (1.27)0.29
(2.19) 0.09 (2.24) 0.11 (3.00)6 0.09 (0.60) 0.11 (0.69)0.15
(1.35)0.26 (1.95) 0.10 (2.30) 0.16 (3.90)7 0.28 (1.82) 0.10
(0.62)0.32 (2.33)0.21 (1.29) 0.11 (2.64) 0.16 (4.55)8 0.30 (1.65)
0.12 (0.61)0.35 (2.18)0.24 (1.31) 0.14 (2.64) 0.26 (5.30)9 0.61
(2.26)-0.05 (-0.28) 0.76 (3.23)0.23 (1.21) 0.15 (2.13) 0.31
(4.81)102.03 (3.56) 0.51 (1.42)2.39 (4.47)1.07 (2.94) 0.70 (2.58)
0.80 (4.31) SpCorr1.00 (0.00)-0.52 (0.13) 0.95 (0.00) -0.36
(0.31)0.03 (0.95) 0.73 (0.02)32 34. Table 4: Post-Ranking One- to
Three-Year Alphas This table lists the post-ranking alphas for
deciles of funds sorted according to the benchmark-adjusted return
(for domestic equity in Panel A and domestic xed income in Panel B)
or raw return (for international equity in Panel C) during the
ranking period of one year. The portfolio deciles are rebalanced at
the end of every year and are held for one to three post-ranking
years. Unconditional alphas are calculated from the factor model K
rp,t = U + p p,k fk,t + p,t , k=1while the conditional alphas are
calculated from the factor modelKLrp,t = C +p 0p,k +l p,k Zl,t1
fk,t + p,t ,k=1l=1where f are K factors and Zs are L instruments.
The list of instruments includes T-bill, dividend price ratio, term
spread and default spread. The factors for domestic equity are the
three Fama and French (1993) factors and the Carhart (1997)
momentum factor. The factors for domestic xed income are the Lehman
Brothers Aggregate Bond Index returns, Term Spread Return, and
Default Spread Return. The factors for international equity are the
international versions of Fama and French (1993) factors. We report
unconditional alphas for domestic equity and international equity
and conditional alphas for domestic xed income. The sample period
is 1991 to 2004. All alphas are in percent per quarter and
t-statistics are reported in parentheses next to alphas. Decile 1
contains the worst performing portfolio and decile 10 contains the
best performing portfolios. Panel A: Domestic Equity (Unconditional
alphas)First-Year Second-Year Third-Year Decile
FFCarhartFFCarhartFFCarhart10.05 (0.15) 0.40 (1.30)1.04 (3.54)0.80
(2.55) 0.73 (2.38)0.31 (1.00) 20.16 (0.56) 0.24 (0.77)0.56
(2.82)0.39 (1.83) 0.40 (2.12)0.20 (1.01) 30.15 (0.64) 0.21
(0.80)0.38 (2.08)0.27 (1.37) 0.14 (0.98)0.03 (0.20) 40.06 (0.31)
0.19 (0.86)0.46 (2.37)0.47 (2.15) 0.13 (0.82)0.18 (1.02) 50.12
(0.77) 0.17 (0.98)0.06 (0.38)0.11 (0.57) 0.17 (0.96)0.23 (1.16)
60.12 (0.77) 0.15 (0.83)0.08 (0.54)0.15 (0.85) 0.11 (0.67)0.18
(1.01) 70.16 (0.95) 0.20 (1.07)0.09 (0.53)0.21 (1.19) 0.18
(1.24)0.18 (1.08) 80.32 (1.74) 0.16 (0.82)0.17 (0.98)0.32 (1.70)
0.09 (0.48)0.12 (0.54) 90.56 (2.51) 0.21 (0.98)0.05 (0.17)0.13
(0.43) 0.32 (1.13)0.22 (0.70) 10 1.44 (3.34) 0.47 (1.36)0.09 (0.23)
-0.22 (-0.53)0.23 (0.58) -0.04 (-0.09)SpCorr 0.64 (0.05) 0.49
(0.15)1.00 (0.00) -0.52 (0.13) 0.81 (0.01) -0.18 (0.63) 33 35.
Panel B: Domestic Fixed Income (Conditional alphas)
DecileFirst-YearSecond-YearThird-Year1 0.45 (3.25)0.37 (3.19) 0.18
(1.44) 2 0.23 (3.14)0.16 (2.55) 0.08 (1.07) 3 0.13 (2.91)0.10
(2.32) 0.10 (1.66) 4 0.14 (3.37)0.10 (2.59) 0.07 (1.61) 5 0.11
(3.54)0.10 (2.61) 0.07 (2.81) 6 0.15 (3.81)0.05 (1.60) 0.09 (2.64)
7 0.10 (3.24)0.09 (2.57) 0.07 (1.86) 8 0.11 (2.01)0.14 (2.40) 0.09
(1.70) 9 0.17 (1.93)0.23 (2.88) 0.11 (1.46) 100.62 (2.80)0.60
(2.91) 0.51 (2.51)SpCorr0.85 (0.00) -0.14 (0.71) 0.84 (0.00)Panel
C: International Equity (Unconditional alphas)
DecileFirst-YearSecond-YearThird-Year1 0.51 (0.48) 1.63 (1.90)-0.03
(-0.03) 2 0.22 (0.27) 0.67 (0.97) 0.39 (0.50) 3 0.04 (0.07) 0.47
(0.84) 0.75 (1.42) 4 0.39 (0.89) 0.31 (0.66) 0.72 (1.46) 5 0.22
(0.51) 0.40 (0.98) 0.77 (2.01) 6 0.96 (1.93) 0.51 (1.07) 0.70
(1.65) 7 0.86 (1.61) 0.52 (0.95) 0.67 (1.45) 8 1.41 (2.30) 1.10
(1.64) 0.96 (1.95) 9 1.98 (2.33) 0.93 (1.10) 0.87 (1.41) 102.76
(2.19) 0.97 (0.95) 0.48 (0.57)SpCorr0.08 (0.84) 0.98 (0.00)0.15
(0.68) 34 36. Table 5: Post-Ranking One- to Three-Year
Benchmark-Adjusted Returns This table shows descriptives on the
post-ranking benchmark-adjusted returns for deciles of funds sorted
according to the benchmark-adjusted return for domestic equity in
Panel A and domestic xed income in Panel B during the ranking
period of one year. The portfolio deciles are rebalanced at the end
of every year and are held for one to three post-ranking years. The
sample period of 1991 to 2004. Mean returns are listed under the
columns Mean while Information Ratios are listed under the column
IR. Means signicant at the 99% level are indicated by three stars,
signicant at the 95% level are indicated by two stars, and
signicant at the 90% level are indicated by one star. Decile 1
contains the worst performing portfolio and decile 10 contains the
best performing portfolios. Panel A: Domestic EquityFirst-Year
Second-YearThird-Year DecileMeanIRMean
IRMeanIR10.120.040.960.230.800.28 20.200.090.63 0.250.50 0.24
30.200.140.430.210.34 0.20 40.30 0.240.43 0.280.380.25
50.340.350.38 0.260.49 0.32 60.47 0.430.320.300.40 0.33 70.52
0.450.330.360.35 0.30 80.65 0.550.46 0.590.270.30 90.86 0.540.49
0.440.440.28 10 1.55 0.400.71 0.410.700.26Panel B: Domestic Fixed
IncomeFirst-YearSecond-Year Third-YearDecile MeanIR MeanIRMeanIR
10.30 0.23-0.18-0.110.030.0320.110.18-0.09-0.12 -0.03
-0.0330.100.29-0.03-0.100.080.3140.08 0.42 0.03 0.130.050.3550.05
0.50 0.050.450.05 0.4160.060.29 0.06 0.320.04 0.2770.050.20 0.09
0.350.070.3780.020.04 0.09 0.290.060.199 -0.04 -0.05
0.220.430.050.0810 0.110.08 0.500.540.080.07 35 37. Table 6:
Post-Ranking One- to Three-Year Fund Flows This table lists the
post-ranking fund ows for deciles of funds sorted according to the
benchmark- adjusted return (for domestic equity in Panel A and
domestic xed income in Panel B) or raw return (for international
equity in Panel C) during the ranking period of one year. The
portfolio deciles are rebalanced at the end of every year and are
held for one to three post-ranking years. Decile 1 contains the
worst performing portfolio and decile 10 contains the best
performing portfolios. The sample period is 1991 to 2004. Flows are
in percent per year.Domestic Equity Domestic Fixed
IncomeInternational EquityDecileY1 Y2 Y3 Y1Y2Y3Y1Y2 Y3 1
-4.282.100.961.315.943.51 5.494.588.822 -2.790.451.682.914.796.09
7.125.747.1532.542.884.016.473.934.73
8.276.267.1646.053.513.395.743.103.4312.24
10.096.1058.685.717.495.735.336.4515.26 13.05
10.9169.329.418.378.259.345.4119.38 18.22 16.237 14.77
12.309.678.436.717.6019.39 18.209.578 19.52 15.70 12.19
10.587.807.4619.25 19.38 16.799 23.49 18.34 12.46 10.09
10.428.6323.65 22.68 21.041031.80 23.01 17.36 17.25 13.42
12.2829.40 23.60 19.03 36 38. Table 7: Descriptives on Fees
Incentives This table lists the descriptives on fees for deciles of
funds sorted according to the benchmark-adjusted return (for
domestic equity and domestic xed income) or raw return (for
international equity) during the ranking period of one year. The
portfolio deciles are rebalanced at the end of every year and are
held for one year. Decile 1 contains the worst performing portfolio
and decile 10 contains the best performing portfolios. Panel A
lists the fraction of funds that either charge or do not charge
performance based fees (the remainder is fraction of funds for
which information is not available). Panel B lists the fraction of
funds that either do or do not have most-favored nation clause in
fee schedules (the remainder is fraction of funds for which
information is not available). The sample period is 1991 to
2004.Panel A: Performance based feesDomestic DomesticInternational
EquityFixed Income EquityDecileYes No Yes No YesNo 1 47.3
41.237.741.964.6 20.22 46.6 41.938.046.462.2 22.73 49.5
39.939.344.365.5 24.74 50.1 38.735.948.360.9 26.15 51.5
37.239.246.061.3 24.96 53.8 34.541.444.266.2 23.07 53.8
35.742.544.559.2 28.48 52.8 37.944.842.161.8 28.39 53.9
36.645.939.964.2 23.11054.9 36.446.335.865.8 20.8Panel B:
Most-favored nation clauseDomestic DomesticInternational
EquityFixed Income EquityDecileYes No Yes No YesNo 1 32.8
32.029.528.842.8 21.02 32.7 32.033.032.042.6 21.13 31.4
33.437.129.745.0 21.04 32.1 32.835.629.643.5 20.45 32.2
31.334.729.349.7 16.96 32.3 32.141.924.151.9 16.97 36.5
29.239.428.948.4 17.08 34.4 32.042.525.048.6 23.39 38.2
30.037.426.248.6 18.51037.8 30.533.122.344.4 22.1 37 39. Table 8:
Fees This table lists the fees for deciles of funds sorted
according to the benchmark-adjusted return (for domestic equity in
Panel A and domestic xed income in Panel B) or raw return (for
international equity in Panel C) during the ranking period of one
year. The portfolio deciles are rebalanced at the end of eve