mutual fund buys and sells of individual and institutional - fep
TRANSCRIPT
WHICH MONEY IS SMART?
MUTUAL FUND BUYS AND SELLS
OF INDIVIDUAL AND INSTITUTIONAL INVESTORS
ABSTRACT Gruber (1996) and Zheng (1999) report that investors channel money towards mutual funds that subsequently perform well (money is “smart”). However, Sapp and Tiwari (2004) show that this result no longer holds after controlling for the momentum factor in stock returns. In this paper, we examine the smart money effect with a unique British dataset. Previous literature has worked with approximate quarterly net money flows for US mutual funds. For our UK mutual funds, we observe not only the exact net flows each month, but also inflows and outflows, and how much of each comes from individual and from institutional investors. We find that the smart money effect is alive and well in the UK. It is driven by buying (but not selling) decisions of both individuals and institutions. Keywords: mutual funds, investor behavior, smart money JEL codes: G11; G23
1
Can investors identify superior mutual funds? The first studies to address this question (Gruber
(1996), Zheng (1999)) answer in the affirmative. Funds which receive greater net money flows
subsequently outperform their less popular peers. This pattern was termed the “smart money”
effect. More recent research questions this finding, however. After fund performance is adjusted
for the momentum factor in stock returns, greater net flows no longer lead to better performance
(Sapp and Tiwari (2004)).
In this paper, we re-examine the smart money issue with UK data. All of the above studies, due
to data constraints, work with aggregate money flows to funds: all investors are aggregated, and
sales are offset by repurchases. Further, not having access to exact net flows, these papers
approximate them using fund net asset values (NAV) and fund returns. Our data allow us to
conduct a stronger test for the smart money effect by using monthly data on exact fund flows, and
to gain greater insight into investors’ decisions by considering separately the sales and purchases
of individual and institutional investors.
The smart money hypothesis states that investor money is “smart” enough to flow to the funds
that will outperform in the next period, i.e. that investors have genuine fund selection ability.
Research into smart money in the mutual fund context was initiated by Gruber (1996). His aim
was to understand the continued expansion of the actively managed mutual fund sector despite
the widespread evidence that on average active fund managers did not add value. To test whether
investors are more sophisticated than simply being chasers of past performance, he examined
whether investors’ money tended to flow to the funds which subsequently outperform. Working
with a subset of US equity funds, he found evidence that the weighted average performance of
funds that receive net inflows was positive on a risk-adjusted basis. Thus money appeared smart.
2
Zheng (1999) further developed the analyses of Gruber (1996), expanding the data set to cover
the universe of all equity funds between 1970 and 1993. She found that funds which enjoyed
positive net flows subsequently performed better on a risk-adjusted basis than funds which had
experienced negative net flows. She also examined whether a trading strategy could be devised
based on the predictive ability of net flows and found evidence that information on net flows into
small funds could be used to make risk-adjusted profits.
However, more recent research of Sapp and Tiwari (2004) has argued that the smart money effect
documented in prior studies is an artefact of these studies’ failure to account for the momentum
factor in stock returns. Their argument can be synthesized as follows. Stocks that performed
well tend to continue doing well (Jegadeesh and Titman (1993)). Investors tend to put their
money into ex-post best performing funds. These funds necessarily have disproportionate
holdings of ex-post best performing stocks. Thus, after buying into winning funds, investors
unwittingly benefit from momentum returns on winning stocks. To test this reasoning, Sapp and
Tiwari (2004) calculate abnormal performance following money flows with and without
accounting for the momentum factor, and find that inclusion of the momentum factor in the
performance evaluation procedure eliminates outperformance of high-flow funds. They
additionally show that investors are not deliberate in seeking to benefit from stock-level
momentum: more popular funds do not have higher exposure to the momentum factor at the time
they are selected. Wermers (2003) further contributes by examining fund portfolio holdings to
establish that fund managers who have recently done well try to perpetuate this performance by
investing a large proportion of the new money they receive in stocks that have recently done well.
3
All of the research work we have surveyed has been conducted with US data. This fact is not
surprising, given that the US mutual fund marketplace is by far the largest in the world (Khorana,
Servaes and Tufano (2005)). There are however, a number of advantages to examining the smart
money effect in fund management using our UK mutual fund data. First, our money flow data
are monthly rather than quarterly. Second, we observe exact flows rather than approximations
based on fund values and fund returns. Third, we can distinguish institutional and individual
money flows. Fourth, we can distinguish between purchases and sales.
A further advantage is that we get to examine mutual fund investor behavior in a different
institutional setting from that of the US. For example, unlike US mutual funds, UK funds
compete within well-defined peer groups, which may facilitate investors’ decision-making. Also,
the tax overhang issue (Barclay, Pearson and Weisbach (1998)) does not apply to UK mutual
funds, which means that investors’ decisions are not complicated by the dependence of their
future tax liability on the interaction of fund flows and fund performance.
We find that portfolios where funds are weighted by their net money flows outperform portfolios
where funds are weighted by NAV: new money beats old money. We also find that high net flow
funds outperform low net flow funds. Thus, within the universe of actively managed funds, new
investors tend to choose the better ones: money is smart. This result holds for both individual
and institutional investors, and is driven by investors’ fund buys rather than sells. The smart
money effect is not explained by the Chen et al. (2004) fund size effect or by performance
persistence, nor is it concentrated in smaller funds. Although the effect is statistically significant,
its economic significance is modest. A researcher without access to actual monthly flows would
not be able to detect investors’ fund picking ability.
4
The remainder of this paper is organized as follows. Section I describes our mutual fund data in
the context of the UK institutional environment. Section II reports on the determinants of the
different components of money flows to funds. Section III examines whether funds favored by
investors generate better performance than those not favored, and establishes the smart money
effect in the UK. Section IV investigates the pervasiveness of the effect and the possible reasons
for it. Section V discusses our findings and their implications. Section VI concludes.
I. Data and Institutional Background
A. The UK Mutual Fund Industry
The first open-ended mutual funds (called “unit trusts”, because formally investors buy units in a
fund) appeared in the United Kingdom in the 1930s, or about a decade later than in the United
States.1 At the end of 2000 (which coincides with the end of our sample period), 155 fund
families ran 1,937 mutual funds managing £261 billion (or $390 billion) in assets,2 making the
UK mutual fund industry one of the largest outside the US (Khorana et al. (2005)). While the US
and UK mutual fund environments are quite similar in many respects, we note two institutional
differences, both of which likely make investor fund choice more complicated in the US than in
the UK.
1 The late 1990s saw the introduction of a new legal structure for UK’s open-ended mutual funds, called open-ended investment company, or OEIC. For our purposes, however, differences between unit trusts and OEICs are unimportant and we refer to both types of funds as mutual funds. 2 From http://www.investmentuk.org/press/2002/stats/stats0102.asp
5
First, in the US there is no single, official classification system for fund objectives. This allows
funds to mislead investors about their objectives (Cooper, Gulen and Rau (2004)), suggesting that
ambiguous classification complicates investors’ fund picking. By contrast, in the UK the
Investment Management Association (IMA) classifies funds into sectors on the basis of the
funds’ asset allocation, and the official IMA classification system is used by the funds
themselves, by information providers, and by brokers.3 This reduces the potential for confusion
on the part of any investors whose fund choosing process requires breaking down the fund
universe into groups of comparable funds.
The second difference has to do with the tax treatment of capital gains. In the UK, the system is
simple: investors only pay capital gains tax when they sell their shares in a fund. In the US,
however, investors face an additional form of capital gains tax. US mutual funds must distribute
net capital gains realized by the fund, and when they do so, their investors are liable for tax on
these distributions. While existing investors prefer their fund managers to defer realization of
capital gains, the resulting tax overhang is likely to deter new investors (Barclay et al. (1998)).
UK investors therefore face a simpler asset allocation problem than their US counterparts, as they
need not be concerned with how any pre-existing fund-level tax liability may affect their own
after-tax returns.
3 The IMA enforces its sector definitions, and if the asset allocation of a fund contravenes the allocation rules of its current sector, the IMA will warn the fund to change its allocation if it does not wish to change sectors. If the fund does not comply, the IMA will move the fund to a new sector reflecting its new asset allocation. The sectors are well-defined and relatively stable over time (although the IMA occasionally revises its sector definitions to reflect the industry’s and investors’ needs). For example, throughout much of the 1990s, UK equity funds were subdivided into Income, Growth and Income, Growth, and Smaller Companies categories. Such diverse information providers as Standard & Poor’s, Hemscott, Money Management and Allenbridge all use the official classification system. By contrast, in the US there is a proliferation of methods for assigning funds to a peer group (e.g. Morningstar, Wiesenberger, Strategic Insight and ICDI each have their own classification).
6
B. The Population of Funds
[Table I here]
Unlike in the US, there unfortunately does not exist a survivorship bias free electronic database
of UK mutual funds. Therefore, to round up the population of funds over the period we study, we
manually collected and linked across years data from consecutive editions of the annual Unit
Trust Year Book corresponding to year-end 1991 through year-end 1999. This dataset
additionally contains fund fees, management style (active or passive), and the fund sector
assignment. Like earlier literature on the smart money effect, we focus on funds investing in
domestic equities. Unlike the earlier papers, which all examine US funds, we can select these
funds unambiguously by retaining only funds whose official sector definitions correspond to a
UK equity mandate. Panel A of Table I shows the evolution of this group of funds. The number
of domestic equity funds grows from 425 at the start of 1992 to 496 at the start of 2000
(averaging 461 per year), while assets under management increase almost four-fold over the same
period to £115 billion. Since our interest lies in whether investors can identify superior funds, we
then drop passively managed (“index tracker”) funds. This leaves us with 432 eligible funds per
year on average.
C. Data on Funds’ Money Flows
Our money flow data come from the IMA and give monthly mutual fund flows over the 1992-
2000 period. Thus, unlike other studies of mutual fund investor behavior, which back out net
flows from data on fund values and fund returns, we observe the exact amount of money injected
7
by investors into each mutual fund. Further, in our dataset these net flows are disaggregated into
their component parts: sales to individual investors, sales to institutional investors, repurchases
from individual investors, and repurchases from institutional investors.
The IMA obtains money flow information directly from its member companies every month.4
Not all management groups report this information; however, since information is collected live
and historical information is not discarded, there is no survivorship bias in the data. We
manually link these money flow data to the dataset constructed from consecutive editions of the
Unit Trust Year Book to obtain our final mutual fund sample. Panel B of Table I shows that our
sample averages 313 funds per year with annual attrition rate of 6.3 percent. Whether on the
basis of assets under management or the number of funds, our sample covers roughly three-
quarters of the population of eligible funds which we identified earlier.
The remainder of Panel B reports total money flows as well as their components parts. The net
aggregate money flow is positive in every year except 2000, and averages £1,981 million
annually. As it turns out, this amount masks an annual inflow of £6,855 million and an outflow of
£4,873 million. This indicates that research based on approximations of net money flows
observes (with noise) only a fraction of investors’ capital moving through mutual funds.
As mentioned earlier, fund management companies report to the IMA not only the total sales and
repurchases for each fund, but also whether these flows took place through retail channels and
thus originated from individual clients, or whether they came from the fund’s institutional clients.
4 The IMA started collecting these data in January 1992. The data available to us stop in 2000 for confidentiality reasons.
8
Over the full sample period, net flows from institutions are £329 million per year, as compared
with £1,652 million from individuals. Even on a year-by-year basis, it is clear that individual and
institutional investors do not behave alike. For example, the year 1999 had the highest net flow
of any year from individuals, but one of the lower annual net flows from institutions.
The remainder of Table I presents a further disaggregation of annual money flows by direction
and by client type. Once again it can be seen that major capital movements are masked by the
netting of sales and repurchases: for example, in 1999 the mere £19 million net flow from
institutions is the result of them buying £3,319 million worth of fund units and selling £3,300
million worth of units.
Before we can start working with our flow data at the fund-month level, we address several data
issues. First, we eliminate fund months without any recorded money flow. This leaves 32,803
fund-months. Next, we set to missing retail (institutional) flows for fund-months without any
retail (institutional) client sales or repurchases. This is because the fund universe we study
includes funds that are open only to retail (institutional) investors, as well as funds that are open
to both investor types. There are 15,527 fund-months with both retail and institutional activity,
15,508 fund-months with retail activity only, and 1,768 fund-months with institutional activity
only. The next step is to “clean” our data, so that highly unusual flows do not drive our results.
In particular, unusual flow activity can take place for very young funds or for funds about to be
closed down. Rather than to set a common relative flow cutoff for all funds, we use a filtering
procedure which takes into account a fund’s flow volatility. First, we drop funds with fewer than
10 months of flow data. Next, we calculate normalized net flows, i.e. we divide the net monetary
9
flow into a fund in a given month by the fund’s size at the start of the month.5 Then we drop
fund-months with relative net flows which are more than five standard deviations away from the
fund’s average.6 We iterate the last two steps until no more fund-months are dropped. This
leaves us with 30,846 fund-months, of which 29,207 experience retail activity, 16,155 experience
institutional activity, and 14,516 experience both institutional and retail activity. Table II reports
on the distribution of net flows and their components for these fund-months.
[Table II here]
In Panel A of Table II we show moments of the distribution of relative flows, averaged across the
108 monthly cross-sections. The first row describes the flow estimate which is implied by fund
net asset values (NAV) and return data alone. This is the variable used in the existing smart
money literature and is calculated as t
ttt
NAVrNAVNAV )1(1 +− − (fund subscripts are suppressed).7 It
is instructive to compare its distribution to that of the actual net money flow. While the mean net
flow is 0.67 percent of fund value, corresponding to roughly 8 percent growth per year, the
growth rate estimate based on implied flows averages 0.49 per month or about 6 percent
annually. The noise in implied flows is also clear from observing that they vary more than actual
net flows: the standard deviation of implied flows is more than ten percent greater than that of
actual flows, and the inter-quartile range for implied flows is over forty percent wider than the
5 Ideally, institutional (retail) flows would be scaled by the amount of institutional (retail) holdings of each fund. Unfortunately, these data are unavailable. 6 Both the average and the standard deviation are estimated excluding the fund-month under consideration. In other words, we regress the net aggregate relative flows for each fund on unity, and drop fund-months for which the value of the externally studentized residual exceeds five in magnitude. 7 The literature additionally applies an adjustment for NAV increase due to fund mergers. To avoid problems due to the quality of our data about fund mergers, we do not include fund months in which mergers take place.
10
one for actual net flows. More evidence on the quality of the implied flow estimate is in Panel B
of Table II, which shows correlations between our flow variables. The table shows that the
average correlation between implied and actual net flows equals 0.853. In an important sense,
this number overestimates how good an approximation the implied measure is. The reason is that
this correlation is heavily influenced by fund-months whose net flows are very large relative to
fund size – whereas smaller money movements are more easily obscured by fund returns within
the month. To illustrate this point, we recalculate the implied/actual correlation after dropping
observations where actual net flows exceed one percent of fund size. This removes just over a
third of our fund-month observations, leaving 19,688 fund-months, yet the implied/actual
correlation drops to 0.297. The practical implication of this effect is that when portfolios are
formed on the basis of implied flows, many funds will be assigned to the wrong portfolios. For
example, in our sample of 30,846 fund-months, implied flows have the wrong sign for 5,335
fund-months, or 17.2 percent of the time.8
The remainder of Panels A and B shows time series averages of moments and correlations for
components of the net aggregate money flow. First and most important, note the low average
correlation between institutional and retail flows. For net flows, the correlation equals 0.247; for
inflows, 0.271; and for outflows, 0.141. This leaves much scope for the possibility – which the
remainder our paper explores in detail – that the behavior of aggregate net flows studied in the
8 Berk and Xu (2004) suggest that ( )( )tt
ttt
rNAVrNAVNAV
++− −
111 is a conceptually superior way to estimate flows. In our
full sample, the distributions of the two measures of implied flows are virtually identical, and the correlation between them is 0.9996. This is also the case if we drop observations with relative net actual flows in excess of one percent in absolute value; and since the Berk-Xu adjustment consists of division by the total return, the sign of the implied flow is unaffected by using their formula. We therefore use the traditional formula for implied flows throughout the paper.
11
existing smart money literature could belie very different behaviors by investors, depending on
whether they are buying into a fund or taking money out, and depending on who the investors
are.
The correlations between inflows and corresponding outflows are also telling. In aggregate (for
both individual and institutional investors), the correlation averages 0.122, and is similar for
individual investors (0.143) and institutional investors (0.112). The fact that these correlations
are positive, albeit small in magnitude, indicates that funds with low sales are not necessarily the
funds with high withdrawals – and vice versa. We will briefly examine the determinants of the
different money flow components in Section II.
D. Performance Measurement
Our fund return data are survivorship-bias free and come from Quigley and Sinquefield (2000),
who collected monthly returns for domestic equity funds over the 1975-1997 period, and
subsequently extended this dataset to the end of 2001. Returns are net of tax (because gross of
tax returns are unavailable for most dead funds) but gross of management fees.
As the debate over the smart money effect in the US shows, proper performance measurement is
paramount. Like Sapp and Tiwari (2004), we measure fund performance using the Carhart
(1997) four-factor model, which we adapt to the UK setting. Namely, we estimate the regression
model ittWMLtHMLtSMBtMKTitit eWMLHMLSMBMKTRFR +++++=− ββββα , where Rit is the
rate of return on investment i in month t, RFt is the risk-free interest-rate in month t, MKTt is
12
the return on the market portfolio in excess of the risk-free rate, and SMBt , HMLt and WMLt are
returns on the size, value, and momentum factor mimicking portfolios, respectively. Our
monthly Fama and French (1992, 1993) size and value factor realizations come from Dimson,
Nagel, and Quigley (2003), who confirm the size and value effects in the UK. Our monthly
momentum factor is constructed following Carhart (1997).9 Specifically, each month we rank all
British firms listed on the London Stock Exchange on their 11-month returns lagged by one
month, and calculate the difference between the average returns of the highest and the lowest 30
percent of firms.10
II. Determinants of Money Flows
In order to understand better how different types of investors make their fund buying and selling
decisions, we present evidence on the determinants of money flows. For the sake of parsimony,
we have included only two explanatory variables which past work has shown to be strong
predictors of net mutual fund flows: past flows and past performance.
[Table III here]
The past flow measure we use for each flow component is the value of that flow component 12
months earlier. This is a simple way to account for seasonalities in investors’ decisions (which
may be due, for example, to regularly scheduled fund purchases). Since using lagged flows costs
9 For evidence on the pervasiveness of the momentum effect internationally, including in the UK, see Rouwenhorst (1998). 10 The only deviation from Carhart’s method is that our averages are value-weighted, to avoid spurious results due to “micro-cap” companies. Monthly returns and market capitalizations are taken from London Business School’s London Share Price Database.
13
us a year of data, there are 96 monthly regressions, corresponding to the period from January
1993 through December 2000. Our results, based on the time series of cross-sectional regression
coefficient estimates (the Fama-Macbeth approach) are shown in Table III. Past performance is
measured either as the Jensen (1968) alpha (Panel A) or as the Carhart (1997) four-factor alpha
(Panel B), averaged over the 12 months preceding the money flow. The reported coefficients are
averages of the monthly coefficient estimates, and p-values are based on the time-series standard
deviations of these estimates.
Panel A indicates that our flow variables are persistent: coefficient estimates for lagged flows are
always positive and significant. The much higher coefficient estimate for actual net aggregate
flow than for implied flow (0.148 vs. 0.066) is clearly due to the noise inherent in estimating the
implied flow. The patterns of coefficient estimates further tell us that retail flows are more
persistent than institutional flows, and that inflows are more persistent than outflows.
There exists overwhelming evidence in US-based work that investors “chase” high returns
(Chevalier and Ellison (1997), Sirri and Tufano (1998), Del Guercio and Tkac (2002)). Our data
show that UK investors do likewise. The coefficient of 1.127 for net aggregate flows suggests
that on the whole, a one percent increase in monthly alpha results in an additional inflow of more
than one percent of fund value. Since the levels of the relative flow variables that we examine
are different, estimates of their sensitivity to past returns are not directly comparable.
Nonetheless, it is clear that inflows increase with past performance while outflows tend to do the
opposite; and that the reaction of inflows to past performance is markedly more pronounced than
that of outflows both for individuals and for institutions. Indeed, the effect of past performance
on sales by institutions is not even significant at conventional confidence levels. The asymmetry
14
in investor reaction to good and bad performance is well known (Sirri and Tufano (1998));
however, previous researchers have not been able to observe this reaction for in- and outflows
directly.
Measuring past performance with the four-factor alpha, as we do in Panel B, does not alter the
above conclusions. While an in-depth study of what factors influence investor decisions is
outside the scope of this paper, comparing the patterns of coefficients on past performance with
those in Panel A is instructive. It appears that investors, both institutional and retail, tend to put
more weight on fund performance calculated after taking size, value and momentum effects into
account, rather than on simple market-adjusted performance. Whether any sophistication in
interpreting past performance translates into actual fund selection ability is the focus of the
remainder of this paper.
III. Performance of Money Flow Based Portfolios
A. Money-Weighted Portfolios
So, do investors benefit from their fund selection process? A simple way to address this question
is to evaluate the performance of all “new money” put into mutual funds by investors. A natural
benchmark against which to measure the success of these new investments is the performance of
“old money”, i.e. of assets already in place before the latest round of investments.
Our data allow us to define what constitutes new money in several ways. First, we can measure it
using the implied net money flow, as would a researcher with access to fund size and return data
15
only. In addition, we can use actual net aggregate flows from our dataset, as well as each of their
components. A hypothetical portfolio of new money is then constituted from all eligible funds
weighted in proportion to their value of the flow measure in the preceding month. Note that as a
result of this portfolio formation scheme, when performance is evaluated on the net money flow
basis, funds with negative net flows are assumed sold short in our hypothetical portfolio (which is
generally a practical impossibility for mutual funds). When portfolios are formed on the basis of
either sale or repurchase activity, there are of course no negative weights. Performance
evaluation of our new money-weighted portfolios gives us the performance of the average pound
(dis)invested in UK mutual funds in the past month. Similarly, we can form a portfolio of funds
on the basis of the funds’ net asset values excluding the money put in during the last month.11
Comparing performance of the new and old money weighted portfolios tells us whether recent
investing decisions outperform the mutual fund industry as a whole. This is the analysis whose
results we report in Tables IV and V.
[Table IV here]
In Table IV, we characterize our fund portfolios using what Zheng (1999) calls the fund-level
approach. Specifically, each month we conduct a Carhart (1997) four-factor regression for every
fund, using the preceding 36 monthly returns to obtain our four estimated factor loadings.12 We
then subtract from that month’s fund return the product of each factor realization and its
11 We approximate this quantity by cumulating a fund’s NAV at the start of the previous month at the fund’s rate of investment return for that month. 12 We require a minimum of 30 monthly returns to estimate the regression. Thus the fund-level results have a slight survivor bias, since extremely short-lived funds are excluded.
16
estimated loading to obtain that month’s alpha for each fund.13 These alphas and the fund-level
regression estimates are used to compute the money-weighted average across funds for each
month. The panel reports the time-series average of the monthly averages. In the last two
columns, it also reports the difference between the money-weighted alpha obtained in this way,
and the similarly obtained fund value-weighted alpha. P-values are computed from the time-
series of the monthly averages.
Before discussing the performance of our new-money weighted portfolios, we first turn to the
value-weighted portfolio in row 11 of the table, where all actively managed domestic equity
funds are represented in proportion to their NAV. This corresponds to the performance of “old”
money (specifically, of assets in place excluding the previous month’s round of investments).14
This portfolio’s four-factor alpha averages -14.5 basis points per month over the full 1992-2000
period. We will use this portfolio as a benchmark with respect to which we measure the
performance of new money. We have additionally evaluated an equally weighted portfolio of
actively managed domestic equity funds, whose four-factor alpha can be seen to average -12.0
basis points per month (the last two columns of the table show this alpha to be insignificantly
different from the value-weighted portfolio’s alpha). These numbers are lower than the
corresponding numbers for the US; for example, Sapp and Tiwari (2004) report the average alpha
to be -1.5 basis points. There are several reasons for this discrepancy. First, our returns are net
of tax. Second, the transaction costs on the London Stock Exchange have historically exceeded
13 This alpha corresponds to the realized return on a strategy of using the estimated factor loadings to determine the size of hedge positions in each factor mimicking portfolio at the start of each month. An alternative alpha estimation method would be to include the current month in the four-factor regression and save the current month’s residual as the alpha. The alpha calculated in this way, although very similar to ours, cannot be captured by an investment strategy because the factor loadings depend on the current month’s performance. 14 To reflect this interpretation, the exact weight we use is the start-of-month NAV cumulated to the end of the month at the fund’s rate of investment return.
17
those on the main US exchanges, outside the largest companies’ stocks. Third, the UK has a 0.5
percent stamp duty on share dealing, which further increases funds’ transaction costs. As a
further reference, in the last row of the table we have summarized the performance of an equally-
weighted portfolio of passively managed domestic equity funds; its alpha, at -13.1 basis points
per month, is insignificantly different from that of the value-weighted portfolio.15
The first row of the table presents results for portfolios based on the implied measure of net new
money flow, i.e. using the approximation one could obtain from monthly fund NAVs and
investment returns. The average monthly alpha of -5.9 basis points is 8.6 basis points higher than
the “old money” alpha; this difference is highly significant (p-value < 0.01). We then form
portfolios on the basis of actual net flows in our dataset. As the next row shows, this increases
the outperformance of new money relative to old money to 9.9 basis points per month, or more
than 1 percent annualized. In other words, there is a statistically significant smart money effect
in the UK.
To ensure that the smart money effect is not period-specific, we repeat our analysis separately for
1992-1995 and for 1996-2000 (results not reported in a table). In the earlier period, actual net
flows outperform old money by 11.9 basis points per month (p-value = 0.02), while in the later
period this difference is 8.3 basis points per month (p-value = 0.01). In other words, the smart
money effect is present in the UK throughout the 1990s.
15 Quigley and Sinquefield (2000) report that the Fama-French three-factor net of tax alpha of UK’s domestic equity mutual funds over the 1978-1997 period is -18 basis points per month, while their estimated gross of tax alpha is -9 basis points per month.
18
In the remainder of the table, we form portfolios using our different net flow components as
weights, the goal being to determine what drives our net flow result. As the third and fourth rows
indicate, the effect is strong for aggregate inflows (7.0 basis points per month, p-value < 0.01) but
there is no evidence that outflows precede poor performance: outflow-weighted alpha averages
0.6 basis points per month more than the old money portfolio alpha and is insignificantly
different from it (p-value = 0.75). Classifying net flows by type of investor, we find that retail
investors achieve an especially high alpha on their recent net fund purchases (11.0 basis points in
excess of the value weighted portfolio) while institutional investors’ net purchases only
outperform the old money portfolio by an insignificant 5.2 basis points per month. Further
partitioning of flows reveals that, consistent with the above results, retail investors’ purchases are
responsible for most of the smart money effect for net aggregate flows: their performance
exceeds that of old money by a highly significant 8.4 basis points per month, whereas the
performance of retail outflows, as well as that of institutional inflows and outflows, is not
significantly different from the performance of old money.
An alternative to the fund-level approach to appraising the smart money effect is what Zheng
(1999) terms the portfolio-level approach. Under this method, a suitably weighted portfolio of
funds is formed first, and then the resulting time series of excess portfolio returns is regressed on
the time series of factor realizations. A potential disadvantage of this method is that the factor
loadings are assumed to be constant over a longer period than under the fund-level approach
(nine years as contrasted with three years in our case). On the other hand, the portfolio-level
approach avoids the survivor bias of the fund-level approach, has a straightforward interpretation,
and has been popular with prior researchers (for example, Sapp and Tiwari (2004) only present
results on the basis of the portfolio-level approach).
19
[Table V here]
The results under the portfolio-level approach are given in Table V, and tell essentially the same
story as our fund-level results. Thus, net new money outperforms old money by 9.6 basis points
per month. One difference is that aggregation across funds prior to risk-adjustment renders the
portfolio-level tests less powerful than the fund-level tests. Of the four types of investor activity
comprising net flow, only the retail inflow-weighted portfolio posts statistically different
performance relative to size-weighted portfolios (8.4 basis points per month, p-value = 0.04).
Nonetheless, the smart money effect for net aggregate flows remains statistically significant (p-
value = 0.02). We also note that point estimates of the smart money effect in individuals’ and
institutions’ net flows are comparable (8.7 and 8.2 basis points, respectively).
Lastly, it is instructive to examine the patterns of factor loadings for our fund portfolios. Like in
the US (Carhart (1997), Sapp and Tiwari (2004)), money invested with UK’s active managers has
a market beta close to one and a positive exposure to the size factor. Contrary to the US, where
value factor exposure tends to be positive and momentum exposure negative, in the UK these
signs are reversed. These results are consistent with prior studies of UK mutual fund
performance (Quigley and Sinquefield (2000), Fletcher and Forbes (2002)). The momentum
result in particular has special significance because Sapp and Tiwari argue that momentum
investing by US funds alone accounts for the previously documented smart money effect. In the
UK, however, Wylie (2005) shows that mutual funds herd out of large stocks with high prior-year
returns.
20
B. Portfolios of Funds Sorted by Money Flow
The results in the preceding section establish that the average pound of new money outperforms
the average pound of old money. This finding leaves the possibility that the smart money effect
may be driven by a small number of funds with large money flow activity. To examine how
pervasive UK investors’ ability to select superior funds is, we partition funds into portfolios
based on their normalized flow activity. Specifically, each month we sort funds using our
measures of normalized money flows into high-flow portfolios (consisting of funds where the
normalized flow measure is above its median value for the month), and low-flow portfolios
(consisting of the remaining funds). We then compare the risk-adjusted performance of high- and
low-flow portfolios.
[Table VI here]
In Table VI, we show the results using the fund-level approach. In Panel A, the high- and low-
flow portfolios are equally weighted. The results prove to be broadly consistent with the findings
of the last section. Whether net normalized flows are measured directly or estimated from fund
NAVs and returns, sorting on this quantity results in a highly statistically significant performance
spread between high- and low-flow portfolios of about 7 basis points per month. A similar
spread is achieved by sorting on aggregate inflows, while high-outflow funds do not perform
worse than low-outflow funds. When classifying flows by investor type, we find institutional net
flows to be highly significant (p-value < 0.01), while retail net flows are significant at the 0.10
level (p-value = 0.06). When we disaggregate money flows further by investor type and by flow
direction, both individual and institutional inflows turn out to be highly significant, with high-
21
flow portfolios beating low-flow ones by over 7 basis points per month. Retail and institutional
outflows, on the other hand, are not significantly associated with future fund performance.
While the above results suggest that the smart money effect is widespread, weighting all funds
equally leaves the possibility that our results can be influenced by a relatively small number of
extreme fund returns. Therefore, in Panel B we re-estimate our results after combining funds in
both the high-flow and low-flow portfolios in proportion to the inverse of their estimation-period
residual return variance. This weighting procedure has the effect of reducing the performance
spread between high and low net aggregate flow portfolios, but at 5.6 basis points this spread
remains highly significant (p-value < 0.01). Interestingly, a researcher with access only to
implied flows would not be able to observe a statistically significant smart money effect, as the
corresponding performance difference shrinks to 2.9 basis points while its p-value increases to
0.13. The other results are also weakened relative to those based on equal weighting of funds,
but both institutional and individual net flows continue to be associated with fund
outperformance at the 0.10 significance level or higher.
[Table VII here]
In Table VII we re-examine performance differences between high- and low-flow fund portfolios
using the portfolio-level approach. In Panel A, where funds are equally weighted, we once again
document the smart money effect based on our net aggregate flow sorts (high minus low
difference = 8.6 basis points, p-value = 0.01). The remaining rows again indicate that this effect
is driven by investor purchases rather than withdrawals. There is some ambiguity about the
relative contributions of individual and institutional flows to the smart money effect, as
22
individual but not institutional purchases are associated with better subsequent fund performance,
while on a net basis, institutional but not retail flows are significant. After we re-weight funds by
the inverse of their estimation period variances in Panel B, the smart money effect is still highly
significant on a net aggregate basis (high minus low difference = 6.9 basis points, p-value <
0.01). There continues to be overwhelming evidence that fund purchases rather than withdrawals
drive the smart money effect. We note that here too, the smart money effect would not be
evident on the basis of implied flows estimated from fund NAVs and returns alone.
C. Results with Quarterly Flows Data
All the existing literature on the smart money effect has worked with quarterly implied flows,
because that is the frequency at which NAVs were available. Our results so far, unlike those of
Sapp and Tiwari, unequivocally point to the presence of the smart money effect – although the
evidence is less strong when we use implied net money flows than when we use actual flows. In
this section, we check whether using data at the quarterly frequency further hampers the
researcher’s ability to detect the smart money effect. There are at least three potential reasons
why this may be so. First, the number of observations relative to monthly frequency is reduced
by a factor of three. Second, implied flows lose accuracy as the span over which they are
measured grows. Lastly, when flows are quarterly and consequently fund portfolios are
rebalanced quarterly as well, flows are in effect used to predict returns between one and five
months ahead, and three months ahead on average – as compared to one month ahead when
monthly data are used.
23
To accomplish our investigation, first we aggregate all our flow measures to the quarterly
frequency. For implied flows, this means subtracting quarterly investment return from quarterly
change in fund value. For actual flows, we add up the flows for the three months comprising
each quarter (while requiring all three monthly flow measures to be non-missing). We obtain
9,209 fund-quarters of data as a result.16
Each quarter, we rank our funds on their actual and implied normalized net aggregate flows, and
then form equally weighted portfolios of funds with above-median flows (high-flow portfolios)
and of the remaining funds (low-flow portfolios). Using the fund-level approach, the
performance difference between high and low actual net flow portfolios is 5.7 basis points per
month (p-value = 0.02), while for portfolios formed using implied net flow, the difference is 4.9
basis points (p-value = 0.04). When we use the portfolio regression approach (as in Sapp and
Tiwari (2004)), the performance difference is insignificant for both actual and implied net flow
portfolios. These results indicate that even when the smart money effect is clearly present in
monthly data, it can be much harder to detect using quarterly data.
IV. Further Evidence on the Smart Money Effect
A. Conditional Performance Evaluation
The results of the last section establish that investors’ money tends to flow to funds which
subsequently deliver better investment performance. By employing the four-factor model we
16 Recall that using monthly data, actual and implied flows had different signs 17.2 percent of the time. Backing out net flows at the quarterly rather than at the monthly frequency further decreases their accuracy – the signs now differ for 18.8 percent of the observations.
24
control for some of the main known stock market regularities, with the intention of isolating fund
returns due to stock selection. There nonetheless remains the possibility that some or even all of
the performance differences between most and least popular funds is due to differences in style
rotation or to market timing rather than to stock picking. Indeed, Ferson and Schadt (1996) show
that allowing for dynamic portfolio strategies which are based on publicly available information –
such as changing a fund’s exposure to the stock market based on predictable variations in the
expected risk premium – can significantly alter conclusions about fund performance. Ferson and
Schadt recommend using conditional performance evaluation to address this problem.
There are several reasons why we do not expect the above problem to have a strong impact on
our sample. First, our funds are all in IMA-defined domestic equity sectors, and in order to
remain in these sectors they must at all times maintain their holdings of UK equities above the 80
percent level. This greatly limits these funds’ ability to execute any dynamic strategies relying
on rotation into other asset classes, such as fixed income securities or international equities.
Second, even the scope for style rotation within the domestic equity category is limited for these
funds, because this, too, is monitored by the IMA. For example, in order to remain in the UK
Equity Income sector, a fund is expected to have a yield in excess of 110 percent of the yield on
the Financial Times All Share (FTA) index. Third, there is little evidence that domestic equity
funds seek to implement market-timing by making significant modifications to their cash
positions.17
17 While we are not aware of a commercially available electronic source for historical fund cash holdings, the Unit Trust Year Book publishes cash percentages at each year end. If actively managed UK equity funds successfully time market movements by changing the size of their cash positions, one could expect these positions to be especially large at the end of 1999 (with the FTA index dropping by eight percent over the course of the following calendar year). Instead, when we examine the first twenty five funds (alphabetically) in our sample, we find that
25
Lastly, our finding that money is smart holds not only for our portfolio-level results, but for fund-
level results as well. Under the fund-level approach, we use three-year rolling regressions, which
has the effect of allowing for some time variation in betas and may therefore approximate
conditional performance evaluation (Ferson and Schadt (1996, footnote 1)).
Nonetheless, to ensure that our findings are not driven by macroeconomic influences, we re-
examine our portfolio-level results using Ferson and Schadt (1996) conditional performance
evaluation. Specifically, we follow Fletcher and Forbes (2002) in implementing the conditional
version of the Carhart (1997) model for the UK with the lagged market dividend yield and risk-
free rate representing the conditioning set of publicly available information.18
[Table VIII here]
Table VIII follows the layout of Table VII, but all estimates are based on conditional
performance evaluation. For brevity, we do not report regression coefficients for the terms
corresponding to the products of factor realizations and the conditioning variables. Since
conditioning variables are de-meaned, the coefficient estimates for our four factors correspond to
the average factor loadings over the sample period. Panel A examines equally-weighted
portfolios of high- and low-flow funds, while in Panel B the funds are weighted by the inverse of
their average end-1999 cash holding is only 2.3 percent, and the largest cash holding is only 6.6 percent. These numbers are close to cash levels that are to be expected in the course of normal fund operations. 18 These variables are obtained from the London Share Price Database as the dividend yield for the FTA index and the 90-day Treasury bill rate, respectively. To facilitate interpretation of regression coefficients in our conditional performance evaluation models, we de-mean both variables.
26
their residual variance of returns. The results are qualitatively similar to those in Table VII. This
suggests that the UK smart money effect is driven by fund-level characteristics, rather than by
publicly available macroeconomic information.
B. Is the Smart Money Effect Concentrated in Small Funds?
As we examine the smart money effect in the UK mutual fund marketplace, an important
consideration is to document how pervasive this effect is. In particular, Zheng (1999) draws
attention to the role of fund size, which may both condition investor choice and influence the
extent to which manager skill translates into fund performance. We therefore replicate our
analyses separately for small funds (those above median fund size in a given month) and large
funds (the others). High and low money flow funds are defined with the respect to the full fund
universe, as before. For brevity, we only show fund-regression results; results under the portfolio
regression approach do not alter the conclusion.
[Table IX here]
Panel A of Table IX is dedicated to small funds, while large funds are considered in Panel B.
Although the splitting of our sample into size groups hurts somewhat our ability to detect
statistical significance, the point estimates of the smart money effect for net flows are of
comparable magnitude for small and large funds (0.070 and 0.093, respectively). Comparing
these analyses with their counterpart for the full sample (Table VI, Panel A), there is no evidence
that either small or large funds are responsible for the bulk of the smart money effect.
27
C. Is the Smart Money Effect Subsumed by Regularities in Mutual Fund Returns?
The literature on mutual funds has long searched for patterns in mutual fund performance. Thus,
extensive research has been dedicated to the issue of performance persistence (e.g. Carhart (1997)
and Wermers (2003) for the US, Fletcher and Forbes (2002) for the UK). More recently, Chen et
al. (2004) have documented a size effect for US mutual funds, whereby larger funds exhibit
poorer performance, presumably due to diseconomies of scale in investment management. An in-
depth analysis of these regularities falls outside the scope of our paper; if, however, such
regularities manifest themselves in our sample, then it is important to check whether they explain
investors’ fund picking ability. To do so, we conduct a multivariate analysis of fund
performance. Specifically, we pool our data across funds and months, and regress the four-factor
alpha on relative flows and fund characteristics measured at the end of the preceding month.
[Table X here]
One of the main challenges in explaining fund performance is the highly non-normal distribution
of fund alphas. To cope with this problem we use least absolute deviation (LAD) regression. In
addition, to control for time fixed effects, we measure all variables as differences from their mean
values for each month. The results are presented in Table X. The first regression confirms the
smart money effect: fund alpha is positively and significantly related to previous month’s net
aggregate flow (coefficient = 2.183, p-value < 0.01). In the second regression, alongside net
aggregate flow we include the logarithm of fund size and the four-factor fund alpha over the
preceding 12 months. The highly significant coefficients on past alpha and fund size indicate,
respectively, that our sample is characterized by both performance persistence and a fund size
28
effect.19 However, the coefficient for the money flow term, at 1.722, remains highly significant
(p-value < 0.01) indicating that neither performance persistence nor the fund size effect subsume
the smart money effect.
Lastly, in the third regression we conduct an additional check of the possibility that the smart
money effect is unevenly distributed across fund sizes. To do so, we add an interaction term for
the logarithm of fund size and money flow. Consistently with the evidence in the previous
subsection, the interaction term is insignificant, while the flow term itself continues to be highly
significant.
D. The Span of the Smart Money Effect
While our analyses up to this point have been based on fund performance in the month
immediately following money flows, an interesting question is for how long the smart money
effect persists. Unfortunately, with only nine years of money flows, our dataset is less than
ideally suited for addressing this question. With this caveat, Table XI examines the performance
of the smart money effect for up to one year ahead. Specifically, it uses the fund-regression
approach to compare performance of high- and low-flow funds where flows are lagged from one
to twelve months.
[Table XI here]
19 As a further check on the persistence result, we compare the performance of high and low performing funds using the methodologies of the preceding sections; these analyses confirm the persistence finding. For example, under the portfolio-level approach, the alpha of the portfolio of “winner” funds is -0.048 percent, while that of the portfolio of “loser” funds is -0.200 percent (p-value for the difference = 0.002). As we will discuss in the next section, however, high initial charges are likely to preclude trading strategies designed to exploit such persistence.
29
In Panel A funds are sorted based on their implied flows. We note that this is the only measure of
flows which investors can actually observe. While the short sample period causes our point
estimates to fluctuate considerably, both the signs and the p-values of the performance difference
between high and low flow funds show the effect to be relatively short-lived, lasting only a few
months following the month in which the flow is measured. Thus, even if mutual fund
investments could be made without a front load (which in the UK typically can only be done by
an investor transferring money within a fund family), and if (counterfactually) low-flow funds
could be sold short, an investor seeking to profit from the smart money effect would on average
gain only a very modest return of a dozen basis points or so. Thus, while investors tend to make
good fund choices on average, this smart money effect does not give rise to what Zheng (1999)
calls an “information effect”.
Panel B shows results for actual flows. While the point estimates are rather imprecise as before,
this time the signs and p-values of the performance differences between high and low flow funds
suggest that the true smart money effect may last up to seven months, i.e. somewhat longer than
the span observed using implied flows. This observation is consistent with investors making
good mutual fund choices and implied flows being an approximation of actual investor behavior.
30
V. Discussion
Having established the existence of the smart money effect in the British mutual fund
marketplace, in this section we seek to put our findings in perspective. Specifically, we now
discuss the interpretation of our results, their economic significance, and their implications.
We start with an important caveat. The term “smart money” in the mutual fund literature has
come to be associated with investors’ ability (or lack thereof) to identify superior future
performers from a group of comparable funds. Our paper follows this convention. There are, of
course, other ways for mutual fund investors to be smart: they can, for example, move their
money among funds with different objectives so as to implement a dynamic asset allocation
strategy. Indeed, the low net flows into domestic equity funds from institutional investors (as
compared to those from individual investors) in 1998 and 1999 as shown in Table 1 may be an
indication of institutional investor “smartness” in this sense. While this is an interesting and
important topic, it falls outside the scope of our study. Although an investor who is good at
sector picking may or may not be good at fund picking, it is the latter skill which is the focus of
this paper and the literature which it extends.
Investor fund picking ability originally attracted research interest as a possible explanation for the
puzzling popularity of actively managed funds despite the availability of passive funds (Gruber
(1996)). Like in the US, UK passive mutual funds are cheaper to own than active funds. Thus, in
1999, the median front-end load for UK’s actively managed domestic equity funds was 5 percent,
while the equivalent for passive funds was only 1 percent (UK funds do not have back-end
loads); the median annual charge was 1.25 percent for active funds, and 1 percent for passive
31
funds. The magnitude of the smart money effect we observe is not nearly enough to equate net-
of-charges returns for new money invested in actively managed funds, taken in aggregate, with
those for passively invested money over any investment horizon. We are therefore inclined to
concur with Sapp and Tiwari’s (2004) assertion that Gruber’s (1996) puzzle is not explained by
smart money – or in any case, smart money can only be part of the answer. Our contribution,
rather, is to demonstrate that – in contrast to Sapp and Tiwari’s findings – within the universe of
actively managed funds, investors consistently find funds which perform better than average in
the future.
What lies behind this future outperformance? While a natural answer is manager skill, Wermers
(2003) and Bernhardt, Davies and Westbrook (2004) suggest another possibility. They argue that
investor money flows are disproportionately used by funds to buy stocks they already own,
exerting upward pressure on these stocks’ prices, and thus contributing to funds’ post-flow
performance. Unfortunately, we do not have UK funds’ portfolio holdings data to test this
alternative explanation. We have nonetheless attempted checking for a testable implication of the
price pressure story. We can expect an individual fund’s influence on the price of stocks it buys
or sells to be greater the less liquid these stocks are. Further, it is well known that generally
liquidity is inversely related to market capitalization. Although we do not have the data to
measure the market capitalization of stocks held by a given mutual fund, a crude measure of the
“smallness” of a fund’s holdings is its estimated coefficient on the SMB factor. Prima facie
evidence on the validity of the price pressure explanation can be gleaned from Table IX. It
shows that large and small funds have similar SMB loadings, suggesting that they hold a similar
mix of small cap and large cap stocks. Since large funds can exert greater price pressure on the
stocks they hold, the price pressure story would predict a stronger smart money effect for these
32
funds. The results in the table, however, contradict this prediction. To investigate this issue
further, we implement regressions paralleling those of section IV.C, but including an interaction
term between the SMB factor and net aggregate flow (results not reported in a table). This
interaction term is insignificantly different from zero (whether we measure the net aggregate flow
as proportion of fund value or as an absolute amount) while the separate relative flow term
continues to be significant. We regard this as inconsistent with the price pressure story being the
explanation for our smart money effect.20
Our interpretation of the smart money effect in the UK is that mutual fund investors (both
institutions and individuals) indeed tend to buy into those actively managed funds which are able
to deliver better than average returns for some time – at least until the market environment
changes to one where the manager’s skills become less valuable and/or until fund size increases
to the point of hurting performance. At the same time, the fact that sales of mutual fund shares
are not associated with poor subsequent performance is easy to reconcile with the notion of
investors possessing fund picking skill. Investors who are unhappy with their funds’
performance prospects would need to weigh any potential performance gain by switching to
another fund against the certainty of paying the initial charge on the money they transfer (unless
they are transferring within the same fund family, in which case all or part of the initial charge
may be waived). A complementary consideration is that, as Lynch and Musto (2003) argue,
investors can rationally expect poor fund managers to be replaced.
20 We note that at least two other “mechanical” factors have a bearing on post-flow fund returns. One is the costs associated with buying or selling of stocks triggered by money inflows. This has the effect of depressing returns so it works against the smart money effect. The other factor is changes in beta occasioned by temporary parking of investor money flows in cash (Ferson and Schadt (1996)). If these beta changes are correlated with subsequent market performance, then a spurious smart money effect may appear if the performance evaluation model does not allow for changes in beta. As we have shown, however, our results are robust to conditional performance evaluation.
33
In recent years, legislators around the world have been considering whether investors should be
protected, in various ways, from hurting themselves with poor mutual fund investment decisions.
Underlying such initiatives appears to be the assumption that investors are unlikely to make good
mutual fund choices on their own. Our empirical findings contradict this assumption. Much
more needs to be done, however, to understand how different categories of investors arrive at
their mutual fund buying and selling decisions. Gaining insight into mutual fund investor
behavior continues to be an exciting area for future research.
VI. Conclusion
Millions of investors around the world place their assets in mutual funds. Thousands of
institutions do likewise. Their common goal, presumably, is to pick the best funds to invest in.
Do they succeed?
If (at least some) investors can spot superior funds in advance, we should see a positive
relationship between investors’ money flows and subsequent abnormal fund performance. Using
this insight, Gruber (1996) and Zheng (1999) documented that investors indeed make good
decisions – money is “smart”. However, Sapp and Tiwari (2004), after applying a more
appropriate performance evaluation procedure which takes the stock return momentum effect into
account, were unable to detect the smart money effect.
34
We revisit the smart money controversy with a unique British dataset. Unlike previous studies,
all of which are US-based and use quarterly flows implied by fund sizes and investment returns,
we have access to exact net flows for our funds. We also observe four key components of these
net flows: investments by individuals; investments by institutions; disinvestments by individuals;
and disinvestments by institutions. These features of the data allow us to conduct a more
powerful test of the smart money effect than previously possible.
We find conclusive evidence that smart money is alive and well in the UK. The performance
difference between new money and old money, or between high and low net flow funds, averages
up to 10 basis points per month and is highly statistically significant. The smart money effect is
driven by fund purchases (but not withdrawals) of both individuals and institutions.
35
REFERENCES
Barclay, Michael J., Neil D. Pearson, and Michael S. Weisbach, 1998, Open-end mutual funds
and capital-gains taxes, Journal of Financial Economics 49, 3-43. Berk, Jonathan B., and Richard C. Green, 2004, Mutual fund flows and performance in rational
markets, Journal of Political Economy 112, 1269-1295. Berk, Jonathan B., and Jing Xu, 2004, Persistence and fund flows of the worst performing
mutual funds, Working Paper, University of California, Berkeley. Bernhardt, Dan, Ryan Davies, and Harvey Westbrook Jr., 2004, Smart fund managers? Stupid
money?, Working Paper, University of Illinois. Carhart, Mark M., 1997, On persistence in mutual fund performance, Journal of Finance 52, 57-
82. Chen, Joseph, Harrison Hong, Ming Huang, and Jeffrey Kubik, 2004, Does fund size erode
mutual fund performance? The role of liquidity and organization, American Economic Review 95, 1276-1302.
Chevalier, Judith, and Glenn Ellison, 1997, Risk taking by mutual funds as a response to
incentives, Journal of Political Economy, 105, 1167-1200. Cooper, Michael J., Huseyin Gulen, and P. Raghavendra Rau, 2004, Changing names with style:
Mutual fund name changes and their effects on fund flows, Journal of Finance, Forthcoming.
Del Guercio, Diane, and Paula Tkac, 2002, The determinants of the flow of funds on managed
portfolios: Mutual funds vs. pension funds, Journal of Financial and Quantitative Analysis 37, 523-557.
Dimson, Elroy, Stefan Nagel, and Garrett Quigley, 2003, Capturing the value premium in the UK
1955-2001, Financial Analysts Journal 59, 35-45. Fama, Eugene F., and Kenneth R. French, 1992, The cross-section of expected stock returns,
Journal of Finance 47, 427-465. Fama, Eugene F., and Kenneth R. French, 1993, Common risk factors in the returns on stocks
and bonds, Journal of Financial Economics 33, 3-56. Ferson, Wayne E., and Rudi W. Schadt, 1996, Measuring fund strategy and performance in
changing economic conditions, Journal of Finance 51, 425-461.
36
Fletcher, Jonathan, and David Forbes, 2002, An exploration of the persistence of UK unit trust
performance, Journal of Empirical Finance 9, 475-493. Gruber, Martin J., 1996, Another puzzle: The growth in actively managed mutual funds, Journal
of Finance 51, 783-810. Jegadeesh, Narasimhan, and Sheridan Titman, 1993, Returns to buying winners and selling
losers: Implications for stock market efficiency, Journal of Finance 48, 65-91. Jensen, Michael C., 1968, The behavior of mutual funds in the period 1945-64, Journal of
Finance 23, 389-416. Khorana, Ajay, Henri Servaes, and Peter Tufano, 2005, Explaining the size of the mutual fund
industry around the world, Journal of Financial Economics, Forthcoming. Lynch, Anthony W., and David K. Musto, 2003, How investors interpret past fund returns,
Journal of Finance 58, 2033-2058. Quigley, Garrett, and Rex A. Sinquefield, 2000, Performance of UK equity unit trusts, Journal of
Asset Management 1, 72-92. Rouwenhorst, K. Geert, 1998, International momentum strategies, Journal of Finance 53, 267-
284. Sapp, Travis, and Ashish Tiwari, 2004, Does stock return momentum explain the "smart money"
effect?, Journal of Finance 59, 2605-2622. Sirri, Erik R., and Peter Tufano, 1998, Costly search and mutual fund flows, Journal of Finance
53, 1589-1621. Wermers, Russ, 2003, Is money really “smart”? New evidence on the relation between mutual
fund flows, manager behavior, and performance persistence, Working Paper, University of Maryland.
Wylie, Sam, 2005, Fund manager herding: A test of the accuracy of empirical results using UK
data, Journal of Business 78, 381-403. Zheng, Lu, 1999, Is money smart? A study of mutual fund investors' fund selection ability,
Journal of Finance 54, 901-933.
37
Table I
Characteristics of the Mutual Fund Sample This table describes our sample of UK mutual funds investing in domestic equities. “Number of funds” and “total assets” counts eligible funds and their assets under management, respectively, at the start of the calendar year. Attrition rate is the proportion of funds in existence at the start of the year which cease to exist (through merger or liquidation) by the end of the year. Money inflow (outflow) is the exact amount of sales to (repurchases from) investors as reported by fund management companies to the Investment Management Association. Fund assets and money flow values are in £1 million.
Panel A: All UK equity funds
1992 1993 1994 1995 1996 1997 1998 1999 2000 average
All funds
Number of funds 425 447 438 436 466 491 480 469 496 461
Total assets 28,278 32,614 43,279 39,834 54,470 64,288 79,894 85,594 115,210 60,385
Actively managed funds
Number of funds 413 430 419 416 443 456 441 425 441 432
Total assets 27,686 31,422 41,676 38,264 52,181 60,985 74,117 77,551 103,263 56,349
Panel B: Funds with flow data
1992 1993 1994 1995 1996 1997 1998 1999 2000 average
Number of funds 258 296 319 316 330 339 337 318 302 313
Total assets 20,338 24,314 35,623 31,925 40,609 47,844 61,837 62,673 78,519 44,853
Average fund size 79 82 112 101 123 141 183 197 260 142
Proportion of funds covered 62.5% 68.8% 76.1% 76.0% 74.5% 74.3% 76.4% 74.8% 68.5% 72.4%
Proportion of assets covered 73.5% 77.4% 85.5% 83.4% 77.8% 78.5% 83.4% 80.8% 76.0% 79.6%
Attrition rate 0.8% 4.4% 6.0% 4.7% 3.6% 10.9% 10.4% 12.6% 3.3% 6.3%
Net aggregate flow 253 3,074 3,276 2,153 2,272 2,734 1,943 2,359 -230 1,981
Aggregate inflow 2,558 5,173 5,622 4,969 6,346 7,940 9,094 9,661 10,329 6,855
Aggregate outflow 2,305 2,099 2,346 2,816 4,075 5,206 7,151 7,302 10,559 4,873
Net individual flow 236 1,463 2,236 1,286 1,498 2,217 2,177 2,340 1,418 1,652
Net institutional flow 17 1,610 1,040 867 774 517 -234 19 -1,648 329
Individual inflow 1,164 2,599 3,548 2,983 3,769 4,951 6,029 6,342 6,307 4,188
Individual outflow 928 1,135 1,312 1,696 2,271 2,734 3,852 4,002 4,889 2,536
Institutional inflow 1,394 2,574 2,074 1,986 2,577 2,989 3,065 3,319 4,022 2,667
Institutional outflow 1,377 964 1,034 1,119 1,803 2,472 3,299 3,300 5,670 2,338
38
Table II
Distribution of Monthly Money Flows
This table shows the distribution of monthly money flows over the 1992-2000 period for 30,846 fund-months. Money flows are expressed as percentage of start-of-month fund size. Moments and correlations in the tables are time-series averages of corresponding quantities calculated for each of the 108 monthly cross-sections.
Panel A: Moments of money flow measures
Standard Percentile
Mean Deviation Minimum 10th 25th Median 75th 90th Maximum
(1) Implied flow 0.49 3.70 -13.87 -2.16 -0.86 -0.02 1.17 3.62 27.09
(2) Net aggregate flow 0.67 3.31 -12.10 -1.18 -0.47 0.01 0.97 3.25 27.05
(3) Aggregate inflow 1.60 3.24 0.00 0.02 0.13 0.54 1.68 4.10 31.07
(4) Aggregate outflow 0.93 1.62 0.00 0.05 0.27 0.58 1.00 1.83 17.64
(5) Net individual flow 0.56 2.84 -9.84 -0.90 -0.37 -0.01 0.63 2.58 24.44
(6) Net institutional flow 0.26 2.18 -8.71 -0.73 -0.16 0.01 0.37 1.43 15.23
(7) Individual inflow 1.28 2.77 0.00 0.01 0.08 0.36 1.26 3.32 26.45
(8) Individual outflow 0.72 1.19 0.00 0.03 0.20 0.47 0.82 1.42 13.13
(9) Institutional inflow 0.74 2.08 0.00 0.00 0.02 0.14 0.59 1.76 17.91
(10) Institutional outflow 0.48 1.36 0.00 0.00 0.01 0.11 0.37 1.13 11.76
Panel B: Correlations between money flow measures
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
(1) Implied flow 1 0.853 0.743 -0.297 0.774 0.719 0.693 -0.219 0.588 -0.310
(2) Net aggregate flow 0.853 1 0.870 -0.348 0.918 0.822 0.818 -0.266 0.675 -0.356
(3) Aggregate inflow 0.743 0.870 1 0.122 0.835 0.643 0.926 0.116 0.798 0.079
(4) Aggregate outflow -0.297 -0.348 0.122 1 -0.256 -0.413 0.104 0.829 0.067 0.827
(5) Net individual flow 0.774 0.918 0.835 -0.256 1 0.247 0.893 -0.283 0.255 -0.054
(6) Net institutional flow 0.719 0.822 0.643 -0.413 0.247 1 0.224 -0.082 0.791 -0.462
(7) Individual inflow 0.693 0.818 0.926 0.104 0.893 0.224 1 0.143 0.271 0.013
(8) Individual outflow -0.219 -0.266 0.116 0.829 -0.283 -0.082 0.143 1 -0.011 0.141
(9) Institutional inflow 0.588 0.675 0.798 0.067 0.255 0.791 0.271 -0.011 1 0.112
(10) Institutional outflow -0.310 -0.356 0.079 0.827 -0.054 -0.462 0.013 0.141 0.112 1
39
Table III
Regression of Components of Money Flows on Lagged Flow and Performance Each row of this table summarizes the results of 96 (January 1993 – December 2000) monthly cross-sectional regressions of different flow variables on their lagged values and past performance. The columns labeled “Intercept”, “Lagged flow” and “Performance” contain the average value of the coefficient estimates for that variable, followed by the p-value from a t-test based on the time-series standard deviation of the coefficient estimates. “Lagged flow” for each flow component is the value of the same flow component from 12 months earlier. “Performance” is either the Jensen (1968) market model alpha (Panel A) or the Carhart (1997) four-factor alpha (Panel B) averaged over the 12 months preceding the flow. “N” is the average number of funds in a cross-section, and “R2” is the average of the cross-sectional regressions’ R-squared values.
Panel A: Past performance is measured as Jensen (1968) alpha
Dependent variable Intercept Lagged flow Performance N R2
(1) Implied flow 0.003 0.000 0.066 0.000 1.252 0.000 242 0.095
(2) Net aggregate flow 0.004 0.000 0.148 0.000 1.127 0.000 242 0.146
(3) Aggregate inflow 0.010 0.000 0.209 0.000 0.977 0.000 242 0.182
(4) Aggregate outflow 0.007 0.000 0.112 0.000 -0.119 0.000 242 0.054
(5) Net individual flow 0.003 0.000 0.209 0.000 0.939 0.000 228 0.196
(6) Net institutional flow 0.001 0.000 0.111 0.000 0.391 0.000 115 0.086
(7) Individual inflow 0.007 0.000 0.275 0.000 0.806 0.000 228 0.232
(8) Individual outflow 0.005 0.000 0.201 0.000 -0.117 0.000 228 0.084
(9) Institutional inflow 0.004 0.000 0.241 0.000 0.311 0.000 115 0.131
(10) Institutional outflow 0.003 0.000 0.202 0.000 -0.054 0.204 115 0.087
Panel B: Past performance is measured as Carhart (1997) four-factor alpha
Dependent variable Intercept Lagged flow Performance N R2
(1) Implied flow 0.003 0.000 0.064 0.000 1.527 0.000 242 0.100
(2) Net aggregate flow 0.004 0.000 0.143 0.000 1.353 0.000 242 0.149
(3) Aggregate inflow 0.010 0.000 0.204 0.000 1.197 0.000 242 0.187
(4) Aggregate outflow 0.008 0.000 0.115 0.000 -0.118 0.000 242 0.049
(5) Net individual flow 0.003 0.000 0.201 0.000 1.139 0.000 228 0.199
(6) Net institutional flow 0.001 0.000 0.112 0.000 0.447 0.000 115 0.084
(7) Individual inflow 0.008 0.000 0.268 0.000 0.993 0.000 228 0.235
(8) Individual outflow 0.005 0.000 0.202 0.000 -0.112 0.000 228 0.083
(9) Institutional inflow 0.004 0.000 0.242 0.000 0.353 0.000 115 0.131
(10) Institutional outflow 0.004 0.000 0.207 0.000 -0.075 0.073 115 0.084
40
Table IV
Comparison of New Money and Old Money Portfolios: Fund Regression Approach This table describes portfolios of UK equity mutual funds formed on the basis of the funds’ money flows in the preceding month. Fund flow data are for 1992-2000. Flows are classified by source as originating from individual investors or from institutional investors; we additionally calculate aggregate flows (individual and institutional flows combined). Flows are also classified by direction as inflows (sales to investors) or outflows (repurchases from investors); we additionally calculate net flows (inflows less outflows). “Implied flow” is obtained as fund NAV at the end of a month, less the product of the fund’s NAV at the start of the month and its total return during the month. Only actively managed UK equity funds are used, with the exception of the last row, which describes an equally-weighted portfolio of passive UK equity funds. Fund portfolios are characterized on the basis of fund-level regression results. Specifically, for each fund-month, we run a Carhart (1997) time-series regression over the preceding 36 months of excess fund returns on the excess market return (MKT), the size factor (SMB), the value factor (HML), and the momentum factor (WML) for the UK stock market. We require a minimum of 30 return observations. The fund alpha is obtained as the fund excess return less the sum of the products of each of the four factor realizations and the corresponding factor loadings. For each month, we then calculate the portfolio alpha, the factor loadings, and the average R2 as a money-weighted average of these measures for the funds comprising the portfolio (except for the last two rows, where the averages are equally-weighted). The table reports time-series averages of these quantities. The average alpha value is followed by the p-value for its difference from zero, where the p-value is based on the time-series standard deviation. The last two columns show the difference between the average portfolio alpha and the alpha of the fund value-weighted portfolio, followed by the p-value for the hypothesis that the difference is zero.
Factor loading on Alpha difference
Portfolio description Alpha MKT SMB HML WML R2 from VW alpha
(1) Weighted by implied flow -0.059 0.439 0.975 0.181 0.034 -0.044 0.915 0.086 0.001
(2) Weighted by net aggregate flow -0.046 0.534 0.971 0.197 0.043 -0.044 0.913 0.099 0.001
(3) Weighted by aggregate inflow -0.074 0.299 0.973 0.201 0.054 -0.049 0.913 0.070 0.001
(4) Weighted by aggregate outflow -0.138 0.085 0.988 0.244 0.080 -0.061 0.914 0.006 0.748
(5) Weighted by net individual flow -0.031 0.682 0.963 0.177 0.042 -0.042 0.907 0.114 0.000
(6) Weighted by net institutional flow -0.093 0.235 0.983 0.228 0.050 -0.046 0.921 0.052 0.166
(7) Weighted by individual inflow -0.060 0.399 0.967 0.175 0.057 -0.049 0.909 0.084 0.000
(8) Weighted by individual outflow -0.121 0.099 0.979 0.193 0.078 -0.059 0.910 0.024 0.196
(9) Weighted by institutional inflow -0.110 0.160 0.985 0.243 0.055 -0.052 0.920 0.035 0.310
(10) Weighted by institutional outflow -0.167 0.072 1.000 0.312 0.079 -0.065 0.920 -0.022 0.560
(11) Weighted by fund value -0.145 0.054 0.983 0.170 0.056 -0.059 0.921 --- ---
(12) Equally-weighted -0.120 0.137 0.982 0.259 0.051 -0.037 0.894 0.025 0.390
(13) Passive funds, equally weighted -0.131 0.054 0.986 0.004 -0.003 -0.025 0.939 0.014 0.785
41
Table V
Comparison of New Money and Old Money Portfolios: Portfolio Regression Approach This table describes portfolios of UK equity mutual funds formed on the basis of the funds’ money flows in the preceding month. Fund flow data are for 1992-2000. Flows are classified by source as originating from individual investors or from institutional investors; we additionally calculate aggregate flows (individual and institutional flows combined). Flows are also classified by direction as inflows (sales to investors) or outflows (repurchases from investors); we additionally calculate net flows (inflows less outflows). “Implied flow” is obtained as fund NAV at the end of a month, less the product of the fund’s size at the start of the month and its total return during the month. Only actively managed UK equity funds are used, with the exception of the last row, which describes an equally-weighted portfolio of passive UK equity funds. Fund portfolios are characterized on the basis of portfolio-level regression results. Specifically, for each month we form money-weighted portfolios of funds (except for the last two rows, where portfolios are equally weighted). Each row of the table reports the results of a Carhart (1997) regression of portfolio excess returns on the excess market return (MKT), the size factor (SMB), the value factor (HML), and the momentum factor (WML) for the UK stock market. “Alpha” is the regression intercept, and is followed by its p-value. The last two columns show the difference between each portfolio’s alpha and the alpha of the value-weighted portfolio, followed by the p-value for the hypothesis that the difference is zero.
Factor loading on Alpha difference
Portfolio description Alpha MKT SMB HML WML R2 from VW alpha
(1) Weighted by implied flow -0.080 0.255 1.004 0.191 0.043 0.003 0.966 0.092 0.010
(2) Weighted by net aggregate flow -0.076 0.280 0.997 0.198 0.035 -0.001 0.966 0.096 0.019
(3) Weighted by aggregate inflow -0.110 0.105 1.005 0.207 0.048 -0.007 0.969 0.062 0.041
(4) Weighted by aggregate outflow -0.197 0.009 1.012 0.249 0.098 -0.024 0.963 -0.024 0.366
(5) Weighted by net individual flow -0.085 0.236 0.989 0.182 0.021 -0.001 0.964 0.087 0.057
(6) Weighted by net institutional flow -0.090 0.271 1.010 0.224 0.074 0.005 0.955 0.082 0.107
(7) Weighted by individual inflow -0.104 0.134 1.004 0.188 0.048 -0.009 0.967 0.068 0.038
(8) Weighted by individual outflow -0.138 0.056 1.017 0.197 0.093 -0.032 0.966 0.034 0.201
(9) Weighted by institutional inflow -0.128 0.096 1.014 0.249 0.059 -0.002 0.961 0.045 0.321
(10) Weighted by institutional outflow -0.270 0.004 1.003 0.314 0.102 -0.017 0.942 -0.098 0.103
(11) Weighted by fund value -0.172 0.011 1.007 0.176 0.059 -0.022 0.970 --- ---
(12) Equally-weighted -0.120 0.081 1.004 0.261 0.039 0.002 0.967 0.052 0.046
(13) Passive funds, equally weighted -0.168 0.008 1.007 -0.008 0.023 0.008 0.975 0.004 0.933
42
Table VI
Performance of Funds Sorted by Money Flows: Fund Regression Approach This table shows the performance of actively managed UK equity mutual funds classified on the basis of their normalized money flows in the preceding month. Fund flow data are for 1992-2000. Flows are classified by source as originating from individual investors or from institutional investors; we additionally calculate aggregate flows (individual and institutional flows combined). Flows are also classified by direction as inflows (sales to investors) or outflows (repurchases from investors); we additionally calculate net flows (inflows less outflows). “Implied flow” is obtained as fund NAV at the end of a month, less the product of the fund’s NAV at the start of the month and its total return during the month. We normalize each flow measure by dividing it by fund NAV at the start of the month. “High money flow funds” refers to a portfolio of funds in the top 50 percent of all funds each month, according to the stated normalized flow measure. “Low money flow funds” refers to a portfolio of the remaining funds. In Panel A, funds are equally weighted, and in Panel B, funds are weighted by the inverse of their estimation-period residual variance. Portfolios are rebalanced monthly. Fund portfolios are characterized on the basis of fund-level regression results. Specifically, for each fund-month, we run a Carhart (1997) time-series regression over the preceding 36 months of excess fund returns on the excess market return (MKT), the size factor (SMB), the value factor (HML), and the momentum factor (WML) for the UK stock market. We require a minimum of 30 return observations. The fund alpha is obtained as the fund excess return less the sum of the products of each of the four factor realizations and the corresponding factor loadings. For each month, we then calculate the portfolio alpha, the factor loadings, and the average R2 as the simple average of these measures for the funds comprising the portfolio. The table reports time-series averages of these quantities. The last two rows show the difference between the average alphas of the high- and low-flow portfolios, followed by the p-value for the hypothesis that the difference is zero.
Panel A: Funds are equally weighted
High money flow funds Low money flow funds
Flow variable Alpha MKT SMB HML WML R2 Alpha MKT SMB HML WML R2 Alpha difference
(1) Implied flow -0.083 0.977 0.259 0.037 -0.033 0.896 -0.153 0.986 0.260 0.063 -0.040 0.893 0.070 0.007
(2) Net aggregate flow -0.084 0.975 0.254 0.039 -0.031 0.898 -0.152 0.988 0.264 0.061 -0.041 0.891 0.068 0.005
(3) Aggregate inflow -0.082 0.976 0.262 0.046 -0.033 0.893 -0.154 0.987 0.257 0.055 -0.040 0.895 0.072 0.010
(4) Aggregate outflow -0.112 0.984 0.273 0.062 -0.038 0.889 -0.128 0.979 0.246 0.039 -0.036 0.900 0.016 0.505
(5) Net individual flow -0.096 0.975 0.245 0.035 -0.030 0.898 -0.145 0.987 0.266 0.058 -0.041 0.889 0.049 0.063
(6) Net institutional flow -0.071 0.988 0.284 0.046 -0.037 0.908 -0.146 0.993 0.261 0.078 -0.047 0.907 0.074 0.006
(7) Individual inflow -0.082 0.974 0.254 0.048 -0.031 0.890 -0.158 0.988 0.258 0.047 -0.040 0.896 0.077 0.013
(8) Individual outflow -0.109 0.981 0.261 0.058 -0.036 0.885 -0.135 0.982 0.252 0.037 -0.036 0.901 0.026 0.303
(9) Institutional inflow -0.071 0.990 0.304 0.060 -0.037 0.906 -0.146 0.991 0.243 0.066 -0.046 0.909 0.074 0.006
(10) Institutional outflow -0.104 0.995 0.297 0.075 -0.041 0.905 -0.116 0.986 0.247 0.050 -0.042 0.910 0.012 0.608
43
Table VI (continued) Performance of Funds Sorted by Money Flows: Fund Regression Approach
Panel B: Funds are weighted by the inverse of estimation period residual variance
High money flow funds Low money flow funds
Flow variable Alpha MKT SMB HML WML R2 Alpha MKT SMB HML WML R2 Alpha difference
(1) Implied flow -0.140 0.963 0.140 0.022 -0.033 0.930 -0.169 0.978 0.151 0.045 -0.037 0.927 0.029 0.126
(2) Net aggregate flow -0.126 0.962 0.138 0.024 -0.032 0.931 -0.182 0.979 0.154 0.043 -0.037 0.926 0.056 0.004
(3) Aggregate inflow -0.137 0.959 0.144 0.033 -0.034 0.927 -0.171 0.980 0.148 0.034 -0.036 0.930 0.035 0.064
(4) Aggregate outflow -0.153 0.971 0.155 0.046 -0.037 0.925 -0.157 0.969 0.137 0.021 -0.033 0.931 0.004 0.843
(5) Net individual flow -0.139 0.962 0.125 0.020 -0.030 0.931 -0.177 0.978 0.152 0.040 -0.038 0.924 0.038 0.058
(6) Net institutional flow -0.120 0.977 0.164 0.026 -0.037 0.936 -0.169 0.985 0.161 0.053 -0.041 0.933 0.049 0.026
(7) Individual inflow -0.140 0.956 0.132 0.036 -0.032 0.925 -0.175 0.981 0.143 0.025 -0.036 0.931 0.035 0.116
(8) Individual outflow -0.158 0.967 0.141 0.043 -0.035 0.923 -0.160 0.972 0.136 0.018 -0.033 0.932 0.002 0.908
(9) Institutional inflow -0.122 0.979 0.191 0.038 -0.037 0.933 -0.168 0.983 0.138 0.041 -0.040 0.936 0.047 0.041
(10) Institutional outflow -0.137 0.984 0.190 0.056 -0.040 0.932 -0.153 0.978 0.134 0.023 -0.038 0.936 0.016 0.428
44
Table VII
Performance of Funds Sorted by Money Flows: Portfolio Regression Approach This table shows the performance of actively managed UK equity mutual funds classified on the basis of their normalized money flows in the preceding month. Fund flow data are for 1992-2000. Flows are classified by source as originating from individual investors or from institutional investors; we additionally calculate aggregate flows (individual and institutional flows combined). Flows are also classified by direction as inflows (sales to investors) or outflows (repurchases from investors); we additionally calculate net flows (inflows less outflows). “Implied flow” is obtained as fund NAV at the end of a month, less the product of the fund’s NAV at the start of the month and its total return during the month. We normalize each flow measure by dividing it by fund NAV at the start of the month. “High money flow funds” refers to a portfolio of funds in the top 50 percent of all funds each month, according to the stated normalized flow measure. “Low money flow funds” refers to a portfolio of the remaining funds. In Panel A, funds are equally weighted, and in Panel B, funds are weighted by the inverse of their estimation-period residual variance. Portfolios are rebalanced monthly. Fund portfolios are characterized on the basis of portfolio-level regression results. Specifically, each row of the table reports the results of a Carhart (1997) regression of portfolio excess returns on the excess market return (MKT), the size factor (SMB), the value factor (HML), and the momentum factor (WML) for the UK stock market. “Alpha” is the regression intercept, and is followed by its p-value. The last two columns show the difference between each portfolio’s alpha and the alpha of the value-weighted portfolio, followed by the p-value for the hypothesis that the difference is zero.
Panel A: Funds are equally weighted
High money flow funds Low money flow funds
Flow variable Alpha MKT SMB HML WML R2 Alpha MKT SMB HML WML R2 Alpha difference
(1) Implied flow -0.081 0.998 0.260 0.019 0.016 0.967 -0.160 1.010 0.262 0.059 -0.011 0.964 0.080 0.026
(2) Net aggregate flow -0.077 0.995 0.260 0.017 0.013 0.967 -0.164 1.013 0.262 0.062 -0.009 0.964 0.086 0.011
(3) Aggregate inflow -0.066 0.997 0.274 0.017 0.013 0.967 -0.175 1.010 0.249 0.061 -0.009 0.964 0.109 0.003
(4) Aggregate outflow -0.108 1.004 0.272 0.051 0.000 0.967 -0.133 1.004 0.250 0.028 0.005 0.965 0.025 0.372
(5) Net individual flow -0.092 0.998 0.257 -0.005 0.014 0.969 -0.148 1.010 0.259 0.075 -0.009 0.961 0.056 0.157
(6) Net institutional flow -0.065 1.017 0.290 0.022 0.010 0.962 -0.155 1.020 0.260 0.063 -0.015 0.967 0.090 0.021
(7) Individual inflow -0.068 0.996 0.272 0.019 0.015 0.966 -0.173 1.012 0.245 0.051 -0.010 0.964 0.105 0.004
(8) Individual outflow -0.105 1.002 0.259 0.056 0.000 0.967 -0.134 1.005 0.257 0.014 0.005 0.965 0.029 0.330
(9) Institutional inflow -0.091 1.020 0.305 0.049 0.011 0.961 -0.128 1.016 0.244 0.036 -0.015 0.967 0.037 0.364
(10) Institutional outflow -0.123 1.025 0.298 0.045 0.004 0.966 -0.096 1.011 0.252 0.040 -0.009 0.965 -0.027 0.343
45
Table VII (continued)
Performance of Funds Sorted by Money Flows: Portfolio Regression Approach
Panel B: Funds are weighted by the inverse of estimation period residual variance
High money flow funds Low money flow funds
Flow variable Alpha MKT SMB HML WML R2 Alpha MKT SMB HML WML R2 Alpha difference
(1) Implied flow -0.154 0.985 0.131 0.041 0.002 0.972 -0.167 1.003 0.139 0.060 -0.018 0.969 0.013 0.611
(2) Net aggregate flow -0.126 0.983 0.133 0.039 -0.003 0.973 -0.195 1.005 0.138 0.064 -0.013 0.969 0.069 0.003
(3) Aggregate inflow -0.135 0.982 0.136 0.054 -0.004 0.973 -0.184 1.004 0.134 0.048 -0.012 0.968 0.049 0.062
(4) Aggregate outflow -0.154 0.994 0.137 0.070 -0.013 0.970 -0.169 0.994 0.133 0.032 -0.003 0.971 0.015 0.524
(5) Net individual flow -0.139 0.988 0.124 0.017 -0.001 0.975 -0.191 1.001 0.135 0.074 -0.013 0.968 0.052 0.052
(6) Net institutional flow -0.123 1.000 0.156 0.032 -0.005 0.969 -0.164 1.013 0.147 0.059 -0.019 0.969 0.041 0.199
(7) Individual inflow -0.141 0.980 0.126 0.059 -0.002 0.974 -0.185 1.006 0.129 0.035 -0.011 0.969 0.044 0.080
(8) Individual outflow -0.154 0.989 0.121 0.080 -0.013 0.972 -0.174 0.998 0.134 0.014 0.000 0.971 0.019 0.436
(9) Institutional inflow -0.147 1.005 0.174 0.045 -0.006 0.968 -0.141 1.007 0.128 0.044 -0.019 0.970 -0.006 0.853
(10) Institutional outflow -0.141 1.013 0.174 0.056 -0.011 0.970 -0.151 1.000 0.126 0.033 -0.013 0.970 0.010 0.680
46
Table VIII
Conditional Performance Evaluation of High and Low Money Flow Fund Portfolios This table shows the performance of actively managed UK equity mutual funds classified on the basis of their normalized money flows in the preceding month. Fund flow data are for 1992-2000. Flows are classified by source as originating from individual investors or from institutional investors; we additionally calculate aggregate flows (individual and institutional flows combined). Flows are also classified by direction as inflows (sales to investors) or outflows (repurchases from investors); we additionally calculate net flows (inflows less outflows). “Implied flow” is obtained as fund size at the end of a month, less the product of the fund’s size at the start of the month and its total return during the month. We normalize each flow measure by dividing it by fund NAV at the start of the month. “High money flow funds” refers to a portfolio of funds in the top 50 percent of all funds each month, according to the stated normalized flow measure. “Low money flow funds” refers to a portfolio of the remaining funds. In Panel A, funds are equally weighted, and in Panel B, funds are weighted by the inverse of their estimation-period residual variance. Portfolios are rebalanced monthly. Fund portfolios are characterized on the basis of portfolio-level regression results. Specifically, each row of the table reports the results of a conditional Carhart (1997) performance evaluation model. Specifically, we regress portfolio excess returns on the excess market return (MKT), the size factor (SMB), the value factor (HML), and the momentum factor (WML) for the UK stock market, as well as the products of these four factors and de-meaned FTA dividend yield, and the products of the four factors and the de-meaned 90-day Treasury bill rate. Coefficient estimates for the product terms are not shown. “Alpha” is the regression intercept, and is followed by its p-value. The last two columns show the difference between each portfolio’s alpha and the alpha of the value-weighted portfolio, followed by the p-value for the hypothesis that the difference is zero.
Panel A: Funds are equally weighted
High money flow funds Low money flow funds
Flow variable Alpha MKT SMB HML WML R2 Alpha MKT SMB HML WML R2 Alpha difference
(1) Implied flow -0.098 1.003 0.265 0.024 0.017 0.968 -0.206 1.025 0.277 0.061 -0.004 0.967 0.108 0.004
(2) Net aggregate flow -0.104 1.004 0.267 0.020 0.016 0.968 -0.201 1.024 0.274 0.065 -0.003 0.967 0.097 0.009
(3) Aggregate inflow -0.089 1.002 0.278 0.023 0.016 0.968 -0.215 1.026 0.263 0.062 -0.002 0.967 0.126 0.001
(4) Aggregate outflow -0.142 1.014 0.282 0.051 0.005 0.969 -0.162 1.014 0.260 0.034 0.008 0.967 0.020 0.516
(5) Net individual flow -0.110 1.002 0.262 0.004 0.017 0.970 -0.193 1.026 0.273 0.073 -0.003 0.965 0.084 0.041
(6) Net institutional flow -0.113 1.030 0.301 0.030 0.017 0.964 -0.175 1.028 0.269 0.067 -0.012 0.969 0.062 0.140
(7) Individual inflow -0.087 1.001 0.276 0.021 0.017 0.968 -0.216 1.027 0.259 0.055 -0.003 0.967 0.129 0.001
(8) Individual outflow -0.139 1.012 0.268 0.056 0.005 0.970 -0.163 1.016 0.267 0.020 0.009 0.966 0.025 0.455
(9) Institutional inflow -0.136 1.032 0.316 0.058 0.016 0.963 -0.151 1.026 0.255 0.038 -0.010 0.969 0.015 0.735
(10) Institutional outflow -0.151 1.037 0.307 0.048 0.008 0.968 -0.136 1.022 0.263 0.048 -0.002 0.967 -0.016 0.611
47
Table VIII (continued)
Conditional Performance Evaluation of High and Low Flow Fund Portfolios
Panel B: Funds are weighted by the inverse of estimation period residual variance
High money flow funds Low money flow funds
Flow variable Alpha MKT SMB HML WML R2 Alpha MKT SMB HML WML R2 Alpha difference
(1) Implied flow -0.162 0.991 0.136 0.038 0.003 0.973 -0.190 1.015 0.148 0.055 -0.014 0.971 0.028 0.290
(2) Net aggregate flow -0.142 0.991 0.140 0.036 -0.001 0.974 -0.210 1.015 0.145 0.059 -0.011 0.970 0.068 0.007
(3) Aggregate inflow -0.141 0.987 0.138 0.048 -0.003 0.974 -0.208 1.017 0.146 0.046 -0.008 0.970 0.067 0.010
(4) Aggregate outflow -0.167 1.003 0.143 0.059 -0.011 0.972 -0.186 1.002 0.141 0.035 0.000 0.972 0.019 0.462
(5) Net individual flow -0.147 0.993 0.127 0.017 0.001 0.975 -0.209 1.012 0.144 0.067 -0.010 0.970 0.062 0.029
(6) Net institutional flow -0.150 1.011 0.164 0.032 -0.001 0.970 -0.169 1.020 0.152 0.051 -0.018 0.971 0.018 0.608
(7) Individual inflow -0.149 0.985 0.129 0.054 -0.002 0.975 -0.204 1.017 0.139 0.031 -0.008 0.971 0.054 0.037
(8) Individual outflow -0.165 0.998 0.126 0.068 -0.012 0.973 -0.189 1.006 0.142 0.016 0.002 0.972 0.025 0.351
(9) Institutional inflow -0.177 1.016 0.182 0.045 -0.002 0.969 -0.147 1.015 0.135 0.036 -0.017 0.972 -0.031 0.391
(10) Institutional outflow -0.164 1.024 0.183 0.054 -0.008 0.971 -0.163 1.007 0.131 0.027 -0.011 0.971 -0.001 0.975
48
Table IX
Performance of High and Low Flow Funds Sorted by Fund Size
This table shows the performance of actively managed UK equity mutual funds classified on the basis of their normalized money flows in the preceding month. Panel A presents results for small funds (i.e. those whose value is below the median fund value in a given month), and Panel B presents results for large funds (i.e. those whose value is above the median). Fund flow data are for 1992-2000. Flows are classified by source as originating from individual investors or from institutional investors; we additionally calculate aggregate flows (individual and institutional flows combined). Flows are also classified by direction as inflows (sales to investors) or outflows (repurchases from investors); we additionally calculate net flows (inflows less outflows). “Implied flow” is obtained as fund NAV at the end of a month, less the product of the fund’s NAV at the start of the month and its total return during the month. We normalize each flow measure by dividing it by fund NAV at the start of the month. “High money flow funds” refers to an equally-weighted portfolio of funds in the top 50 percent of all funds each month, according to the stated normalized flow measure. “Low money flow funds” refers to an equally-weighted portfolio of the remaining funds. Portfolios are rebalanced monthly. Fund portfolios are characterized on the basis of fund-level regression results. Specifically, for each fund-month, we run a Carhart (1997) time-series regression over the preceding 36 months of excess fund returns on the excess market return (MKT), the size factor (SMB), the value factor (HML), and the momentum factor (WML) for the UK stock market. We require a minimum of 30 return observations. The fund alpha is obtained as the fund excess return less the sum of the products of each of the four factor realizations and the corresponding factor loadings. For each month, we then calculate the portfolio alpha, the factor loadings, and the average R2 as the simple average of these measures for the funds comprising the portfolio. The table reports time-series averages of these quantities. The last two rows show the difference between the average alphas of the high- and low-flow portfolios, followed by the p-value for the hypothesis that the difference is zero.
Panel A: Small funds
High money flow funds Low money flow funds
Flow variable Alpha MKT SMB HML WML R2 Alpha MKT SMB HML WML R2 Alpha difference
(1) Implied flow -0.041 0.977 0.303 0.020 -0.030 0.869 -0.127 0.982 0.302 0.057 -0.031 0.867 0.086 0.051
(2) Net aggregate flow -0.047 0.977 0.297 0.020 -0.027 0.872 -0.117 0.982 0.306 0.055 -0.033 0.865 0.070 0.063
(3) Aggregate inflow -0.027 0.979 0.295 0.026 -0.027 0.869 -0.142 0.981 0.308 0.053 -0.033 0.867 0.115 0.009
(4) Aggregate outflow -0.079 0.987 0.306 0.050 -0.031 0.865 -0.101 0.972 0.298 0.029 -0.029 0.871 0.022 0.559
(5) Net individual flow -0.054 0.979 0.302 0.017 -0.025 0.870 -0.112 0.983 0.318 0.053 -0.034 0.863 0.058 0.160
(6) Net institutional flow -0.018 0.985 0.305 0.014 -0.029 0.887 -0.116 0.986 0.289 0.063 -0.027 0.884 0.098 0.076
(7) Individual inflow -0.027 0.980 0.301 0.027 -0.026 0.866 -0.145 0.982 0.319 0.049 -0.034 0.866 0.118 0.012
(8) Individual outflow -0.087 0.986 0.310 0.045 -0.030 0.863 -0.088 0.976 0.314 0.028 -0.030 0.871 0.001 0.985
(9) Institutional inflow -0.028 0.986 0.318 0.036 -0.025 0.885 -0.106 0.986 0.279 0.042 -0.030 0.886 0.078 0.132
(10) Institutional outflow -0.055 0.993 0.310 0.056 -0.019 0.882 -0.085 0.979 0.285 0.020 -0.038 0.891 0.030 0.569
49
Table IX (continued)
Performance of High and Low Flow Funds Sorted by Fund Size
Panel B: Large funds
High money flow funds Low money flow funds
Flow variable Alpha MKT SMB HML WML R2 Alpha MKT SMB HML WML R2 Alpha difference
(1) Implied flow -0.014 0.986 0.296 0.007 0.025 0.960 -0.125 0.998 0.282 0.056 -0.010 0.952 0.110 0.030
(2) Net aggregate flow -0.023 0.986 0.283 -0.005 0.016 0.960 -0.116 0.997 0.294 0.070 0.001 0.953 0.093 0.033
(3) Aggregate inflow 0.017 0.987 0.294 -0.005 0.016 0.957 -0.157 0.998 0.285 0.065 0.000 0.956 0.174 0.000
(4) Aggregate outflow -0.040 0.993 0.297 0.043 0.009 0.957 -0.093 0.989 0.281 0.016 0.006 0.958 0.053 0.172
(5) Net individual flow -0.043 0.989 0.295 -0.013 0.020 0.962 -0.089 0.995 0.301 0.070 0.001 0.947 0.046 0.382
(6) Net institutional flow 0.035 1.005 0.283 -0.049 0.018 0.950 -0.124 1.004 0.276 0.073 0.017 0.942 0.159 0.009
(7) Individual inflow 0.020 0.986 0.309 -0.001 0.018 0.957 -0.164 1.001 0.287 0.061 0.002 0.954 0.184 0.000
(8) Individual outflow -0.040 0.992 0.300 0.044 0.006 0.955 -0.095 0.994 0.298 0.010 0.015 0.957 0.055 0.199
(9) Institutional inflow 0.006 1.007 0.307 0.008 0.026 0.946 -0.081 1.002 0.250 -0.015 0.007 0.946 0.087 0.174
(10) Institutional outflow -0.033 1.022 0.315 -0.009 0.042 0.938 -0.024 0.990 0.248 0.001 -0.004 0.955 -0.009 0.874
50
Table X
Regression of Fund Performance on Money Flows and Fund Attributes
In this table, performance of actively managed UK equity mutual funds is regressed on previous month’s money flows and other fund attributes. Data are pooled across funds and months. Fund performance for each month is measured using the Carhart (1997) model. Specifically, for each fund-month, we run a Carhart (1997) time-series regression over the preceding 36 months of excess fund returns on the excess market return (MKT), the size factor (SMB), the value factor (HML), and the momentum factor (WML) for the UK stock market. We require a minimum of 30 return observations. The fund alpha is obtained as the fund excess return less the sum of the products of each of the four factor realizations and the corresponding factor loadings. Net aggregate flows are expressed as proportion of fund NAV at the start of the month. Log(Size) is the logarithm of fund NAV, and prior year’s average alpha is based on the Carhart (1997) four-factor model. All variables are measured as differences from each month’s cross-sectional average. Results are based on least absolute deviation regressions.
Independent variable
Intercept -0.025 a (0.009) -0.020 a (0.009) -0.018 a (0.008)
Net aggregate flow 2.183 a (0.364) 1.722 a (0.356) 2.005 a (0.365)
Log(Size) -0.015 a (0.005) -0.014 a (0.005)
Net aggregate flow * Log(Size) 0.300 (0.184)
Prior year's average alpha 0.187 a (0.017) 0.183 a (0.016)
Number of observations 27722 26243 26243
Pseudo R-squared 0.001 0.003 0.004
(1) (2) (3)
51
Table XI
Performance of High and Low Money Flow Funds up to 12 Months Ahead
This table shows the performance of actively managed UK equity mutual funds classified on the basis of their normalized net aggregate money flows lagged from one to twelve months. Fund flow data are for 1992-2000. We normalize flows by dividing them by fund NAV at the start of the month. “High money flow funds” refers to a portfolio of funds in the top 50 percent of all funds each month, according to our flow measure. “Low money flow funds” refers to a portfolio of the remaining funds. Panel A is based on “implied” flows, and Panel B is based on actual flows. Implied flows are obtained as fund NAV at the end of a month, less the product of the fund’s NAV at the start of the month and its total return during the month. Actual flows are obtained directly from our dataset. Fund portfolios are characterized on the basis of fund-level regression results. Specifically, for each fund-month, we run a Carhart (1997) time-series regression over the preceding 36 months of excess fund returns on the excess market return (MKT), the size factor (SMB), the value factor (HML), and the momentum factor (WML) for the UK stock market. We require a minimum of 30 return observations. The fund alpha is obtained as the fund excess return less the sum of the products of each of the four factor realizations and the corresponding factor loadings. Each panel shows the time-series average of average alphas for high-flow and low-flow funds, followed by the difference between the two quantities and (in italics) the p-value for the hypothesis that the difference is zero.
Panel A: Implied net aggregate flows Panel B: Actual net aggregate flows
Performance measure High Low Difference High Low Difference
1 month ahead -0.083 -0.153 0.070 0.007 -0.084 -0.152 0.068 0.005
2 months ahead -0.109 -0.131 0.022 0.346 -0.107 -0.133 0.026 0.266
3 months ahead -0.105 -0.159 0.054 0.030 -0.100 -0.163 0.062 0.009
4 months ahead -0.129 -0.168 0.039 0.163 -0.128 -0.169 0.041 0.129
5 months ahead -0.148 -0.133 -0.015 0.570 -0.132 -0.147 0.015 0.541
6 months ahead -0.165 -0.159 -0.007 0.763 -0.153 -0.170 0.017 0.477
7 months ahead -0.168 -0.167 0.000 0.988 -0.142 -0.191 0.049 0.044
8 months ahead -0.160 -0.162 0.002 0.932 -0.168 -0.154 -0.015 0.544
9 months ahead -0.180 -0.171 -0.009 0.730 -0.162 -0.188 0.027 0.366
10 months ahead -0.176 -0.155 -0.021 0.497 -0.169 -0.161 -0.008 0.817
11 months ahead -0.174 -0.168 -0.005 0.866 -0.168 -0.174 0.005 0.876
12 months ahead -0.168 -0.186 0.018 0.586 -0.155 -0.199 0.044 0.178