Ali Klasa Yueng RFS 2009

The Limitations of Industry Concentration
Measures Constructed with Compustat Data:

Implications for Finance Research
Ashiq Ali
School of Management, University of Texas at Dallas
Eric Yeung
J. M. Tull School of Accounting, University of Georgia
Industry concentration measures calculated with Compustat data, which cover only the
public firms in an industry, are poor proxies for actual industry concentration. These
measures have correlations of only 13% with the corresponding U.S. Census measures,
which are based on all public and private firms in an industry. Also, only when U.S. Census
measures are used is there evidence consistent with theoretical predictions that moreconcentrated industries, which should be more oligopolistic, are populated by larger and
fewer firms with higher price-cost margins. Further, the significant relations of Compustatbased industry concentration measures with the dependent variables of several important
prior studies are not obtained when U.S. Census measures are used. One of the reasons
for this occurrence is that Compustat-based measures proxy for industry decline. Overall,
our results indicate that product markets research that uses Compustat-based industry
concentration measures may lead to incorrect conclusions. (JEL G10, G30, L10)
A growing number of studies that consider the effects of product markets on

financial economics-related phenomena use industry concentration measures
calculated with Compustat data, which cover only the public firms in an industry. These studies examine issues related to asset pricing (Hou and Robinson
2006), informed trading (Tookes 2008), idiosyncratic stock return volatility
(Gaspar and Massa 2006), mergers and acquisitions (Song and Walkling 2000;
Fee and Thomas 2004; Shahrur 2005), corporate governance (DeFond and Park
1999; Engel, Hayes, and Wang 2003; Rennie 2006; Karuna 2007), capital structure (Lang and Stulz 1992; Kale and Shahrur 2007), corporate disclosure policy
We appreciate helpful comments from two anonymous referees, Michelle Sovinsky Goeree, Eric Kelley,
Chris Lamoureux, Bill Maxwell, Hernan Ortiz-Molina, David Robinson, Janet Smith, Matthew Spiegel, Mark
Trombley, and Harold Zhang. We also thank Ed Altman for providing us with the Altman-NYU Salomon
Center Bankruptcy list. Send correspondence to Sandy Klasa, Department of Finance, Eller College of Management, University of Arizona, Tucson, AZ 85721-0108; telephone: (520) 621-8761; fax: (520) 621-1261. E-mail:
sklasa@eller.arizona.edu.

C The Author 2008. Published by Oxford University Press on behalf of The Society for Financial Studies.
All rights reserved. For Permissions, please e-mail: journals.permissions@oxfordjournals.org.
doi:10.1093/rfs/hhn103
Advance Access publication December 23, 2008
Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014
Sandy Klasa
Department of Finance, University of Arizona
The Review of Financial Studies / v 22 n 10 2009
3840
(Harris 1998; Botosan and Harris 2000; Botosan and Stanford 2005; Rogers
and Stocken 2005; Verrecchia and Weber 2006), income-increasing accounting
choices (Zmijewski and Hagerman 1981), and the determinants of corporate
earnings (Cheng 2005).
We consider the empirical implications of using industry concentration
measures that are based on only a firms publicly traded rivals. To do so,
we compare Compustat-based industry concentration measures with industry concentration measures collected from 19632002 Census of Manufactures publications provided by the U.S. Census Bureau, which are based
on all public and private firms in an industry. The Census of Manufactures data have also been used to examine the effect of product market
factors on a wide spectrum of finance issues: corporate takeover decisions
(Eckbo 1985, 1992; Maksimovic and Phillips 2001), capital structure decisions (Phillips 1995; Kovenock and Phillips 1997; Mackay and Phillips
2005; Campello 2006), corporate investment patterns (Akdogu and Mackay
2008), chief executive officer (CEO) compensation contracts (Aggarwal
and Samwick 1999), and risk management decisions (Haushalter, Klasa, and
Maxwell 2007). These studies argue that it is preferable to use concentration
measures calculated by the U.S. Census because the measures based on Compustat data are subject to measurement error due to the exclusion of private
firms, which often account for a nonnegligible percentage of industry sales.
MacKay and Phillips (2005, p. 1439) point out that because industry concentration measures calculated by the U.S. Census are used by regulatory agencies
such as the Department of Justice, these measures are likely to be the most
appropriate to study product market issues.
Our empirical evidence indicates that Compustat-based industry concentration measures are poor proxies for actual industry concentration. The correlation between the Compustat and U.S. Census-based Herfindahl indexes
is only 13%. Moreover, U.S. Census-based concentration measures are positively related to industry price-cost margins and to firm size measures such
as net sales, total assets, and market capitalization. However, these relations
are not obtained using Compustat-based industry concentration measures. Further, we show that the total number of private and public firms in an industry
markedly drops between the highest and lowest quintiles of U.S. Census-based
industry concentration measures. In contrast, this number changes very little
if Compustat-based industry concentration measures are used instead. Thus,
only when U.S. Census-based industry concentration measures are used are the
results consistent with theoretical predictions that more-concentrated industries
that should be more oligopolistic are populated by fewer and larger firms that
enjoy higher price-cost margins due to their greater market power.
Next, we use the U.S. Census data to reexamine several important results
obtained in prior studies that use Compustat-based industry concentration measures. First, we consider the Hou and Robinson (2006) finding that firms in
more-concentrated industries earn lower future stock returns. They argue that
The Limitations of Industry Concentration Measures Constructed with Compustat Data
In contrast to his earlier prediction, Schumpeter (1942) claims that there is more innovation in less-competitive
industries because firms in such industries can enjoy economic profits resulting from their innovation, instead of
having these profits competed away.
3841
their results indicate that barriers to entry in highly concentrated industries

insulate firms from undiversifiable distress risk, which is priced in equity returns. Hou and Robinson (2006) also report that firms in less-concentrated
industries spend more on research and development. They contend that this
result supports the Schumpeter (1912) proposition that innovation, which is a
form of creative destruction, is more likely to occur in competitive industries.1
Also, they posit that higher innovation risk in less-concentrated industries contributes to the higher cost of equity capital in such industries. In contrast, we
document that industry concentration measures calculated by the U.S. Census
are not related to future stock returns. Further, we find that the U.S. Census
measures are positively rather than negatively associated with research and development expenses. These differences in results are not driven by our sample
being confined to the manufacturing sector, because using Compustat-based
industry concentration measures and limiting our analysis to this sector, we are
able to replicate the Hou and Robinson (2006) findings.
Second, we reexamine the Lang and Stulz (1992) result that the effect of
bankruptcy announcements on the equity values of competitors is more positive
in more-concentrated industries and that this effect is amplified in industries
with low leverage. Lang and Stulz (1992) argue that in industries that are more
concentrated and have lower leverage, competitors are more likely to benefit
from the difficulties faced by a bankrupt firm. We obtain the same results as
in Lang and Stulz (1992) with our sample of manufacturing firms, when we
use Compustat-based industry concentration measures. However, using U.S.
Census-based industry concentration measures, we are unable to replicate the
Lang and Stulz (1992) findings.
Third, we reexamine the Harris (1998) result that firms are less likely to
provide segment disclosures for operations in more-concentrated industries,
measured using Compustat data. She argues that to protect their abnormal
profits and market share, firms in less-competitive industries are less likely
to disclose commercially valuable information to competitors. We obtain the
same results as in Harris (1998) with our sample of manufacturing firms, when
we use Compustat-based industry concentration measures. However, we find
that the decision to provide segment disclosures for operations in a particular
industry is not associated with the U.S. Census-based industry concentration
measures of that industry.
Finally, we consider the Defond and Park (1999) result that CEO turnover is
negatively associated with Compustat-based industry concentration measures.
They argue that in more-competitive industries in which there is greater homogeneity across firms and in which CEOs are likely to have more peers, it
is easier to identify and replace poorly performing CEOs. We obtain results
To construct Compustat-based industry concentration measures, studies such as that by Hou and Robinson (2006)
often require that firms are included on both CRSP and Compustat.
3842
similar to those in Defond and Park (1999) with our sample of manufacturing
firms when we use Compustat-based industry concentration measures. However, we find an insignificant relationship between CEO turnover and U.S.
Census-based industry concentration measures.
Our finding that using U.S. Census-based industry concentration measures
we are unable to replicate the Hou and Robinson (2006); Lang and Stulz (1992);
Harris (1998); and Defond and Park (1999) results suggests that Compustatbased industry concentration measures capture other industry characteristics
that happen to be correlated with the dependent variables of these studies. To
provide evidence on this issue, we examine what drives the Hou and Robinson
(2006) result that Compustat-based industry concentration measures are negatively related to future stock returns and the Harris (1998) finding that firms
are less likely to provide segment disclosures for operations in industries with
higher values for Compustat-based measures.
We find that Compustat-based industry concentration measures are significantly negatively related to the change in industry shipments reported by the
Census of Manufactures during the prior five years. However, U.S. Censusbased industry concentration measures are not related to past shipment growth.
Thus, for some reason other than the actual concentration of an industry, industries with high Compustat-based measures experience poor growth in the
recent past. An explanation for these findings is that a declining industry is
left with only a few large, public firms relative to private firms. Consequently,
there are only a few companies in the Compustat database for the industry,
and this results in high Compustat-based industry concentration values. Consistent with this explanation, we find a significant negative relationship between
the Compustatbased industry concentration measures and the change over
the prior five years in the number of firms in an industry included in both the
Center for Research in Security Prices (CRSP) and Compustat databases.2
Our finding that industries with high Compustat-based industry concentration measures tend to be declining industries explains why these industries
spend less on research and development, as reported in Hou and Robinson
(2006) and confirmed in our study. Given that prior work suggests that firms
that spend more on research and development have higher future stock returns (e.g., Chan, Lakonishok, and Sougiannis 2001; Chambers, Jennings, and
Thompson 2002; Eberhart, Maxwell, and Siddique 2004), we examine whether
the association between Compustat-based industry concentration measures and
future stock returns is sensitive to controlling for research and development expenses. After controlling for current research and development expenses, which
we find are positively related to future stock returns, the negative association
between Compustat-based industry concentration measures and future stock
1. Description of Compustat- and U.S. Census-based industry concentration

measures
1.1 Compustat-based industry concentration measures
Compustat-based industry concentration measures are calculated using the
sales data of firms included in the Compustat database. Because Compustat
excludes private firms, Compustat-based industry concentration measures can
potentially provide an inaccurate picture of the actual degree of concentration
in an industry. In particular, in industries in which private firms account for
a nonnegligible percentage of industry sales, it is problematic to rely on data
3843
returns becomes insignificant. This finding suggests that the negative relationship between research and development expenses and Compustat-based
industry concentration measures drives the negative association between these
measures and future stock returns.
Next, we show that after controlling for prior growth in industry shipments,
the Harris (1998) documented negative relationship between a firms decision to
provide segment disclosures of its operations in an industry and the Compustatbased concentration measures of that industry becomes insignificant. Further,
we find that a firms decision to provide segment disclosures for its operations
in an industry is positively related to that industrys prior shipment growth.
The latter result is consistent with the Miller (2002) and Kothari, Shu, and
Wysocki (forthcoming) evidence that firms with weak (strong) prior operating
performance provide less (more) informative disclosures. Overall, these findings suggest that the Harris (1998) result is driven by Compustat-based industry
concentration measures proxying for the prior performance of a firm in one of
its segments that in turn determines the firms decision to provide a separate
disclosure for that segment.
Our study makes the following contributions. First, we document that
Compustat-based industry concentration measures, which exclude data on private firms, are poor measures of actual industry concentration. Second, we
show that researchers who use Compustat data to construct industry concentration measures can arrive at results that lead to incorrect conclusions. Finally,
our findings suggest that the significant results obtained in prior studies that
use Compustat-based industry concentration measures could be due to these
measures proxying for other industry characteristics that are correlated with
the dependent variables of these studies.
The remainder of this article is organized as follows. Section 1 describes the
Compustat- and U.S. Census-based industry concentration measures used in the
study. Section 2 provides evidence that indicates that Compustat-based industry
concentration measures are poor proxies for actual industry concentration.
Section 3 reexamines results of four prior studies that use Compustat-based
industry concentration measures. Section 4 concludes.
1.2 U.S. Census-based industry concentration measures

The Census of Manufactures publications provided by the U.S. Census Bureau
report concentration ratios for hundreds of industries in the manufacturing
sector. We hand-collect data on the U.S. Census-based Herfindahl index and
four-firm concentration ratio from Census of Manufactures publications for
the years 1963, 1966, 1967, 1970, 1972, 1977, 1982, 1987, 1992, 1997, and
2002. The data are for four-digit SIC industries (SIC codes between 2000
and 3999) for the years 19631992 and for six-digit North American Industry
Classification System (NAICS) industries (NAICS codes between 311111 and
339999) for the years 1997 and 2002.5
3
Kahle and Walkling (1996) report that over long sample periods there are advantages to using historical CRSP
SIC codes instead of Compustat SIC codes. Further, because over the past fifty years the U.S. Census Bureau has
revised the SIC system a number of times, it is advantageous to use historical CRSP SIC codes when constructing
Compustat-based industry concentration measures.
Most work that uses Compustat- or U.S. Census-based industry concentration measures assumes that a firm
competes in only the industry represented by the industry classification code assigned to the firm. Because the
aim of this article is to compare results obtained with these two types of industry concentration measures, we
make the same assumption.
For nonmanufacturing industries, the alternative to using Compustat data and consequently excluding data on
private firms to construct industry concentration measures is to use concentration measures collected from
industry specific publications. Work that uses such measures examines issues such as the interaction between
3844
that exclude these firms (Hay and Morris 1991, p. 210). However, there are
two advantages of using Compustat data to construct industry construction
measures. First, such measures can easily be calculated by extracting from
Compustat total sales for each firm in a particular industry. Therefore, they can
provide a long and continuous time series of concentration measures. Second,
using the Compustat database to calculate industry concentration measures
allows researchers to construct these measures for a wide spectrum of industries.
The Compustat-based Herfindahl index is calculated by adding the squares
of the sales market shares of all the firms in an industry that have sales data
on Compustat. Similarly, the Compustat-based four-firm ratio is calculated by
adding the sales market shares of the four largest firms in an industry in terms of
market share. We refer to the Compustat-based Herfindahl index and four-firm
ratio as HI-Compustat and FFR-Compustat.
For the univariate results presented in this and the second section of the
article, HI-Compustat and FFR-Compustat are calculated in a manner similar
to that in Hou and Robinson (2006). HI-Compustat and FFR-Compustat are
calculated using the sales market shares of all the firms in an industry with
sales data on Compustat and are averaged over the past three years. Industry is
defined using historical CRSP Standard Industrial Classification (SIC) codes.3,4
For the reexamination of the Hou and Robinson (2006); Lang and Stulz (1992);
Harris (1998); and Defond and Park (1999) results, we calculate Compustatbased industry concentration measures using the methodology employed by
the specific study.
1.3 Descriptive statistics of U.S. Census- and Compustat-based industry

concentration measures
Table 2 provides descriptive statistics of the industry concentration measures.
Panel A compares HI-Census and HI-Compustat for four-digit SIC industries
industry concentration (collected from the annual publication Supermarket News Distribution Study of Grocery
Store Sales) and capital structure decisions in the supermarket industry (Chevalier 1995a, 1995b; Chevalier
and Scharfstein 1996), the relationship between industry concentration (collected from the American Trucking
Association) and firm survival after deregulation of the trucking industry (Zingales 1998), and how industry
concentration (collected from Discount Merchandiser) interacts with ownership structure, capital structure, and
corporate focus in the discount department industries (Khanna and Tice 2000).
3845
Unlike Compustat-based industry concentration measures, U.S. Censusbased measures are constructed using data from all public and private firms
in an industry and hence should better capture actual industry concentration.
The use of U.S. Census-based measures by government regulatory agencies
suggests that these measures should be quite reliable. For instance, these measures are often used by the Federal Trade Commission when it decides whether
to challenge mergers on antitrust grounds. Another factor that suggests the
U.S. Census-based industry concentration measures should be reliable is that
all firms in the United States are required by federal law to respond to U.S.
Census surveys (under Title 13 of the U.S. code). Further, Sections 213 and
224 of Title 13 of the U.S. code state that employees of the U.S. Census who
collect data on its behalf and who knowingly furnish false information are
subject to imprisonment and that agents of companies who willfully provide
false answers to questions about their company are subject to hefty fines.
The Census of Manufactures calculates the Herfindahl index of an industry
as the sum of the squares of the individual company market shares of all
the companies in an industry or the fifty largest companies in the industry,
whichever is lower. The four-firm ratio of an industry is the sum of the market
shares of the four largest firms in the industry in terms of market share. We
refer to these measures as HI-Census and FFR-Census, respectively.
The Census of Manufactures is published only during years when a U.S.
Census takes place. We use the U.S. Census data for a given year as a proxy
for industry concentration not only for that year but also for the one or
two years immediately before and after it. This approach is similar to that
used in several prior studies (e.g., Aggarwal and Samwick 1999; MacKay and
Phillips 2005; Campello 2006; Haushalter, Klasa, and Maxwell 2007). Table
1 provides information on the time periods to which we apply a given years
U.S. Census data. For example, we use the 1992 Census of Manufactures data
as a proxy for industry concentration for the period 19901994. Data for FFRCensus are available in all Census of Manufactures publications, resulting in a
sample period of 19632005. However, data for HI-Census are available only
from Census of Manufactures publications from 1982 on, resulting in a sample
period of 19802005.
Table 1
Sample periods to which a particular years Census of Manufactures data on industry concentration are
applied
Years for which the Census of
Manufactures reports data
Sample periods to which Census of Manufactures

data on industry concentration are applied
19631964
19651966
19671968
19691970
19711974
19751979
19801984
19851989
19901994
19951999
20002005
The Census of Manufactures is published by the U.S. Census Bureau.
Table 2
Descriptive statistics of industry concentration measures
Mean
Median
STD
20%
40%
60%
80%
0.032
0.596
0.058
0.857
0.104
1.000
0.290
1.000
0.411
1.000
0.560
1.000
0.027
0.272
0.053
0.402
0.103
0.559
Panel A: 19802005 sample period

Industry at four-digit SIC level
HI-Census
HI-Compustat
0.064
0.696
0.043
0.714
0.062
0.278
0.015
0.410
Panel B: 19632005 sample period

FFR-Census
FFR-Compustat
0.382
0.969
0.350
1.000
0.208
0.079
0.195
0.970
Panel C: 19952005 sample period

HI-Census
FFR-Census
0.061
0.368
0.040
0.327
0.063
0.217
0.012
0.159
HI-Census is the Herfindahl index for four-digit SIC industries as reported by the Census of Manufactures. The
1997 and 2002 U.S. Censuses define industry at the six-digit NAICS level. Over the 19952005 period, we use
HI-Census values for six-digit NAICS industries to calculate HI-Census values for four-digit SIC industries by
weighting the HI-Census values of component six-digit NAICS industries by the square of their share of the
broader four-digit SIC industry. To determine what are the component six-digit NAICS industries of a broader
four-digit SIC industry, we use NAICS correspondence tables provided by the U.S. Census. HI-Compustat is the
sum of the squares of the sales market shares of all firms in a CRSP four-digit SIC industry that have sales data
on Compustat. For each year t, HI-Compustat is averaged over a three-year period, from year t 2 to year t.
FFR-Census is the sum of the market shares of the four largest firms in terms of market share in a four-digit
SIC industry as defined by the Census of Manufactures. Over the 19952005 period, FFR-Census values for
six-digit NAICS industries are used to approximate FFR-Census for broader four-digit SIC industries by first
determining which component six-digit NAICS industry of a broader four-digit SIC industry has the largest sales
as measured by the sales of its top four firms. Next, we divide the sales of the top four firms of this six-digit
NAICS industry by the total sales of the firms in all the component six-digit NAICS industries within the broader
four-digit SIC industry. FFR-Compustat is the sum of the market shares of the four largest firms in terms of
market share in a CRSP four-digit SIC industry. A firms market share is measured as sales divided by total sales
of all CRSP firms in that industry that have sales data on Compustat. For each year t, FFR-Compustat is averaged
over a three-year period from year t 2 to year t. Descriptive statistics for industry concentration measures
are calculated by pooling Compustat- and U.S. Census-based industry-year observations. Observations used to
calculate HI-Census or HI-Compustat are taken from all years within specific sample periods.
3846
1963
1966
1967
1970
1972
1977
1982
1987
1992
1997
2002
For instance, if there are J six-digit NAICS industries that belong to one four-digit SIC industry, the six-digit
NAICS Herfindahl index values are weighed by the square of the shares of each component six-digit NAICS
industry. Thus, if the broader four-digit SIC industry is called p, then the Herfindahl index of industry p,
H H Ip =
J

j=1
2j H H I j ,
where j =
total shipments of component 6 digit NAICS industryj

.
total shipments of broader 4 digit SIC industryp
To determine what are the component six-digit NAICS industries of a broader four-digit SIC industry, we use
NAICS correspondence tables provided by the U.S. Census.
We refer to this method as an approximation method because it is possible that a component six-digit NAICS
industry that does not have the largest sales as measured by the sales of its top four firms has a firm whose sales
is greater than at least one of the top four firms of the component six-digit NAICS industry with the largest value
for the sales of its top four firms.
3847
over the 19802005 period. The difference between these two measures is striking. The mean values of HI-Census and HI-Compustat are 0.064 and 0.696,
respectively. Panel B compares FFR-Census and FFR-Compustat for four-digit
SIC industries over the 19632005 period. The mean values of FFR-Census
and FFR-Compustat are 0.382 and 0.969, respectively. The large differences
between the values of the U.S. Census- and Compustat-based concentration
measures indicate that on average private firms account for a significant percentage of industry sales and that it is therefore problematic to exclude data
on these firms when calculating industry concentration measures. The fortieth
percentile value for FFR-Compustat is 1.000, showing that the majority of
four-digit SIC industries have four or fewer firms with sales data available on
Compustat.
As shown in Table 1, for the period 19952005, we use concentration ratios
from the 1997 and 2002 Census of Manufactures publications in which industry is defined using six-digit NAICS codes. HI-Census for six-digit NAICS
industries and the total shipments for these industries reported in the Census
of Manufactures can be used to calculate HI-Census for broader four-digit
SIC industries. We do this by weighting HI-Census of the component six-digit
NAICS industries by the square of their share of the shipments of the broader
four-digit SIC industry.6,7 To calculate FFR-Census for four-digit SIC industries using FFR-Census of component six-digit NAICS industries, we use an
approximation method. We first determine the component six-digit NAICS industry of a broader four-digit SIC industry that has the largest value for the
sales of its top four firms. Next, we divide the sales of the top four firms of
this six-digit NAICS industry by the total sales of all the component six-digit
NAICS industries within the broader four-digit SIC industry.8
Panel C of Table 2 provides descriptive statistics for the 19952005 period
for HI-Census and for FFR-Census at the four-digit SIC level. The statistics
for these measures are very similar to those reported in panels A and B. These
results suggest that our methods for converting the six-digit NAICS level concentration measures to four-digit SIC level measures are reasonable.
2. Evidence that Compustat-based industry concentration measures are poor

proxies for actual industry concentration
Industry Markup =
(Value of Sales + Inventories Payroll Cost of Materials)
.
(Value of Sales + Inventories)
We collect annual industry data at the four-digit SIC level from 1993, 1994,
1995, and 1996 Annual Survey of Manufacturers publications and calculate
industry markups for the period from 1993 to 1996. For this period, we form
quintiles based on HI-Census and HI-Compustat and calculate median industry
markups for each of the quintiles.
The results in panel A of Table 3 show that industry markups are higher in
industries with higher values of HI-Census. In industries that are in the highest
quintile of HI-Census, industry markups are almost 25% larger than they are
in industries in the lowest quintile of HI-Census. In contrast, panel B shows
that industry markups are not systematically related to HI-Compustat. These
results suggest that U.S. Census-based measures are better proxies for actual
industry concentration than are Compustat-based measures.
Next, we compute for each quintile the median number of public and private
firms per industry, based on U.S. Census data, Nfirms-Census. Panel A reports
that for HI-Census-based quintiles 1 and 5, the median Nfirms-Census values
are 1385 and 88, respectively. Panel B shows that for HI-Compustat-based
3848
Table 3 reports firm and industry characteristics for quintiles sorted by HICensus and HI-Compustat. The panel A quintiles are based on HI-Census. The
median value of HI-Census for quintile 1 is 0.009 and for quintile 5 is 0.153,
about fifteen times larger. However, the corresponding values of HI-Compustat
are quite similar, 0.659 and 0.891, respectively. The panel B quintiles are based
on HI-Compustat. The median value of HI-Compustat for quintile 1 is 0.311 and
for quintile 5 is 1.000, about three times larger. However, the corresponding
values of HI-Census are quite similar: 0.041 and 0.059, respectively. These
results suggest that the exclusion of data on private firms not only leads to
large differences between Compustat- and U.S. Census-based concentration
measures but also leads to a low correlation between the two types of measures.
Next, we determine the relation of industry markups with HI-Census and
HI-Compustat. Industry markups represent average price-cost margins in an
industry. Industrial organization theory predicts that in more-concentrated industries there is less intense competition and price is consequently set further
away from marginal cost. Thus, a positive relation is expected between industry
concentration and price-cost margins. We follow Allayannis and Ihrig (2001)
and calculate industry markups using aggregate industry-level data from Annual Survey of Manufacturers publications. We also use their definition for
industry markups, which is as follows:
HI-Census
HI-Compustat
Industry Markup
Nfirms-Census
1
2
3
4
5
0.009
0.025
0.045
0.079
0.153
0.659
0.680
0.590
0.744
0.891
0.299
0.314
0.315
0.321
0.369
1385
518
356
164
88
1
2
3
4
5
0.041
0.042
0.044
0.046
0.059
0.311
0.516
0.706
0.929
1.000
0.319
0.313
0.305
0.351
0.328
535
423
413
287
211
Nfirms-Compustat
Nfirms-Compustat as a
percentage of
Nfirms-Census
Net Sales
Market Capitalization
Book Assets
217
304
298
379
911
94
141
181
188
651
152
264
253
318
691
388
274
306
470
304
200
138
178
225
187
328
206
272
387
219
Panel A: Quintiles based on HI-Census

2
3
4
2
2
0.19%
0.56%
0.82%
1.46%
2.20%
Panel B: Quintiles based on HI-Compustat

9
4
3
2
1
1.79%
1.09%
0.78%
0.69%
0.55%
3849
The sample period is 19802005. Median values are reported. HI-Census is the Herfindahl index for four-digit SIC industries as reported by the Census of Manufactures. The 1997 and
2002 U.S. Censuses define industry at the six-digit NAICS level. Over the 19952005 period, we use HI-Census values for six-digit NAICS industries to calculate HI-Census values for
four-digit SIC industries by weighting the HI-Census values of component six-digit NAICS industries by the square of their share of the broader four-digit SIC industry. To determine what
are the component six-digit NAICS industries of a broader four-digit SIC industry we use NAICS correspondence tables provided by the U.S. Census. HI-Compustat is the sum of the
squares of the sales market shares of all firms in a CRSP four-digit SIC industry that have sales data on Compustat. For each year t, HI-Compustat is averaged over a three-year period from
year t 2 to year t. Industry Markup is calculated using data collected from 1993, 1994, 1995, and 1996 Annual Survey of Manufactures publications and is calculated as (Value of Sales +
Inventories Payroll Cost of Materials) / (Value of Sales + Inventories). For the analysis of Industry Markup we form HI-Census and HI-Compustat based quintiles over the
19931996 period. Nfirms-Census is the number of firms per four-digit SIC industry as reported by the Census of Manufactures. Over the 19952005 period we use Nfirms-Census values
for six-digit NAICS industries to calculate Nfirms-Census values for four-digit SIC industries by summing the Nfirms-Census values of component six-digit NAICS industries in a broader
four-digit SIC industry. Nfirms-Compustat is the total number of CRSP firms in a four-digit SIC industry that are included on Compustat. Net Sales is defined as net sales in millions in year t.
Market Capitalization is defined as market value of equity in millions at the end of year t. Book Assets is the book value of total assets at the end of year t. Net Sales, Market Capitalization,
and Book Assets are inflation adjusted. Descriptive statistics are calculated by pooling firm-year observations.
Table 3
Firm characteristics of portfolios sorted by industry concentration measures
3850
quintiles 1 and 5, the median Nfirms-Census values are 535 and 211, respectively. Thus, the difference in Nfirms-Census between quintiles 1 and 5 is far
greater when the quintiles are based on HI-Census than when they are based
on HI-Compustat. Given that more-concentrated industries, which are presumably less competitive, should be populated with a smaller number of firms,
the above results further suggest that HI-Census is a better indicator of true
industry concentration than is HI-Compustat.
Table 3 also reports for each of the quintiles the median value of NfirmsCompustat, defined as the number of firms per industry that have data available
on CRSP and Compustat. This number represents the number of firms per industry used to calculate Compustat-based industry concentration measures. In
all cases, median Nfirms-Compustat is markedly lower than is median NfirmsCensus. For example, for quintile 1 of panel A, median Nfirms-Census is 1385,
but median Nfirms-Compustat is only 2. This finding suggests that an important
contributing factor to why HI-Compustat could be a poor indicator of industry
concentration is that this measure is based on partial data. Furthermore, median
Nfirms-Compustat does not vary systematically across the HI-Census quintiles,
suggesting that there is no relationship between actual industry concentration
and Nfirms-Compustat. However, panel B shows that median Nfirms-Compustat
markedly decreases from HI-Compustat-based quintiles 15. This result indicates that Compustat-based industry concentration measures are more likely
to proxy for the number of firms in an industry that are covered by CRSP and
Compustat than proxy for true industry concentration.
Additionally, Table 3 presents for each of the quintiles the median percentage
of firms in an industry reported by the U.S. Census that are included on CRSP
and Compustat (Nfirms-Compustat as a percentage of Nfirms-Census). Panel
A shows that this percentage increases substantially from HI-Census-based
quintiles 15. This is consistent with the expectation that in more-concentrated
industries there should be a greater percentage of large, public firms, the type
that are likely to be included on both CRSP and Compustat. Panel B, in contrast,
shows that the percentage of U.S. Census firms that are included on CRSP and
Compustat does not increase from HI-Compustat-based quintiles 15. Instead,
this percentage decreases over these quintiles. This finding provides further
support to the notion that Compustat-based industry concentration measures
are poor indicators of actual industry concentration.
Next, we examine how firm size varies across quintiles sorted by HI-Census
and by HI-Compustat. We consider three measures of size: net sales, market
capitalization, and book assets. Each of these measures is inflation adjusted.
Data for these variables are obtained from Compustat. Thus, the median values
of each of the firm size variables reported in Table 3 are not the medians across
all U.S. Census firms but are the medians for the subset of firms covered by
the CRSP and Compustat databases. Consequently, inferences based on these
variables should be viewed with caution. Panel A shows that median net sales
(market capitalization, book assets) for firms in the HI-Census-based quintiles
Table 4
Correlations between industry concentration measures and industry markup
HI-Census
HI-Census
HI-Compustat
Industry Markup
0.111
(0.000)
0.183
(0.000)
HI-Compustat
Industry Markup
0.129
(0.000)
0.220
(0.000)
0.016
(0.586)
0.019
(0.519)
1 and 5 are $217m ($94m, $152m) and $911m ($651m, $691m), respectively.
These results indicate that, consistent with theoretical expectations, in lessconcentrated industries, which are likely to be more competitive, firm size is
substantially smaller than it is in highly concentrated less-competitive industries. Panel B shows that median net sales (market capitalization, book assets)
for firms in HI-Compustat-based quintiles 1 and 5 are $388m ($200m, $328m)
and $304m ($187m, $219m), respectively. These findings show that firm size
is actually smaller in the highest HI-Compustat quintile than it is in the lowest
quintile, further suggesting that the Compustat-based industry concentration
measures are poor proxies for actual industry concentration.
Table 4 reports correlations between HI-Census, HI-Compustat, and Industry
Markup. The table first confirms the result in Table 3 that the correlation
between HI-Census and HI-Compustat is low. Specifically, the Spearman and
Pearson correlations between the two variables are only 0.111 and 0.129.9
The table also confirms the results in Table 3 that the correlation between HICensus and Industry Markup is positive and that HI-Compustat and Industry
Markup are not systematically related. These results are further evidence that
We find that the Spearman and Pearson correlations between HI-Census and HI-Compustat for the period from
1980 to 1994 are 0.126 and 0.135, respectively. These results rule out the possibility that the low correlation
between HI-Census and HI-Compustat reported in Table 4 is due to the fact that for the period from 1995 to
2005, we convert HI-Census values for six-digit NAICS industries into values for four-digit SIC industries.
3851
The sample period is 19802005. This table presents Pearson (above the diagonal) and Spearman (below the
diagonal) correlations among selected variables. Numbers in parentheses are significance levels. HI-Census
is the Herfindahl index for four-digit SIC industries as reported by the Census of Manufactures. The 1997
and 2002 U.S. Censuses define industry at the six-digit NAICS level. Over the 19952005 period we use
HI-Census values for six-digit NAICS industries to calculate HI-Census values for four-digit SIC industries by weighting the HI-Census values of component six-digit NAICS industries by the square of their
share of the broader four-digit SIC industry. To determine what are the component six-digit NAICS industries of a broader four-digit SIC industry we use NAICS correspondence tables provided by the U.S.
Census. HI-Compustat is the sum of the squares of the sales market shares of all firms in a CRSP fourdigit SIC industry that have sales data on Compustat. For each year t, HI-Compustat is averaged over a
three-year period from year t 2 to year t. Industry Markup is calculated using data collected from 1993,
1994, 1995, and 1996 Annual Survey of Manufactures publications and is calculated as (Value of Sales +
Inventories Payroll Cost of Materials) / (Value of Sales + Inventories). Correlations involving Industry
Markup are calculated over the 19931996 period.
Table 5
Industries sorted by U.S. Census based Herfindahl index values
Two-digit SIC code
HI-Census
HI-Compustat
Lumber & Wood Products

Apparel
Fabricated Metal Products
Furniture and Fixtures
Miscellaneous Manufacturing
Printing and Publishing
Machinery and Computer Equipment
Rubber and Plastics
Petroleum Refining
Measuring Instruments
Paper
Primary Metal
Textile Product Mills
Electronic Equipment
Food and Kindred Products
Stone, Clay, Glass, and Concrete
Chemicals
Leather
Transportation Equipment
Tobacco
0.027
0.029
0.037
0.039
0.048
0.053
0.054
0.054
0.058
0.059
0.060
0.063
0.063
0.078
0.078
0.087
0.091
0.092
0.096
0.153
0.691
0.731
0.724
0.697
0.753
0.617
0.648
0.596
0.636
0.590
0.700
0.659
0.787
0.709
0.763
0.793
0.630
0.738
0.669
0.837
This table reports mean values of HI-Census and HI-Compustat for four-digit SIC industries within a particular
two-digit SIC industry. The industries are listed in ascending order of HI-Census. HI-Census is the Herfindahl
index for four-digit SIC industries as reported by the Census of Manufactures. The 1997 and 2002 U.S. Censuses
define industry at the six-digit NAICS level, and consequently for these two years we use HI-Census values
for six-digit NAICS industries to calculate HI-Census values for four-digit SIC industries by weighting the HICensus values of component six-digit NAICS industries by the square of their share of the broader four-digit SIC
industry. To determine what are the component six-digit NAICS industries of a broader four-digit SIC industry
we use NAICS correspondence tables provided by the U.S. Census. HI-Compustat is the sum of the squares of
the sales market shares of all firms in a CRSP four-digit SIC industry that have sales data on Compustat. For
each year t, HI-Compustat is averaged over a three-year period from year t 2 to year t. The observations used
to calculate mean HI-Census or HI-Compustat for two-digit SIC industries are taken only from years when a
U.S. Census takes place (1982, 1987, 1992, 1997, and 2002).
Compustat-based industry concentration measures are not good proxies for

actual industry concentration.10
In Table 5, we report average values of HI-Census and HI-Compustat for
four-digit SIC industries within particular two-digit SIC industry groups. These
data allow us to provide information on what the typical Herfindahl index value
is for a four-digit SIC industry within a broader two-digit SIC industry. Also,
this way of reporting our findings makes it easier to comprehend the data given
the large number of four-digit SIC industries within the manufacturing sector.
The industries are listed in ascending order of HI-Census. There is a large
difference between HI-Census and HI-Compustat for every industry. Further,
our results show that although HI-Census increases significantly as one moves
10
We also examine the correlation between industry markups and the four-firm ratios. We find that the Pearson
and Spearman correlations between Industry Markup and FFR-Census are positive and significant and equal
0.198 and 0.179, respectively. However, the Spearman and Pearson correlations between Industry Markup and
FFR-Compustat are negative and significant and equal 0.082 and 0.123, respectively.
3852
24
23
34
25
39
27
35
30
29
38
26
33
22
36
20
32
28
31
37
21
Industry name
from less- to more-concentrated industries based on this measure, there is no

systematic variation in HI-Compustat along these industries, reflecting the low
correlation between the two measures.
3. Reexamination of results obtained in prior studies that use
Compustat-based industry concentration measures
3.1 Hou and Robinson (2006)

Hou and Robinson (2006) find that firms in more-concentrated industries earn
lower future stock returns. They argue that barriers to entry in highly concentrated industries insulate firms from undiversifiable distress risk, which is
priced in equity returns. We examine the sensitivity of their results to using
U.S. Census-based industry concentration measures in place of the Compustat
measures.
Following Hou and Robinson (2006), we estimate firm-level Fama-MacBeth
cross-sectional regressions of the model in their panel B of Table 4. Specifically,
we regress future monthly stock returns from July of year t + 1 to June of year
t + 2 on HI-Compustat and other firm characteristics. As did Hou and Robinson
(2006), we calculate HI-Compustat at the three-digit SIC level and measure it
as the mean value of HI-Compustat over the three prior years, t 2, t 1,
and t. Hou and Robinson (2006) use the sample period from 1963 to 2001.
Because data on HI-Census from the Census of Manufactures are available
starting in 1982 and, as shown in Table 1, the earliest year to which we can
apply these data is 1980, we study the period from 1980 to 2001. Also, data
on HI-Census are available only for manufacturing firms. So that we can make
better comparisons, we examine only manufacturing firms when estimating
results with the HI-Compustat measure.
The regression results in the first column of Table 6 document a significant
negative relationship between HI-Compustat and future stock returns. This
result is similar to that reported in Hou and Robinson (2006) and is the basis of
their conclusion that firms in more-concentrated industries earn lower returns
3853
Results that we have presented so far suggest that U.S. Census-based industry
concentration measures are superior to Compustat-based industry concentration measures in measuring actual industry concentration and that there is a
very low correlation between the two measures. It is therefore important to examine whether the results of prior empirical studies that use Compustat-based
industry concentration measures are robust to the use of U.S. Census-based industry concentration measures. We reexamine the results in Hou and Robinson
(2006); Lang and Stulz (1992); Harris (1998); and Defond and Park (1999). We
reexamine these four papers for the following reasons. First, we are able to collect the necessary data to replicate these studies. Second, the issues addressed in
these studies are from four different finance-related areas: asset pricing, capital
structure, corporate disclosure policy, and corporate governance.
Table 6
The relationship between stock returns and industry concentration
1
1.408 (3.70)
0.342 (1.93)
1.518 (4.45)
0.115 (0.24)
2.170 (6.67)
1.840 (6.92)
0.342 (2.00)
0.014 (0.24)
0.016 (0.29)
0.113 (2.65)
0.054 (0.47)
0.111 (2.61)
0.258 (2.57)
0.660 (3.26)
0.028 (0.16)
0.025 (0.09)
0.032
0.299 (3.22)
0.644 (3.59)
0.098 (0.61)
0.059 (0.23)
0.032
0.220 (2.99)
0.573 (3.28)
0.128 (0.97)
0.072 (0.35)
0.049
0.218 (2.95)
0.570 (3.27)
0.124 (0.94)
0.053 (0.26)
0.049
991
991
936
936
This table presents results from Fama-MacBeth cross-sectional regressions explaining monthly stock returns,
estimated monthly between July 1980 and June 2001 for models 1 and 2 and between July 1963 and June 2001
for models 3 and 4. Monthly returns are the one-year-ahead twelve monthly returns from July of year t + 1 to
June of year t + 2. Industry concentration ratios and accounting data are measured at the end of year t. HI-Census
in this table represents the Herfindahl index for three-digit SIC industries calculated using data collected from
Census of Manufactures publications. Prior to 1997, the U.S. Census defines industry using four-digit SIC codes,
while from 1997 onward industry is defined using six-digit NAICS codes. Over the 19801994 period, we use
HI-Census values for four-digit SIC industries to calculate HI-Census values for three-digit SIC industries by
weighting the HI-Census values of component four-digit SIC industries by the square of their share of the broader
three-digit SIC industry. Over the 19952001 period, we use HI-Census values for six-digit NAICS industries to
calculate HI-Census values for three-digit SIC industries by weighting the HI-Census values of component sixdigit NAICS industries by the square of their share of the broader three-digit SIC industry. To determine what are
the component six-digit NAICS industries of a broader three-digit SIC industry we use NAICS correspondence
tables provided by the U.S. Census. HI-Compustat is the sum of the squares of the sales market shares of all
firms in a CRSP three-digit SIC industry that have sales data on Compustat. For each year t, HI-Compustat is
averaged over a three-year period from year t 2 to year t. FFR-Census in this table represents the four-firm ratio
for four-digit SIC industries calculated using data collected from Census of Manufactures publications. Over the
19952001 period FFR-Census values for six-digit NAICS industries are used to approximate FFR-Census for
broader four-digit SIC industries by first determining which component six-digit NAICS industry of a broader
four-digit SIC industry has the largest sales as measured by the sales of its top four firms. Next, we divide the
sales of the top four firms of this six-digit NAICS industry by the total sales of the firms in all the component
six-digit NAICS industries within the broader four-digit SIC industry. FFR-Compustat is the sum of the market
shares of the four largest firms in terms of market share in a CRSP four-digit SIC industry. A firms market share
is measured as sales divided by total sales of all CRSP firms in that industry that have sales data on Compustat.
For each year t, FFR-Compustat is averaged over a three-year period from year t 2 to year t. Ln(Market
Capitalization) is defined as the natural logarithm of the market value of equity in millions at the end of year t.
Ln(B/M) is defined as the natural logarithm of book value of equity divided by market value of equity at the end
of year t. Momentum for each month is prior one-year stock returns. Beta is market model beta estimated using
the prior thirty-six monthly equally weighted CRSP index returns. Leverage is defined as total debt divided by
market value of total assets (i.e., market value of equity plus debt) at the end of year t and is trimmed at the 1%
level. Time-series average values of the monthly regression coefficients are reported with time-series t-statistics
in parentheses. , , and represent significance at the 1%, 5%, and 10% levels, respectively, for a two-tailed
test.
because these industries are less risky. The second column of the table presents
the regression results of the same model as in the first column, except that the
three-digit SIC level HI-Compustat is replaced with the three-digit SIC level
HI-Census. We find that future stock returns are not associated with HI-Census.
Thus, this result does not support the Hou and Robinson (2006) conclusion that
industry concentration is related to future stock returns.
3854
Intercept
HI-Compustat
HI-Census
FFR-Compustat
FFR-Census
Ln(Market
Capitalization)
Ln(B/M)
Momentum
Beta
Leverage
Average adjusted
R-square
Average number of
observations
11
Our results remain qualitatively the same when we use FFR-Compustat for three-digit industries and approximate
FFR-Census for three-digit SIC industries, using a methodology similar to that discussed in Section 1.3 of the
article. Further, if HI-Compustat and HI-Census are defined at the four-digit SIC level instead of the three-digit
SIC level, as in the first two columns of Table 6, the results remain qualitatively the same. We also examine
whether our conclusions are sensitive to defining industry at the six-digit NAICS level. Since the U.S. Census
measures for six-digit NAICS industries are available for 1997 and 2002 and we can apply these data for the
period from 1995 to 2001, we use this sample period in our analysis. We continue to find that future stock
returns are significantly negatively associated with HI-Compustat and FFR-Compustat, but are not significantly
associated with HI-Census and FFR-Census.
12
We also examine the sensitivity of the results to applying a given years U.S. Census industry concentration
data to the surrounding one or two years. We estimate the models in columns 2 and 4 of Table 6 only for those
years to which the U.S. Census data belong and find that the coefficients on HI-Census and FFR-Census remain
insignificant.
13
In footnote 6 of their paper, Hou and Robinson (2006) report that for a smaller sample of observations their
univariate analysis results suggest that U.S. Census-based industry concentration measures are negatively related
to future stock returns. However, as we have shown in this article, our finding that these measures are not
significantly related to future stock returns is quite robust.
3855
The third and fourth columns of Table 6 provide results of regression models
that are similar to the models in the first two columns of this table, except that
the four-firm ratio is used as the measure of industry concentration rather than
the Herfindahl index. This analysis provides evidence on the sensitivity of our
conclusions to the use of an alternative definition of industry concentration.
Another benefit of this analysis is that we have data on FFR-Census for the
19632001 period, which is the sample period used in Hou and Robinson
(2006). We use the four-firm ratio at the four-digit SIC level because FFRCensus is reported at the four-digit SIC level and converting this measure to
the three-digit SIC level involves approximation.11 The results show that FFRCompustat is significantly negatively associated with future stock returns. In
contrast, there is no association between FFR-Census and future stock returns,
further suggesting that the Hou and Robinson (2006) conclusion that industry
concentration is related to future stock returns may not be valid.12,13
Hou and Robinson (2006) also document that firms in less-concentrated
industries have higher research and development expenses. They posit that this
result is consistent with the Schumpeter (1912) proposition that innovation as
a form of creative destruction is more likely to occur in competitive industries.
Also, they argue that if innovation risk is priced into stock returns and this risk
is higher in more-competitive industries, it contributes to the higher cost of
equity capital in such industries.
We examine whether the relationship between industry concentration and
research and development expenses is sensitive to using Compustat or U.S.
Census data to measure industry concentration. The model we estimate is
from panel B of Table 2 of Hou and Robinson (2006), which relates industry
concentration to certain industry characteristics. The dependent variables for
the models in columns 1 and 2 of Table 7 are HI-Compustat and HI-Census,
and the sample period is 19802001.
Table 7
The relationship between Compustat- and U.S. Census-based industry concentration measures with
industry average research and development expenses and industry averages of other firm characteristics
Dependent variable
2
HI-Census
3
FFR-Compustat
4
FFR-Census
0.782 (12.03)
1.574 (4.17)
0.024 (4.23)
0.269 (4.98)
0.980 (48.54)
0.232 (6.28)
0.087 (6.25)
0.734 (8.16)
0.021 (7.03)
0.016 (15.37)
0.004 (10.36)
0.060 (27.13)
0.607 (2.13)
0.544 (1.80)
0.081 (2.69)
0.069 (1.09)
0.065 (2.99)
0.170 (3.80)
0.124 (1.44)
0.355 (3.20)
0.037 (2.23)
0.072 (3.16)
0.089 (1.75)
0.057
0.010 (2.82)
0.011 (3.04)
0.013 (1.16)
0.134
0.004 (1.61)
0.010 (6.79)
0.019 (3.10)
0.005
0.036 (6.62)
0.013 (1.99)
0.073 (4.27)
0.137
269
269
118
118
Models 1 and 2 present results from Fama-MacBeth cross-sectional regressions estimated using annual data
from 19802001. Models 3 and 4 present results from Fama-MacBeth cross-sectional regressions estimated
using annual data from 1963 to 2001. Industry concentration ratios and firm characteristics are measured at the
end of year t. HI-Census in this table represents the Herfindahl index for three-digit SIC industries calculated
using data collected from Census of Manufactures publications. Prior to 1997, the U.S. Census defines industry
using four-digit SIC codes, while from 1997 onward industry is defined using six-digit NAICS codes. Over
the 19801994 period, we use HI-Census values for four-digit SIC industries to calculate HI-Census values
for three-digit SIC industries by weighting the HI-Census values of component four-digit SIC industries by the
square of their share of the broader three-digit SIC industry. Over the 19952001 period, we use HI-Census values
for six-digit NAICS industries to calculate HI-Census values for three-digit SIC industries by weighting the HICensus values of component six-digit NAICS industries by the square of their share of the broader three-digit SIC
industry. To determine what are the component six-digit NAICS industries of a broader three-digit SIC industry,
we use NAICS correspondence tables provided by the U.S. Census. HI-Compustat is the sum of the squares of
the sales market shares of all firms in a CRSP three-digit SIC industry that have sales data on Compustat. For
each year t, HI-Compustat is averaged over a three-year period from year t 2 to year t. FFR-Census in this
table represents the four-firm ratio for four-digit SIC industries calculated using data collected from Census of
Manufactures publications. Over the 19952001 period, FFR-Census values for six-digit NAICS industries are
used to approximate FFR-Census for broader four-digit SIC industries by first determining which component
six-digit NAICS industry of a broader four-digit SIC industry has the largest sales as measured by the sales of
its top four firms. Next, we divide the sales of the top four firms of this six-digit NAICS industry by the total
sales of the firms in all the component six-digit NAICS industries within the broader four-digit SIC industry.
FFR-Compustat is the sum of the market shares of the four largest firms in terms of market share in a CRSP
four-digit SIC industry. A firms market share is measured as sales divided by total sales of all CRSP firms in
that industry that have sales data on Compustat. For each year t, FFR-Compustat is averaged over a three-year
period from year t 2 to year t. All independent variables are industry averages of firm-level characteristics.
Ln(Market Capitalization) is defined as the natural logarithm of the market value of equity in millions at the end
of year t. Ln(B/M) is defined as the natural logarithm of book value of equity divided by market value of equity
at the end of year t. Beta is market model beta estimated using the prior thirty-six monthly equally weighted
CRSP index returns. Leverage is defined as total debt divided by market value of total assets (i.e., market value
of equity plus debt) at the end of year t. R&D expense/book assets, Earnings/book assets, and Dividends/book
equity are as of year t and are trimmed at the 1% level. Leverage is also trimmed at the 1% level. Time-series
average values of the monthly regression coefficients are reported with time-series t-statistics in parentheses. ,
, and represent significance at the 1%, 5%, and 10% levels, respectively, for a two-tailed test.
The first column of Table 7 shows the replication of the Hou and Robinson
(2006) finding that HI-Compustat is significantly negatively associated with
research and development expenses. The second column presents estimates of
the model that replaces the dependent variable HI-Compustat with HI-Census.
We find a significant positive association between HI-Census and research and
3856
Intercept
R&D expense/book
assets
Ln(Market
Capitalization)
Earnings/book assets
Dividends/book
equity
Ln(B/M)
Beta
Leverage
Average adjusted
R-square
Average number of
observations
1
HI-Compustat
3.2 Lang and Stulz (1992)

Lang and Stulz (1992) examine the intra-industry effects of bankruptcy announcements. They show that in industries with high leverage that are less
concentrated, there is a significant and important negative price reaction to a
bankruptcy announcement in the industry. They argue that this price reaction
reflects the loss experienced by other firms in the industry, because the announcement conveys information about lower future cash flows for these firms.
They refer to this effect as a contagion effect. They also find that industries
with low leverage that are more concentrated exhibit significantly positive price
reactions to a bankruptcy announcement in the industry. They contend that this
price reaction reflects the benefits to competitors that result from the difficulties
faced by the bankrupt firm and refer to it as a competitive effect.
We investigate whether the Lang and Stulz (1992) results, reported in their
Tables 3 and 4, are robust to the use of U.S. Census-based industry concentration measures. They examine fifty-nine bankruptcies from 1970 to 1989.
Our sample of bankrupt firms comes from the current Altman-NYU Salomon
Center Bankruptcy list, which is an updated version of the data used by Lang
and Stulz (1992). We examine eighty-six bankruptcies that took place in the
manufacturing sector from 1980 to 2004. As in Lang and Stulz (1992), we
3857
development expenses. If U.S. Census-based industry concentration measures

are more appropriate measures of actual industry concentration, then this result
indicates that innovation risk is actually higher in more-concentrated industries.
This finding is consistent with the claim made in Schumpeter (1942) that there is
more innovation in less-competitive industries because firms in such industries
can enjoy economic profits resulting from their innovation, instead of having
these profits competed away. However, this finding is not supportive of the Hou
and Robinson (2006) claim that higher innovation risk in less-concentrated
industries raises the overall cost of capital in these industries.
There is another notable difference in the results in columns 1 and 2 in Table 7.
Similar to Hou and Robinson (2006), we find a significant negative association
between HI-Compustat and firm size. However, HI-Census is significantly
positively associated with firm size. Given that more-concentrated industries
are expected to have on average larger firms, these results provide further
support to the arguments made in section 2 of the article that U.S. Census-based
measures of industry concentration are more meaningful than are Compustatbased measures.
Columns 3 and 4 of Table 7 present results of the models that use FFRCompustat and FFR-Census, respectively, as the dependent variables. Also,
these results are for the longer sample period from 1963 to 2001, the sample
period used in Hou and Robinson (2006). The relations of FFR-Compustat
and FFR-Census with research and development expenses and firm size are
qualitatively the same as those reported in columns 1 and 2 of Table 7.
Table 8
Abnormal returns for subsamples of industry portfolios for the eleven days around a bankruptcy
announcement
Industry portfolio
characteristics
No. of industry portfolios

with industry
characteristics below/above
the sample median
Average abnormal returns for the subsample

of industry portfolios with the value of the
industry portfolio characteristics below/above
the sample median
Below
Above
Panel A: Compustat-based industry concentration measures

43/43
0.140
(0.101)
HI-Compustat
43/43
2.891
(1.920)
HI-Compustat (subsample of
industry portfolios with
below-median leverage)
HI-Compustat (subsample of
above-median leverage)
19/24
2.700
(0.931)
24/19
3.042
(2.149)
2.319
(2.587)
[2.278]
0.432
(0.646)
[1.869]
1.887
(2.038)
[2.736]
1.406
(1.466)
[1.457]
Panel B: U.S. Census-based industry concentration measures

HI-Census
43/43
1.328
(1.385)
HI-Census (subsample of
below-median leverage)
HI-Census (subsample of
above-median leverage)
18/25
0.670
(0.428)
25/18
1.802
(1.498)
1.131
(0.844)
[0.154]
0.242
(0.116)
[0.513]
3.038
(2.267)
[0.403]
Cumulative abnormal returns (market model errors) are calculated from day 5 to +5 relative to bankruptcy
announcements. Bankruptcy announcements are from the Altman-NYU Salomon Center Bankruptcy list and
include bankruptcies between January 1980 and December 2004 with liabilities greater than $100 million for
firms in the manufacturing sector (SIC codes between 2000 and 3999) with a primary SIC code that is available
from Compustat (eighty-six bankruptcies). An industry portfolio is a value-weighted portfolio of firms with
the same primary four-digit SIC code as the bankrupt firm for which announcement returns are available from
the CRSP files. Except for HI-Census, industry characteristics are obtained from Compustat for the fiscal year
preceding the announcement. HI-Census is the Herfindahl index for four-digit SIC industries as reported by the
Census of Manufactures. The 1997 and 2002 U.S. Censuses define industry at the six-digit NAICS level. Over
the 19952004 period, we use HI-Census values for six-digit NAICS industries to calculate HI-Census values
for four-digit SIC industries by weighting the HI-Census values of component six-digit NAICS industries by
the square of their share of the broader four-digit SIC industry. To determine what are the component six-digit
NAICS industries of a broader four-digit SIC industry, we use NAICS correspondence tables provided by the
U.S. Census. HI-Compustat is defined as the sum of the squares of the sales market shares of all firms in a
Compustat four-digit SIC industry. Leverage is the debt-to-total assets ratio. The numbers in parentheses are
z-statistics and the numbers in square brackets are z-statistics for differences between subsamples. , , and
represent significance at the 1%, 5%, and 10% levels, respectively, for a two-tailed test.
define industry using four-digit SIC codes and calculate announcement returns
from days 5 to +5 relative to bankruptcy announcements.
Panel A of Table 8 reports our replication of the Lang and Stulz (1992)
univariate results. First, the announcement returns for portfolios of industry
peers are significantly more negative in industries with leverage above the sample median. Second, the announcement returns are significantly more positive
3858
Leverage
3.3 Harris (1998)

Harris (1998) shows that firms are less likely to disclose separate segment
information for operations in more-concentrated industries. She argues that
firms behave in this manner to protect the abnormal profits and market shares
related to their operations. She uses Compustat-based industry concentration
measures in her empirical analyses. We reexamine the sensitivity of her results
to using U.S. Census-based industry concentration measures. Specifically, we
reestimate the multivariate logit model in her Table 3 on a sample of manufacturing firms. To create the dependent variable in these models, we follow
3859
for industries with HI-Compustat values above the sample median. Third, for
industries with high HI-Compustat and low leverage, the announcement returns
are significantly positive (1.887%). Finally, for industries with high leverage
and low HI-Compustat values, the announcement returns are significantly negative (3.042%). These results are similar to those reported in Lang and Stulz
(1992). Panel B presents the results with HI-Compustat replaced with HICensus. None of the significant relationships involving industry concentration
that are observed in panel A are found to be significant in panel B. Thus, results
based on the U.S. Census-based industry concentration measures are not consistent with those reported in Lang and Stulz (1992) and do not support their
conclusions.
Lang and Stulz (1992) use multivariate regression models to control for other
factors that might be related to the announcement returns. We reexamine their
regression results and report our findings in Table 9. The first four columns in
this table present the results from estimating the four multivariate models of
Table 4 of Lang and Stulz (1992). In the first three models, the explanatory
variables of interest are the three dummy variables representing high debt/high
HI-Compustat, low debt/high HI-Compustat, and low debt/low HI-Compustat.
Consistent with the Lang and Stulz (1992) results, in all three models the coefficients on the dummy variable representing low debt/high HI-Compustat are
positive and significant. They conclude that this result reflects the competitive
effect. The coefficients on the other variables in these models are also similar to
those in Lang and Stulz (1992). In column 4, the dummy variables are replaced
by HI-Compustat and leverage. Once again, consistent with the Lang and Stulz
(1992) results, the coefficient on HI-Compustat is positive and significant, and
the coefficient on leverage is insignificant. The coefficients on the remaining
variables are also consistent with Lang and Stulz (1992).
Columns 58 of Table 9 present results from estimating the announcement
return models after replacing HI-Compustat with HI-Census. In columns 57,
the coefficient on the dummy variable representing low debt/high HI-Census
is not significant. Also in the last column, the coefficient on HI-Census is not
significant. In sum, the Lang and Stulz (1992) results and conclusions are not
robust to using U.S. Census-based industry concentration measures instead of
Compustat-based measures.
Intercept
0.034 (2.19)
1 if high debt/high
0.016 (0.68)
HI-Compustat; 0 otherwise
1 if low debt/high
0.053 (2.41)
1 if low debt/low
0.006 (0.24)
1 if high debt/high HI-Census; 0
otherwise
1 if low debt/high HI-Census; 0
otherwise
1 if low debt/low HI-Census; 0
otherwise
HI-Compustat
HI-Census
Returns correlation
Leverage
Log of average price
Distress cumulative return
0.020 (0.62)
Predistress cumulated return
0.024 (1.03)
Adjusted R-square
0.031
Number of observations
86
0.157 (4.79)
0.028 (1.34)
0.156 (4.70)
0.029 (1.37)
0.061 (3.01)
0.060 (2.95)
0.017 (0.79)
0.016 (0.77)
0.100 (2.68)
0.023 (1.43) 0.137 (4.29)
0.139 (4.30)
0.009 (0.37) 0.019 (0.85)
0.020 (0.89)
0.024 (1.08)
0.025 (1.26)
0.024 (1.18)
0.016 (0.66)
0.013 (0.58)
0.012 (0.55)
0.053 (2.02)
0.058 (0.36)
0.016 (0.27)
0.043 (4.25)
0.016 (0.53)
0.199
0.044 (4.27)
0.197
86
0.023 (0.40)
0.004 (0.05)
0.042 (3.92)
0.007 (0.21)
0.157
86
86
0.120 (3.14)
0.043 (4.14)
0.009 (0.28) 0.005 (0.15)
0.024 (0.97)
0.022
0.149
86
86
0.043 (4.16)
0.150
0.022 (0.27)
0.040 (3.65)
0.002 (0.07)
0.116
86
86
Cumulative abnormal returns (market model errors) are calculated from day 5 to +5 relative to bankruptcy announcements. Bankruptcy announcements are from the Altman-NYU
Salomon Center Bankruptcy list and include bankruptcies between January 1980 and December 2004 with liabilities greater than $100 million for firms in the manufacturing sector (SIC
codes between 2000 and 3999) with a primary SIC code that is available from Compustat (86 bankruptcies). An industry portfolio is a value-weighted portfolio of firms with the same
primary four-digit SIC code as the bankrupt firm for which announcement returns are available from the CRSP files. The industry characteristics are obtained from Compustat for the
fiscal year preceding the announcement except for HI-Census and the returns correlation variable. HI-Census is the Herfindahl index for four-digit SIC industries as reported by the Census
of Manufactures. The 1997 and 2002 U.S. Censuses define industry at the six-digit NAICS level. Over the 19952004 period we use HI-Census values for six-digit NAICS industries to
calculate HI-Census values for four-digit SIC industries by weighting the HI-Census values of component six-digit NAICS industries by the square of their share of the broader four-digit
SIC industry. To determine what are the component six-digit NAICS industries of a broader four-digit SIC industry, we use NAICS correspondence tables provided by the U.S. Census.
HI-Compustat is defined as the sum of the squares of the sales market shares of all firms in a Compustat four-digit SIC industry. High/low debt, HI-Compustat, and HI-Census are as defined
in Table 8. Returns correlation is the correlation between the industry portfolio and the bankrupt firm returns for the year preceding the announcement. Leverage is the average debt-to-total
assets ratio in a firms industry. Log of average price is the natural logarithm of average stock price in a firms industry. Distress cumulative return is the industry portfolio cumulative return
in excess of the market return from five days before the first distress announcement to five days before the bankruptcy announcement. Predistress cumulated return is the industry portfolio
cumulative return in excess of the market return from 800 to 50 days before the first distress announcement. As in Lang and Stulz (1992), we identify first distress announcements using the
methodology employed by Gilson, John, and Lang (1990). t-statistics are in parentheses. , , and represent significance at the 1%, 5%, and 10% levels, respectively, for a two-tailed
test.
3860
Table 9
Weighted least squares regressions of industry portfolio market model cumulative residuals at bankruptcy announcements on industry characteristics
3.4 Defond and Park (1999)

Finally, we reexamine the Defond and Park (1999) result that CEO turnover
is negatively associated with industry concentration. They argue that in
more competitive industries there is greater homogeneity across firms and
CEOs are likely to have more peers, making it easier to identify and replace
poorly performing CEOs.
Defond and Park (1999) use the Lexis/Nexis news database for the period
from 1988 to 1992 to construct their CEO turnover sample. Their control sample
14
We reestimate all the models in Table 10 using FFR-Compustat and FFR-Census calculated for three-digit SIC
industries and obtain results similar to those reported in this table.
3861
her approach and use the Compustat Multiple SICs Tape, which reports all the
SIC codes for a firm in a given year. As in Harris (1998), we compare these
SICs to those appearing in the Compustat Industry Segment File, which reports
the segments actually disclosed in a firms annual report. If a three-digit SIC
in which a firm has operations is reported as a primary or secondary SIC for
one of the firms business segments, the dependent variable takes a value of 1;
otherwise, the dependent variable equals 0. This definition allows for multiple
firm-industry observations in a given year. The Compustat Multiple SICs Tape
was discontinued in 1998. However, we were able to locate the 1997 Compustat
Multiple SICs Tape and we used it in our analysis. This tape allows us to study
the period from 1995 to 1997 rather than the period from 1987 to 1991 studied
by Harris (1998). We do not study the years examined in Harris (1998) because
we have data for the SIC codes in which a firm has operations only for 1997.
Table 10 reports the results of our logit regressions. The model in the first
three columns of the table uses three-digit SIC level HI-Compustat as the
measure of industry concentration, although Harris (1998) measures industry
concentration as FFR-Compustat at the three-digit SIC level. Our reason for
using HI-Compustat as an explanatory variable instead of FFR-Compustat
in the regression models is that the 1997 Census of Manufactures reports
concentration measures at the six-digit NAICS level, and it is possible to
use these data to calculate HI-Census precisely at the three-digit SIC level,
whereas we can compute only approximate FFR-Census values for three-digit
SIC industries.14 The other explanatory variables in the model are the same as
in Harris (1998).
We follow Harris (1998) and report results for each of the sample years
separately. The coefficients on HI-Compustat are negative and significant for
each of the individual sample period years examined, consistent with the Harris
(1998) results. However, in models 46 in Table 10, in which we replace HICompustat with HI-Census, the coefficients on HI-Census are insignificant for
each of the sample years. Thus, the Harris (1998) findings are sensitive to the
use of the U.S. Census-based industry concentration measures in place of the
Compustat-based measures.
Sample period
Intercept
HI-Compustat
HI-Census
ROA persistence
Earnings persistence across the SICs in which a firm operates
Median industry sales scaled by firm sales
Number of SICs in which the firm operates
Likelihood ratio
1
1995
2
1996
3
1997
1.778 (0.000)
0.536 (0.043)
1.900 (0.000)
0.494 (0.056)
1.959 (0.000)
0.486 (0.064)
0.152 (0.015)
1.112 (0.000)
0.841 (0.000)
0.718 (0.000)
1.896 (0.000)
5109
0.106 (0.087)
1.115 (0.000)
0.839 (0.000)
0.735 (0.000)
2.034 (0.000)
5384
0.143 (0.021)
1.067 (0.000)
0.838 (0.000)
0.769 (0.000)
2.121 (0.000)
5468
4
1995
5
1996
6
1997
1.678 (0.000)
1.827 (0.000)
1.917 (0.000)
1.215 (0.458)
0.164 (0.009)
1.194 (0.000)
0.836 (0.000)
0.720 (0.000)
1.893 (0.000)
5109
0.074 (0.964)
0.115 (0.064)
1.182 (0.000)
0.841 (0.000)
0.736 (0.000)
2.030 (0.000)
5384
1.551 (0.338)
0.144 (0.020)
1.135 (0.000)
0.841 (0.000)
0.770 (0.000)
2.119 (0.000)
5468
The sample period is 19951997 and includes firms in the manufacturing sector (SIC codes between 2000 and 3999) covered by Compustat. The dependent variable takes a value of 1 if
during the current year firm i decides to provide a segmental disclosure of its operations for a three-digit SIC industry j in which it operates, and equals 0 otherwise. This allows for multiple
firm-industry observations in a given year. HI-Compustat is defined as the sum of the squares of the sales market shares of all firms in a Compustat three-digit SIC industry. HI-Census is
the Herfindahl index for three-digit SIC industries as reported by the Census of Manufactures. The 1997 U.S. Census defines industry at the six-digit NAICS level. Over the 19951997
period we use HI-Census values for six-digit NAICS industries to calculate HI-Census values for three-digit SIC industries by weighting the HI-Census values of component six-digit
NAICS industries by the square of their share of the broader three-digit SIC industry. To determine what are the component six-digit NAICS industries of a broader three-digit SIC industry
we use NAICS correspondence tables provided by the U.S. Census. ROA persistence represents the speed of adjustment for firm-level positive abnormal return on assets in industry j and
is measured as in Harris (1998) as the slope coefficient B2j from the following regression model, Xijt = B0j + B1j (Dn Xijt 1 ) + B2j (Dp Xijt 1 ) + eijt , where Xijt = the difference between
firm is ROA and mean ROA for its industry j, in year t, Dn = 1 if Xijt1 is less than or equal to zero, zero otherwise, and Dp = 1 if Xijt 1 is greater than zero, zero otherwise. Earnings
persistence across the SICs in which a firm operates is measured as the maximum value minus the minimum value of ROA persistence for a firm in the three-digit SICs in which the firm
has operations during the current year. Median industry sales scaled by firm sales is measured as median sales for single-segment firms in a three-digit SIC industry j divided by firm is
sales. Number of SICs in which the firm operates is measured as the number of three-digit SIC industries in which the firm operates during the current year. Significance levels for Wald
chi-square test statistics are in parentheses.
3862
Table 10
Logit analysis of managers business segment reporting decisions
15
Given that comprehensive data in the ExecuComp database is available from 1993 onward and the requirement
that we have information on the identity of the CEO in year t 1, 1994 is the earliest year in which our sample
period can begin.
3863
consists of firms in the Compact Disclosure database without any CEO turnover
during their sample period. Their final sample has 301 firm-year observations
with CEO turnovers and a control sample of 2429 firm-year observations with
no CEO turnovers. For our sample, we consider manufacturing firms in the
ExecuComp database over the period from 1994 to 2000 and treat firm-years
for which the CEO for year t differs from the CEO for year t 1 as a CEO
turnover observation. Like Defond and Park (1999), we consider all instances
of CEO turnover regardless of the reason for the turnover. Our control sample
consists of manufacturing firms included on the ExecuComp database that do
not experience a CEO turnover during the period from 1994 to 2000. Our
final sample consists of 203 firm-year observations with CEO turnovers and
a control sample of 2267 firm-year observations with no CEO turnovers. We
point out that we do not study a longer sample period so as to not impose
the condition that control firms have no CEO turnover over this longer period.
Also, we study the period from 1994 to 2000 rather than a more recent period
so that our sample period can be closer in time to the Defond and Park (1999)
sample period.15
The first three models in Table 11 report the results of replications of the three
logit models of panel A of Table 4 of Defond and Park (1999). The dependent
variable takes a value of 1 for firm-year observations with a CEO turnover,
and a value of 0 for firm-years belonging to the control sample. The main
independent variable of interest is the square root of HI-Compustat calculated
at the two-digit SIC level. All the models have the same set of control variables,
except that the first model does not include analysts earnings forecast errors,
the second model does not include industry-relative earnings, and the third
model includes both of these variables. The results in the first three columns of
Table 11 show that there is a significantly negative association between CEO
turnover and the square root of HI-Compustat, consistent with the Defond and
Park (1999) results. The results for the control variables are also consistent
with those in Defond and Park (1999), except that our coefficients on analysts
earnings forecast errors are not statistically significant.
The models in the fourth to the sixth columns of Table 11 replace the square
root of HI-Compustat, as an explanatory variable, with the square root of
HI-Census calculated at the two-digit SIC level. We do not find significant
associations between CEO turnover and the square root of HI-Census in any of
the three models. Consequently, the Defond and Park (1999) findings and their
conclusions are not robust to using U.S. Census-based industry concentration
measures instead of Compustat-based measures.
Intercept
Square root of HI-Compustat
Square root of HI-Census
Industry-relative earnings
Analysts earnings forecast errors
Market-adjusted stock returns
Age of CEO
Dummy variable for if CEO age
= 63, 64, or 65
Industry market-to-book ratio
Stock return volatility
Pseudo R-square
6.695 (7.08)
6.013 (2.82)
6.758 (7.12)
5.023 (2.43)
6.703 (7.09)
6.033 (2.82)
8.155 (10.56)
8.010 (10.29)
8.174 (10.60)
2.490 (0.94)
2.696 (3.43)
2.168 (0.81)
1.835 (1.00)
0.579 (2.63)
0.093 (8.24)
0.573 (2.81)
2.883 (3.73)
1.654 (0.90)
0.415 (1.88)
0.097 (8.47)
0.607 (2.97)
2.930 (3.77)
0.403 (1.84)
0.097 (8.47)
0.597 (2.93)
0.021 (0.15)
3.046 (2.44)
0.162
2470
0.120 (0.93)
4.389 (2.53)
0.153
2470
0.013 (0.10)
2.879 (2.44)
0.163
2470
0.436 (1.99)
0.097 (8.51)
0.613 (3.01)
1.754 (0.96)
0.593 (2.70)
0.094 (8.32)
0.591 (2.91)
2.432 (0.92)
2.653 (3.39)
1.571 (0.86)
0.445 (2.03)
0.097 (8.51)
0.623 (3.05)
0.174 (1.45)
3.291 (2.58)
0.156
2470
0.024 (0.22)
4.399 (2.57)
0.149
2470
0.168 (1.40)
3.122 (2.58)
0.157
2470
The sample period is 19942000 and includes firms in the manufacturing sector (SIC codes between 2000 and 3999) covered by ExecuComp. We define our CEO turnover sample as
firm-years for which the CEO for year t differs from the CEO for year t 1. We define the control sample of firms that do not experience CEO turnover as firms that do not change their
CEO over the 19942000 period. The dependent variable equals 1 if a firm-year is included in the CEO turnover sample and equals 0 if a firm-year is instead included in the control sample.
HI-Compustat is the sum of the squares of the sales market shares of all firms in a CRSP two-digit SIC industry that have sales data on Compustat. For each year t, HI-Compustat is averaged
over a five-year period from year t 6 to year t 1. HI-Census represents the Herfindahl index for two-digit SIC industries calculated using data collected from Census of Manufactures
publications. Prior to 1997, the U.S. Census defines industry using four-digit SIC codes while from 1997 onward industry is defined using six-digit NAICS codes. For the 1994 year, we
use HI-Census values for four-digit SIC industries to calculate HI-Census values for two-digit SIC industries by weighting the HI-Census values of component four-digit SIC industries
by the square of their share of the broader two-digit SIC industry. Over the 19952000 period, we use HI-Census values for six-digit NAICS industries to calculate HI-Census values for
two-digit SIC industries by weighting the HI-Census values of component six-digit NAICS industries by the square of their share of the broader two-digit SIC industry. To determine what
are the component six-digit NAICS industries of a broader four-digit SIC industry, we use NAICS correspondence tables provided by the U.S. Census. Industry-relative earnings is net
earnings scaled by assets during the fiscal year prior to the event year minus median net earnings for the firms two-digit SIC industry. Analysts earnings forecast errors is actual reported
earnings scaled by assets immediately preceding the event year minus analysts forecast of the fiscal years earnings scaled by assets immediately preceding the event year, made nine
months prior to the fiscal year end. Market-adjusted stock returns is the cumulative market-adjusted stock returns over twelve months prior to the CEO turnover for the CEO turnover
sample. Market-adjusted stock returns for the control sample are computed over the twelve months prior to a randomly assigned pseudo-event month. Industry market-to-book ratio is the
mean value of the market-to-book ratio in a two-digit SIC industry over the prior five years where the yearly industry market-to-book ratio is computed as the median value for the industry.
Stock return volatility is the variance of returns during the twenty-four months prior to the event year. Industry-relative earnings, Analysts earnings forecast errors, and Market-adjusted
stock-returns are trimmed at the 1% level. Year dummies are included in all the models. t-statistics are in parentheses. , , and represent significance at the 1%, 5%, and 10% levels,
respectively, for a two-tailed test.
3864
Table 11
Logit analysis predicting CEO turnover
16
Because U.S. Census data are not available every year, we apply a given years U.S. Census data on shipment
values to the one or two years surrounding that year, as is shown in Table 1.
17
We also find significant negative Spearman and Pearson correlations between HI-Compustat and the change in
industry shipments over the next five years, as reported by the Census of Manufactures (0.100, p < 0.001, and
0.063, p < 0.001). These results suggest that industries with high Compustat-based concentration measures
also exhibit a decline in future years. However, we find that HI-Census is not related to the future change in
industry shipments which indicates that actual industry concentration is not related to future industry decline.
18
We examine how the number of firms per four-digit SIC industry that have data available on CRSP and Compustat
changes during the prior five years for industries in the highest quintile of HI-Compustat. We find that the average
number of firms per industry drops from approximately five to three firms. This further suggests that industries
with high values for Compustat-based industry concentration measures are likely to be declining industries.
3865
3.5 What drives the results obtained with Compustat-based industry

concentration measures?
Our results show that there is no significant relationship between U.S. Censusbased industry concentration measures and the dependent variables of the four
studies we examined. An interesting question that arises is why the Compustatbased industry concentration measures have significant associations with the
dependent variables of these studies. A likely explanation is that the Compustatbased measures capture some other industry characteristics that happen to
be correlated with the dependent variables of these studies. To examine this
possibility, we investigate what drives the Hou and Robinson (2006) finding
that Compustat-based industry concentration measures are negatively related
to future stock returns and the Harris (1998) result that firms are less likely to
provide segment disclosures for operations in industries with higher values for
Compustat-based concentration measures.
We find that Compustat-based industry concentration measures are related
to industry decline. The Spearman and Pearson correlations between HICompustat and the change in industry shipments over the prior five years,
as reported in the Census of Manufactures, are significantly negative (0.122,
p < 0.001, and 0.076, p < 0.001).16,17 Likewise, the Spearman and Pearson
correlations between HI-Compustat and the change over the prior five years in
the number of firms per industry that have data available on CRSP and Compustat are negative and significant (0.128, p < 0.001, and 0.135, p < 0.001).
However, we do not find these significant relations when we use HI-Census
instead of HI-Compustat. These results show that for some reason other than
the actual concentration of an industry, industries with high Compustat-based
concentration measures experience poor growth in the recent past, leaving them
with only a few large, public companies relative to private companies. Consequently, the Compustat database covers only a few firms for these industries,
which results in high values for the Compustat-based industry concentration
measures.18
Our finding that industries with high values for Compustat-based industry concentration measures are likely to be declining industries explains the
result documented by Hou and Robinson (2006) and confirmed in our Table 7
Table 12
The effect of research and development expenses on the relationship between stock returns and
Compustat-based industry concentration measures
1
1.943 (7.29)
0.247 (2.16)
0.117 (2.74)
0.208 (2.83)
0.568 (3.29)
0.141 (1.06)
0.094 (0.48)
0.049
985
1.796 (6.97)
0.177 (1.57)
2.170 (6.67)
1.706 (5.69)
0.342 (2.00) 0.190 (1.17)

5.702 (4.69)
4.202 (4.17)
0.113 (2.68) 0.113 (2.65) 0.076 (1.91)
0.266 (3.80)
0.220 (2.99)
0.277 (3.89)
0.547 (3.18)
0.573 (3.28)
0.552 (3.15)
0.175 (1.35)
0.128 (0.97)
0.153 (1.17)
0.213 (1.06)
0.072 (0.35)
0.247 (1.15)
0.051
0.049
0.052
985
936
936
This table presents results from Fama-MacBeth cross-sectional regressions explaining monthly stock returns,
estimated monthly between July 1963 and June 2001. Monthly returns are the one-year ahead twelve monthly
returns from July of year t + 1 to June of year t + 2. Industry concentration ratios and accounting data are
measured at the end of year t. HI-Compustat is the sum of the squares of the sales market shares of all firms in
a CRSP three-digit SIC industry that have sales data on Compustat. For each year t, HI-Compustat is averaged
over a three-year period from year t 2 to year t. FFR-Compustat is the sum of the market shares of the four
largest firms in terms of market share in a CRSP four-digit SIC industry. A firms market share is measured
as sales divided by total sales of all CRSP firms in that industry that have sales data on Compustat. For each
year t, FFR-Compustat is averaged over a three-year period from year t 2 to year t. R&D expense/book assets
is as of year t. Ln(Market Capitalization) is defined as the natural logarithm of the market value of equity in
millions at the end of year t. Ln(B/M) is defined as the natural logarithm of book value of equity divided by
market value of equity at the end of year t. Momentum for each month is prior one-year stock returns. Beta is
market model beta estimated using the prior thirty-six monthly equally weighted CRSP index returns. Leverage
is defined as total debt divided by market value of total assets (i.e., market value of equity plus debt) at the end
of year t. R&D expense/book assets and Leverage are trimmed at the 1% level. Time-series average values of the
monthly regression coefficients are reported with time-series t-statistics in parentheses. , , and represent
significance at the 1%, 5%, and 10% levels, respectively, for a two-tailed test.
that such industries spend less on research and development.19 Furthermore, prior studies (e.g., Chan, Lakonishok, and Sougiannis 2001; Chambers,
Jennings, and Thompson 2002; Eberhart, Maxwell, and Siddique 2004) provide
evidence suggesting that firms that spend more on research and development
have higher future stock returns. Thus, we investigate whether the association
between Compustat-based industry concentration measures and future stock
returns is robust to controlling for current research and development expenses.
Specifically, we reestimate the relationship between Compustat-based industry
concentration measures and future stock returns after adding R&D/book assets
as an independent variable in the Table 6 models and report our findings in
Table 12. For this analysis, we use the sample period from 1963 to 2001, as
in Hou and Robinson (2006), since we are able to calculate HI-Compustat and
FFR-Compustat over this period.
19
Consistent with the conjecture that firms in declining industries spend less on research and development, we find
significant positive Spearman and Pearson correlations between the change in shipments in a firms industry over
the prior five years and the firms current value for research and development expenses scaled by book assets
(0.096, p < 0.001, and 0.144, p < 0.001).
3866
Intercept
HI-Compustat
FFR-Compustat
R&D expense/book assets
Ln(Market Capitalization)
Ln(B/M)
Momentum
Beta
Leverage
Average adjusted R-square
Average number of
observations
20
This model is the same as the third model in Table 6, and the sample period analyzed is also the same used in
that model. We include this model in Table 12 so that the results of this model can be more easily compared with
the results of the fourth model of this table.
21
Given that Table 7 shows that research and development expenses are positively associated with HI-Census and
FFR-Census, we examine the effect of controlling for R&D/book assets in the second and fourth models in Table 6
that provide evidence on the association of future stock returns with HI-Census and FFR-Census. We find that
controlling for R&D/book assets does not affect the result that future stock returns are unrelated to U.S. Censusbased industry concentration measures.
22
Specifically, Miller (2002) shows that after a period of sustained superior earnings performance, firms increase
their disclosure level and then when earnings decline firms decrease their disclosure level. Also, Kothari, Shu,
and Wysocki (forthcoming) present evidence that suggests managers delay reporting bad news but reveal good
news to investors promptly. Kothari, Shu, and Wysocki (forthcoming) argue that their findings are consistent with
managers career concerns motivating them to withhold bad news and gamble that subsequent corporate events
will allow them to bury the bad news. They also claim that their findings are consistent with the Graham,
Harvey, and Rajgopal (2005) survey evidence that some chief financial officers contend that they delay disclosing
bad news in the hope they may never have to actually disclose this information if the firms status subsequently
improves.
3867
The first column in Table 12 reports a significantly negative relationship between future stock returns and HI-Compustat. The model in the second column
of Table 12 includes R&D/book assets as an additional explanatory variable.
Consistent with the findings of prior work, we find a significantly positive relationship between the R&D variable and future stock returns. Interestingly,
after controlling for R&D/book assets, the association between future stock
returns and HI-Compustat becomes insignificant. The third column of Table 12
documents a significantly negative relationship between future stock returns
and FFR-Compustat for the period from 1963 to 2001.20 However, the fourth
model in Table 12 shows that after controlling for R&D/book assets, the association between future stock returns and FFR-Compustat becomes insignificant.
In sum, the Table 12 results suggest that the negative relationship between
research and development expenses and Compustat-based industry concentration measures drives the negative association between the Compustat-based
measures and future stock returns.21
Next, we examine whether the correlation between prior shipment growth
and Compustat-based industry concentration measures can explain the Harris
(1998) result that firms are less likely to provide segment disclosures for operations in industries that have higher values for Compustat-based concentration
measures. This examination is motivated by the Miller (2002) and Kothari, Shu,
and Wysocki (forthcoming) studies, which show that firms with weak (strong)
prior operating performance provide lower (higher) levels of disclosure.22 Their
findings suggest that the Harris (1998) result could be driven by Compustatbased industry concentration measures proxying for the prior operating performance of a firm in one of its industry segments, and that performance level in
turn affecting the firms decision to provide segment disclosures of its operations in that industry.
Table 13 presents the results of our analysis of what drives the Harris (1998)
result. The models in this table are the same as the first three models in Table
10, except that we also include as a control variable the change in industry
Table 13
The effect of industry decline on the association between Compustat based industry concentration
measures and managers business segment reporting decisions
Sample period
2
1996
3
1997
1.330 (0.000)
0.412 (0.132)
0.456 (0.056)
0.163 (0.009)
1.114 (0.000)
1.258 (0.000)
0.310 (0.246)
0.656 (0.005)
0.123 (0.049)
1.114 (0.000)
1.263 (0.000)
0.270 (0.323)
0.709 (0.004)
0.160 (0.010)
1.073 (0.000)
0.850 (0.000)
0.850 (0.000)
0.847 (0.000)
0.712 (0.000)
0.728 (0.000)
0.762 (0.000)
1.900 (0.000)
5109
2.041 (0.000)
5384
2.129 (0.000)
5468
The sample period is 19951997 and includes firms in the manufacturing sector (SIC codes between 2000
and 3999) covered by Compustat. The dependent variable takes a value of 1 if during the current year firm i
decides to provide a segmental disclosure of its operations for a three-digit SIC industry j in which it operates,
and equals 0 otherwise. This allows for multiple firm-industry observations in a given year. HI-Compustat is
defined as the sum of the squares of the sales market shares of all firms in a Compustat three-digit SIC industry.
Shipment-Growth (5,0) is the mean value of percentage shipment growth over the prior five years as reported
by the Census of Manufactures for three-digit SIC industries. Shipment growth is then transformed into ranks
in the regressions taking on a value between 0 and 1. We use shipment values for six-digit NAICS industries
to calculate shipment values for three-digit SIC industries by summing the shipments of component six-digit
NAICS industries of broader three-digit SIC industries. To determine what are the component six-digit NAICS
industries of a broader three-digit SIC industry, we use NAICS correspondence tables provided by the U.S.
Census. ROA persistence represents the speed of adjustment for firm-level positive abnormal ROA in industry
j and is measured as in Harris (1998) as the slope coefficient B2j from the following regression model, Xijt =
B0j + B1j (Dn Xijt 1 ) + B2j (Dp Xijt1 ) + eijt , where Xijt = the difference between firm is ROA and mean ROA
for its industry j, in year t, Dn = 1 if Xijt1 is less than or equal to zero, zero otherwise, and Dp = 1 if Xijt1 is
greater than zero, zero otherwise. Earnings persistence across the SICs in which a firm operates is measured as
the maximum value minus the minimum value of ROA persistence for a firm in the three-digit SICs in which the
firm has operations during the current year. Median industry sales scaled by firm sales is measured as median
sales for single-segment firms in a three-digit SIC industry j divided by firm is sales. Number of SICs in which
the firm operates is measured as the number of three-digit SIC industries in which the firm operates during the
current year. Significance levels for Wald chi-square test statistics are in parentheses.
shipments over the prior five years, Shipment-Growth (5,0). After controlling
for this variable, the relationship between firms decisions to provide segment
disclosures for their operations in a particular industry and HI-Compustat for
that industry becomes insignificant. We also find a significant positive association between firms decisions to provide segment disclosures of their operations in a particular industry and Shipment-Growth (5,0) for that industry,
consistent with the findings of Miller (2002) and Kothari, Shu, and Wysocki
(forthcoming) that firms with weaker (stronger) prior operating performance
provide less (more) informative disclosures. Overall, our findings suggest that
the Harris (1998) result that firms are less likely to provide segment disclosures
for operations in industries with higher HI-Compustat values is driven by the
negative correlation between prior industry shipment growth and Compustatbased industry concentration measures. Finally, we note that our results in this
section also suggest that the findings of the other two studies that we consider
in this article, Lang and Stulz (1992) and Defond and Park (1999), may also
3868
Intercept
HI-Compustat
Shipment-Growth (5,0)
ROA persistence
Earnings persistence
across the SICs in
which a firm operates
Median industry sales
scaled by firm sales
Number of SICs in which
the firm operates
Likelihood ratio
1
1995
be driven by Compustat-based industry concentration measures proxying for

other industry characteristics that are correlated with the dependent variables
of these studies.
4. Conclusion
References
Aggarwal, R. K., and A. A. Samwick. 1999. Executive Compensation, Strategic Competition, and Relative
Performance Evaluation: Theory and Evidence. Journal of Finance 54:19992043.
Akdogu, A., and P. MacKay. 2008. Investment and Competition. Journal of Financial and Quantitative Analysis
43:299320.
Allayannis, G., and J. Ihrig. 2001. Exposure and Markups. Review of Financial Studies 14:80535.
Botosan, C. A., and M. S. Harris. 2000. Motivations for a Change in Disclosure Frequency and Its Consequences:
An Examination of Voluntary Quarterly Segment Disclosures. Journal of Accounting Research 38:32953.
Botosan, C. A., and M. Stanford. 2005. Managers Motives to Withhold Segmental Disclosures and the Effect
of SFAS No. 131 on Analysts Information Environment. Accounting Review 80:75171.
Campello, M. 2006. Debt Financing: Does It Hurt or Boost Firm Performance in Product Markets? Journal of
Financial Economics 82:13572.
Chambers, D., R. Jennings, and R. B. Thompson. 2002. Excess Returns to R&D-Intensive Firms. Review of
Accounting Studies 7:13358.
3869
We compare the properties of industry concentration measures calculated with

Compustat data, which cover only the public firms in an industry, with Census of
Manufactures industry concentration measures, which are based on all public
and private firms in an industry. We find that the correlation between the
two measures is only 13%. Also, only for the U.S. Census measures do we
find evidence consistent with theoretical predictions that more-concentrated
industries, which should be more oligopolistic in nature, are populated by
larger and fewer firms with higher price-cost margins.
Given that Compustat-based industry concentration measures seem to be
poor proxies for actual industry concentration, we examine whether results of
empirical studies that use Compustat-based industry concentration measures
are robust to the use of U.S. Census-based industry concentration measures.
We reexamine the results found in Hou and Robinson (2006); Lang and Stulz
(1992); Harris (1998); and Defond and Park (1999). We show that Compustatbased industry concentration measures are significantly related to the dependent
variables of each of these studies, consistent with the results of these studies.
However, the U.S. Census-based industry concentration measures are not significantly related to any of these dependent variables. Further analysis of the
Hou and Robinson (2006) and Harris (1998) findings suggests that the significant relations obtained in prior studies that use Compustat-based measures
could be due to these measures proxying for other industry characteristics,
such as industry decline or the associated decline in research and development
spending, that are related to the dependent variables of these studies.
Chan, L. K. C., J. Lakonishok, and T. Sougiannis. 2001. The Stock Market Valuation of Research and Development
Expenses. Journal of Finance 56:243156.
Cheng, Q. 2005. What Determines Residual Income? Accounting Review 80:85112.
Chevalier, J. A. 1995a. Do LBO Supermarkets Charge More? An Empirical Analysis of the Effects of LBOs on
Supermarket Pricing. Journal of Finance 50:1095112.
Chevalier, J. A. 1995b. Capital Structure and Product-Market Competition: Empirical Evidence from the Supermarket Industry. American Economic Review 85:41535.
Chevalier, J. A., and D. S. Scharfstein. 1996. Capital-Market Imperfections and Countercyclical Markups: Theory
and Evidence. American Economic Review 86:70325.
Eberhart, A. C., W. F. Maxwell, and A. R. Siddique. 2004. An Examination of Long-Term Stock Returns and
Operating Performance Following R&D Increases. Journal of Finance 59:62350.
Eckbo, E. E. 1985. Mergers and the Market Concentration Doctrine: Evidence from the Capital Market. Journal
of Business 58:32549.
Eckbo, E. E. 1992. Mergers and the Value of Antitrust Deterrence. Journal of Finance 47:100529.
Engel, E., R. M. Hayes, and X. Wang. 2003. CEO Turnover and Properties of Accounting Information. Journal
of Accounting and Economics 36:197226.
Fee, C. E., and S. Thomas. 2004. Sources of Gains in Horizontal Mergers: Evidence from Customer, Supplier,
and Rival Firms. Journal of Financial Economics 74:42360.
Gaspar, J., and M. Massa. 2006. Idiosyncratic Volatility and Product Market Competition. Journal of Business
79:3125152.
Gilson, S. C., K. John, and L. H. P. Lang. 1990. Troubled Debt Restructurings: An Empirical Study of Private
Reorganization of Firms in Default. Journal of Financial Economics 27:31553.
Graham, J., C. Harvey, and S. Rajgopal. 2005. The Economic Implications of Corporate Financial Reporting.
Journal of Accounting and Economics 40:373.
Harris, M. H. 1998. The Association between Competition and Managers Business Segment Reporting Decisions. Journal of Accounting Research 36:11190.
Haushalter, D., S. Klasa, and W. Maxwell. 2007. The Influence of Product Market Dynamics on a Firms Cash
Holdings and Hedging Behavior. Journal of Financial Economics 84:797825.
Hay, D. A., and D. J. Morris. 1991. Industrial Economics and Organization, Theory and Evidence. Oxford, UK:
Oxford University Press.
Hou, K., and D. T. Robinson. 2006. Industry Concentration and Average Stock Returns. Journal of Finance
61:192756.
Kahle, K. M., and R. A. Walkling. 1996. The Impact of Industry Classifications on Financial Research. Journal
of Financial and Quantitative Analysis 31:30935.
Kale, J. R., and H. Shahrur. 2007. Corporate Capital Structure and the Characteristics of Suppliers and Customers.
Journal of Financial Economics 83:32165.
Karuna, C. 2007. Industry Product Market Competition and Managerial Incentives. Journal of Accounting and
Economics 43:27597.
Khanna, N., and S. Tice. 2000. Strategic Responses of Incumbents to New Entry: The Effect of Ownership
Structure, Capital Structure, and Focus. Review of Financial Studies 13:74979.
Kothari, S. P., S. Shu, and P. Wysocki. Forthcoming. Do Managers Withhold Bad News? Working Paper,
Massachusetts Institute of Technology, Journal of Accounting Research.
3870
DeFond, M. L., and C. W. Park. 1999. The Effect of Competition on CEO Turnover. Journal of Accounting and
Economics 27:3556.
Kovenock, D., and G. M. Phillips. 1997. Capital Structure and Product Market Behavior: An Examination of
Plant Exit and Investment Decisions. Review of Financial Studies 10:767803.
Lang, L. H. P., and R. M. Stulz. 1992. Contagion and Competitive Intra-industry Effects of Bankruptcy Announcements: An Empirical Analysis. Journal of Financial Economics 32:4560.
MacKay, P., and G. M. Phillips. 2005. How Does Industry Affect Firm Financial Structure? Review of Financial
Studies 18:143366.
Maksimovic, V., and G. M. Phillips. 2001. The Market for Corporate Assets: Who Engages in Mergers and Asset
Sales and Are There Efficiency Gains? Journal of Finance 56:201965.
Miller, G. S. 2002. Earnings Performance and Discretionary Disclosure. Journal of Accounting Research 40:173
204.
Rennie, C. G. 2006. Governance Structure Changes and Product Market Competition: Evidence from U.S.
Electric Utility Deregulation. Journal of Business 79:19892017.
Rogers, J., and P. Stocken. 2005. Credibility of Management Forecasts. Accounting Review 80:123360.
Schumpeter, J. 1912. The Theory of Economic Development. Cambridge, MA: Harvard University Press.
Schumpeter, J. 1942. Capitalism, Socialism, and Democracy. New York: Harper & Row.
Shahrur, H. 2005. Industry Structure and Horizontal Takeovers: Analysis of Wealth Effects on Rivals, Suppliers,
and Corporate Customers. Journal of Financial Economics 76:6198.
Song, M. H., and R. A. Walkling. 2000. Abnormal Returns to Rivals of Acquisition Targets: A Test of the
Acquisition Probability Hypothesis. Journal of Financial Economics 55:14371.
Tookes, H. E. 2008. Information, Trading, and Product Market Interactions: Cross-Sectional Implications of
Informed Trading. Journal of Finance 63:379413.
Verrecchia, R. E., and J. Weber. 2006. Redacted Disclosure. Journal of Accounting Research 44:791814.
Zingales, L. 1998. Survival of the Fittest or the Fattest? Exit and Financing in the Trucking Industry. Journal of
Finance 53:90538.
Zmijewski, M. E., and R. L. Hagerman. 1981. An Income Strategy Approach to the Positive Theory of Accounting
Standard Setting/Choice. Journal of Accounting and Economics 3:12949.
3871
Phillips, G. M. 1995. Increased Debt and Industry Product Markets: An Empirical Analysis. Journal of Financial
Economics 37:189238.

Ali Klasa Yueng RFS 2009

Enviado por

Dados do documento

Descrição original:

Direitos autorais

Formatos disponíveis

Compartilhar este documento

Compartilhar ou incorporar documento

Opções de compartilhamento

Você considera este documento útil?

Este conteúdo é inapropriado?

Direitos autorais:

Formatos disponíveis

Ali Klasa Yueng RFS 2009

Enviado por

Direitos autorais:

Formatos disponíveis

The Limitations of Industry Concentration

Measures Constructed with Compustat Data:

A growing number of studies that consider the effects of product markets on

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Review of Financial Studies / v 22 n 10 2009

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

their results indicate that barriers to entry in highly concentrated industries

The Review of Financial Studies / v 22 n 10 2009

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

1. Description of Compustat- and U.S. Census-based industry concentration

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Review of Financial Studies / v 22 n 10 2009

1.2 U.S. Census-based industry concentration measures

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

1.3 Descriptive statistics of U.S. Census- and Compustat-based industry

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Review of Financial Studies / v 22 n 10 2009

Sample periods to which Census of Manufactures

The Census of Manufactures is published by the U.S. Census Bureau.

Panel A: 19802005 sample period

Panel B: 19632005 sample period

Panel C: 19952005 sample period

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

total shipments of component 6 digit NAICS industryj

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Review of Financial Studies / v 22 n 10 2009

2. Evidence that Compustat-based industry concentration measures are poor

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

Panel A: Quintiles based on HI-Census

Panel B: Quintiles based on HI-Compustat

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

The Review of Financial Studies / v 22 n 10 2009

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Review of Financial Studies / v 22 n 10 2009

Lumber & Wood Products

Compustat-based industry concentration measures are not good proxies for

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

from less- to more-concentrated industries based on this measure, there is no

3.1 Hou and Robinson (2006)

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Review of Financial Studies / v 22 n 10 2009

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Review of Financial Studies / v 22 n 10 2009

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

The Limitations of Industry Concentration Measures Constructed with Compustat Data

3.2 Lang and Stulz (1992)

Downloaded from http://rfs.oxfordjournals.org/ at Univ of Southern California on April 17, 2014

development expenses. If U.S. Census-based industry concentration measures

The Review of Financial Studies / v 22 n 10 2009

No. of industry portfolios

Average abnormal returns for the subsample

Panel A: Compustat-based industry concentration measures