+ All Categories
Home > Documents > Acquisitions and the Information Environment of Firms

Acquisitions and the Information Environment of Firms

Date post: 27-Sep-2016
Category:
Upload: ravi-bhushan
View: 221 times
Download: 1 times
Share this document with a friend
21
THE FINANCIAL REVIEW VOL. 31 No. 1 FEBnuARY 1996 PP. 105-125 Acquisitions and the Information Environment of Firms Ravi Bhushan* and Jang Y. ChP Abstract This paper investigates how acquisitions affect ana- lyst following of firms. Analyst following increases as a result of a merger. However, all of that increase can be attributed to the changes in firm-specific characteristics resulting from the merger. Changes in analyst following around mergers are positively related to changes in firm size, expenditures on R&D, and the ratio of book to the market value of equity. Finally, the relatedness of merger appears to be an important determinant of analyst follow- ing of firms engaged in acquisitions. Introduction Over the last two decades, a considerable body of research has accumulated suggesting that financial ana- lysts perform a useful role in capital markets. There are studies (Barefield and Comiskey [41, Brown and Rozeff [121, Collins and Hopwood [131, Fried and Givoly [MI, Brown, Griffin, Hagerman, and Zmijewski [9,101, Brown, Richardson, and Schwager [ill, and O'Brien [27]) which document that financial analysts' earnings forecasts are superior to those based on time-series models. Others (e.g., Cragg and Malkiel [14]) have *Southern Methodist University, Dallas, TX 75275, deceased **University of Nebraska, Lincoln, NE 68588-0488 The authors wish to thank Andrew Alford, Paul Asquith, Paul Healy, and two anonymous reviewers of this journal for many helpful comments. Thanks are also due to I/B/E/S Inc. for the analyst data. Financial support from the Cox School of Business at the Southern Methodist University and the College of Business Administration at the University of Nebraska, Lin- coln, is gratefully acknowledged. 105
Transcript

THE FINANCIAL REVIEW VOL. 31 No. 1 FEBnuARY 1996 PP. 105-125

Acquisitions and the Information Environment of Firms

Ravi Bhushan* and Jang Y. C h P

Abstract

This paper investigates how acquisitions affect ana- lyst following of firms. Analyst following increases as a result of a merger. However, all of that increase can be attributed to the changes in firm-specific characteristics resulting from the merger. Changes in analyst following around mergers are positively related to changes in firm size, expenditures on R&D, and the ratio of book to the market value of equity. Finally, the relatedness of merger appears to be an important determinant of analyst follow- ing of firms engaged in acquisitions.

Introduction Over the last two decades, a considerable body of

research has accumulated suggesting that financial ana- lysts perform a useful role in capital markets. There are studies (Barefield and Comiskey [41, Brown and Rozeff [121, Collins and Hopwood [131, Fried and Givoly [MI, Brown, Griffin, Hagerman, and Zmijewski [9,101, Brown, Richardson, and Schwager [ill, and O'Brien [27]) which document that financial analysts' earnings forecasts are superior to those based on time-series models. Others (e.g., Cragg and Malkiel [14]) have

*Southern Methodist University, Dallas, TX 75275, deceased

**University of Nebraska, Lincoln, NE 68588-0488

The authors wish to thank Andrew Alford, Paul Asquith, Paul Healy, and two anonymous reviewers of this journal for many helpful comments. Thanks are also due to I/B/E/S Inc. for the analyst data. Financial support from the Cox School of Business at the Southern Methodist University and the College of Business Administration at the University of Nebraska, Lin- coln, is gratefully acknowledged.

105

106 Bhushan and Cho

documented that the prices of common stocks are more closely associated with analysts’ growth rate and risk projections than with the alternatives (e.g., those based on history). Finally, there is some recent research that suggests that analysts play an important role in efficient pricing of securities.’

The importance of financial analysts for the effi- ciency of securities markets has, in turn, generated con- siderable research interest in understanding the factors that affect a company’s analyst following.2 The objective of this paper is to examine how mergers affect the ana- lyst following of firms. Mergers and acquisitions often result in almost instantaneous changes in the underly- ing business of a firm, the nature of its business environ- ment, its scale of operations, its long-term strategy, goals, and competitiveness, etc. Such changes can have a significant impact on the information environment of merging firms, and, in particular, on their analyst fol- lowing. For example, Bhushan [6] suggests that analyst following should be a decreasing function of the costs of information acquisition. Since a merger can result in sig- nificant changes in the nature of business of a firm, it can also considerably affect such costs. If one firm ac- quires another in a completely different line of business, the cost of collecting information for the analysts is likely to increase and, consequently, one would expect, ceteris paribus, lower analyst following after such a merger. This paper examines how such changes arising from merger-related activity affect a firm’s analyst following.

Even though prior research on analyst following has suggested several potentially important determi- nants of analyst following, there is still considerable dis- agreement on which factors are important, empirically. Bhushan [61, for example, argues for the importance of return volatility, which measures the idiosyncratic risk of the firm, while Pearson [29] argues for the importance of beta. Brennan and Hughes [8] argue that analyst fol- lowing should be inversely related to share price. This paper is also intended to provide further evidence on the importance of these different factors.

Acquisitions and Information Enuironment of Firms 107

The activity of a merger generates considerable public and media interest in both the target and acquir- ing companies. Merton [251 proposes a model where investors invest only in companies they know about. This model suggests the possibility that firms may undertake acquisitions to increase their investor bases and to become better-known in the marketplace. This study is also intended to examine the question that firms may undertake mergers to generate additional analyst following.

The evidence examined in this paper suggests that analyst following of companies engaged in mergers in- creases after the merger. However, all of that increase can be explained by changes in the underlying funda- mentals of the company, and the pure act of merger causes no increase in analyst coverage. This suggests that the heightened media or public interest around the time of acquisition does not by itself translate into addi- tional analyst interest in the company. The results also indicate that firms engaged in related mergers experi- ence an increase in analyst following compared to those engaged in diversifying mergers. This is consistent with the idea that analysts follow industries and that costs of information acquisition go up for firms undertaking unrelated acquisitions.

Finally, analyst following is found to increase with firm size, expenditures on research and development, and the ratio of book to market value of equity. No sig- nificant relation is documented between beta, share price, or volatility and analyst following. These findings are generally consistent with the information production and monitoring roles of analysts. However, they do not support some of the specific predictions of the formal models in this area.

The rest of the paper is organized as follows. The role of analysts as information intermediaries and the factors that affect analyst following are discussed next. This is followed by a section presenting the data and the sample selection criteria. The empirical results are discussed next and the conclusions are presented in the final section.

108 Bhushan and Cho

Analysts as Information Intermediaries

Under the view that analysts serve as information intermediaries, analyst following of a firm is deter- mined by the demand for such information and the costs of obtaining the information. Bhushan [61, Bren- nan and Hughes [8], and Pearson [29] have developed formal models of analyst following, and most of the hypotheses developed here are related to the arguments in those paper^.^

The models presented in the above papers suggest that an increase in firm size should result in increased analyst following. An increase in firm size increases the total benefits (or expected profits) from information ac- quisition about the firm, and hence leads to higher demand for information, or increased analyst following.

Using similar reasoning, Bhushan [6] argues that analyst following should be positively related to return variability since expected profits conditional on firm- specific information will increase with the company’s re- turn variability. Pearson [291, on the other hand, adopts a portfolio perspective on the value of information and argues that analyst following should be positively re- lated to the beta of the firm. He argues that information is more valuable on a high-beta firm, since an investor’s wealth is more highly correlated with such information. If analysts’ information acquisition activities deal with both firm-specific as well as market-wide information, then it is possible that both return variability and beta affect analyst following.

Fama and French [16] argue that the ratio of book to market value of equity (BEIME) can be considered as a surrogate for risk. A firm with poor prospects will have a high book value of equity compared to its market value. Such a firm is riskier, may involve higher agency costs, and may require more analyst scrutiny to resolve agency problems. Under this interpretation, as a firm’s prospects worsen, the ratio BEIME will increase with a corresponding increase in analyst following.

Brennan and Hughes [8] argue that analyst follow- ing should be inversely related to share price. Commis- sion schedules for trading are such that for the same

Acquisitions and Information Environment of Firms 109

dollar amount of trade, commission revenues decrease as share price increases. A lower share price, by in- creasing the commission revenues on the firm, there- fore, attracts more analysts into the business of following the company.

Moyer et al. [261 suggest that demand for analyst following should be an increasing function of potential agency costs in the firm. This is because analysts aid in the monitoring role by bridging the information gap be- tween managers and owners. Agency costs are likely to be high for highly leveraged firms and ‘frms that under- take substantial R&D. Such firms are also likely to face higher information asymmetry between managers and insiders on the one hand and the outsiders on the other. These arguments imply that demand for information in- termediation should be an increasing function of lever- age as well as the expenditures on R&D.

Bhushan [6] argues that analyst following should be a decreasing function of the costs of information acquisi- tion. This is because, ceteris paribus, a decrease in the cost of collecting information increases the supply of in- formation. Since mergers can result in significant changes in the nature of business of the firm, they can also affect information acquisition costs. Consider the case of firm A acquiring firm B. If A and B are in the same line of business, information acquisition costs would not change much for the analysts. The larger size of the merged entity implies increased demand for infor- mation, and hence higher analyst following. If, however, B is in a completely different line of business from A, then the cost of collecting information for analysts is likely to increase owing to the increased complexity, which will reduce the supply of information. Thus, even though the increase in firm size is same for the unrelated and related mergers so that the increase in demand for information is same in both cases, one would expect a lower increase in analyst following after an unrelated merger. Ceteris paribus, analyst following would increase less for unrelated mergers than for related ones.

In summary, an information intermediation per- spective on the role of analysts suggests that mergers can afTect analyst following through changes in firm

110 Bhushan and Cho

size, return volatility, beta, book to market value of eq- uity, share price, leverage, expenditures on R&D, and the degree of diversification of the firm. In the sections that follow, tests are designed to assess the effect of each these factors on the analyst following of firms engaged in acquisitions.

Sample Selection and Data

Initial Sample

The source for the sample of acquiring firms is the publication Mergers and Acquisitions, which lists all the large deals completed during a year. The sample in- cludes all firms which were engaged in acquisitions over the period 1977 to 1988, and also met the following addi- tional criteria: (1) The transaction is valued at more than $10 million. (2) The merger results in a significant increase in the company’s scale of operations. This crite- rion is operationalized by requiring the pre-merger sales of the acquired firm to be at least 10 percent of the pre- merger sales of the acquiring company. (3) The comple- tion date of the merger is available in the Wall Street Journal. (4) The acquiring firm is on the Compustat tapes and the CRSP NYSE/AMEX files. The require- ment that the firms be on the CRSP NYSE/AMEX files eliminates all the OTC firms and biases the sample to- ward large firms. This requirement was imposed to re- duce the burden of hand-collecting some of the data. To the extent that the relations between analyst following and firm characteristics are different for OTC firms ver- sus listed firms, the results are not generalizable to smaller or OTC firms.

The sample selection criteria resulted in a sample of 196 acquiring firms, and Table 1 presents the distribu- tion of these firms over time.4 The acquiring firms, in general, are quite evenly distributed over time. The year 1986, however, has a considerably higher concentration of the sample (about 20 percent). This observation is consistent with the heightened merger activity in that year (see, e.g., Table 24.2, Scholes and Wolfson [301).

Acquisitions and Information Environment of Firms 111

TABLE 1

Distribution of Sample Firms by Year of Acquisition

This table describes how the acquiring firms are distributed over the various years.

Year Number of Firms Percent

1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988

14 12 15 18 10 12 8

14 17 38 21 17

196 -

7.1 6.1 7.7 9.2 5.1 6.1 4.1 7.1 8.7

19.4 10.7 8.7 100.0

An analysis of the industry distribution of acquiring firms did not indicate much industry clustering. Defin- ing a firm's industry as the primary two-digit SIC code as reported on the Compustat tapes results in 45 indus- tries being represented by the sample of acquiring firms. The transportation equipment industry (SIC code 37) and the banking industry (SIC code 601, representing 8.4 percent and 5.6 percent of the acquiring firms respec- tively, are the two highest industry concentrations in the sample. About 60 percent of the acquiring firms in the sample are from the manufacturing sector (SIC codes

Since changes in analyst coverage can result from temporal factors or macro-economic phenomena that af- fect firms in different industries differentially, it is nec- essary to control for such effects. This is done by including control firms in the sample, which were ob- tained as follows.

For each acquiring firm in the sample in a given year, an attempt was made to find up to three matches in that year from the pool of firms on the Compustat tapes and the CRSP NYSE/AMEX files.5 Control firms

20-39).

112 Bhushan and Cho

selected were restricted to be no less than one-third and no more than three times the size of the acquiring firm, where firm size is measured in terms of the market value of equity of the firm. Matches were attempted first at the four-digit SIC level and, if those could not be found, then matches were attempted at the three-digit level. A total of 328 firms satisfied the matching criteria and this sample of 328 firms is referred to as the “con- trol firms” sample. Descriptive statistics on both the ac- quiring and control firms are presented in the next sub-section.

Variable Selection and Definitions The data for analyst following (#ANLST) were ob-

tained from I/I3/E/S tapes. As in O’Brien and Bhushan [271, who observe that the number of analysts reporting forecasts tends to increase as fiscal year progresses and levels off in the tenth or eleventh month of the fiscal year, #ANLST was obtained from the eleventh month of the fiscal year. Analyst following before merger was ob- tained as the #ANLST for the fiscal year (Year -1) be- fore the merger consummation year (Year 0). #ANLST in Year +1 and Year +2 constitute two alternative prox- ies for the analyst following after the merger. The differ- ence between #ANLST in Year +1 or Year +2 and #ANLST in Year -1 form two alternative proxies for the change in analyst coverage around merger.

Data for the explanatory variables for years -1, +1, and +2 were obtained as follows. Day 0 denotes the fiscal year-end-date for a firm in a given year. Using the daily returns on CRSP tapes, a market model regression using the CRSP equally-weighted index was run for days -249 to day 0. The slope coefficient from this regression pro- vides an estimate of beta in that year. The standard de- viation of the daily returns over these 250 days serves as the proxy for return volatility. These variables were computed only if there were at least 30 daily returns available in the year.

The annual Compustat tapes are the source for data on price per share (Price), the market value of common equity (MVquty), book value to market value of equity

Acquisitions and Information Environment of Firms 113

(BE /ME) , leverage, and R&D expenditures. The proxy chosen for leverage is the ratio of the sum of the short- term and long-term debt to the market value of common equity. Both the raw total R&D expenditures as well as those normalized by sales are considered.

Given that analysts tend to specialize by industries, a diversifying merger may imply that the analyst has to learn about a new industry, and consequently, may re- sult in an increase in information acquisition costs for the analysts. The proxy for relatedness of merger is similar to that in Kaplan and Weisbach [241 and is de- veloped as follows. The lines of business of a firm are considered to be the SIC codes listed in the Million Dol- lar Directory of Dun and Bradstreet, which lists up to six SIC codes per firm. If the acquiring and the acquired firms share a four-digit SIC code, the merger is related at the four-digit level. If the two firms share a three- digit, a two-digit, or a one-digit SIC code, it is a related merger at the three-digit, two-digit, or a one-digit level respectively. Otherwise, it is an unrelated merger.

The data on the relatedness of merging firms is summarized in Table 2. Approximately 20 percent of the firms undertake unrelated or diversifying acquisitions. However, a significant fraction (45 percent) of firms

TABLE 2

Distribution of Acquiring Firms by Relatedness of Merger

This table classifies the sample of acquiring firms by the relatedness of merger.

Merger Type Number of Firms Percent

Unrelated 36

Related at: 1-Digit level 20 2-Digit level 22 3-Digit level 11 4-Digit level 80

179*

20.1

16.8 12.3 6.1

44.7 100.0

*For 17 acquisitions, data on lines of business could not be found in the Millions Dollar Directory.

114 Bhushan and Cho

acquire another with which they share a four-digit SIC code

The median acquiring firm had $364 million of eq- uity outstanding in year -1. In comparison, the median for the market value of equity in year -1 for the acquired firms was $62 million. The ratio of these two medians is 0.17, while the median value for the ratio of MV,,,, for the acquired and acquiring firms is 0.22. Thus, in the sample, the acquired firm on average is about one-fifth of the acquiring firm. The median number of analysts following an acquiring firm in year -1 was found to be eight, while most of the acquired firms were not followed by analysts (or, at least, not covered on the I/B/E/S tapes). Only four of the acquired firms had any analyst following data on the I/B/E/S tapes.

Tables 3 and 4 respectively present some summary statistics for the acquiring and control firms. Data are presented for the years -1, +1, and +2. The number of observations listed is the minimum number of observa- tions available in any of the three years. The actual number of observations for the other years is very close to this number, however. These tables show that the me- dian control firm is slightly larger than the median con- trol firm in year -1 (which is by design), and in years +1 and +2, the differences in the market values are much smaller. The data on analyst following are also quite similar for the acquiring and control firms.

Overall, Tables 3 and 4 suggest that the acquiring and control firms are quite similar on most dimensions. Two exceptions appear to be leverage and R&D. In com- parison to the control firms, the acquiring firms are more highly levered before the merger. Their median leverage grows from 31 percent before merger to 38 per- cent after merger, while median leverage of control firms stays at about 27-29 percent during that period. A com- parison of the means with the medians suggests that the distribution of the R&D expenditures is considerably right-skewed. Comparing the median values for the ac- quiring and control firms, it appears that the acquiring firms spent much less on R&D in year -1, but their ex- penditures on R&D grew substantially from year -1 to +l; however, for control firms the R&D expenditures are

TA

BL

E 3

.-

n.

2.

g. D

escr

iptiv

e St

atis

tics

for F

irm

s E

ngag

ed in

Acq

uisi

tions

z A

& 3 2

A 2.

Thi

s ta

ble

pres

ents

som

e de

scri

ptiv

e st

atis

tics

for

acq

uiri

ng fi

rms.

#ANLST

is th

e nu

mbe

r of

ana

lyst

s fo

llow

ing

the firm a

s li

sted

on

the

J.BL

E/S

tape

s. h

4Vqt

y an

d B

Ve,

ty ar

e re

spec

tivel

y th

e m

arke

t an

d bo

ok v

alue

s of

com

mon

equ

ity

(in

$ m

illio

ns).

Pric

e is

the

pric

e pe

r sh

are

(in

$) o

f com

mon

sto

ck. L

ever

age

is th

e ra

tio

of th

e su

m o

f the

boo

k va

lues

of t

he s

hort

- and

long

-ter

m d

ebt d

ivid

ed b

y th

e m

arke

t val

ue o

f co

mm

on e

quity

. R&

D e

xp. i

s th

e to

tal

expe

ndit

ure

on R

&D

(in

$ m

illio

n). R

&D

ISaL

es i

s th

e ra

tio

of R

&D

exp

endi

ture

s to

sal

es. R

etur

n V

olat

ility

is th

e st

anda

rd d

evia

tion

of th

e da

ily r

etur

ns a

nd B

eta

is th

e m

arke

t ris

k (w

ith re

spec

t to

the

CR

SP e

qual

ly-w

eigh

ted

inde

x).

s s V

aria

ble

NO

BS

YR

-1

YR

+1

YR

+2

YR

-1

YR

+1

YR

+2

YR

-1

YR

+1

YR

+2

h s

#AN

LST

196

9.7

11.5

12

.1

8 9

10

8.3

9.0

9.5

-.

BV

qtyD

fVqt

y 16

8 0.

84

0.80

0.

56

0.70

0.

74

0.78

0.

58

0.80

3.

57

Mea

n M

edia

n St

d. D

ev.

W,

tY

16

8 1,

476.

2 2,

038.

3 2,

204.

9 36

3.8

637.

1 71

5.5

3,46

6.4

4,77

0.1

5,37

0.6

$! Pr

ice

173

30.9

28

.7

27.9

29

.0

26.0

25

.4

17.2

15

.8

16.0

3

Leve

rage

16

8 0.

34

0.38

0.

40

0.31

0.

38

0.38

0.

22

0.22

0.

23

5 R

&D

exp

. 79

84

.5

122.

2 14

4.4

5.5

12.7

19

.2

264.

8 33

5.1

376.

6 r,

R&

DIS

aLes

(%

) 79

1.

88

1.97

2.

15

1.00

1.

19

1.38

2.

57

2.41

2.

65

3 R

etur

n V

olat

ility

(%)

165

2.11

2.

33

2.37

1.

97

2.06

2.

01

0.75

1.

15

1.53

B

eta

165

1.13

1.

19

1.26

1.

10

1.14

1.

20

0.45

0.

45

0.59

2

TAB

LE 4

Des

crip

tive

Stat

isti

cs fo

r C

ontr

ol F

irm

s

Thi

s ta

ble

pres

ents

som

e de

scri

ptiv

e st

atis

tics

for

the

cont

rol f

irm

s. #

AN

LST

is th

e nu

mbe

r of

ana

lyst

s fo

llow

ing

the firm a

s li

sted

on

the

I/B/E

/S ta

pes.

IMV,

t, an

d M

V,,t,

are

resp

ectiv

ely

the

mar

ket a

nd b

ook

valu

es o

f com

mon

equ

ity (

in $

mill

ions

). Pr

ice

is th

e pr

ice

per

shar

e (i

n $)

of c

omm

on s

tock

. Lev

erag

e is

the

rati

o of

the

sum

of t

he b

ook

valu

es o

f the

sho

rt- a

nd lo

ng-t

erm

deb

t div

ided

by

the

mar

ket v

alue

of

com

mon

equ

ity. R

&D

exp

. is

the

tota

l ex

pend

itur

e on

R&

D (

in $

mill

ion)

. R&

D/S

ales

is th

e ra

tio

of R

&D

exp

endi

ture

s to

sal

es. R

etur

n V

olat

ility

is th

e st

anda

rd d

evia

tion

of th

e da

ily r

etur

ns a

nd B

eta

is th

e m

arke

t ris

k (w

ith

resp

ect t

o th

e C

RSP

equ

ally

-wei

ghte

d in

dex)

.

Mea

n M

edia

n St

d. D

ev.

YR-1

Y

R+

1 m

+2

~

Var

iabl

e N

OB

S Y

R-1

m

+1

m

+2

Y

R-1

m

+1

Y

R+

2

#AN

LST

328

10.5

11

.9

12.0

9

10

10.5

8.

3 8.

8 9.

1 M

Veqt

, 28

9 1,

648.

1 1,

982.

5 2,

146.

4 52

0.3

722.

2 76

7.5

3,23

3 3,

872

4,31

9 Pr

ice

290

33.1

32

.1

32.5

27

.9

28.4

28

.4

23.4

21

.3

21.9

B

Veq

tWV

qt,

289

0.87

0.

86

0.59

0.

75

0.76

0.

72

0.59

0.

66

2.85

Le

vera

ge

288

0.31

0.

33

0.35

0.

27

0.28

0.

29

0.22

0.

24

0.25

R

&D

exp.

18

5 10

4.9

142.

8 15

0.7

18.3

20

.4

23.4

32

2.0

443.

6 15

0.7

R&

D lS

ales

18

5 2.

43

2.59

2.

66

1.50

1.

45

1.62

3.

00

3.01

3.

08

Ret

urn

Vol

atili

ty (%

) 26

9 1.

95

2.18

2.

19

1.81

2.

04

1.98

0.

70

0.80

B

eta

269

1.11

1.

10

1.16

1.

07

1.02

1.

10

0.44

0.

44

:::: 6

2

R 3

R

n

Acquisitions and Information Environment of Firms 117

almost flat. Both the raw and the normalized R&D ex- penditures provide similar evidence. The median BVe,,,,IWequi, (BE / M E ) ratios indicate a slight increase for the sample firms while they are almost flat for the control firms.

It appears that the acquiring firms may have a slightly higher systematic risk than the market. The mean beta for the control firms is also greater than one. This suggests the possibility that firms in industries that are more sensitive to market movements are more likely to engage in acquisitions. Also, given that manu- facturing firms are more heavily represented in the Sam- ple, this finding of mean beta being greater than one most likely reflects the fact that manufacturing firms have higher systematic risk. This is consistent with the evidence in Foster “171, Table 10.1).

Another interesting observation appears to be that the cross-sectional standard deviation of the volatility for the acquiring firms is significantly higher in years +1 and +2 compared to year -1, while the control firms do not show any such trend. However, the inter-quartile ranges for the volatility of the acquiring firms are 0.95 percent, 0.95 percent, and 0.91 percent for years -1, +1 and +2 and thus do not show any significant trend. This suggests that the increase in standard deviation in the later years is being driven by some outliers. An exami- nation of the individual observations confirms that in years +1 and +2, there are a few firms with extremely high volatility that are responsible for this result.

Empirical Results The change in analyst following over years -1 to +1

(or +2) is the primary variable of interest. Denote this variable as M L S T . For the sample of acquiring firms, the mean values of AANLST are 1.8 (t = 5.93) and 2.3 (t = 5.73) over years -1 to +1 and to +2. Both the median values of AANLST are 2. For the control firms, the corre- sponding means are 1.3 (t = 6.77) and 1.5 (t = 5.67) and both the medians are 1. These data suggest that analyst following increases for acquiring as well as control firms. The increase in analyst coverage cannot be merely

118 Bhushan and Cho

attributed to the idea that some analysts who were cov- ering acquired firms continue to cover the merged entity. This is because, as noted earlier, 95 percent of the ac- quired firms had no analyst coverage. Also, the results remain virtually unchanged when the subsample involv- ing followed acquired firms is excluded. The t-statistics for the differences in the means for AANLST for acquir- ing and control firms are 1.47 and 1.82 for years -1 to +1 and to +2.

To assess what part of the increase in analyst fol- lowing is due to changes in firm-specific characteristics and what part is due to the mere activity of merger, OLS regressions were run with AANLST as the depend- ent variable. The explanatory variables are based on the earlier discussion of factors affecting analyst follow- ing. Since the dependent variable is in the form of a change, most of the explanatory variables are also des- ignated in that form. The basic model that is estimated is the following:

AANLST ( i ) = b, + b, A ln[MVeqUi,(i )I + b2A Price ( i )

+ b, ABE/ME(i) + b, ALeuerage (i)

+ b, AU [R&D/Sales(i)l + b, AVolatility ( i )

+ b, ABETA(i) + b, SIZE(i) + b, Merg(i)

+ b,, Rel.Siz4i) + b,, ReZated(i) + e(i) . (1)

In equation (l), for example, AAiVLST(i) is the change in analyst following from year -1 to +1 (or +2) for firm i. There are four dummy variables in the regression: SIZE, Relative Size, Merg, and Related. The dummy variables SIZE takes the value 1 if MVequi, in year -1 is greater than $400 million (which is roughly the median value) and 0, otherwise. The dummy variable Merg takes the value 1 for acquiring firms and 0 for control firms. Rela- tive Size dummy takes the value 1 if firm is engaged in an acquisition and the ratio of MVe,,, for the acquired and acquiring firms is greater than the median (=0.22).

Acquisitions and Information Environment of Firms 119

Else, it takes the value 0. The dummy Related takes the value 1 if the firm is engaged in a merger that is related at the one-digit level or higher. It takes the value 0, otherwise.

The dummy variable Size allows AAjVLSTto be dif- ferent for large and small firms. Likewise, Relative Size dummy allows WLSTto be different for firms engaged in large versus small acquisitions relative to their size. Alternative specifications allowing these dummy vari- ables to interact with the other explanatory variables were also considered. The results from such regressions are qualitatively similar to those reported for the basic specification in equation (1).

For the regressions, observations with negative val- ues of BEIME are deleted. Very few (at most three or four) observations are lost because of this restriction. Also, given that many firms have missing R&D data, there would be a considerable loss in the number of ob- servations if all such firms were deleted. To keep such firms in the sample, the following procedure is adopted. For all firms with some missing R&D data, if the firm has missing R&D data in both years -1 and +I (+2), AR&DISalesis coded as 0. Otherwise, it is assigned a missing value. The procedure thus assumes that for a company not reporting R&D expenditures, the changes from year to year are not significant. This is probably a reasonable assumption since if R&D expenditures were significant for the company, it would probably report them.7

The residual plots from the regressions suggest that they are well-specified and the specification does not generate many outliers. Table 5 presents the regression results. The reported results are not materially affected when influential observations are deleted.' The results are presented for changes in #ANLST from year -1 to year +1 as well as to year +2.

The F-statistics are significant for both regressions indicating that the regressions have significant explana- tory powers. This is consistent with the adjusted R2 of about 13-14 percent in both cases. The results of the re- gressions for both periods are generally quite similar. The variables R&D expenditures, MV,,,,, BE / M E , and

120 Bhushan and Cho

TABLE 5

Regression Results on Analyst Following for the Whole Sample This table presents the results of estimating the following regression:

AAiVLST(i) = bo + bl A In[MVequit,(i)1 + b2 A Price ( i ) + ba A BEIME(i)

+ b4 A Leuerage(i) + bs AU [R&D/SaZes(i)] + be A Volatility(i) + b7 ABetNi)

+ ba SIzE(i) + bs Merg(i) + blo Relative Size(i) + bll ReZated(i) + e(i) .

The dummy variable SIZE takes the value 1 for large firms and 0 for small firms. The dummy variable Merg equals 1 for acquiring firms and 0 for control firms. The dummy variable ReEative Size equals 1 for firms where the acquired firm represents a relatively large part of the acquiring firm, otherwise, it equals 0. The dummy variable Related takes the value 1 if the firm is engaged in a related merger, 0 otherwise. See Table 3 for the definitions of other variables. Year 0 is the year in which merger is consummated. All the variables (except the dummies) are in the form of changes. For example, AANLST over year -1 to Year +1 is the change in analyst following from one year before merger to one year after merger.

Year -1 to Year +1 Year -1 to Year +2 Explanatory Variable Coefficient T-Statistic Coefficient T-Statistic

Intercept

Pr i ce ABEIME ALeverage AR&DISales AVolatility ABeta Size Merg Relative Size Related NOBS F-Stat Adj. R2

W e q U l t Y

0.387 2.497 0.006 1.865 0.697

60.595 -78.216 -0.161

1.217 -1.315

0.589 1.357

358 5.86 0.130

1.15 4.20 0.48 2.99 0.42 2.54

-3.36 -0.45 3.44

-1.89 1.05 1.94

0.534 2.726 0.009 1.238 2.925

81.100

0.134 1.043

0.719 2.438

5.91 0.136

-16.489

-1.754

343

1.30 4.85 0.55 2.22 1.81 3.30

0.29 2.32

0.98 2.73

-0.53

-1.95

the SIZE and Related dummies have t-statistics of about 2.0 or greater (suggesting significance at 5 percent level) in both regressions. Volatility is significant at the 5 per- cent level in the first regression, but not the second. The variables Leverage, BETA, and share price are insignifi- cant at the 5 percent level in both the regressions. Most of the significant coefficients increase in magnitude as attention is directed from first regression to the second.

Acquisitions and Information Environment of Firms 12 1

This suggests that the full adjustment of analyst follow- ing to a change in the underlying firm characteristic takes time and the first regression which goes only up to year +1 may not capture this adjustment fully.

The significance of the SIZE dummy indicates that analyst following adjusts differently for large versus small firms. However, the Relative Size dummy is insig- nificant, which suggests that after controlling for the level of firm size and other explanatory variables, the relative size of the acquisition does not affect analyst fol- lowing.

Interestingly, the estimated coefficient on the dummy variable Merg is negative, with t-values of about -1.9 in both the regressions. This suggests that after controlling for the other factors, the activity of merger per se certainly does not result in any increase in ana- lyst following. Thus, even though analyst following in- creases around mergers, all of that increase can be attributed to changes in firm-specific characteristics, e.g., size. The decrease in analyst following of merged firms, after controlling for other factors, is, however, consistent with a merger making the organization more complex and hence increasing the costs that analysts face in following a firm. The coefficient on the Related dummy is positive and significant. This implies that analyst following for firms engaging in related mergers increases in comparison to those undertaking diversify- ing mergers. The negative coefficient on the Merg dummy together with the positive coefficient on the Related dummy are consistent with the notion that on average a merger makes a firm more complex leading to lower analyst following, ceteris paribus. But, if the ac- quisition is related, these costs do not increase as much so that analyst following for related mergers is higher than that for unrelated ones.

Discussion The regressions document that analyst following

around mergers is positively related to changes in firm size, R&D expenditures, and the ratio of the book to market value of equity. No significant relation exists

122 Bhushan and Cho

between analyst following and beta, share price, lever- age, or volatility. A positive relation between analyst fol- lowing around mergers and firm size is consistent with the arguments of Bhushan [61 and Pearson [29] that benefits from information-based trading are increasing in firm size. A positive relation between analyst follow- ing and R&D expenditures is consistent with the moni- toring role of analysts: the demand for monitoring increases as the company invests more in R&D. A posi- tive relation between analyst following and the ratio of book to market value of equity is also consistent with the monitoring role of analysts. An increase in BEIME is equivalent to a firm’s prospects getting worse, so that the firm may require more monitoring. The positive coef- ficient on BE / M E is consistent with the demand for ana- lyst following going up as firms’ prospects worsen.

The lack of significant relations of analyst following with either beta, volatility, or share price is inconsistent with the formal models in this area. In summary, the evidence is not very consistent with some of the specific predictions of the formal models in this area. However, the empirical work does support the general notion that analysts act as information intermediaries.

Conclusions This paper has investigated the factors that affect

analyst following of firms using a sample of firms en- gaged in acquisitions and control firms within those in- dustries over the period 1977-1988. To the extent the acquisition activity is not representative of the whole economy, the results are not generalizable. Keeping this caveat in mind, it is found that analyst following of firms engaged in acquisitions increases as a result of the merger. However, all of that increase can be attributed to the changes in firm-specific characteristics resulting from the merger. Changes in analyst following around mergers are positively related to changes in firm size, expenditures on R&D, the ratio of book to the market value of equity. No strong support for the relations be- tween analyst following and beta, share price, or volatil- ity is found. Finally, the results also suggest that the

Acquisitions and Information Environment of Firms 123

relatedness of merger is an important determinant of analyst following of firms engaged in acquisitions. The overall results of this paper are consistent with predic- tions derived from the general notion that analysts act as information intermediaries.

Notes

1. For example, Alford and Zmijewski [31 observe that the negative stock price reactions for firms that delay earnings announcements are mainly concentrated in firms with low levels of analyst following. Bhushan [7] examines the effect of analyst following on post-earnings- announcement drift.

2. See, for example, Bhushan [61, Moyer et al. [261, O’Brien and Bhushan [28], Brennan and Hughes [8], and Pearson [291.

3. A large body of theoretical literature exists on the subject of infor- mation acquisition and its aggregation into stock prices (Grossman [19, 201, Grossman and Stiglitz [21, 221, Hellwig [23], Diamond and Verrec- chia [15], Admati [l], and Admati and Pfleiderer [21). Most of these papers, however, do not directly address the issue of analyst following.

4. There are some firms that are engaged in multiple acquisitions over this period. For such a firm, if the acquisitions occur in the same year then all those acquisitions correspond to only one observation; other- wise, they are treated as separate observations. There are only two in- stances of the former type in the sample.

5. The acquiring firms will grow in size through acquisitions. Hence, to ensure that the sizes of the sample and control firms would be similar in the years following the acquisition, control firms are chosen to be slightly bigger than sample firms in the acquisition year. This is opera- tionalized by attempting to select two matching firms that are bigger than the acquiring firm and one that is smaller.

6. Attention is focused on the medians here since the means and the standard deviations are significantly influenced by outliers. (In both the control and sample groups, there are one-three f m s that have very large negative values of BEIME in years +1 and +2).

7. The results for the subsample of firms with non-missing R&D data are qualitatively similar to those reported here for the whole sample.

8. To assess the impact of influential observations, the Influence op- tion in SAS was used. (This option requests the statistics proposed by Belsley, Kuh, and Welsch [5] to measure the influence of each observation on the estimates.)

References

Admati, Anat R. “A Noisy Rational Expectations Equilibrium for Multi-Asset Securities Market.” Econornetrica 53(May 1985): 629- 657.

Admati Anat R., and Paul Pfleiderer. “A Monopolistic Market for Information.” Journal of Economic Theory 39(August 1986): 400- 438.

Bhushan and Cho

[31 Alford, Andrew, and Mark Zmijewski. “An Empirical Analysis of the Anomalous Stock Market Reaction to Delayed Earnings Announcements.” Working Paper, M.I.T. and the University of Chi- cago, 1992.

141 Barefield, Russell M., and Eugene E. Comiskey. “The Accuracy of Analysts’ Forecasts of Earnings per Share.” Journal of Business Research 3(July 1975): 241-252.

[5] Belsley, David, Edwin Kuh, and Roy Welsch. Regression Diagnos- tics: Identifying Influential Data and Sources of Collinearity. New York: John Wiley & Sons, 1980.

[6] Bhushan, Ravi. “Firm Characteristics and Analyst Following.” Journal of Accounting and Economics 1 l(Ju1y 1989): 255-274.

[7] Bhushan, Ravi. “An Informational Efficiency Perspective on the Post-Earnings Announcement Drift.” Journal of Accounting and Economics 18(1994), 45-66.

[8] Brennan, Michael J., and Patricia J. Hughes. “Stock Prices and the Supply of Information.” Journal of Finance 46(December 1991): 1665-1692.

[9] Brown, Lawrence D., Paul A. Griffin, Robert L. Hagerman, and Mark E. Zmijewski. “Security Analyst Superiority Relative to Uni- variate Time Series Models in Forecasting Quarterly Earnings.” Journal of Accounting and Economics 9(April 1987a): 61-87.

[lo] Brown, Lawrence D., Paul A. Griffin, Robert L. Hagerman, and Mark E. Zmijewski. “An Evaluation of Alternate Proxies for the Market Assessment of Unexpected Earnings.” Journal of Account- ing and Economics 9(July 198713): 159-193.

1111 Brown, Lawrence D., Gordon D. Richardson, and Steven J. Schwager. “An Information Interpretation of Financial Analyst Su- periority in Forecasting Earnings.” Journal of Accounting Research 25(Spring 1987): 49-67.

[12] Brown, Lawrence D., and Michael S. Rozeff. “The Superiority of Analysts’ Forecasts as Measures of Expectations: Evidence from Earnings.” Journal ofFinance 33(March 1978): 1-6.

[13] Collins, William A., and William S. Hopwood. “A Multivariate Analysis of Annual Earnings Forecasts Generated from Quarterly Forecasts of Financial Analysts and Univariate Time-Series Models.” Journal of Accounting Research 18(Autumn 1980): 390- 406.

[14] Cragg, John, and Burton Malkiel. Expectations and the Structure of Share Prices, Chicago: The University of Chicago Press, 1982.

[15] Diamond, Douglas W., and Robert E. Verrecchia. “Information Ag- gregation in a Noisy Rational Expectations Economy.” Journal of Financial Economics 9(September 1981): 221-235.

Acquisitions and Information Environment of Firms 125

[161 Fama, Eugene F., and Kenneth R. French. “The Cross-Section of Ex- pected Stock Returns.” Journal of Finance 47(June 1992): 427-465.

[171 Foster, George. Financial Statement Analysis. Second Edition. Englewood Cliffs, New Jersey: Prentice Hall, 1986.

[181 Fried, Dov, and Dan Givoly. “Financial Analysts’ Forecasts of Earn- ings: A Better Surrogate for Market Expectations.” Journal of Ac- counting and Economics 4(0ctober 1982): 85-107.

[191 Grossman, Sanford. “On the Efficiency of Competitive Stock Prices Where Traders Have Diverse Information.” Journal of Finance 31(May 1976): 573-585.

[20] Grossman, Sanford. “Further Results on the Informational Effi- ciency of Competitive Stock Markets.” Journal of Economic Theory 18(June 1978): 81-101.

[21] Grossman, Sanford J., and Joseph E. Stiglitz. “Information and Competitive Price Systems.” American Economic Review 66(May 1976): 246-253.

[22] Grossman, Sanford J., and Joseph E. Stiglitz. “On the Impossibility of Informationally Efficient Markets.” American Economic Review 70(June 1980): 393-408.

[23] Hellwig, Martin F. “On the Aggregation of Information in Competi- tive Markets.” Journal of Economic Theory 22(June 1980): 477-498.

[241 Kaplan, Steven N., and Michael S. Weisbach. “The Success of Ac- quisitions: Evidence from Divestitures.” Journal of Finance 47(March 1992): 107-138.

[251 Merton, Robert C. “A Simple Model of Capital Market Equilibrium with Incomplete Information.” Journal of Finance 42(July 1987): 483-510.

[261 Moyer, R. Charles, Robert E. Chatfield, and Phillip M. Sisneros. “Security Analyst Monitoring Activity: Agency Costs and Informa- tion Demands.” Journal of Financial and Quantitative Analysis 24(December 1989): 503-512.

[271 O’Brien, Patricia C. “Analysts’ Forecasts as Earnings Expectations.” Journal of Accounting and Economics lO(January 1988): 53-83.

[28] OBrien, Patricia C., and Ravi Bhushan. “Analyst Following and In- stitutional Ownership.” Journal of Accounting Research 28(Supple- ment 1990): 55-76.

[291 Pearson, Neil. “Determinants of the Production of Information.” Working Paper, University of Rochester, 1991.

[301 Scholes, Myron, and Mark Wolfson. Taxes and Business Strategy: A Planning Approach. Englewood Cliffs, New Jersey: Prentice Hall, 1992.


Recommended