Asset Management and

Affiliated Analysts Forecasts

Paul Irvine, Paul J. Simko, and Siva Nathan
The interaction between the various departments of a full-service brokerage
firm can have positive effects. An asset management department operates
the firms mutual funds and has a need for high-quality information. Thus,
the firms sell-side analysts in the research department are motivated to
gather information about particular securities and to use the asset
management department as an additional information source. The outcome
can be to raise the quality of sell-side analysts earnings forecasts. In the
study reported here, affiliated analysts earnings forecasts for a security
became significantly more accurate as the fund familys percentage
ownership of that security increased. Also, the study found a significant
positive relationship between the accuracy of an affiliated analyst before the
new investment and the percentage of the companys shares acquired by
that fund family. The implication is that fund families make the largest
purchases of stocks for which their affiliated analysts provided the most
accurate forecasts ex ante. These results have implications for investors
who are using analysts earnings forecasts.

F
ull-service brokerage firms typically
engage in investment banking, broker/
dealer services, and fundamental research.
The activities of brokerage firms research
departments are ostensibly independent of other
operations, but the evidence is clear that relation-
ships exist among departments. The influence of
lucrative corporate finance relationships on sell-
side analysts opinions has been particularly well
documented in both academic studies and the busi-
ness press (Dugar and Nathan 1995; Lowenstein
1996; Lin and McNichols 1998; Laderman 1998;
Michaely and Womack 1999). Recent lawsuits in
Massachusetts (Craig 2002) and New York (Beck
2002; Silverman 2002) support the notion that
relationships between sell-side research and
other brokerage firm departments adversely influ-
ence research quality. Although the business press
recognizes that investment banking is not the only
potential source of conflict for sell-side analysts
(Sidel and Craig 2002; Opdyke 2002), commentary
Paul Irvine is assistant professor of finance at the Uni-
versity of Georgia, Athens. Paul J. Simko is Robert
Vandell Research Associate Professor at the Darden
Graduate School of Business, University of Virginia,
Charlottesville. Siva Nathan is associate professor of
accounting at Georgia State University, Atlanta.
related to the investment banking litigation has
been particularly negative:
When internal communications reveal Merrill
Lynch research analysts disparaging the very
stocks they are recommending to millions of
hard-working families seeking guidance on
how to invest their life savings and their chil-
drens college funds, we must ask: do their
investment recommendations serve the inves-
tors to whom they owe a duty of loyalty and
honesty or their investment banking clients.
(Spitzer 2002)
What is frequently not made clear in this
debate is that positive externalities (positive effects)
can come from bundling different functions within
a full-service broker. An example is in-house asset
management, which produces considerable
demand for high-quality financial information
within the brokerage firm. Asset management has
been one of the fastest growing areas in financial
service, with many brokerage firms participating
through the development of mutual fundbased
asset management departments.1 To meet these
departments demands for information, the buy
side can generate information internallyfrom its
own buy-side analysts and portfolio managersor
obtain it externally from sell-side analysts. This
situation raises several questions. For instance, to
what extent does the buy side communicate with

May/June 2004 67
Financial Analysts Journal
its own brokers affiliated sell-side research depart-
ments? Does the potential for pooling information
from the research and asset management depart-
ments have meaningful implications for sell-side
analysts forecasts or for the actions of their affili-
ated asset management departments?
Although communication between affiliated
buy-side and sell-side analysts is not explicitly pro-
hibited in the Investment Company Act of 1940, we
found that practitioners were divided on whether
the buy side and the sell side within the same firm
share information. Conversations with buy-side
and sell-side analysts at several full-service broker-
age firms revealed polarized opinions. Some
claimed that such communication was rare or non-
existent; others suggested that it is a common activ-
ity on the Street. To determine whether the firms
themselves had made any explicit references to
intrafirm communication, we examined N-SAR
item 20, which lists the top 10 soft dollar brokers for
each fund. The N-SAR filings did not provide a
definitive answer. For example, the Dean Witter
Mid-Cap Growth Funds January 1995 N-SAR lists
Dean Witter as the primary supplier of soft dollar
services but the Smith Barney Aggressive Growth
Fund April 1995 N-SAR does not include Smith
Barney among the funds top-10 soft dollar brokers.
Our conclusion was that whether intrafirm com-
munication has any important effects is essentially
an open empirical question.
The study reported here evaluated this issue
by examining the association between two vari-
ables that capture the information environments of
asset management and researchthe asset man-
agement departments level of stock ownership
and the earnings forecast accuracy of affiliated sell-
side analysts.
In contrast to the generally negative commen-
tary surrounding the influence of the corporate
finance department on sell-side research, we
hypothesized that bundling asset management and
research might be benign or even beneficial for the
quality of both the sell-side analysts forecasts and
the fund managers investments. These benefits
could arise in two ways. First, fund managers might
closely examine stocks for which their affiliated sell-
side analysts have particular expertise relative to
other analysts. Such expertisearising from analyst
talent, experience, or effort or from differences in
resources among brokerage firmswould provide
a compelling reason for asset managers to follow
these analysts views. Second, a managers fidu-
ciary and performance incentives motivate them to
closely monitor stocks that are particularly impor-
tant to the performance of the managers portfolio.
Given sell-side analysts incentives to gather infor-
68
mation about the stocks they cover, if the asset
management department of the analysts own firm
owns a large position in a stock, then that depart-
ment is a clear source of information on that stock.
Affiliated sell-side analysts would thus also have
incentives to increase their focus on securities that
constitute a significant share of the fund managers
investments (Sidel and Craig).
We predicted that if a brokerage firms sell-side
analysts gather high-quality information about
stocks owned by their affiliated asset management
departments, either directly from their own exper-
tise or indirectly from buy-side analysts and port-
folio managers, then that affiliated analysts
earnings forecasts on these stocks will be more
accurate than those of a control group of unaffili-
ated analysts. Furthermore, we hypothesized that
affiliated analysts forecast accuracy, relative to that
of unaffiliated analysts, increases in relation to the
significance of the funds investment.
To test our predictions, we studied new secu-
rity purchases by the mutual fund families of 17
full-service brokerage firms in the years 19942001.
We examined purchases because we posited that
they would provide an unambiguous positive sig-
nal about the information environment within the
fund family by showing that the portfolio manag-
ers were sufficiently confident in these stocks to
add them to the fund familys portfolios.
Data and Methodology
Our empirical analysis required data of two types
detailed portfolio data for mutual funds stock
holdings and affiliated and unaffiliated sell-side
analysts earnings forecasts for these same securi-
ties. The objective of our sampling procedure was
to identify a comprehensive sample of securities
with data available from both sources.
We first identified a sample of full-service bro-
kerage firms with both a research and a mutual
fundbased asset management department. Infor-
mation on these firms portfolio compositions was
then obtained from Morningstars OnDisc database.
Analysts earnings forecast data were obtained from
I/B/E/S.
Identification of Mutual Fund Family. From
the 199495 Nelsons Directory of Investment
Research, we identified 310 U.S.-based brokerage
firms that provided security research services in
the sample period. We then matched these firms to
the mutual fund ownership information in the
Morningstar Report on Equity Mutual Funds for
19911995. Thirty-five brokerage firms had fund
data available in Morningstar, and seventeen of
2004, AIMR

these firms also had earnings forecasts available on
I/B/E/S for the study period. These 17 firms con-
stitute our sample of mutual fund families owned
by full-service brokerages.2
Morningstar collects mutual fund information
from the N-SAR form mandated by the U.S. SEC.
The semiannual N-SAR filing includes information
on trades, 12b-1 fees, portfolio turnover rates, sales
charges, and selected financial information. Every
six months, in April and October, Morningstar sum-
marizes the mutual fund stock ownership data from
the N-SAR forms in its OnDisc database. We used
the N-SAR filing in lieu of other SEC-mandated
forms because it has the advantage of being the most
inclusive and comprehensive filing available.3 Our
sample of mutual fund stock holdings was gathered
from the October 1994 through October 2001
OnDisc reporting periods (i.e., 15 total periods). We
ended the Morningstar data collection in 2001
because, as discussed later, to measure analyst accu-
racy, we needed to allow significant time for com-
panies to report actual EPS.
For each OnDisc reporting period, we deter-
mined the individual funds managed by the 17
full-service brokers in our sample. For these funds,
we collected the net asset value (NAV), identified
the individual securities held by each fund, and
calculated the market value of each equity security
held by the fund. We excluded from our analysis
any fund that did not hold U.S. equities (e.g., a
bond fund).
Identification of Fund Familys Security
Investments. We focused on stock ownership at
the fund-family level, as opposed to the individual-
fund level, because the fund-family level best
reveals the aggregate importance of the ownership
position to the asset management department. Fur-
thermore, we used only purchases of new securities
because we expected them to represent an unam-
biguous positive signal from the fund manager.
For each security, we computed the NAV
weight of the holding as a percentage of all fund
holdings (NAVWGHT) and the percentage of stock
owned by the fund family (%OWN). Specifically,
NAVWGHT for a fund familys portfolio was mea-
sured (for each semiannual Morningstar report
period t) as the summed market value of security i
across all k funds, f, for each fund family F. Scaling
by the fund familys total NAV yielded the fund-
family allocation attributable to security i:4
kt
f=1 MV i,f,F,t
NAVWGHT i,F,t = ------------------------------------- 100,
f=1 kt
NAV f,F,t
May/June 2004
Asset Management and Affiliated Analysts Forecasts
where MV is market value, defined as price per
common share multiplied by total common shares
outstanding.5
The variable %OWN was calculated as the total
common shares held by the fund family as of the
end of the OnDisc reporting period divided by the
total common shares outstanding:
f=1
k t Shares owned i,f,F,t
%OWN i,F,t = ---------------------------------------------------------------- 100. (2)
Shares outstanding
i,t
Ideally, we would like to have identified the
specific date on which the investment by a fund
actually took place. Unfortunately, pinpointing this
date was not possible because funds report only
semiannual holdings and funds within a family
have different semiannual report dates. For
instance, for a fund whose semiannual reporting
period ends 30 November, an actual trade inferred
from comparing the 31 May and 30 November port-
folio holdings may have occurred as early as 1 June
or as late as 30 November.
Because information can flow from the analyst
to the portfolio manager or from the portfolio man-
ager to the analyst, we examined earnings forecasts
both before and after the dates at which we could
unequivocally state that ownership existed. (In the
aforementioned example, the dates would be 1 June
and 30 November.)
Table 1 presents descriptive data for selected
characteristics of the 17 fund families in the sample.
For each fund family, Table 1 presents the average
number of funds and net assets under management
over all Morningstar reporting periods. (All fund
families were represented in multiple periods, but
only the large families had the requisite data for the
full 15 reporting periods.) For example, Merrill
Lynch is the largest fund family in the sample, with
an average $35 billion under management in an
average of 74 funds in the sample period. The size
and total investments for these families vary con-
siderably. The last columns of Table 1 report the
composition of the sample used in the empirical
tests. Specifically, we gathered analyst forecast
data from I/B/E/S for periods before and after a
fund familys new security purchase. The larger
fund families tend to have more observations in the
final sample, but no single family dominates.
An interesting conclusion that can be readily
inferred from the data in the last columns is that
sell-side analysts have a tendency to initiate cov-
erage of a stock after purchase of the stock by the
affiliated portfolio manager. This conclusion is
reflected in a 57 percent increase in the after-
(1)
purchase sample size from the before-purchase
sample size.6
69
Financial Analysts Journal

Table 1. Fund Family Characteristics and Sample Composition, October 1994October 2001
Final Sample: New Investments with
I/B/E/S Forecast Data Available
Average No. Average NAV
Fund Family of Funds (millions) Before Fund Purchase After Fund Purchase
No. of % of No. of % of
Obs. Securities Held Obs. Securities Held
Blair William 3.6 $ 739 85 5 147 9

Burnham Group 2.4 145 19 3 23 3

Dean Witter 34.4 28,853 148 2 189 2

Goldman Sachs 17.2 4,971 566 7 992 12

Kemper 42.4 14,469 79 1 97 1

Legg Mason 8.3 8,727 44 1 74 1

Merrill Lynch 73.5 35,152 994 4 1,565 7

Montgomery 14.1 2,724 268 6 444 10

Morgan Keegan 1.0 56 43 6 56 8

Morgan Stanley 52.6 29,247 856 4 1,459 7

Oppenheimer 49.3 32,589 494 3 807 5

PaineWebber 23.2 3,294 503 3 680 5

Piper Jaffray 6.3 760 47 2 66 3

Prudential 46.6 14,456 710 4 1,042 6

Robertson Stephens 10.2 2,012 223 8 311 12

Salomon Brothers 7.9 1,037 97 4 126 5

Smith Barney 38.8 14,698 727 6 1,189 10

Total 5,903 9,267

Note: For Morningstar OnDisc semiannual reporting periods.

Estimation of Analysts Relative Accuracy.
We gathered earnings forecasts and EPS data from
I/B/E/S. To be included in the sample, a security
had to (1) be actively followed by at least three
unaffiliated analysts during the quarter immedi-
ately before or immediately after the end of the
OnDisc reporting period, (2) have two-year-ahead
earnings forecasts available for both the affiliated
analyst and a control group of unaffiliated analysts,
and (3) have actual EPS data available for each
earnings forecast. We imposed these criteria
because our tests relied on measures of relative
forecast accuracy. The benchmark against which
we evaluated the level of an affiliated analysts
earnings forecast was a control sample of analysts
without direct affiliation with the fund family.
The first criterion (at least three unaffiliated
analysts) provides a reasonable comparison by mit-
igating the influence of outliers that may arise
solely because a security is infrequently followed.7
The second sampling criterion (two-year-ahead
earnings forecasts) arose from the nature of the
presumed investment horizon of mutual fund
managers. Given restrictions on short-term trading
activities during our study period, we reasonably
assumed that the mutual fund managers private
information and investment horizon probably per-
tained to a time period beyond the current fiscal
year of the company.8 Our final requirement (that
actual EPS data be available) was necessary if we
were to compute relative forecast accuracy. This
criterion restricted our sampling period to no later
than October 2001.
Although our main focus was on the relation-
ship between sell-side analyst accuracy and a fund
familys investments, our initial tests also exam-
ined forecast optimism of in-house analysts relative
to a control group of unaffiliated analysts. McNi-
chols and OBrien (1997) defined earnings forecast
relative optimism as the difference between an
observed forecast and a relevant benchmark. We
adopted this approach by specifically measuring
affiliated analysts relative optimism, OPTM, as

70 2004, AIMR
 Asset Management and Affiliated Analysts Forecasts

AFF
FOREF,i,t FORE F,i,t
OPTM F,i,t = ------------------------------------------------------------------ ,
i,t
AFF
where FORE F,i,t is a two-year-ahead earnings fore-
cast by an affiliated analyst of fund family F for
security i, with date t as the earliest date in the
quarter after the end of the OnDisc reporting period.
Superscript UNAFF indicates the most recent
forecast made by an unaffiliated analyst within 30
days of the affiliated analysts forecast.9 We scaled
the numerator by the standard deviation of all fore-
casts for security i to control for cross-sectional vari-
ation in the dispersion of analysts forecasts.10
Next, we examined relative forecast accuracy,
a measure distinct from optimism, in that accu-
racy captures the quality of forecasts made by the
affiliated analyst whereas optimism reflects any
potential bias in the forecast. We defined relative
accuracy for security i, ACCURi , as the difference
in standardized unaffiliated and affiliated analyst
forecast errors (forecasted EPS less actual EPS)
scaled by their respective standard deviations:
UNAFF
FORE i,t EPS i
ACCURF,i,t = ---------------------------------------------------------- -------------------------------------------------- .
UNAFF
Table 2 summarizes the pooled distributions
of the components of OPTM and ACCUR parti-
tioned by whether the affiliated analyst forecast
was measured before or after the fund purchase.
The mean forecasts of affiliated analysts are gener-
ally comparable to the group of unaffiliated fore-
UNAFF casts (FOREUNAFF), although the affiliated analysts
(3) forecast is higher at both the mean and median.
Both affiliated and unaffiliated analysts forecasts
are, on average, approximately 20 percent larger
than actual reported EPS, indicating that the previ-
ously documented phenomenon of forecast opti-
mism exists in all our sample analysts. Table 2 also
summarizes the NAV weight of the stock holding
as a percentage of all fund holdings, the percentage
of stock owned by the fund family, the market value
of common equity (in millions), and number of
analysts following the stock (NUMEST).11 The dis-
tribution of NAVWGHT, because its median is less
than 1/10 of 1 percent, shows that managers are not
committing a significant portion of fund assets to
new investments.
Table 3 provides descriptive evidence on the
distribution of relative optimism and relative accu-
racy partitioned by whether the forecast occurred
before or after fund purchase. We did not find
widespread evidence that the distributions of affil-
iated analyst forecasts are significantly different
AFF
from those of unaffiliated analysts; relative opti-
FORE i,t EPS i
(4) mism and relative accuracy before and after fund
AFF purchase have medians of zero. A difference-in-
means test indicated that the optimism of affiliated
analyst forecasts was significantly higher than the
optimism of unaffiliated analysts before (t = 3.74)
and after (t = 1.86) the new investment, but these
results were not confirmed for the period after the
investment by a nonparametric sign-rank test.

Table 2. Distribution Characteristics for Components of Various Variables,

October 1994October 2001
Standard
Variable Mean Deviation 90th Percentile Median 10th Percentile
Before fund purchase (n = 5,903)
FOREAFF ($) 1.650 1.215 3.200 1.480 0.370
FOREUNAFF ($) 1.630 1.206 3.169 1.470 0.370
($) 0.144 0.208 0.330 0.077 0.018

After fund purchase (n = 9,267)

FOREAFF ($) 1.642 1.199 3.800 1.520 0.350
FOREUNAFF ($) 1.639 1.190 3.800 1.520 0.355
($) 0.143 0.205 0.322 0.077 0.019
NAVWGHT (%) 0.171 0.464 0.415 0.038 0.001
NUMEST 12.8 8.6 24 11 4
EPS ($) 1.237 1.336 2.960 1.190 0.220
%OWN 0.370 1.108 0.905 0.088 0.003
MV ($ millions) 14,134.0 32,698.9 33,063.3 3,830.0 526.2
Notes: Variables measured before fund purchase relate I/B/E/S analyst forecast data gathered during
the quarter preceding the fund familys N-SAR reporting date. Variables measured after fund purchase
are for those measured as of the end of the reporting period or, in the case of forecast data, during the
quarter immediately after the reporting period. NUMEST = number of analysts following the security.

May/June 2004 71
Financial Analysts Journal
Table 3. Affiliated Analyst Relative Earnings Forecast Optimism and
Accuracy, October 1994October 2001
Variable N Positive Negative
Before fund purchase
OPTM 5,903 46.8% 44.5%
ACCUR 5,903 46.7 44.5
After fund purchase
OPTM 9,267 45.0% 44.7%
ACCUR 9,267 44.4 45.1
aDenotes a positive sign rank for the distribution.
b
Denotes a negative sign rank.
Note: p-values below 0.05 indicate significance at the 5 percent level.
Consistent with the descriptive evidence in
Table 2, Table 3 shows that relative accuracy dete-
riorated slightly after the fund purchased the stock;
affiliated analysts initiating coverage after the
funds purchase forecasted somewhat worse, on
average, than affiliated analysts who covered the
firm prior to the funds purchase. As a whole, at this
full-sample univariate level, the only inference that
is consistent from parametric and nonparametric
tests at conventional significance levels is that affil-
iated analysts are more optimistic than unaffiliated
analysts before fund purchase.
In the analysis of the tests that follow, we
explore whether conditioning on the analyst fund
familys percentage ownership of the stock affects
the relative accuracy of affiliated analysts forecasts.
Ownership Level and Relative
Forecast Accuracy
We report our main results in two stages. We first
examine the cross-sectional association between
the relative accuracy of affiliated analysts forecasts
and the fund familys ownership level in the stock.
We then estimate multivariate regression models
that control for other factors that could affect the
relationship between accuracy and ownership.
After establishing these main results, we provide a
summary of robustness checks.
Relative Forecast Optimism and Accuracy
Conditional on Ownership. Our primary empiri-
cal tests examined whether affiliated analysts
forecasts exhibit a systematic association with the
fund familys percentage ownership of a stock
(%OWN). Our preliminary examination of this
issue involved the use of portfolios. We formed 10
portfolios by ranking all observations based on the
values of %OWN, then examined the coincident
values of affiliated analyst relative accuracy for
each of these deciles.
72
p-Value Standard
Median (sign rank) Mean Deviation t-Statistic
$0.000 0.021a $0.079 $1.622 3.74
0.000 0.600a 0.028 1.500 1.41
$0.000 0.651a $0.030 $1.553 1.86
0.000 0.072b 0.039 1.449 2.55
Our primary focus was on the quality of affili-
ated analyst output as reflected by relative accu-
racy, but we also included relative optimism for
comparison. In this analysis, relative optimism and
accuracy in the ownership deciles are reported as
the percentage of affiliated forecasts that were
above the samplewide median of the respective
metric. Thus, under the null hypothesis that there
is no relationship between affiliated forecasts and
percentage ownership, relatively optimistic
(%OPTM) and relatively accurate (%ACCUR) fore-
casts in any given decile is an expected 50 percent.
Note that, by construction, when affiliated fore-
casts are more accurate for one portfolio, they must
be, on average, less accurate for another.
Table 4 presents relative accuracy and relative
optimism by ownership decile for forecasts made
after the fund family purchased the stocks. The
results are striking. We found no consistent pattern
between relative optimism and percentage owner-
ship, but the percentage of more-accurate affiliated
analysts forecasts rises dramatically to 61.5 percent
in the 10th decile of security ownership. A likeli-
hood ratio test appropriate for percentages (Greene
1997, p. 886) indicates that the 2 statistic against
the null of no difference in forecast accuracy is 21.50
(p-value < 0.01). Although we found that the higher
the percentage ownership, the more accurate the
affiliated analysts forecast, we found for the entire
sample little systematic relationship between own-
ership and accuracy; we observed no other signifi-
cant test statistics across ownership deciles.
Figure 1 provides a histogram of the results in
Table 4. It is readily apparent that the top decile of
%OWN, where ownership approached 1 percent of
the companies shares outstanding (see Table 2),
yielded the highest percentage of relatively supe-
rior forecasts, with a monotonic increase observ-
able beginning in Decile 8.
2004, AIMR
 Asset Management and Affiliated Analysts Forecasts

Table 4. Percentage of Observations above the Median of Relative Forecast Optimism and
Accuracy for Percentage Ownership Deciles: Each Measured after Fund Family Purchase,
October 1994October 2001
%OWNt Decile
1 2 3 4 5 6 7 8 9 10
Relative accuracy
%ACCURt above median 49.8 51.5 52.2 50.3 48.3 49.9 49.7 52.4 52.8 61.5
No. of observations 930 930 915 930 930 929 920 930 928 925
Likelihood ratio 2 0.004 0.37 0.79 0.01 0.45 0.001 0.01 0.90 1.31 21.50
p-Value 0.95 0.54 0.37 0.91 0.50 0.98 0.91 0.34 0.25 0.00

Relative optimism
%OPTMt above median 50.0 49.6 50.3 51.0 51.5 49.7 51.4 48.6 49.0 50.2
No. of observations 930 930 915 930 930 929 920 930 928 925
Likelihood ratio 2 0.001 0.03 0.01 0.17 0.37 0.02 0.32 0.32 0.15 0.01
p-Value 0.98 0.86 0.91 0.68 0.54 0.90 0.57 0.57 0.70 0.93
Note: -statistic tests each percentage against the population mean of 50 percent; p-values below 0.05 indicate significance at the
2

5 percent level.

Taken together, the results reported in Table 4
are consistent with the overall distribution of opti-
mism and accuracy presented in Table 3: No sam-
plewide differences are apparent between
affiliated and unaffiliated analysts for most deciles.
Untabulated results indicate a similar distribu-
tion in relative accuracy before fund purchase, with
60.9 percent of affiliated analysts more accurate
than their peers in the largest decile of percentage
ownership. Thus, fund managers are making the
largest new investments in the stocks for which
their affiliated analysts appear to have the highest
quality information, as measured by accuracy of
their earnings forecasts before fund purchase.
Multivariate Regression Analysis. T h e
portfolio-based results presented in Table 4 sup-
port our hypothesis that affiliated analyst forecast

Figure 1. Affiliated Analyst Forecast Optimism and Accuracy (after Fund

Purchase) by Ranked Percentage Ownership, October 1994
October 2001
Relative to Median
0.65

0.60

0.55

0.50

0.45

0.40
1 2 3 4 5 6 7 8 9 10
Portfolios Ranked by Percentage Ownership

Accuracy Optimism

May/June 2004 73
Financial Analysts Journal
accuracy is related to percentage ownership. The
tests in Table 4 did not control for other factors,
however, that could have an impact on relative
accuracy. To evaluate the robustness of our conclu-
sions, we conducted two regression analyses. The
specification of each model depended on whether
we were examining forecasts made before or after
the new investment. The regressions differ because
we expected that the direction of influence would
differ between the two periods. We expected asset
managers before purchase to be more inclined to
use information from affiliated sell-side analysts
that they believed was relatively more accurate. We
expected affiliated sell-side forecasts after purchase
to benefit from the externalities generated by the
asset management ownership position. Thus, the
models tested for two causal relationships between
ownership level and relative accuracypresented
first for the after-purchase period when relative
accuracy would depend on the asset managers
new investment position and presented second for
the before-purchase period when ownership
would depend on the quality of sell-side analyst
information (measured by the forecast relative
accuracy before fund purchase). We viewed these
models as complementary, in that each was
designed to evaluate the association between fund
ownership and sell-side analyst output.
The following regressions were estimated
(security subscripts omitted):
R ( ACCUR t+1 ) = 0 + 1 R ( %OWN t ) + 2 R ( NAVWGHTt )
+ 3 R ( NUMEST t ) + 4 R ( ASSETS t )
(5)
+ 5 R ( MVt ) + F=116 F FAM F + t ACCURt1, 1, would be positive because asset
and
R ( %OWN t ) = 0 + 1 R ( ACCUR t1 ) + 2 R ( OPTM t1 )
+ 3 R ( NUMEST t ) + 4 R ( ASSETS t )
(6)
+ 5 R ( MVt ) + F=1F FAMF + vt ,
16
where R(X) represents the standardized ranked
value of variable X, t and t are residuals, and 16
fund family fixed-effects coefficients (FAM) were
included to control for any potential effects across
fund families. We specified Equations 5 and 6 as
rank-transform regressions largely to overcome
potential violations of the ordinary least-squares
(OLS) linearity assumption in our data.12 The pri-
mary coefficients of interest are those on %OWNt1
in Equation 5 and ACCURt1 in Equation 6. We
expected %OWN t to be positively related to
ACCURt+1 in Equation 5 and expected lagged accu-
74
racy, ACCURt1, to be positively related to subse-
quent ownership (%OWNt) in Equation 6.
We expected the coefficient on %OWNt in
Equation 5 to be positive because we believed that
the direct and indirect channels of information on
which the accuracy of the affiliated analysts fore-
cast relied would be stronger the greater the per-
centage of the stock owned by the fund. An
alternative hypothesis for the results in Table 4 is
that analysts exert greater effort in forecasting earn-
ings for stocks that are important to their brokers
asset management department. To control for this
alternative, we included NAVWGHT, the net new
investments relative importance to the fund port-
folio. We posited that the higher the NAV weight,
the greater the incentive for the analyst to forecast
accurately; therefore, we expected a positive coef-
ficient on NAVWGHT. Because the dependent vari-
able in the regression is the forecast accuracy of the
affiliated analyst relative to that of an unaffiliated
analyst following the same company, company-
specific variables affecting analyst forecast accu-
racy did not need to be included in the regression
(see Clement 1999). Characteristics explaining
cross-sectional differences in a stocks information
environment, however, could be important con-
trols. For this reason, we included the number of
analysts forecasting earnings, NUMEST, and the
market value of the firm, MV. In addition, we con-
trolled for potential resource differences among
fund families with the variable ASSETS, which rep-
resents total assets under management.
Equation 6 presents a model in which the level
of ownership depends on the sell-side analyst. Our
expectation here was that the coefficient on
managers are, on the margin, more likely to invest
in securities in which the firms sell-side analysts
have revealed their expertise by producing more-
accurate forecasts than their competitors. As con-
trol variables in Equation 6, we included a ranked
OPTM variable to control for the possibility that
fund managers would invest in securities for which
the firms affiliated analysts had higher expecta-
tions than other analysts. We also included
NUMEST, MV, and ASSETS, for control as noted in
the preceding paragraph.
Table 5 and Table 6 report the results from
estimating restricted specifications and full speci-
fications of, respectively, Equations 5 and 6.13 The
results shown in Table 5 reinforce our inferences
from the portfolio analysis reported in Table 4:
Consistent with the general pattern of increasing
accuracy in Table 4, in the restricted regression,
%OWN is positive and significant (1 = 0.038, t =
2.60). In the unrestricted regression, these results
2004, AIMR
 Asset Management and Affiliated Analysts Forecasts

Table 5. Rank Regression of Affiliated Analysts Relative Forecast Accuracy

after Fund Family Purchase on Various Variables, October 1994
October 2001
R(%OWNt) R(NAVWGHTt) R(NUMESTt) R(ASSETSt) R(MVt) F-Statistic
Restricted regression
Coefficient 0.038 0.073 11.63
t-Statistic 2.60 5.04

Unrestricted regression
Coefficient 0.110 0.041 0.042 0.008 0.092 15.15
t-Statistic 4.89 1.64 3.17 4.16 4.77
Note: Total number of observations is 9,267.

Table 6. Rank Regression of Percentage of Security Owned by the Fund

Family before Fund Family Purchase on Various Variables,
October 1994October 2001
R(ACCURt1) R(OPTMt1) R(NUMESTt) R(ASSETSt) R(MVt) F-Statistic
Restricted regression
Coefficient 0.195 0.003 48.84
t-Statistic 15.63 2.83

Unrestricted regression
Coefficient 0.103 0.002 0.004 0.593 0.132 105.19
t-Statistic 8.49 1.56 3.05 32.77 9.28
Note: Total number of observations is 5,903.

are even stronger. In other words, the more of a
particular security the affiliated fund family owns,
the more accurate the affiliated analyst forecasts
are subsequent to the purchase. In the restricted
regression, the slope coefficient on NAVWGHT is
also positive and significant, indicating that larger
stocks in the fund family are associated with accu-
rate forecasts. This result is insignificant, however,
in the unrestricted regression. Thus, the positive
coefficient on NAVWGHT in the restricted
regression appears to be explained by the fact that
larger, heavily covered firms, which also tend to
have a high NAVWGHT, tend to have higher rela-
tive forecast accuracy among their analysts. Also,
all else being equal, larger fund families tend to be
associated with more-accurate affiliated analysts.
In Table 6, we document a significant positive
relationship between ownership and the relative
accuracy of affiliated analysts forecasts before the
fund purchase. The coefficient on relative accuracy
is significantly positive in both the restricted and
unrestricted regressions. This result is notably
strong and potentially indicative that more-accurate
forecasts (i.e., analyst expertise) influence fund
managers to purchase larger positions.
Unlike the results for accuracy, the results for
optimism are not robust to the inclusion of the con-
trol variables. Although significant in the restricted
regression, optimism is insignificantly positive in
the unrestricted regression. Not surprisingly,
because %OWN represents funds committed to a
particular stock, this variable is positively related to
the size of the fund family and negatively related to
market value and the number of analysts covering
the stock.14
Taken as a whole, the results in Tables 4, 5, and
6 support our contention that a positive externality
exists between fund managers investment deci-
sions and affiliated sell-side analysts forecasts.
Fund family percentage ownership of a companys
stock, which measures incentives driving the buy-
sides information collection, is positively related
to the relative accuracy of the fund familys affili-
ated sell-side analysts forecastsboth before and
after the initial fund investment. The relationship
appears to be much stronger, however, when own-
ership is measured subsequent to the forecast. We
infer from the regression results in Tables 5 and 6
that the greater the ownership in the stock and the
relative accuracy of the affiliated analyst, the
greater the overall information quality within the
brokerage firm.

May/June 2004 75
Financial Analysts Journal
Robustness Checks. We performed a num-
ber of (untabulated) robustness checks to ensure
that the reported results are not an artifact of our
research design. Our tests focused on whether rel-
ative forecast accuracy improves as ownership
increases. We examined whether this result holds
for forecast accuracy without netting out the con-
trol group of unaffiliated analysts. Replication of
Equation 5 but with analyst absolute forecast errors
used as the dependent variable yielded results con-
sistent with those reported in Table 5. Other alter-
native specifications were tested, including (1)
computation of OPTM and ACCUR with the con-
sensus forecast as an alternative deflator, (2) using
the consensus forecast rather than a single unaffil-
iated forecast as the control, and (3) including
reporting period intercepts in Equations 5 and 6.
Each alternative research design produced infer-
ences that were qualitatively the same as those
reported here.
Finally, in an analysis of the 199596 subperiod,
we tested whether reputation and investment bank-
ing relationships affected our results. Stickel (1992)
showed that analysts who are members of the Insti-
tutional Investor All-American Research Team sup-
ply more-accurate forecasts than other analysts
who follow the same company. Thus, we included
as a control in the regression a variable, IIAA, if an
affiliated analyst had All-American Research Team
status during the 12-month window surrounding
the forecast date. The slope coefficients on IIAA
were positive, which is consistent with Stickel, but
our results were not statistically significant.
The papers on the influence of investment
banking on analyst activity cited in the introduc-
tion suggested that underwriting activity may bias
analyst forecasts and reduce accuracy. We con-
trolled for investment banking activity by includ-
ing a dummy variable in the regression if the
brokerage firm engaged in any investment banking
activity, including underwriting stock issues and
mergers and acquisition advisory services, on the
stock in question in the year prior to a forecast.
Overall, we found only 7 percent of the forecasts
connected with investment banking activity in the
SDC Platinum database provided by Thomson
Financial. The investment banking dummy vari-
able was not significant in any regression specifica-
tion in the 199596 subperiod.15
Conclusion
We examined the relationship between holdings of
mutual funds operated by full-service brokerage
firms and the relative earnings forecast accuracy of
76
the firms affiliated sell-side analysts. The purpose
was to provide evidence on the interaction between
the asset management side and research depart-
ment of a full-service firm. We measured informa-
tion quality by using relative earnings forecast
accuracy, defined as the difference in absolute earn-
ings forecast errors between affiliated analysts and
a control group of unaffiliated analysts. Relative
accuracy captures differences in expertise from
such sources as talent, experience, or effort or from
differences in resources among brokerage firms.
Using this measure, we tested whether (1) portfolio
managers tend to follow the advice of affiliated
analysts who are relatively more accurate and (2)
whether affiliated analyst forecast accuracy
improves in relation to the percentage ownership of
a security after the fund family invests in the stock.
We found that fund families largest new
investments are significantly associated with affil-
iated analysts forecast accuracy before the fund
purchase decision, with the relationship particu-
larly strong when ownership is measured subse-
quent to the forecast. This finding suggests that
managers can identify affiliated analysts that have
demonstrated expertise in a particular stock and
that, to some degree, they base their new invest-
ment decisions on their beliefs in these analysts
expertise. After the fund familys purchase of a new
stock, affiliated analyst coverage increased and
affiliated analyst forecast accuracy was signifi-
cantly positively related to the percentage of the
stock owned by the fund family.
Our findings contribute to ongoing research
regarding the properties of analysts earnings fore-
casts. We have documented that affiliated analysts
earnings forecasts vary from others in a predictable
fashion and that these differences are related to the
actions of affiliated asset management depart-
ments. In contrast to the research emphasizing the
negative influence of underwriting on analysts
forecasts, our results suggest that positive external-
ities can exist between the different departments of
a full-service brokerage firm.
The recent introduction of Regulation FD
could have a significant impact on our results. On
the one hand, if Regulation FD effectively curtails
private communications between company man-
agers and analysts, affiliated sell-side analysts may
lose any informational advantage they obtain from
communications with affiliated buy-side analysts
or from access to managers that accrues to institu-
tions that own large positions. On the other hand,
because private communications with all, affiliated
and unaffiliated, sell-side analysts could be
restricted by Regulation FD, the presence of affili-
ated buy-side analysts examining the same stock
2004, AIMR

could continue to help affiliated sell-side analysts
to provide more-accurate forecasts. Preliminary
analysis based on our post-FD (August 2000) sam-
ple suggests both effects are present. On the one
hand, the relationship between affiliated forecast
accuracy before the purchase and the size of the
fund purchase is less positive than in the remaining
sample, which suggests that portfolio managers
could be relying less on affiliated analysts since
Regulation FD was instituted. On the other hand,
the relationship between ownership and accuracy
after the fund purchase is more positive in the post-
FD period, which suggests that affiliated analysts
and portfolio managers may be effectively sharing
more information now that both parties access to
management forecasts has been restricted. The
long-run impact of Regulation FD on the relation-
ships we have documented is an interesting topic
for future research.
Notes
1. Throughout this article, the term asset management
refers to mutual fundbased asset management.
2. In the sample period, I/B/E/S stopped collecting broker
codes after mergers (e.g., Dean Witter or Salomon Brothers).
We followed I/B/E/S when identifying active brokers.
When a broker code was dropped by I/B/E/S, we retained
the fund family in our sample but stopped any further
collection of Morningstar data.
3. A limitation of the N-SAR form is that it provides only
semiannual data for a given fund. Less inclusive but more
frequently required by the SEC is Form 13-F, which must
be filed quarterly by any institutional investment manager
having equity assets under management greater than $100
million. In addition, Form 13-D must be filed within 10 days
of any transaction that results in the fund beneficially own-
ing more than 5 percent of the outstanding securities of any
particular stock.
4. The variable k (the number of funds in a family) is time
dependent because families frequently create new funds
and delete existing funds.
5. Price and share data were gathered from CRSP.
6. Although our study involved only new investments, our
predictions should also hold for the divestiture of a secu-
rity by the asset manager. Empirically, we found that when
an asset manager significantly decreases an investment in
a security, about one-third of the affiliated analysts tend to
drop coverage of that security. This finding is consistent
with the results reported by McNichols and OBrien (1997),
who found that when analysts have negative information
about a stock, they tend to drop coverage of that stock
rather than revise their earnings forecast downward or
issue a sell recommendation.
7. Tightening this constraint so that five forecasts were
required did not materially influence the results.
8. During most of our study period, tax law specified that if a
fund desired to remain tax exempt, no more than 30 percent
of the funds net income could involve short-term trading
activity. Violation of this requirement rendered all current-
year income taxable during that year.
May/June 2004
Asset Management and Affiliated Analysts Forecasts
We thank George Benston, Bruce Branson, Larry Brown,
Kirsten Ely, Shehzad Mian, Kumar Sivakumar, Greg
Waymire, Richard Willis, workshop participants at
Emory University, the University of Florida, Georgia
State University, the University of Houston, and Wake
Forest University, and participants at the Southeast
Regional Meeting of the American Accounting Asso-
ciation, the 10th annual Financial Economics and
Accounting meetings, and the Financial Management
Association Meeting for helpful comments. Ron Harris
and Kathryn Epps provided very capable research
assistance. We also thank Thomson Financial for the use
of I/B/E/S earnings forecast data and for releasing the
identity of brokerage firms issuing earnings forecasts.
These data were provided as part of a broad academic
program to encourage earnings expectations research.
The Goizueta Business School provided financial support
for the purchase of Morningstar OnDisc. The Darden
School Foundation is also gratefully acknowledged.
9. Fewer than 3 percent of the unaffiliated analysts were affil-
iated, in terms of purchases of new securities, with other
brokerage firms at some time during the sample period.
Elimination of these observations had no material effect on
any inference we report.
10. For this calculation, we used all active unaffiliated analysts
forecasts released in the 30-day window centered on the
affiliated analysts forecast date. If the same analyst made
multiple forecasts during the 30-day window, we kept in
the distribution the earnings forecast made nearest, in
absolute days, to the affiliated analysts forecast. See
Comiskey, Walkling, and Weeks (1987) and Ajinkya,
Atiase, and Gift (1991) for a discussion of standardized
mean earnings forecasts.
11. The descriptive statistics for all variables for each OnDisc
reporting period are similar to those reported in aggregate.
12. Rank transformation provides the ability to correct any
nonlinearities in the data. Table 4 and Figure 1 show that
the right tail of %OWN contains a clustering of more rela-
tively accurate forecasts. In addition, very small deflators
in the calculations of ACCUR and OPTM would lead to
extreme observations. The estimation procedure used in
rank-transform regressions is to replace the continuous
variables with their corresponding ranks. Specifically,
when there are n observations for a sample variable, a rank
of 1 is assigned to the variables smallest observation and a
rank of n to the largest. Taking logs before the ranking
procedure is not necessary because another monotonic
transformation will not change the assignment of ranks.
Average ranks are used in case of ties. OLS regression is
performed on the ranked data. See Iman and Conover
(1979) for further details.
13. Coefficients for the intercept and fund family fixed effects
were not tabulated. As presented in Tables 5 and 6, we
estimated restricted regressions [i.e., we limited the inde-
pendent variables to R(%OWN) and R(NAVWGHT)] and
unrestricted regressions to ensure that the significant
effects we documented were not caused by a particular
regression specification. For each regression, we also esti-
mated variance inflation factors to evaluate the influence
77
Financial Analysts Journal
of multicolinearity. Although multicolinearity does not
bias reported coefficients, its presence can result in an
incorrect inference of insignificance. We found no undue
influence of colinearity across regressors.
14. We also estimated Equation 6 with NAVWGHT as the
dependent variable (because this variable might capture the
importance of the stock to the fund family better than
References
Ajinkya, Bipin, Rowland Atiase, and Michael Gift. 1991.
Volume of Trading and the Dispersion in Financial Analysts
Earnings Forecasts. Accounting Review, vol. 66, no. 2
(April):389401.
Beck, Rachel. 2002. Changes Are Only One Step towards
Analyst Objectivity. Associated Press report in Newark Star-
Ledger (5 September):47.
Clement, Michael. 1999. Analyst Forecast Accuracy: Do
Ability, Resources and Portfolio Complexity Matter? Journal of
Accounting and Economics, vol. 27, no. 3 (July):285303.
Comiskey, Eugene, Ralph Walkling, and Michael Weeks. 1987.
Dispersion of Expectations and Trading Volume. Journal of
Business Finance and Accounting, vol. 14, no. 3 (Summer):229237.
Craig, Stephen. 2002. Massachusetts Claims CSFB Stock Reports
Led Investors Astray. Wall Street Journal (22 October):C1.
Dugar, Amitabh, and Siva Nathan. 1995. The Effect of
Investment Banking Relationships on Financial Analysts
Earnings Forecasts and Investment Recommendations.
Contemporary Accounting Research, vol. 12, no. 1 (Fall):131160.
El-Gazzar, Samir. 1998. Predisclosure Information and
Institutional Ownership: A Cross-Sectional Examination of
Market Revaluations during Earnings Announcement Periods.
Accounting Review, vol. 73, no. 1 (January):119129.
Greene, William. 1997. Econometric Analysis. 3rd ed. Upper
Saddle River, NJ: Prentice Hall.
Iman, Ronald, and William Conover. 1979. The Use of the Rank
Transform Regression. Technometrics, vol. 21, no. 4:499509.
Laderman, Jeffrey. 1998. Wall Streets Spin Game: Stock
Analysts Often Have a Hidden Agenda. Business Week (5
October):148156.
78
%OWN). That model produced similar conclusions to those
reported here, however, so the results are not tabulated here.
15. We did not expand this robustness test to the full sample
because reputation and investment banking relationships
were insignificant in our estimation of Equation 5 for the
subperiod and because of the costs of data collection.
Lin, Hsiou-Wei, and Maureen McNichols. 1998. Underwriting
Relationships, Analysts Earnings Forecasts and Investment
Recommendations. Journal of Accounting and Economics, vol. 25,
no. 1 (February):101127.
Lowenstein, Roger. 1996. Todays Analyst Often Wears Two
Hats. Wall Street Journal (30 August):C1.
McNichols, Maureen, and Patricia OBrien. 1997. Self-Selection
and Analyst Coverage. Journal of Accounting Research
(Supplement):167199.
Michaely, Roni, and Kent Womack. 1999. Conflict of Interest
and the Credibility of Underwriter Analyst Recommendations.
Review of Financial Studies, vol. 12, no. 4 (Special):653686.
Nelsons Directory of Investment Research. 199495. Port Chester,
NY: Nelson Publications.
Opdyke, Jeff. 2002. Stock Advice You Can Trust? Wall Street
Journal (31 October):D1.
Sidel, Robert, and Stephen Craig. 2002. Does Independent
Research Mean Better Stock Picks? Wall Street Journal (31
October):C1.
Silverman, Gary. 2002. Investors Positive about Spitzers
Citigroup Action. Financial Times (2 October):25.
Spitzer, Elliott. 2002. The Crisis of Accountability. Law Day
Remarks, Office of the New York State Attorney General (1 May):
www.oag.state.ny.us/press/statements/speeches_indexpage.
html.
Stickel, Scott. 1992. "Reputation and Performance among
Security Analysts. Journal of Finance, vol. 47, no. 5
(December):181736.
2004, AIMR