1. Art and Diseño

Journal of Business Finance & Accounting, 000, 1–45, January 2015, 0306-686X
doi: 10.1111/jbfa.12106
SEGMENT DISCLOSURE AND COST
OF CAPITAL
BELEN BLANCO, JUAN M. GARCIA LARA AND JOSEP A. TRIBO∗
Abstract: We investigate whether segment disclosure influences cost of capital. Improved
segment reporting is expected to decrease cost of capital by reducing estimation risk. However,
in a competitive environment segment disclosure may also generate uncertainties about future
prospects and lead to a larger cost of capital. Asset-pricing tests confirm that segment disclosure
is a price risk factor. Also, segment disclosure reduces ex-ante estimates of cost of equity capital
and other measures connected to risk. These results suggest a negative relation between segment
disclosure and cost of capital. Our results also show that competition reduces, but does not
eliminate, the previous relationship.
Keywords: segment disclosure, earnings quality, forecast error, cost of capital
1. INTRODUCTION
There is an ongoing debate as to whether and how accounting quality decreases
the cost of capital. One stream of the literature suggests that accounting quality
reduces information asymmetries, which, in turn, decreases the cost of capital (Easley
and O’Hara, 2004). More recently, several studies demonstrate that information
differences across investors affect a firm’s cost of capital through information precision
(Hughes et al., 2007; Lambert et al., 2007, 2012). Standard setters (SFAS 131) and
practitioners (e.g., Ernst and Young, 2004) hold the view that segment reporting is
key to improving information precision, permitting a better evaluation of firm future
∗ The
first author is from the University of Melbourne. The second and third authors are from the Universidad Carlos III de Madrid. We appreciate the comments and suggestions contributed by an anonymous
reviewer, Peter Pope (the editor), Alicia Barroso, Daniel Cohen, Beatriz Garcia Osma, Joachim Gassen,
Miles Gietzmann, Javier Gil Bazo, Bego˜
na Giner, Paul Healy, Araceli Mora, Per Olsson, Fernando Pe˜
nalva,
Rosa Rodriguez, David Smith, Jeroen Suijs, Laurence Van Lent, Steve Young and seminar participants at
the 2014 JBFA Capital Markets Conference, 2010 AAA Financial Accounting and Reporting Section MidYear Meeting, the 2009 American Accounting Association annual meeting, the 2009 annual congress of the
European Accounting Association, the 2009 European Accounting Association Doctoral Colloquium, the
V Symposium for accounting academics (2009, Leeds Business School), the VII Workshop on Empirical
Research in Financial Accounting (2010, Universidad Polit´ecnica de Cartagena), Universidad Carlos III
de Madrid, Universidad de Valencia, Universidad de Navarra, Universitat Pompeu Fabra de Barcelona,
Universitat Aut`onoma de Barcelona, Tilburg University, The University of Melbourne and IE University. We
acknowledge financial assistance from the Spanish Ministry of Economy and Competitiveness (ECO2013–
48328-C3–3-P, ECO2010–19314, ECO2009–10796 and, ECO2012–36559), the European Commission IN´ Ramon
´ Areces, and the
TACCT Research Training Network (MRTN-CT-2006–035850), the Fundacion
government of the Autonomous Community of Madrid (Grant # 2008/00037/001).
Address for correspondence: Belen Blanco, University of Melbourne, Department of Accounting, Level 7,
198 Berkeley Street, Building 110, Carlton 3010, VIC, Australia. E-mail: [email protected].
C
2014 John Wiley & Sons Ltd
1
2
BLANCO, GARCIA LARA AND TRIBO
prospects. These views of regulators and practitioners are also supported by prior
research, as there is plenty of evidence that a commitment to provide comprehensive segment disclosure leads to a rich information environment characterized by:
(1) better predictive ability and increased precision of accounting numbers (e.g.,
Kinney, 1971; Collins, 1976; Silhan, 1983; Baldwin, 1984; Balakrishnan et al., 1990;
Swaminathan, 1991; Berger and Hann, 2003; and Ettredge et al., 2005), (2) reduced
information asymmetries (Greenstein and Sami, 1994), and (3) improved monitoring
ability over managerial decision making (Bens and Monahan, 2004; Berger and Hann,
2007; and Hope and Thomas, 2008). All of these effects of segment disclosure are
expected to make capital markets more efficient (Collins, 1975) and to facilitate firms
the access to external financing (Ettredge et al., 2006).1
However, the effect of segment information on cost of capital is not obvious. While
segment information improves the information environment, decreasing estimation
risk, it is likely that it also favors competitors, creating uncertainties about future
earnings and cash flows (Hayes and Lundholm, 1996; Stanford Harris, 1998). For
example, Bugeja et al. (2014) find evidence consistent with firms being reluctant to
provide segment information whenever most of their segments are profitable. Several
studies provide evidence consistent with increased competition leading to increased
uncertainty about future profitability that in turn leads to increases in the costs of
financing (Gaspar and Massa, 2006; Valta, 2012). Given these two expected opposite
effects of improved segment disclosure on cost of capital, we expect that the benefits
of segment disclosure, in terms of lower cost of capital, will be less pronounced and
might even disappear for firms subject to tougher competition.
To test our main hypothesis, that is, that segment disclosure leads to lower cost
of capital and that this effect is bound to be less pronounced or even disappear for
firms subject to competitive pressures, we construct a proxy for voluntary segment
disclosures based on the counting of the number of items of segment reporting
disclosed on a voluntary basis. In our proxy for voluntary segment disclosures we
control, following Clinch and Verrecchia (2013), for exogenous factors that affect
segment disclosure and also cost of equity capital. In particular, we take the residual
of a specification explaining the number of voluntarily disclosed items in terms of
these exogenous factors. That is, our segment disclosure score captures whether the
firm discloses more or less than what one would predict given its characteristics
(diversification, different types of risks, growth, etc.). We posit that firms that disclose
more than analysts expect ex-ante given their characteristics will be rewarded with
a lower cost of capital. This approach of using the residual of an estimation of
the number of voluntarily disclosed items on the determinants of the disclosure
decision tackles endogeneity issues related to spurious correlations among common
determinants of the cost of capital and voluntary disclosure. For example, Hann
et al. (2013) show that more diversified firms have a lower cost of capital. Given
that diversified firms will provide more segment disclosures, it is crucial that in our
segment disclosure proxy we control for diversification. Similarly, Gietzmann and
Ostaszewski (2014) show that firms for which estimating future earnings is more
difficult, which arguably will affect the cost of capital, disclose more. Our segment
1 Most prior literature assumes that increased disclosure increases the overall amount of information
available in the economy. However, in an analytical study, Tang (2014) shows that increased disclosure could
also discourage private information production and might also lead to a decreased amount of information
overall. While theoretically sound, prior empirical evidence does not support this argument.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
3
disclosure proxy also controls for the determinants of the difficulty in estimating
future earnings. Also, from the findings in Johnstone (2013), one could argue that
improved disclosure could lead to higher cost of capital because of the unveiling of
certain operational or financial risks. Again, we define our segment disclosure score
to control for different types of risk. Once we control for the previous variables in
the construction of our proxy for voluntary disclosure, we expect, first, the score
generated to be a sticky variable that captures a pre-specified disclosure policy.2 Firms
with a higher score are firms for which estimating future profitability measures (like
earnings or cash flows) is easier. Second, our segment disclosure score controls for
the main determinants of the decision to disclose segment information: business and
geographic diversification, information asymmetries, operating risk, growth options,
performance, etc. Therefore, our results capture information effects of segment
disclosure, and not any of the effects that the aforementioned variables could have
on the cost of capital that may generate endogeneity problems of spurious correlation
between cost of capital and voluntary segment disclosure.
Using a sample of non-regulated and non-financial firms for the period 2001–2006,
and our proxy for voluntary segment disclosure, we show that firms providing better
segment disclosure enjoy lower costs of equity capital. We find, however, that such
decrease in the cost of equity capital is less pronounced in the presence of larger
competitive pressures. These results are robust to the use of tests based on asset pricing
and implied cost of equity capital. In addition, we provide empirical evidence that
segment disclosure improves investors’ ability to estimate the firm’s future earnings
by showing that better segment disclosure reduces analysts’ forecast errors. Finally, we
show that the provision of better segment information leads to a reduction in the firm’s
covariance with other firms’ returns, which is also consistent with improved segment
disclosure reducing estimation risk.
In all of our tests we control for the effects of earnings (accruals) quality on cost
of capital. It is noteworthy that in the asset pricing tests we find that accruals quality
is not a priced factor. This finding is in line with the results reported by Core et al.
(2008). We do not interpret our results, though, as segment reporting being as or
more important than accruals quality. On the contrary, a more plausible explanation
is that segment disclosures arise from an attempt at improving financial reporting in
all of its dimensions.
We provide robust results that add to prior research on the links between voluntary
disclosure and cost of capital (Botosan, 1997; Botosan and Plumlee, 2002; Gietzmann
and Ireland, 2005; Francis et al., 2008; and Dhaliwal et al., 2011). These studies offer
mixed evidence on the impact of voluntary disclosure on cost of capital. These mixed
results could be, in part, explained by the types of voluntary disclosures studied in
these prior papers. Our study differs from this prior research in that we use a type of
disclosure, segment reporting, with a clear predictive ability over future earnings and
cash flows (i.e., Kinney, 1971; Collins, 1976; Silhan, 1983; Baldwin, 1984; Balakrishnan
et al., 1990; Swaminathan, 1991; Berger and Hann, 2003; and Ettredge et al., 2005),
while prior studies have either used very wide and aggregate proxies for disclosure
(like AIMR scores in Botosan and Plumlee, 2002) or relative softer information
2 Prior research shows that firms commit to disclosure strategies and that there is little variation in
disclosure quality across time. In particular, Leung and Verriest (2014) show that firms do not alter their
geographic segment disclosures in response to the passage of IFRS 8, which reduced substantially the
required disclosures on geographic segments.
C
2014 John Wiley & Sons Ltd
4
BLANCO, GARCIA LARA AND TRIBO
types (like discussion of corporate strategies, or number of employees, or industry
trends, among many other items, as in Botosan, 1997, and Francis et al., 2008; or
environmental disclosures, as in Dhaliwal et al., 2011) without a clear link with the
prediction of future earnings and cash flows.
Also, prior studies show that the effect of disclosures on cost of capital is contingent
upon the firm reporting choices. In particular, Gietzmann and Ireland (2005) find
cost of capital effects only for firms making aggressive accounting choices, and Francis
et al. (2008) find that any cost of capital effects disappear once they control for
earnings quality. We extend the evidence reported by Gietzmann and Ireland (2005)
and Francis et al. (2008). Our study differs from theirs in that we find that voluntary
disclosure (in our case, segment disclosure) leads to lower cost of capital, and that
this effect is not contingent upon the quality of the reported earnings. This result
confirms our expectation that the particular disclosure type that we analyze (segment
information) is particularly relevant for market participants, while the disclosure of
other information items, without a clear link with the estimation of future earnings
and cash flows, might be less relevant and contingent-like.
Given this informational importance of segment reporting, even beyond earnings
quality, the concern that proprietary costs might reduce the overall benefits of
disclosure is especially relevant in this case. In fact, prior evidence on the cost of
capital effects of segment disclosures provides mixed results (Dhaliwal et al., 1979; Yoo
and Semenenko, 2012; Saini and Herrmann, 2013; and Leung and Verriest, 2014).
We provide a more complete picture of the role of segment disclosure in reducing
cost of capital by considering the moderating effects of competition on the ability of
segment disclosure to reduce cost of capital. We also contribute to prior literature
on segment reporting and cost of capital tackling, as discussed previously, endogeneity
concerns. As described by Clinch and Verrecchia (2013), failing to properly control for
endogenous disclosure decisions might lead to erroneous inferences in the literature
linking voluntary disclosure and cost of capital.
The remainder of the paper is structured as follows. In Section 2 we discuss the
relation between segment disclosure and cost of capital. In Section 3 we present the
research design. In Section 4 we describe the results. Finally, Section 5 summarizes and
concludes.
2. SEGMENT DISCLOSURE AND COST OF CAPITAL
Companies are increasingly international and diversified. The valuation of an international or a diversified firm requires information not only about overall firm activity, but
also about segments of the firm because the performance, risk and potential growth
of different business or geographic lines vary appreciably (SFAS 131; Ernst and Young,
2004; Palepu et al., 2004). Without this disaggregation in segments, predicting the
firm’s future cash flows becomes more difficult (AIMR, 1993), and financial statements
are less useful (Prodhan and Harris, 1989). Regulators require segment disclosure
with the objective of providing “information about the different types of business
activities in which a firm engages and the different economic environments in which it
operates to help users of financial statements to a. Better understand the enterprise’s
performance, b. Better assess its prospects for future net cash flows and c. Make more
informed judgments about the enterprise as a whole” (SFAS 131, paragraph 3).
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
5
A wealth of academic research has focused on segment reporting, showing several
benefits of improved segment information. First, segment information is expected
to help current and potential investors to improve their capital allocation decisions.
Previous literature finds that segment characteristics are useful in equity valuation
and that the value relevance of accounting numbers is higher in firms that provide
disaggregated segment information, especially when operating segments have increasingly different profitability and growth opportunities (Foster, 1975; Tse, 1989; and
Chen and Zhang, 2003). Similarly, segment information is expected to facilitate the
estimation of future earnings and cash flows. Early evidence on the use of segment
information shows that financial analysts issue more accurate earnings forecasts if they
have access to segment data (Kinney, 1971; Collins, 1976; Emmanuel and Pick, 1980;
Silhan, 1983; Baldwin, 1984; Balakrishnan et al., 1990; and Lobo et al., 1998), and
that there is less dispersion in their forecasts (Swaminathan, 1991). More recently,
Botosan and Stanford (2005) show that after the implementation of SFAS 131, firms
withhold less segment information and they argue that this improves the information
environment of the firm. Consistent with this improved information environment,
Hope et al. (2008) show that improved segment information after the passage of SFAS
131 reduces the mispricing of foreign earnings, and that, more generally, segment
information helps predicting future asset write-downs (Collins and Henning, 2004).
Second, segment information decreases information asymmetries between managers and equity holders (Greenstein and Sami, 1994), and permits better monitoring
of managers (Bens and Monahan, 2004). This reduction in information asymmetries
and improved monitoring over managerial decisions constrains managerial possibilities of engaging in inefficient investments that decrease firm value (Hope and Thomas,
2008), and, more generally, leads to a reduction in agency costs (Berger and Hann,
2007).
Taking into consideration the previous arguments, we posit that segment information, particularly when it goes beyond what is compulsory according to regulation
(voluntary segment information),3 reduces the cost of capital through two channels.
First, it contributes to improving the firm information environment. This improved information environment permits a more precise estimation of future earnings and cash
flows, leading to a decrease in estimation risk. We expect this increased information
precision and reduced estimation risk to lead to a lower cost of capital (Gietzmann
and Ireland, 2005; Lambert et al., 2007, 2012). And second, improvements in segment
information, particularly when it is voluntary, allow investors as well as financial
analysts to monitor managerial actions more accurately and with lower costs.4 The
consequence is a generation of value for shareholders through an improvement in the
quality of managerial investment decisions accompanied by a reduction in monitoring
costs. In these circumstances, investors will be more willing to lend capital to the firm
demanding lower returns, which reduces a firm’s cost of capital. This indirect link
between the role of disclosure for monitoring managerial decisions and cost of capital
is also described by Lambert et al. (2007).
3 Cheynel (2013) provides analytical evidence showing how voluntary disclosures reduce the cost of capital
of disclosing firms.
4 Prior research shows that improved segment disclosure leads to lower agency costs in the form of improved
investment decisions (Hope and Thomas, 2008; Blanco et al., 2014).
C
2014 John Wiley & Sons Ltd
6
BLANCO, GARCIA LARA AND TRIBO
Consequently, the central hypothesis in this paper is as follows:
H1: Firms that disclose better segment information are rewarded with a lower cost
of capital.
However, firms subject to competitive pressures face increasing proprietary costs
as they increase segment disclosures (Hayes and Lundholm, 1996; Stanford Harris,
1998).5 We argue that these proprietary costs make segment disclosure less effective
in decreasing cost of capital, and might even lead to segment reporting leading
to increased cost of capital. Gaspar and Massa (2006) and Valta (2012) show that
increased competition leads to higher financing costs. Thus, our second hypothesis
is as follows:
H2: Proprietary costs moderate the relation between segment disclosures and cost
of capital. The decrease in cost of capital driven by segment disclosure is
expected to be less pronounced (or even become an increase in cost of capital)
for firms that suffer larger competitive pressures.
3. RESEARCH DESIGN
(i) Measuring Segment Disclosure
In our main tests, we use a segment disclosure proxy that captures whether a firm
is disclosing more or less than what one would consider normal given its characteristics
(diversification, risk, growth, etc.). For simplicity, throughout the text, we refer to this
proxy as segment disclosure quality (Qlt Seg). To elaborate Qlt Seg we first compute
an index for the quantity of voluntarily reported segment disclosure (Qtt Seg), based
on segment information contained in Compustat. Then, we take the decile ranks of
the residuals of a regression of Qtt Seg on the determinants of segment disclosure as
our final proxy for segment disclosure quality (Qlt Seg). When a firm discloses more
(less) information than the average given its characteristics, this enhances (constrains)
investors’ ability to estimate firm’s future cash flows. This, in turn, reduces (increases)
estimation risk and leads to lower (higher) cost of capital. That is, given two firms
that are identical in terms of equal diversification, facing the same types of risks,
with the same growth, etc., we expect Qlt Seg to identify the one that discloses more
information, and we expect the one that discloses more information to be rewarded
with a lower cost of capital because estimating future earnings (and monitoring
managers) will become easier.
To compute Qtt Seg, we implement the following procedure: First, we take the
segment data available in Compustat, and for every reported business/geographic
segment for each firm, we analyze whether the segment is reported on a compulsory
5 There is a large stream of literature showing that competition decreases disclosure. For example, Burks
et al. (2013) analyze how banks’ disclosure policies are affected by increases in bank competition. There
is also evidence for non-financial firms on how product market competition affects disclosure (Board,
2009). More related to the effects of disclosure and proprietary costs on financing, Yosha (1995) shows that
there are financing consequences connected to the costs of disclosing private information to competitors.
Hence, competition affects disclosure policies given the higher proprietary costs of information disclosure
in competitive environments.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
7
or a voluntary basis. Second, for the compulsory segments, we distinguish between the
items reported compulsorily, as required by SFAS 131, and the items reported on a
voluntary basis. Third, we create the business (geographic) segment score, Qtt Seg Bus
(Qtt Seg Geo), by adding 1 point for every voluntarily disclosed item in every mandatory
business (geographic) segment, and 1 point for every item in each voluntary business
(geographic) segment. Finally, we create the overall index of quantity of voluntary
segment disclosure (Qtt Seg) by summing the business and geographic segment scores.
In the Appendix we describe the criteria that we use to distinguish between mandatory
vs. voluntary segments, and mandatory vs. voluntary items.6
Once we have calculated the voluntary segment disclosure index (Qtt Seg), in
order to obtain the index for quality of segment disclosure (Qlt Seg) we estimate the
following industry fixed effect regression:
Q tt Se g j,t = a + β1 Business Diversification j,t
+ β2 Geographic Diversification j,t + β3 Inform. Asymmetries j,t−1
+ β4 Size j,t + β5 Growth j,t + β6 Earnings Qlt j,t + β7 Leverage j,t
+ β8 Audit Firm j,t + β9 Listing Status j,t + β10 Proprietary Costs j,t
+ β11 New Financing j,t + β12 Profitability j,t + β13 Age j,t
+
βk Control year k, j,t + ε j,t
(1)
k
We take the decile ranks of the residuals of Model (1) and interpret firms in the
lowest decile (more negative residuals) as firms that provide less segment information
than expected given their characteristics (diversification, etc.) and, consequently,
these are firms for which the estimation of future cash flows becomes more difficult
than investors expected it to be. That is, firms in the lowest decile of the residuals
of Model (1) are expected to bear more estimation risk. On the other hand, firms
in the top decile disclose more segment information than expected given their
characteristics, and thus these are firms for which estimating future cash flows becomes
easier. That is, firms in the top decile of the residuals of Model (1) bear lower
estimation risk. For expositional purposes we refer to these decile ranks as segment
reporting quality (Qlt Seg).
The three main determinants of segment disclosure in Model (1) are business and
geographic diversification and information asymmetries.7 More diversified firms and
with larger information asymmetries will be penalized with a larger cost of capital if
they do not provide additional segment information to facilitate the estimation of
their future cash flows. We measure business diversification using the primary and
6 We have studied the convergent validity (internal consistency) and discriminant validity of this index. For
internal consistency we rely on Cronbach’s alpha (Saini and Herrmann, 2013), which has a value of 0.8071
for the index. This figure is above the threshold of 0.7 considered as acceptable. Regarding discriminant
validity, we have investigated whether the items that we consider reported on a voluntary basis show larger
correlations among them than with the items that we consider reported on a mandatory basis (see Appendix
1). The correlations found are significantly larger within the group of voluntary items than between this
group and the group of mandatory items (intra-group correlations are higher than inter-group ones).
Hence, our index satisfies the convergent and discriminant validity tests.
7 We take information asymmetries in t−1 to prevent potential endogeneity problems of reverse causality
given that disclosure has a direct impact on information asymmetries.
C
2014 John Wiley & Sons Ltd
8
BLANCO, GARCIA LARA AND TRIBO
secondary SIC codes that Compustat assigns to each firm. For every firm, we create
the business diversification score by assigning 1 point for every different 2-digit SIC
code assigned by Compustat to the firm as forming part of its primary or secondary
activities. We define our geographic diversification index as the number of different
countries where the firm has subsidiaries.8 For example: if a given company X has four
subsidiaries, one in Spain, one in Italy and two in Croatia, we assign to this company a
geographic diversification score of 3, as the company has subsidiaries in three different
countries. Finally, we include, as a proxy for information asymmetries, the bid-ask
spread.9
As controls in Model (1) we include:
(a) Firm Size (the natural logarithm of firm’s market value of equity), as Buzby
(1975), Diamond and Verrecchia (1991) and Leuz (2004) find that larger firms
disclose more.
(b) Growth (the logarithm of the firm’s book-to-market ratio), as in Nagar et al.
(2003). Growth is an important control as growing firms disclose more and
measures of cost of capital can be noisy and influenced by growth (Easton and
Monahan, 2005).
(c) Earnings Qlt, a proxy for earnings quality (the absolute value of discretionary
accruals calculated using the modified Jones model, as in Dechow et al. (1995),
multiplied by −1 so that it increases with earnings quality),10 as Francis et al.
(2008) show a positive relation between earnings quality and overall disclosure
levels.
(d) Leverage (the ratio of total debt to total assets), as Jensen and Meckling (1976)
argue that leveraged firms incur larger monitoring costs and firms are expected
to disclose more to decrease these costs.
(e) Audit Firm, a dummy variable taking the value of 1 if the auditor is a Big Four
audit firm and 0 otherwise, as Hope (2003) finds that large audit firms pressure
their clients for increased disclosure.
(f) Listing Status, a dummy variable taking the value of 1 if the firm is listed on
the New York Stock Exchange (NYSE) or NASDAQ and 0 otherwise, given that
Leuz and Verrecchia (2000) find a link between disclosure and the firm listing
status.
8 We use subsidiaries information from Osiris. We take into account subsidiaries with a minimum of 25
percent of control by the company under analysis.
|bid −as k |
9 Measured as: Sp r e ad j,t = (bid j,tj,t+as k j,tj,t)/2 , where: bidj,t is firm j’s annual mean of the monthly bid prices for
year t, and askj,t is firm j’s annual mean of the monthly ask prices for year t.
10 Given the discussion regarding the usefulness of earnings quality proxies (Dechow et al., 2010), we also
run Model (1) using as alternative proxies for earnings quality the absolute values of the residuals of the
Jones (1991), Dechow and Dichev (2002) and McNichols (2002) models, and also the standard deviation
of the residuals of the Jones, modified Jones, Dechow and Dichev and McNichols models, calculated at the
firm level using rolling windows of five years, as in Francis et al. (2008). Our main results and inferences
do not vary when we use these other earnings quality proxies. Results are also robust to the use of signed
measures of earnings management.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
9
(g) Proprietary Costs, a proxy for industry concentration based on the Herfindahl
index,11 as there is plenty of evidence showing that proprietary costs affect
segment disclosure levels (Hayes and Lundholm, 1996; Stanford Harris, 1998;
Botosan and Stanford, 2005; and Bugeja et al., 2014).
(h) New Financing, a dummy variable taking the value of 1 if the firm raised new
capital funds or increased debt in a given year, and 0 otherwise, as Ettredge
et al. (2006) show that firms providing better segment disclosure are more likely
to access external funds with better conditions.
(i) Profitability (the percentile ranks of return on assets) is expected to affect firm’s
disclosure policy (Raffournier, 1995).
(j) Firm Age, measured as the difference between the current year and the first year
in which the firm appears in CRSP (Center for Research in Security Prices).
Lang (1991) and Chen et al. (2002) argue that younger firms are subject
to greater uncertainty and they can achieve greater benefits from additional
disclosure. On the other hand, age is taken as a proxy for reputation (Roberts
and Dowling, 2002), and firms with better reputation are expected to disclose
more to maintain such reputation (Armitage and Marston, 2008). Hence, the
expected relationship is ambiguous.
(ii) Testing the Relation between Cost of Equity Capital and Segment Disclosure
We use four different sets of tests to analyze the relation between cost of equity
capital and segment information. First, we study whether better segment information
facilitates predictions about the firm’s future performance. To do so, we analyze
whether it reduces analysts’ forecast errors. Second, we test whether better segment
information reduces the covariance of firm’s returns with the returns of all firms in
the same industry, which, in the Lambert et al. (2007) setting, leads to a reduction in
the cost of capital. Third, we study whether segment reporting is related to implied
cost of equity capital measures. Finally, we investigate whether market participants
price segment information quality. To do so, we use asset pricing tests as in Core
et al. (2008). We explain each of the four sets of tests next.
(a) Analysts’ Forecast Errors
As an initial test of H1, we study whether better segment disclosure reduces analysts’
forecast errors. To do so, we use the following model, estimated with industry fixed
effects (2-digit SIC code):
Forecast error j,t = a + β1 Qlt Seg j,t + β2 Earnings Q lt j,t
+ β3 Number analysts j,t + β4 DevForecast j,t
+ β5 Size j,t +
βk Control sector year k, j,t + ε j,t
(2)
k
N
11 Herf j = i=1
(Si j /S j )2 , where Sij = firm i’s sales in industry j, as defined by the two-digit SIC code; Sj
= the sum of sales for all firms in industry j; N = the number of firms in industry j. Greater values of Herf
indicate larger levels of industry concentration.
C
2014 John Wiley & Sons Ltd
10
BLANCO, GARCIA LARA AND TRIBO
The coefficient of interest in Equation (2) is β 1 . A negative coefficient implies a
reduction in forecast errors for firms disclosing better segment information. Given
that Francis et al. (2008) find that, when they control for earnings quality, the effect
of disclosure on cost of capital is substantially reduced or disappears completely, we
include earnings quality as explanatory variable. Also, we control for the number of
analysts following the firm and firm size, as prior research (Lim, 2001) shows that these
two variables are associated with forecast errors. Finally, prior finance literature (e.g.,
Stickel, 1990) shows a link between the deviation of forecasts and forecast errors. We
therefore also include the deviation of forecasts (DevForecast) as an additional control
variable. However, DevForecast in t can be correlated with Qlt Seg in t, as both of them
capture different aspects of the information environment of the firm. To avoid this
correlation interfering with our inferences regarding segment reporting, we estimate
equation (2) with three different definitions of DevForecast. In the first, we just take
DevForecast in t, as described in the model. In the second, we orthogonalize DevForecast
with respect to Qlt Seg. Finally, we use DevForecast in t−1 to instrument this variable
and prevent a problem of reverse causality given that forecast errors can affect the
dispersion of the errors.
(b) Firm’s Returns Covariance with other Firms’ Returns
In the Lambert et al. (2007) setting, the quality of accounting information can
influence the cost of capital. They show that better information (improved disclosure)
reduces the firm’s cash flows assessed covariance with other firms’ cash flows, and that
this disclosure effect is not diversifiable. In their framework, this leads to improved
disclosure reducing the cost of capital. To test this, we use current returns as a proxy
for the firm’s expected cash flows and study whether better segment disclosure reduces
the firm’s returns covariance with other firms’ returns in the same industry. We use the
following model, estimated with industry fixed effects (2-digit SIC code):
Cov(r j , r s e c tor ) j,t = a + β1 Qlt Seg j,t + β2 Earnings Q lt j,t
+ β3 Size j,t + β4 BM j,t + β5 Proprietary Costs j,t
+ β6 Leverage j,t + β7 Business Diversification j,t
+ β8 Geographic Diversification j,t
+ β9 Listing Status j,t + β10 Profitability j,t + β11 Age j,t
+
βk Control sector year k, j,t + ε j,t
(3)
k
where Cov(rj , rsector ) is the mean of the covariance between the monthly returns of the
firm and the monthly returns of all other firms in the industry where the firm operates,
over a one-year period. The coefficient of interest in Equation (3) is β 1 . If high quality
segment information reduces the covariance of the firm’s returns with the returns of
the firms in the same industry, β 1 will be significantly negative.
As before, we include earnings quality as a control. Regarding the other control
variables, previous studies find that beta – closely connected to the explained variable
of specification (3) – decreases with firm size, industry concentration, age, listing
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
11
status and diversification (Subrahmanyam and Thomadakis, 1980; Caves, 1982; Kim
et al., 1989, 1993; Lubatkin and Rogers, 1989; Harvey, 1991; Fama and French, 1992,
1993; and De Andres et al., 2014), and increases with book-to-market, leverage and
profitability (Fama and French, 1992, 1993). We include all of these variables as
controls in Equation (3).
(c) Implied Cost of Equity Capital Tests
The third set of tests consists of regressing a measure of implied cost of equity capital
on segment disclosure, earnings quality, and control variables commonly used in the
cost of capital literature: size, book-to-market ratio, beta, leverage and diversification.
To explore whether segment information quality reduces cost of equity capital we
estimate the following model using industry fixed effects (2-digit SIC code):
r E X −AN T E j,t = a + β1 Qlt Seg j,t + β2 Earnings Qlt j,t + β3 Size j,t
+ β4 BM j,t + β5 Beta j,t + β6 Leverage j,t
+ β7 Business Diversification j,t + β8 Geographic Diversification j,t
+
βk Control sector year k, j,t + ε j,t
(4)
k
The coefficient of interest in Equation (4) is β 1 . A reduction in cost of equity capital
related to increases in segment information quality will lead to a significantly negative
β 1 . As a proxy for implied cost of equity capital we use the mean of four different
proxies: the first two are based on the PEG and MPEG ratios proposed by Easton
(2004), the third is based on the Ohlson and Juettner-Nauroth (2005) abnormal
earnings growth model, and the fourth is based on the model proposed by Gode and
Mohanram (2003). These four measures are based on data from analysts’ forecasts.
We describe how we calculate each of these four proxies in Appendix 2.12
There is an ongoing debate on the validity of implied cost of capital measures
derived from analysts’ forecasts. On the one hand, Pastor et al. (2008) find that
implied cost of capital proxies based on analysts’ forecasts capture variation in cost
of capital well. Also, Botosan and Plumlee (2005) find that the PEG ratio proposed
by Easton (2004) is positively related to risk measures, and conclude that it is a
good proxy for implied cost of capital. On the other hand, Easton and Monahan
(2005) show that these implied cost of equity estimates are unreliable, unless forecast
errors and growth forecasts are low. Given the relation that Easton and Monahan
(2005) find between the estimates of implied cost of capital and growth, and given
that growing firms are expected to disclose more, one could be concerned that we
might find spurious correlations (growth-driven) between disclosure and the cost of
equity measures. However, it is important to highlight that our proxy for the quality
of voluntary disclosure (Qlt Seg) is orthogonal to growth expectations. This occurs as
in Equation (1), which we use to obtain Qlt Seg, the market-to-book ratio, a proxy for
growth, is included among the explanatory variables. Still, to circumvent any problems
12 We do not use the Claus and Thomas (2001) measure as Hail and Leuz (2006) and Botosan et al. (2011)
show that it is highly correlated with the proxy based on the Ohlson and Juettner-Nauroth (2005) model. In
particular, Hail and Leuz (2006) report a correlation of 0.945 between the two.
C
2014 John Wiley & Sons Ltd
12
BLANCO, GARCIA LARA AND TRIBO
related to the use of implied cost of equity capital proxies based on analysts’ forecasts,
we also estimate Equation (4) using a different proxy for risk, returns volatility, as the
dependent variable.
Regarding the control variables in Equation (4), we include a proxy for earnings
quality, to be certain that our segment reporting proxy is not just capturing the
overall quality of earnings. Also, previous literature finds that increases in size lead
to a decrease in the cost of capital (Fama and French, 1992, 1993; and Hail and
Leuz, 2006). We measure size as the logarithm of market equity value. The market
perceives high-growth firms as riskier, consistent with the asset pricing theory. Thus, we
include the log of the book-to-market ratio (Fama and French, 1992, 1993; Gebhardt
et al., 2001; and Hail and Leuz, 2013). Also, the capital asset pricing model (CAPM)
suggests that market beta should be associated with the cost of equity. Given this,
we include beta, measured as the coefficient from firm-specific CAPM regressions of
the firm’s returns, using the 60 months preceding fiscal year t, and a value-weighted
NYSE/AMEX/NASDAQ return as market index return.13 Additionally, we include
leverage, as it drives cost of capital upwards (Modigliani and Miller, 1958; Fama and
French, 1992, 1993). Finally, we include firm’s diversification, as it is associated with
lower risk (Caves, 1982; Kim et al., 1989, 1993; Lubatkin and Rogers, 1989; and Harvey,
1991), and with lower ex-ante cost of capital proxies (Hann et al., 2013).
(d) Asset Pricing Tests
The last set of tests we use rely on the Fama and French (1992, 1993) three-factor
model, augmented with a momentum factor (Carhart, 1997) and a liquidity factor
(Liu and Du, 2014). This model has recently been applied in the accounting literature
to study the relation between cost of capital and proxies for the quality of accounting
information (Core et al., 2008; Francis et al., 2008; McInnis, 2010; and Garcia Lara
et al., 2011). If segment information is a priced risk factor, then it should be related to
average stock returns. To test this, we group firms into portfolios, as realized returns
employing cross-sectional tests are noisy at the firm level (Black et al., 1972). Each
month, from January 1, 2001 to December 31, 2006, we create a hedge segment
information quality portfolio (HEDGE Qlt Seg). All firm-years in the sample are ranked
into 10 deciles according to segment reporting quality, and we take a long position
in the decile with better segment information (higher Qlt Seg) and a short position
in the decile with worse segment information (poorer Qlt Seg). Then we estimate the
following time series regression for the hedge segment information quality portfolio:
R j t − Rt f = α + β1 RMRF t + β2 SMB t + β3 HML t + β4 UMD t + β5 LIQ t + ε j t
(5)
We include the excess market return (RMRF), size (SMB) and value versus growth
stocks (high minus low book-to-market, HML) as in Fama and French (1992, 1993).
We also include momentum portfolios (UMD) as in Carhart (1997), and liquidity
portfolios (LIQ) as in Liu and Du (2014).
In Model (5), α represents the average return in excess of the return predicted
by the portfolio sensitivity to the risk factors in the model. If the model is properly
13 It is important to note that by including beta in the specification, we are capturing incremental effects of
voluntary segment information beyond what is internalized in financial markets through the beta parameter.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
13
specified (i.e., if it includes all risk factors that affect the firm), the estimated α should
be zero (Black et al., 1972). However, if the model omits a risk factor, then portfolios
with greater exposure to that factor will have higher α, because they have greater
average excess return unexplained. If segment information quality is a risk factor,
and it is orthogonal to beta, size, momentum, book-to-market and liquidity effects,
then we should observe decreasing estimates of α in Model (5) as we move from
bad to good segment information quality portfolios (long on firms providing the best
segment information quality and short on those with the worst segment information
quality).
As an additional set of tests, we investigate whether segment disclosure has an
influence on the firm’s realized returns. To do so, we create a HILO Qlt Seg factor.
This factor is the return of the value-weighted factor-mimicking portfolio for segment
disclosure.14 We rank Qlt Seg into quintiles and we take a long position on the two
quintiles with the best segment information and a short position on the two quintiles
with the worst segment information. As in previous tests, we control for Earnings Qlt. To
do so, we create a HILO Earnings Qlt factor, taking a long position on the two quintiles
with the best accrual quality and a short position on the two quintiles with the worst
accrual quality. We then use a two-stage cross-sectional regression approach, where
excess returns are regressed on risk factor betas.
In the first stage, we estimate multivariate betas from 25 value-weighted portfolios
sorted by Size and B/M15 using a time-series regression of excess returns for a portfolio
on the contemporaneous returns to the Fama–French, momentum and liquidity
factors, the segment disclosure quality factor and the earnings quality factor:
Rp ,t − Rt f = α + β1 RMRF t + β2 SMB t + β3 HML t + β4 UMD t + β5 LIQ t
+ β6 HILO Qlt Seg t + β7 HILO Earnings Qlt t + εt
(6)
where Rp,t is the return of portfolio p for month t. RMRF is the excess return on the
value-weighted market portfolio. Excess returns equal the value-weighted return on
the value-weighted NYSE/AMEX market index return from CRSP less the risk-free
rate. SMB (Small minus Big) is the monthly return of small firms over big firms, and
HML (High minus Low) is the monthly return of high B/M firms over low B/M firms.
HILO Qlt Seg is the monthly return of good quality segment disclosure firms over poor
quality segment disclosure firms and HILO Earnings Qlt is the monthly return of good
earnings quality firms over poor earnings quality firms.
In the second stage, we collect the portfolio-specific loadings from Equation (6) and
estimate the factor premium conditional on the first-stage loadings with cross-sectional
regressions using the Fama and MacBeth (1973) procedure to mitigate concerns about
cross-sectional dependence in the data. The model is as follows:
Rp ,t − Rt f = α + δ1 βRM RF + δ2 βSM B + δ3 βH M L + δ4 βU M D + δ5 βLI Q
+ δ6 βH I LO
Q lt Se g
+ δ7 βH I LO
E ar nings Q lt
+ εt
(7)
14 Results are robust if we use equally weighted portfolios.
15 The portfolios are the intersections of five portfolios formed based on the book-to-market ratio and five
portfolios formed based on size.
C
2014 John Wiley & Sons Ltd
14
BLANCO, GARCIA LARA AND TRIBO
If firms providing better segment information enjoy a lower cost of capital, then δ 6
should be negative.
(e) Analysis of the Moderating Effect of Competition and Proprietary Costs
As we pointed out in the previous sections, segment information could favor competitors, creating uncertainty about future earnings and the viability of the firm. This
increased uncertainty would lead to a higher cost of capital (Gaspar and Massa, 2006;
Valta, 2012). To study the moderating effect of competition and proprietary costs on
the relation between segment disclosure and cost of capital, we modify our tests based
on ex-ante estimates of cost of equity capital, and our tests based on the Fama–French
five-factor model, to also capture the effects of proprietary costs. Regarding the tests
based on ex-ante estimates of cost of capital, we estimate the following model using
industry fixed effects (2-digit SIC code):
r E X −AN T E j,t = a + β1 Qlt Seg j,t + β2 Competition j,t + β3 Q lt Se g j,t × Competition j,t
+ β4 Earnings Qlt j,t + β5 Size j,t + β6 BM j,t + β7 Beta j,t
+ β8 Leverage j,t + β9 Business Diversification j,t
+ β10 Geographic Diversification j,t +
βk Control sector year k, j,t + ε j,t (8)
k
The coefficient of interest in Equation (8) is β 3 . A positive coefficient implies a
positive moderation in the variation of implied cost of equity capital when segment
information quality increases and the firm suffers competitive pressures. To measure
Competition, we use two different sets of measures. In the first set of measures, we use
proxies that identify industries where competition is more pronounced. In the second
set of measures, we use firm-specific measures to capture firm-specific situations where
additional disclosure might penalize a given firm because of competitive pressures,
regardless of whether it operates in a more or less competitive industry.
Regarding the first set of measures, which identify industries, we use seven industrybased measures of competition (HiTech, IndustrySalesGrowth, IndustryMarkUp, NumFirms, Herf, Exist-Comp and SPA). First, we focus on firms operating in a high technology
industry, as the competition in these industries can be higher than in other industries.
We create a dummy variable, HiTech, taking the value 1 for firms with two-digit SIC
codes equal to 28, 35, 36, 73 or 87, and 0 otherwise (Gong et al., 2008; Kimbrough
and Louis, 2011). Second, we use the industry sales growth. More dynamic industries,
with larger growth in sales, tend to be more competitive (Escribano et al., 2009).
Third, we calculate the industry markups following Allayannis and Ihrig (2001).16
The lower the markups, the larger the degree of competition. The fourth measure
is the number of firms in the industry (NumFirms). We expect larger competition
in industries with a larger number of firms (Bresnahan and Reiss, 1991). The fifth
measure is the Herfindahl index, which has been widely used to measure competition
(recent examples are Dhaliwal et al., 2011, Valta, 2012, and Markarian and Santalo,
16 This is defined as Industry Markup = (Value of Sales + Inventories − Pension & Retirement Expenses − Cost
of Materials)/(Value of Sales + Inventories). We employ pension and retirement expenses instead of payrolls
because of data availability in Compustat.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
15
2014). Our sixth measure is Exist-Comp, the first principal component extracted from
principal component analysis of four competition variables: IND-HHI, IND-CON4,
IND-NUM and IND-MKTS, where IND-HHI is the negative of the Herfindahl index;
IND-CON4 is the negative of the four-firm concentration ratio, measured as the sum
of market shares of the four largest firms in an industry; IND-NUM is the total number
of firms in the industry; and IND-MKTS is the product market size, measured as the
natural log of industry aggregate sales. Li (2010) and Dhaliwal et al. (2014) use this
measure as a proxy for existing competition, and provide a thorough discussion of why
they consider these four components, and not others.
Finally, our last measure is the Speed of Profit Adjustment (SPA) metric (Stanford
Harris, 1998). To estimate SPA we run the following regression separately for each
three-digit SIC and year:
Xi, j,t = β0, j + β1, j (Dn Xi, j,t−1 ) + β2 j (Dp Xi, j,t−1 ) + εi, j,t
(9)
where Xi,j,t = the difference between firm i’s return on assets and the mean return on
assets for its industry j, in year t; Dn = 1 if Xi,j,t is less than or equal to 0, and 0 otherwise;
and Dp = 1 if Xi,j,t is greater than 0, and 0 otherwise. A positive and significant β 2j
indicates that firms in industry j with a return above the median are able of maintain
this situation over time, suggesting less competition. We give SPA a value of 0 for firms
in industries where β 2j is positive and significant, and 1 otherwise.
Throughout all the industry-based measures of competition, except for HiTech and
SPA, which are defined as described above at the industry level, we give each of the
other five proxies a value of 1 whenever the industry-year value for each measure is
above or below (depending on how each measure is calculated) the median of the
sample-year.17
Regarding our second set of measures, which capture firm-specific aspects of
competition, we look at three proxies: FirmSalesGrowth, FirmMarkUp and R&D intensity
(HiR&D). Regarding sales growth, we expect that firms with large growth in sales
suffer larger competitive pressures and would be penalized if they give away information on how they have been able to sustain their growth rates. FirmSalesGrowth is
a dummy variable taking a value of 1 if firm’s growth in sales is above the median
for the industry-year, and 0 otherwise. Our second proxy is the markup defined in
Allayannis and Ihrig (2001) as mentioned before, calculated at the firm level. Firms
with a markup lower than the industry-year median are assigned a value of 1, and 0
otherwise. Finally, we focus on R&D intensity, assuming that firms more intensive in
R&D face tougher competition. Franko (1989) shows that R&D intensity is linked to
future growth in sales, and that more intensive R&D firms will therefore be subject
to greater competitive pressures. We therefore assume that these firms might suffer
costs in case they provide additional information about their activities. An additional
explanation as to how R&D intensity captures competitive pressures is provided by
Vives (2008), who shows that firms that suffer competitive pressures are more likely to
resort to investments in R&D to survive. We create a dummy variable, HiR&D, taking a
17 We follow Valta (2012) and use dummy variables because this allows an easier economic interpretation
of the coefficients and because with the dummy variable we reduce problems introduced by measurement
errors in the construction of each of the variables.
C
2014 John Wiley & Sons Ltd
16
BLANCO, GARCIA LARA AND TRIBO
value of 1 for firms with R&D levels above the industry-year median, and 0 otherwise.
We measure R&D as the ratio of R&D expenses to sales.18
Finally, we also introduce the measures of competition in the asset pricing tests.
In particular, we add an additional factor that captures competition (in the reported
tables, we use the Herfindahl index), and analyze whether after including this
additional factor the quality of segment disclosure is still a priced risk factor.19
(iii) Sample Selection
We extract all available non-financial and non-regulated firms from the Compustat
annual files for the period 2001 to 2006, with the necessary data to calculate the
earnings quality measures and all variables needed for our disclosure tests.20 The
number of subsidiaries used to calculate our proxy for geographic diversification
is extracted from BvD Osiris.21 Market data are extracted from CRSP and analysts’
data from I/B/E/S. Our final sample comprises 10,002 firm-year observations (1,667
unique firms) with data on all variables to run all of our tests. We exclude observations
with missing data from any of the variables needed. To mitigate the undesirable effect
of outliers, we delete the top and bottom percentile of the distribution of all variables.
The mean (median) number of items of voluntary segment information (Qtt Seg)
reported by our sample firms is 42.25 (39), with a standard deviation of 20.29 (Table 1).
Note that the standard deviation is high, but it is mainly due to the different number
of reported segments among firms. Regarding our proxy for segment disclosure
quality (Qlt Seg), firms in the 90th percentile (better disclosure) disclose 19.15
more segment items than predicted by Model (1), while firms in the 10th percentile
(worse disclosure) disclose 17.52 fewer items than predicted by the model. The
standard deviation (15.18) is also substantial. Throughout our tests we use the decile
ranks of Qlt Seg, but we include in Table 1 the raw values so that we have a clearer
view of how the quality of segment disclosure varies, in terms of reported items, across
deciles. Mean leverage is 20.19 percent, indicating that our sample firms are relatively
low leveraged, but they are issuing new debt or equity to finance their projects (mean
value of New Financing = 0.88). Also, most of our sample firms are audited by Big-4
firms, and are listed on the NYSE or NASDAQ.
4. RESULTS
In Table 2 we show the pairwise correlations between the quantity of voluntary segment
information score (Qtt Seg) and firm characteristics. Much as expected, business and
geographic diversification are very strongly correlated with Qtt Seg (31.2 percent and
18 When R&D is a missing value, we give it a value of 0. If it is the case that some of these firms conduct
R&D, we are then underestimating the differential moderating effect of competition in the connection from
voluntary disclosure to the cost of capital. Thus, if anything, this would bias results against our hypotheses.
19 We do not consider the effects of competition on our other two sets of tests (analysts’ forecast errors and
the covariance of returns) as they both are intended to capture mostly the positive informational effects of
disclosure, and thus are not expected to be substantially affected by competition.
20 The sample finishes in 2006 as the data on the subsidiaries (needed to create Qlt Seg) cannot be collected
in a standardized fashion after that year.
21 We assume the number of subsidiaries does not change if the data are not available for one year (i.e., if
a firm has no data for 2004, we assume that the number of subsidiaries is equal to that of 2005). Results are
C
2014 John Wiley & Sons Ltd
C
Qtt Seg
Qlt Seg
Qtt Seg Bus
Qlt Seg Bus
Qtt Seg Geo
Qlt Seg Geo
Earnings Qlt
Business Diversification
Geographic Diversification
Information Asymmetries
Size
B/M
Leverage (%)
Audit Firm
Listing Status
Proprietary Costs (%)
New Financing
Profitability
Age
Forecast Error
DevForecast
Number of Analysts
Mean
42.2542
0.0000
23.3837
0.0000
18.8705
0.0000
−0.0743
1.9679
3.0755
0.1504
4,771
0.5211
20.1953
0.9153
0.9696
10.4644
0.8802
0.0140
13.5585
0.1270
0.0690
8.4469
N
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
10,002
20.2906
15.1765
17.4141
13.3659
9.0299
7.6832
0.1118
1.1461
4.2204
0.1475
11,669
0.4632
18.8044
0.2784
0.1716
9.5428
0.3247
0.1497
10.2824
0.8513
0.1297
6.2905
Std. Dev.
20
−17.5237
8
−14.0958
11
−8.8721
−0.2097
1
1
0.0225
121
0.162
0
1
1
4.0059
0
−0.1346
4
−0.1724
0.01
3
10%
Table 1
Descriptive Statistics
27
−10.2192
9
−9.1850
11
−5.3393
−0.0970
1
2
0.0458
317
0.262
1.633
1
1
4.9259
1
−0.0042
6
−0.0592
0.02
4
25%
39
−1.8707
18
−2.5236
17
−0.7478
−0.0349
2
2
0.1041
923
0.419
17.971
1
1
7.8234
1
0.0444
10
−0.0058
0.03
6
Median
53
8.6758
33
6.7440
24
4.1548
−0.0000
2
3
0.2076
3,162
0.635
31.9905
1
1
11.1554
1
0.0846
18
0.0833
0.07
11
75%
2014 John Wiley & Sons Ltd
(Continued)
69
19.1472
47
17.6357
31
10.1924
−0.0000
3
5
0.3416
11,670
0.943
45.5055
1
1
20.8409
1
0.1284
32
0.4444
0.15
18
90%
SEGMENT DISCLOSURE AND COST OF CAPITAL
17
0.1253
1.1765
0.0161
0.0161
0.0133
10,002
10,002
10,002
102,024
102,024
0.0416
0.9368
0.1668
0.2394
0.2394
Std. Dev.
0.0936
0.3599
−0.0513
−0.2022
−0.2052
10%
0.1023
0.9133
−0.0137
−0.0814
−0.0841
25%
0.1131
1.0462
0.0000
0.0017
−0.0008
Median
0.1632
1.3680
0.0192
0.0806
0.0780
75%
0.1921
2.2228
0.0739
0.2145
0.2117
90%
The sample consists of 10,002 firm-year observations for the period 2001–2006. Qtt Seg = voluntarily disclosed business and geographic segment reporting items.
Qlt Seg = the regression residuals obtained from a regression of the firm’s year t. Qtt Seg on controls and determinants of segment disclosure as defined in the main
text. Qtt Seg Bus = the number of voluntary disclosure elements found in the sample firms for business segment disclosure. Qlt Seg Bus = the regression residuals
obtained from a regression of the firm’s year t Qtt Seg Bus on control and determinants of segment disclosure. Qtt Seg Geo = the number of voluntary disclosure
elements found in the sample firms for geographic segment disclosure. Qlt Seg Geo = the regression residuals obtained from a regression of the firm’s year t Qtt Seg Geo
on controls and determinants of segment disclosure. Earnings Qlt = the absolute value, multiplied by −1, of discretionary accruals calculated as the residual of the
modified version of the Jones (1991) accruals model (Dechow et al., 1995), as applied to total accruals. Business Diversification = number of different sectors in which
|bid−as k|
the firm operates. Geographic Diversification = number of different countries where the firm operates. Information Asymmetries = bid-ask spread, calculated as (bid+as
k)/2
measured in t-1. Size = the firm’s market value of equity measured at the beginning of fiscal year 2001–2006. B/M = the firm’s book-to-market ratio measured at the
beginning of fiscal year 2001–2006. Leverage = debt-to-total assets ratio in percentage. Audit Firm = 1 if auditor firm is a Big-Four firm and 0 otherwise. Listing Status
= 1 if firm is listed on the NYSE or NASDAQ and 0 otherwise. Proprietary Costs = Herfindahl index in percentage, as described in the text. New Financing = 1 if the
firm has issued new debt or equity and 0 otherwise. Profitability = return on assets. Age = the difference between the first year when the firm appears in CRSP and the
current year. Forecast Error = Analysts’ forecast errors. It is calculated as forecast in t-1 of eps for year t – actual eps of year t, scaled by the absolute value of actual eps in
year t. DevForecast = deviation of analysts’ forecasts. It is calculated as the standard deviation of analysts’ forecasts of eps for the year t. Number of Analysts = number of
eps forecasts of the firm in year t. rEX-ANTE is the mean of the four estimates for the implied cost of equity capital, as described in Section 3.ii.c. Beta = coefficient from
firm-specific CAPM regression using the 60 months preceding fiscal year 2001–2006. Cov (ri , rsector ) = mean annual covariance of the monthly return of a firm with
the monthly return of the sector in which the firm operates. Realized Returns = monthly realized returns. Excess Realized Returns = monthly excess realized returns over
the risk free rate.
rEX-ANTE
Beta
Cov (ri , rsector )
Realized Returns
Excess Realized Returns
Mean
N
Table 1
Continued
18
BLANCO, GARCIA LARA AND TRIBO
C
2014 John Wiley & Sons Ltd
C
a
b
c
d
e
f
g
H
i
j
k
l
m
n
2014 John Wiley & Sons Ltd
The sample consists of 10,002 firm-year observations for the period 2001–2006. Bold numbers are significant at p < 0.05. Qtt Seg = voluntarily disclosed segment
reporting items. Earnings Qlt = the absolute value, multiplied by −1, of discretionary accruals calculated as the residual of the modified version of the Jones (1991)
accruals model (Dechow et al., 1995), as applied to total accruals. Business Diversification = number of different sectors in which the firm operates. Geographic
|bid−as k|
Diversification = number of different countries where the firm operates. Information Asymmetries = bid-ask spread, calculated as (bid+as
k)/2 measured in t−1. Size = the
logarithm of the firm’s market value of equity measured at the beginning of fiscal year 2001–2006. B/M = the logarithm of the firm’s book-to-market ratio measured
at the beginning of fiscal year 2001–2006. Leverage = debt-to-total assets ratio in percentage. Audit Firm = 1 if auditor firm is a Big-Four and 0 otherwise. Listing Status
N
S
= 1 if firm is listed on the NYSE or NASDAQ and 0 otherwise. Proprietary Costs = Herfindahl index in percentage, calculated as Herf j = i−1
( Sijj )2 , where Sij =
business i’s sales in industry j, as defined by two-digit SIC code; Sj = the sum of sales for all firms in industry j. New Financing = 1 if the firm has issued new debt or
equity and 0 otherwise. Profitability = percentile rank of return on assets. Age = the difference between the first year when the firm appears in CRSP and the current
year.
Qtt Seg (a)
1
0.2178
1
Earnings Qlt (b)
Business Diversification (c)
0.3122 0.1299
1
Geographic Diversification (d) 0.1053 0.0488 0.0536
1
Information Asymmetries (e)
0.0283 0.0221 −0.0038 0.0172
1
Size (f)
0.3399 0.2932 0.2234 0.0269 −0.0183
1
B/M (g)
0.0482 0.0703 0.0625 0.0147 0.0212 −0.2792
1
Leverage (h)
0.1178 0.1494 0.1368 0.0419 0.0525 0.1125 0.0082
1
Audit Firm (i)
0.1318 0.0659 0.0708 0.0049 0.0172 0.1903 −0.0170 0.0287
1
Listing Status (j)
0.1180 0.2703 0.0464 0.0207 0.0195 0.1131 −0.0232 0.0199 0.0382
1
Proprietary Costs (k)
−0.0674 0.0869 0.0807 0.0010 −0.0012 −0.0080 0.0843 0.0993 0.0101 0.0403
1
New Financing (l)
0.1096 0.1080 0.0786 0.0221 0.0094 0.1918 −0.0861 0.1344 0.0548 0.0262 0.0453
1
Profitability (m)
0.0504 0.1350 0.0232 0.0092 −0.0184 0.3834 −0.3257 −0.1637 0.0240 0.0569 0.0670 0.3164
1
Age (n)
0.1943 0.1597 0.1821 0.0152 −0.0190 0.2420 0.0002 0.0697 0.0521 −0.0228 0.0080 0.0785 0.1472 1
Variable
Table 2
Correlation Matrix
SEGMENT DISCLOSURE AND COST OF CAPITAL
19
20
BLANCO, GARCIA LARA AND TRIBO
Table 3
Industry Fixed Effect Regression of the Quantity of Voluntary Segment
Disclosures (Qtt Seg) on its Determinants and Controls
Variable
Expected sign
Coefficient (p-value)
Business Diversification
+
Geographic Diversification
+
Information Asymmetries
+
Size
+
B/M
+
Earnings Qlt
+
Leverage
+
Audit Firm
+
Listing Status
+
Proprietary Costs
-
New Financing
+
5.2727
(0.000)
0.2532
(0.000)
2.9581
(0.013)
3.9419
(0.000)
3.9496
(0.000)
8.0612
(0.000)
0.0844
(0.000)
2.3874
(0.000)
6.4297
(0.000)
−0.4776
(0.000)
1.6418
(0.000)
−0.0301
(0.000)
0.1173
(0.000)
2.4034
(0.173)
0.2958
Profitability
+/−
Age
+/−
Cons
R2
The sample consists of 10,002 firm-year observations for the period 2001–2006. Qtt Seg = voluntarily
disclosed segment reporting items. Business Diversification = number of different sectors in which the firm
operates. Geographic Diversification = number of different countries where the firm operates. Information
|bid−as k|
Asymmetries = bid-ask spread, calculated as (bid+as
k)/2 measured in t−1. Size = the logarithm of the firm’s
market value of equity measured at the beginning of fiscal year 2001–2006. B/M = the logarithm of the
firm’s book-to-market ratio measured at the beginning of fiscal year 2001–2006. Earnings Qlt = the absolute
value, multiplied by −1, of discretionary accruals calculated as the residual of the modified version of the
Jones (1991) accruals model (Dechow et al., 1995), as applied to total accruals. Leverage = debt-to-total assets
ratio in percentage. Audit Firm = 1 if auditor firm is a Big-Four firm and 0 otherwise. Listing Status = 1 if
firm is listed on the NYSE or NASDAQ and 0 otherwise. Proprietary Costs = Herfindahl index in percentage,
N
S
calculated as Herf j = i−1
( Sijj )2 , where Sij = business i’s sales in industry j, as defined by two-digit SIC
code; Sj = the sum of sales for all firms in industry j. New Financing = 1 if the firm has issued new debt or
equity and 0 otherwise. Profitability = percentile rank of return on assets. Age = the difference between the
first year when the firm appears in CRSP and the current year.
10.5 percent respectively). Also, information asymmetries are positively correlated with
Qtt Seg. Earnings quality is significantly and positively related to Qtt Seg (21.8 percent).
robust to the use of a smaller sample including only firms with available data on subsidiaries for all the years
of the period under study.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
21
In Table 3 we show the results of an industry fixed effect regression of Qtt Seg,
our proxy for the quantity of segment disclosure, on the determinants of segment
disclosure. We take the decile ranks of the residuals of this regression as our proxy for
the quality of segment disclosure (Qlt Seg). The results show that firms operating in a
higher number of industry sectors and in more countries provide more comprehensive
segment information. Also, we find that firms with higher information asymmetries
provide more segment information. This is consistent with firms providing more
segment information to reduce uncertainties about the firm. Results also show that
the quantity of segment disclosure (Qtt Seg) increases with firm size, the book-tomarket ratio, earnings quality, leverage, being audited by a Big-Four firm, being listed
on the NYSE or NASDAQ, issuing new financing and firm age, and decreases with
profitability and proprietary costs. This latter result conforms to the underpinnings
behind H2, that is, that more competition diminishes firms’ incentives for disclosing
information. All of the controls are significantly associated with quantity of segment
information. Regarding the fitness of the model, the results show that the determinants of disclosure that we consider explain a significant amount of the variation in
Qtt Seg (29.58 percent). These results suggest that our index of voluntary segment
information (Qtt Seg) is a valid measure of disclosure. Our results are robust to the
use of geographic and business segment quantity measures separately, instead of the
aggregate measure Qtt Seg.
(i) Forecast Errors and Segment Disclosure
In Table 4 we show the results of the estimation of Equation (2), on whether segment
disclosure reduces analysts’ forecast errors. In the first column of the table we show
the results of the regression of forecast errors on quality of segment information
(Qlt Seg), analysts’ following, deviation of forecasts and size. The results show a mean
estimate of the coefficient on segment disclosure equal to −0.0074 (p = 0.006) when
we do not control for earnings quality and −0.0073 (p = 0.000) when we control for
earnings quality (second column).22 Thus, we find that better segment information
reduces forecast errors. The economic effect is such that, if the firm improves segment
reporting and moves to the next decile of segment reporting quality, the forecast error
decreases by 0.73 percentage points. The average forecast error for the whole sample
is 12.7 percent (see Table 1). In columns 3 and 4 we re-estimate Equation (2) using as
a control variable deviation of forecasts orthogonalized with respect to Qlt Seg, and in
columns 5 and 6 we use the deviation of forecasts in t−1, instead of in t. Results are in
line with those reported in columns 1 and 2. Inferences do not change either if we just
drop DevForecast from the model.
22 We do not find that earnings quality reduces forecast errors (the coefficient on earnings quality is not
significant at conventional levels). However, in an additional unreported sensitivity test we use a dummy
variable to capture earnings quality. We assign a value of 1 to firms with earnings quality above the median,
and 0 otherwise. We do this in an attempt to reduce measurement error in the earnings quality proxy. When
we use the dummy instead of the continuous variable the coefficient on earnings quality becomes negative
and significant, consistent with earnings quality decreasing forecast errors. The use of this alternative
earnings quality proxy does not affect the results of the other variables of interest. That is, the coefficient on
segment disclosure quality remains negative and significant at conventional levels.
C
2014 John Wiley & Sons Ltd
+
DevForecast
−0.0687
(0.000)
0.7549
(0.000)
0.0368
−0.0687
(0.000)
0.7492
(0.000)
0.0368
0.0607
(0.000)
−0.0018
(0.285)
−0.0197
(0.011)
Coef.
(p-value)
−0.0692
(0.000)
0.7549
(0.000)
0.0386
0.0608
(0.000)
−0.0196
(0.011)
0.0286
(0.696)
−0.0017
(0.292)
Coef.
(p-value)
0.5556
(0.000)
−0.0731
(0.000)
0.7947
(0.000)
0.0379
−0.0021
(0.256)
−0.0072
(0.015)
Coef.
(p-value)
0.5564
(0.000)
−0.0739
(0.000)
0.8026
(0.000)
0.0395
−0.0072
(0.015)
0.0389
(0.641)
0.0389
(0.266)
Coef.
(p-value)
The sample consists of 10,002 firm-year observations for the period 2001–2006. Forecast error = analysts’ forecast errors. It is calculated as the absolute value of forecast
in t−1 of eps for year t – actual eps of year t, scaled by actual eps in year t. Qlt Seg = the decile-ranked residuals from a regression of the firm’s year t. Qtt Seg on
determinants of segment disclosure and controls (Business Diversification, Geographic Diversification, Information Asymmetries, Size, Growth, Leverage, Earnings Qlt, Audit
Firm, Listing Status, Proprietary Costs, New Financing, Profitability, Age). Earnings Qlt = the absolute value, multiplied by −1, of discretionary accruals calculated as the
residual of the modified version of the Jones (1991) accruals model (Dechow et al., 1995), as applied to total accruals. Number of analysts = number of eps forecasts
of the firm in year t. DevForecast = deviation of analysts’ forecasts. It is calculated as the standard deviation of analysts’ forecasts. Ortho(DevForecast) is the residual of a
regression of DevForecast on Qlt Seg. Size = the logarithm of the firm’s market value of equity measured at the beginning of fiscal year 2001–2006.
R2
Cons
Sizet
DevForecastt-1
−0.0692
(0.000)
0.7605
(0.000)
0.0386
–
Number of analystst
–
–
Earnings Qltt
Ortho(DevForecast)t
−0.0073
(0.000)
0.0286
(0.696)
−0.0017
(0.292)
0.5974
(0.000)
−0.0074
(0.006)
–
Qlt Segt
−0.0017
(0.285)
0.5968
(0.000)
Coef.
(p-value)
Coef.
(p-value)
Expected
sign
Table 4
Industry Fixed Effect Regressions of Analysts’ Forecast Errors on Segment Disclosure Quality (Qlt Seg) and Controls
22
BLANCO, GARCIA LARA AND TRIBO
C
2014 John Wiley & Sons Ltd
23
SEGMENT DISCLOSURE AND COST OF CAPITAL
Table 5
Industry Fixed Effect Regressions of Firm’s Returns Covariance with the Returns
of the Rest of Firms in the Same Industry on Segment Disclosure Quality (Qlt Seg)
and Controls
Expected sign
Coef. (p-value)
Coef. (p-value)
Coef. (p-value)
Qlt Seg
−
−0.0010
(0.000)
Earnings Qlt
−
−0.0010
(0.000)
Size
−
B/M
+
Proprietary Costs
−
Leverage
+
Business Diversification
−
Geographic Diversification
−
Listing Status
−
Profitability
+
Age
−
0.0573
(0.000)
0.0072
−0.0032
(0.000)
0.0140
(0.000)
−0.0825
(0.091)
0.0001
(0.048)
−0.0016
(0.160)
−0.0002
(0.474)
−0.0152
(0.027)
0.0001
(0.005)
−0.0002
(0.067)
0.1133
(0.000)
0.0272
−0.0011
(0.006)
−0.0242
(0.038)
−0.0028
(0.001)
0.0146
(0.000)
−0.0825
(0.091)
0.0001
(0.027)
−0.0016
(0.161)
−0.0001
(0.502)
−0.0110
(0.122)
0.0001
(0.003)
−0.0002
(0.093)
0.1046
(0.000)
0.0278
Cons
R2
The sample consists of 10,002 firm-year observations for the period 2001–2006. Cov (ri , rsector ) = mean
annual covariance of the monthly return of a firm with the monthly return of the firms in the industry
sector in which the firm operates. Qlt Seg = the decile-ranked residuals from a regression of the firm’s year t.
Qtt Seg on determinants of segment disclosure and controls (Business Diversification, Geographic Diversification,
Information Asymmetries, Size, Growth, Leverage, Earnings Qlt, Audit Firm, Listing Status, Proprietary Costs, New
Financing, Profitability, Age). Earnings Qlt = the absolute value, multiplied by −1, of discretionary accruals
calculated as the residual of the modified version of the Jones (1991) accruals model (Dechow et al.,
1995), as applied to total accruals. Size = the logarithm of the firm’s market value of equity measured at
the beginning of fiscal year 2001–2006. B/M = the logarithm of the firm’s book-to-market ratio measured
at the beginning of fiscal year 2001–2006. Proprietary Costs = Herfindahl index in percentage, calculated
2
N
S
as Herf j = i=1
( Sijj ) , where Sij = business i’s sales in industry j, as defined by two-digit SIC code; Sj
= the sum of sales for all firms in industry j. Leverage = debt-to-total assets ratio in percentage. Business
Diversification = number of different sectors in which the firm operates. Geographic Diversification = number
of different countries where the firm operates. Listing Status = 1 if firm is listed on the NYSE or NASDAQ
and 0 otherwise. Profitability = percentile rank of return on assets. Age = the difference between the first
year when the firm appears in CRSP and the current year.
(ii) Firm’s Returns Covariance with other Firms’ Returns and Segment Disclosure
In Table 5 we show the results of regressing the covariance between the firm’s
returns and returns of the firms operating in the same industry sector on the
quality of segment information and controls. Once we focus on the last column of
the table, where we use the full model, including all controls, the coefficient on
segment information quality is negative and significant (coeff. = −0.0011, p = 0.006).
C
2014 John Wiley & Sons Ltd
24
BLANCO, GARCIA LARA AND TRIBO
Thus, results are consistent with better segment information reducing the covariance
between the firm’s returns and returns of the rest of the firms in the same industry.
In terms of economic significance, if the firm moves to the next decile of better
segment reporting quality, the covariance between the firm returns with the returns
of all other firms in the same industry decreases by 0.11 percentage points (see the full
model in the last column). Bearing in mind that the average covariance for all firms
in the sample is 1.61 percent (see Table 1), the reduction introduced by improving
segment reporting is substantial. In the Lambert et al. (2007) setting, these results
are consistent with firms providing high quality segment disclosure enjoying a lower
cost of capital. Results are qualitatively similar if we do not control by earnings quality
(second column), or if we estimate the model without controls (first column).
(iii) Ex-ante Cost of Equity Capital Estimates and Segment Disclosure
In Table 6 we show the pairwise correlations between the main variables of interest for
the tests where we study the relation between segment disclosure and ex-ante estimates
of firms’ cost of equity capital. We find that quality of segment information (Qlt Seg)
is negatively and significantly (p < 0.05) associated with rEX-ANTE (−12.81 percent). This
suggests that firms with better quality segment disclosure, which reduces estimation
risk, enjoy a lower cost of equity capital, as we predict in H1. Earnings quality is also
negatively correlated with rEX-ANTE (−8.23 percent).
In Table 7, we show results of regressions of implied cost of equity capital on
segment disclosure scores, earnings quality and controls – Equation (4). In the first
column of this table we validate our proxy for the ex-ante cost of equity capital (rEX-ANTE ).
We find that the cost of equity capital decreases with firm size and diversification, and
increases with the book-to-market ratio, beta and leverage. These results, consistent
with the evidence in Botosan and Plumlee (2005), suggest that rEX-ANTE is a valid
proxy for cost of equity capital given that the expected relations with all of the
aforementioned proxies for risk hold.
In the second column we show the results of the regression of ex-ante cost of
equity capital on quality of segment information, without controls. The coefficient on
segment disclosure quality is negative and significant (−0.0023, p = 0.000), consistent
with an association between better segment information and lower cost of equity
capital. In the third column we include the common set of controls, and results for
segment disclosure quality do not change (−0.0024, p = 0.000). Finally, in the last
column of the table, which shows the complete model, the coefficient on Qlt Seg is
still negative, −0.0025, and significant at conventional levels. We also find a negative
coefficient on earnings quality (−0.0321, p = 0.000). Hence, the inclusion of earnings
quality in the model does not eliminate the negative association between segment
disclosure quality and cost of equity capital.
In terms of economic significance, firms with the best segment reporting quality
(in the top decile of segment reporting quality) enjoy a 225 basis point23 lower cost
of equity capital than firms with the worst segment reporting quality (in the bottom
decile of segment reporting quality). To put this result in perspective, the average cost
of equity capital for all sample firms is 12.53 percent. This economic effect is in line
with the results in Botosan (1997) and Francis et al. (2008). In both papers they find
23 0.0025 times 9 decile differences and times 100.
C
2014 John Wiley & Sons Ltd
C
2014 John Wiley & Sons Ltd
1
−0.0000
−0.0032
0.0047
0.0004
−0.0071
−0.0094
0.0002
1
−0.1281
−0.0823
−0.1863
0.1808
0.0982
0.1287
−0.0821
−0.0503
1
0.2932
0.0703
0.0080
0.1494
0.1299
0.0488
EarQlt
1
−0.2792
−0.0413
0.1125
0.2234
0.0269
Size
1
0.0320
0.0082
0.0625
0.0147
B/M
1
0.0149
−0.0137
0.0342
Beta
1
0.1368
0.0419
Leverage
1
0.0536
BusDiversif
1
GeoDiversif
The sample consists of 10,002 firm-year observations for the period 2001–2006. Bold numbers are significant at p < 0.05. rEX-ANTE is the mean of the four estimates for
the implied cost of equity capital, as described in Section 3.ii.c. Qlt Seg = the decile-ranked residuals from a regression of the firm’s year t. Qtt Seg on determinants of
segment disclosure and controls (Business Diversification, Geographic Diversification, Information Asymmetries, Size, Growth, Leverage, Earnings Qlt, Audit Firm, Listing Status,
Proprietary Costs, New Financing, Profitability, Age). Earnings Qlt = the absolute value, multiplied by −1, of discretionary accruals calculated as the residual of the modified
version of the Jones (1991) accruals model (Dechow et al., 1995), as applied to total accruals. Size = the logarithm of the firm’s market value of equity measured at
the beginning of fiscal year 2001–2006. B/M = the logarithm of the firm’s book-to-market ratio measured at the beginning of fiscal year 2001–2006. Beta = coefficient
from firm-specific CAPM regression using the 60 months preceding fiscal year 2001–2006. Leverage = debt-to-total assets ratio in percentage. Business Diversification =
number of different sectors in which the firm operates. Geographic Diversification = number of different countries where the firm operates.
rEX-ANTE
Qlt Seg
Earnings Qlt
Size
B/M
Beta
Leverage
Business Diversification
Geographic Diversification
Qlt Seg
rEX-ANTE
Table 6
Pairwise Correlations between rPEG , Segment Disclosure Quality (Qlt Seg), Earnings Quality and Control Variables
SEGMENT DISCLOSURE AND COST OF CAPITAL
25
26
BLANCO, GARCIA LARA AND TRIBO
Table 7
Industry Fixed Effect Regressions of Implied Cost of Equity Capital (rEX-ANTE ) on
Segment Disclosure Quality (Qlt Seg) and Controls
Variable
Expected
sign
Qlt Seg
−
Earnings Qlt
−
Size
−
B/M
+
Beta
+/−
Leverage
+
Business Diversification
−
Geographic Diversification
−
Cons
R2
Coef.
(p-value)
−0.0035
(0.000)
0.0080
(0.000)
0.0052
(0.000)
0.0001
(0.000)
−0.0015
(0.000)
−0.0007
(0.000)
0.1699
(0.000)
0.0716
Coef.
(p-value)
Coef.
(p-value)
Coef.
(p-value)
−0.0023
(0.000)
−0.0024
(0.000)
0.1588
(0.000)
0.0513
−0.0036
(0.000)
0.0080
(0.000)
0.0052
(0.000)
0.0001
(0.000)
−0.0015
(0.003)
−0.0007
(0.000)
0.1829
(0.000)
0.0896
−0.0025
(0.000)
−0.0321
(0.000)
−0.0034
(0.000)
0.0080
(0.000)
0.0051
(0.000)
0.0001
(0.000)
−0.0011
(0.002)
−0.0007
(0.000)
0.1930
(0.000)
0.1061
The sample consists of 10,002 firm-year observations for the period 2001–2006. rEX-ANTE is the mean of the
four estimates for the implied cost of equity capital, as described in Section 3.ii.c. Qlt Seg = the decile-ranked
residuals from a regression of the firm’s year t. Qtt Seg on determinants of segment disclosure and controls
(Business Diversification, Geographic Diversification, Information Asymmetries, Size, Growth, Leverage, Earnings Qlt,
Audit Firm, Listing Status, Proprietary Costs, New Financing, Profitability, Age). Earnings Qlt = the absolute value,
multiplied by −1, of discretionary accruals calculated as the residual of the modified version of the Jones
(1991) accruals model (Dechow et al., 1995), as applied to total accruals. Size = the logarithm of the firm’s
market value of equity measured at the beginning of fiscal year 2001–2006. B/M = the logarithm of the
firm’s book-to-market ratio measured at the beginning of fiscal year 2001–2006. Beta = coefficient from
firm-specific CAPM regression using the 60 months preceding fiscal year 2001–2006. Leverage = debt-to-total
assets ratio in percentage. Business Diversification = number of different sectors in which the firm operates.
Geographic Diversification = number of different countries where the firm operates.
that moving from the best to the worst decile of voluntary disclosure increases cost of
capital by around 2 percent. Overall, our results suggest that firms providing better
quality segment information enjoy a lower cost of equity capital.
Regarding the moderating effect of competition, in Table 8 we show results of regressions of implied cost of equity capital on segment disclosure quality, competition,
earnings quality and controls. In Table 8, Panel A, we report the results using industrybased measures of competition, and in Panel B, we show the results using the measures
of competition at the firm level. Remarkably, regardless of the measure of competition
that we use, the effect of segment disclosure on cost of equity capital is negative. Also,
consistent with our expectations, this effect is less negative in industries and for firms
that operate in a more competitive environment (positive coefficient of Qlt Seg ×
Competition in all specifications of Panels A and B). Hence, the cost of equity capital
effect of segment disclosure is smaller for firms subject to competitive pressures.24
24 When we use Herf and Exist-Comp to measure competition (columns 5 and 6 in Table 8, Panel A), we
exclude the Herfindahl index from the estimation of Equation (1), where we determine Qlt Seg. We do so
C
2014 John Wiley & Sons Ltd
C
2014 John Wiley & Sons Ltd
R2
p-value (β 1 +β 3 )
Cons
−
Geographic Diversification
+/−
Beta
−
+
B/M
Business Diversification
−
Size
+
−
Earnings Qlt
Leverage
+
+/−
−
−0.0024
(0.000)
−0.0016
(0.055)
0.0005
(0.158)
−0.0267
(0.000)
−0.0032
(0.000)
0.0028
(0.000)
0.0050
(0.000)
0.0002
(0.000)
−0.0017
(0.000)
−0.0003
(0.000)
0.1658
(0.000)
0.1065
0.000
−0.0029
(0.000)
−0.0082
(0.035)
0.0008
(0.043)
−0.0306
(0.000)
−0.0032
(0.000)
0.0079
(0.000)
0.0052
(0.000)
0.0001
(0.002)
−0.0016
(0.000)
−0.0006
(0.000)
0.1731
(0.000)
0.1083
0.000
(0.000)
−0.0004
(0.001)
0.1701
(0.000)
0.1081
0.000
0.0002
(0.000)
−0.0016
0.0006
(0.086)
−0.0286
(0.000)
−0.0031
(0.000)
0.0056
(0.000)
0.0050
(0.000)
−0.0028
(0.000)
0.0015
(0.198)
−0.0004
(0.056)
0.1670
(0.000)
0.1096
0.000
0.0001
(0.000)
−0.0024
(0.000)
0.0013
(0.000)
−0.0306
(0.000)
−0.0035
(0.000)
0.0032
(0.000)
0.0049
(0.000)
−0.0015
(0.000)
0.0098
(0.000)
−0.0001
(0.034)
0.1743
(0.000)
0.1089
0.000
0.0002
(0.000)
−0.0022
(0.000)
0.0008
(0.048)
−0.0294
(0.000)
−0.0032
(0.000)
0.0042
(0.000)
0.0051
(0.000)
−0.0025
(0.000)
0.0103
(0.000)
−0.0004
(0.049)
0.1758
(0.000)
0.1092
0.000
0.0002
(0.000)
−0.0014
(0.001)
0.0009
(0.057)
−0.0308
(0.000)
−0.0033
(0.000)
0.0080
(0.000)
0.0051
(0.000)
−0.0032
(0.000)
0.0019
(0.121)
(Continued)
−0.0007
(0.000)
0.1838
(0.000)
0.1094
0.000
0.0001
(0.000)
−0.0011
(0.002)
0.0010
(0.023)
−0.0318
(0.000)
−0.0034
(0.000)
0.0079
(0.000)
0.0051
(0.000)
−0.0031
(0.000)
0.0016
(0.096)
HiTech
Industry Sales Growth Industry MarkUp NumFirms
Herf
Exist-Comp
SPA
Expected sign Coef. (p-value)
Coef. (p-value)
Coef. (p-value) Coef. (p-value) Coef. (p-value) Coef. (p-value) Coef. (p-value)
Qlt Seg × Competition
Competition
Qlt Seg
Variable
Panel A: Industry-level competition and controls
Table 8
Industry Fixed Effect Regressions of Implied Cost of Equity Capital (rEX-ANTE ) on Segment Disclosure Quality (Qlt Seg)
SEGMENT DISCLOSURE AND COST OF CAPITAL
27
C
R2
p-value (β 1 +β 3 )
Cons
−
Geographic Diversification
+/−
Beta
−
+
B/M
Business Diversification
−
Size
+
−
Earnings Qlt
Leverage
+
Qlt Seg × Competition
+/−
−
Qlt Seg
Competition
Expected sign
Variable
Panel B: Firm-level competition and controls
−0.0022
(0.000)
−0.0011
(0.093)
0.0004
(0.047)
−0.0277
(0.000)
−0.0032
(0.000)
0.0044
(0.000)
0.0050
(0.000)
0.0002
(0.000)
−0.0017
(0.000)
−0.0004
(0.000)
0.1634
(0.000)
0.1074
0.000
Firm Sales Growth
Coef. (p-value)
Table 8
Continued
−0.0028
(0.000)
0.0016
(0.135)
0.0006
(0.063)
−0.0275
(0.000)
−0.0031
(0.000)
0.0068
(0.000)
0.0051
(0.000)
0.0002
(0.000)
−0.0016
(0.000)
−0.0004
(0.001)
0.1663
(0.000)
0.1083
0.000
Firm MarkUp
Coef. (p-value)
(Continued)
−0.0028
(0.000)
−0.0059
(0.105)
0.0007
(0.089)
−0.0298
(0.000)
−0.0031
(0.000)
0.0080
(0.000)
0.0053
(0.000)
0.0001
(0.000)
−0.0016
(0.000)
−0.0005
(0.001)
0.1683
(0.000)
0.1083
0.000
HiR&D
Coef. (p-value)
28
BLANCO, GARCIA LARA AND TRIBO
2014 John Wiley & Sons Ltd
C
2014 John Wiley & Sons Ltd
The sample consists of 10,002 firm-year observations for the period 2001–2006. rEX-ANTE is the mean of the four estimates for the implied cost of equity capital:
PEG and MPEG ratios proposed by Easton (2004), Ohlson and Juettner-Nauroth’s (2005) abnormal earnings growth model, and the model proposed by Gode and
Mohanram (2003). Qlt Seg = the decile-ranked residuals from a regression of the firm’s year t. Qtt Seg on determinants of segment disclosure and controls (Business
Diversification, Geographic Diversification, Information Asymmetries, Size, Growth, Leverage, Earnings Qlt, Audit Firm, Listing Status, Proprietary Costs, New Financing, Profitability,
Age). HiTech = 1 for firms with two-digit SIC codes equal to 28, 35, 36, 73 or 87, and 0 otherwise. IndustrySalesGrowth = 1 for firms where (industry sales – lagged
industry sales)/lagged industry sales is above the median of the sample by year, and 0 otherwise. IndustryMarkUp = 1 for firms where (Sales + Inventories – Pension
& Retirement Expense – COGS)/(Sales + Inventories), is below the sample median by year, and 0 otherwise. NumFirms = 1 for firms where the total number of firms
in the industry is above the sample median by year, and 0 otherwise. Herf is dummy variable taking a value of 1 if the Herfindahl index is below the median of all firms
by year, and 0 otherwise. When using this measure as a proxy for competition, Proprietary Costs is excluded in the calculation of Qlt Seg – equation (1). Exist-Comp is a
dummy variable taking a value of 1 if the first principal component extracted from principal component analysis of four competition variables: IND-HHI, IND-CON4,
IND-NUM, and IND-MKTS, is above the sample median by year, and 0 otherwise. IND-HHI is the negative of the Herfindahl index; IND-CON4 is the negative of the
four-firm concentration ratio, measured as the sum of market shares of the four largest firms in an industry; IND-NUM is the total number of firms in the industry;
and IND-MKTS is the product market size, measured as the natural log of industry aggregate sales. When using this measure as a proxy for competition, Proprietary
Costs is excluded in the calculation of Qlt Seg. SPA is the Speed of Profit Adjustment model proposed by Stanford Harris (1998). It is 1 for industries where firms are
able to maintain persistent profits, and 0 otherwise. Firm Sales Growth is a dummy variable taking a value of 1 if (firm sales – lagged firm sales)/lagged firm sales is
above the industry-year median, and 0 otherwise. Firm Markup is a dummy variable taking a value of 1 if the markup is below the median of the industry-year, and 0
otherwise. HiR&D = 1 for firms with levels of R&D above the median of the industry-year, and 0 otherwise. Earnings Qlt = the absolute value, multiplied by −1, of
discretionary accruals calculated as the residual of the modified version of the Jones (1991) accruals model (Dechow et al., 1995), as applied to total accruals. Size =
the logarithm of the firm’s market value of equity measured at the beginning of fiscal year 2001–2006. B/M = the logarithm of the firm’s book-to-market ratio. Beta
= coefficient from firm-specific CAPM regression using the preceding 60 months. Leverage = debt-to-total assets ratio in percentage. Business Diversification = number
of different sectors in which the firm operates. Geographic Diversification = number of different countries where the firm has subsidiaries.
Table 8
Continued
SEGMENT DISCLOSURE AND COST OF CAPITAL
29
30
BLANCO, GARCIA LARA AND TRIBO
For example, when we use as a proxy for competitive pressure HiTech (Column 1,
Panel A), the reduction in the cost of equity capital as we move from the lowest decile
of quality of voluntary segment reporting to the highest decile is 72 basis points less
intense once we compare competitive industries (HiTech = 1) versus non-competitive
industries (HiTech = 0).25 Also, it is noteworthy that the overall effect is still negative
and statistically significant (p < 0.001 for all the competition measures). That is, once
we control for competition the effect of segment disclosure on cost of equity capital is
less pronounced, but it is still negative. This result conforms to H2.
Finally, given the measurement problems of implied estimates of cost of capital
identified by Easton and Monahan (2005), we use a different risk proxy as dependent
variable in Specification (4): returns volatility. We show the results in Table 9. In
the first column we just include the controls, and we can see that returns volatility
is associated with other risk proxies and that the signs of these associations are as
expected. In the second column we include Qlt Seg as the only explanatory variable,
and, as expected, it is negative and significant (−0.0001, p = 0.079). In the third
column we include all controls, except earnings quality, and the coefficient on
segment disclosure quality is still negative and significant at conventional levels. Finally,
in the last column we also control for earnings quality, which has a negative coefficient,
and the result for Qlt Seg does not change qualitatively.
(iv) Asset-Pricing-Based Tests
In Table 10, Panel A shows the descriptive statistics for the variables used in the asset
pricing tests. We can see that the excess market return is positive in our sample, and
that smaller and/or high book-to-market firms have higher returns than larger and/or
low book-to-market firms. Firms presenting better segment reporting quality, and also
firms with better earnings quality have lower returns (cost of equity capital). In Panel
B, we show the pairwise correlations between the factors of the model. The segment
reporting quality factor is positively correlated with the excess return on the market
portfolio, the HML, the momentum, the liquidity, and the Earnings Qlt factors, and
negatively correlated with the SMB factor.
In Table 11 we provide evidence as to whether segment information quality (Qlt Seg)
is a risk factor and, consequently, whether it is expected to impact cost of equity
capital. If segment disclosure quality is a risk factor, we should observe a negative
and statistically significant α of the hedge portfolio (long on firms providing good
voluntary segment reporting quality and short on those providing poor voluntary
segment reporting quality). The results (first column) show that α equals −0.1724
(p = 0.000). This implies that taking a long position in the firms in the top decile of
segment information quality and taking a short position in the firms in the bottom
decile would yield a monthly abnormal return of −0.1724 percent. In annual terms,
this corresponds to a negative abnormal return of 2.069 percent. This is consistent
with segment information quality being an omitted factor in the Fama–French four
to avoid any potential mechanical relations that might arise. However, results do not change qualitatively
regardless of whether we include the Herfindahl index or not in the estimation of Equation (1).
25 This is the result of multiplying the coefficient of Qlt Seg × Competition (β 3 ) times 9 deciles = 0.0008 ×
9 = 0.0072 = 72 basis points.
C
2014 John Wiley & Sons Ltd
C
2014 John Wiley & Sons Ltd
−
−
−
+
+
+
−
−
Qlt Seg
Earnings Qlt
Size
B/M
Beta
Leverage
Business Diversification
Geographic Diversification
−0.0039
(0.000)
0.0018
(0.000)
0.0003
(0.002)
0.0001
(0.000)
−0.0010
(0.000)
−0.0000
(0.424)
0.0534 (0.000)
0.4323
Coef. (p-value)
0.0230 (0.000)
0.2287
−0.0001
(0.079)
Coef. (p-value)
−0.0039
(0.000)
0.0018
(0.000)
0.0003
(0.010)
0.0001
(0.000)
−0.0011
(0.000)
−0.0000
(0.460)
0.0543 (0.000)
0.4549
−0.0002
(0.024)
Coef. (p-value)
−0.0002
(0.016)
−0.0130
(0.000)
−0.0037
(0.000)
0.0015
(0.000)
0.0003
(0.016)
0.0001
(0.000)
−0.0011
(0.000)
−0.0000
(0.668)
0.0517 (0.000)
0.4632
Coef. (p-value)
The sample consists of 10,002 firm-year observations for the period 2001–2006. Returns volatility is the standard deviation of daily returns over a 12 month window.
Qlt Seg = the decile-ranked residuals from a regression of the firm’s year t Qtt Seg on determinants of segment disclosure and controls (Business Diversification, Geographic
Diversification, Information Asymmetries, Size, Growth, Leverage, Earnings Qlt, Audit Firm, Listing Status, Proprietary Costs, New Financing, Profitability, Age). Earnings Qlt = the
absolute value, multiplied by −1, of discretionary accruals calculated as the residual of the modified version of the Jones (1991) accruals model (Dechow et al. 1995),
as applied to total accruals. Size = the logarithm of the firm’s market value of equity measured at the beginning of fiscal year 2001–2006. B/M = the logarithm of the
firm’s book-to-market ratio measured at the beginning of fiscal year 2001–2006. Beta = coefficient from firm-specific CAPM regression using the 60 months preceding
fiscal year 2001–2006. Leverage = debt-to-total assets ratio in percentage. Business Diversification = number of different sectors in which the firm operates. Geographic
Diversification = number of different countries where the firm operates.
Cons
R2
Expected sign
Variable
Table 9
Industry Fixed Effect Regressions of Returns Volatility on Segment Disclosure Quality (Qlt Seg) and Controls
SEGMENT DISCLOSURE AND COST OF CAPITAL
31
32
BLANCO, GARCIA LARA AND TRIBO
Table 10
Asset Pricing Based Tests: Relation between Realized Returns and Segment
Disclosure, Descriptive Statistics
Panel A: Summary statistics
Variable
RMRF
SMB
HML
UMD
LIQ
HILOQlt Seg
HILOEarnings Qlt
Mean
Std. Dev.
Annualized return
0.0032
0.0073
0.0085
0.0093
0.0063
−0.0058
−0.0031
0.0418
0.0296
0.0294
0.0431
0.0325
0.0263
0.0431
3.9083%
9.1204%
10.6906%
11.7489%
7.8275%
−7.1864%
−3.7841%
Panel B: Correlation Matrix
RMRF
SMB
HML
UMD
LIQ
HILOQlt Seg
HILOEarnings Qlt
RMRF
SMB
HML
UMD
LIQ
HILO
QltSeg
HILO
Earnings Qlt
1
0.3318
−0.4658
0.0128
−0.5893
0.0314
0.0621
1
−0.2417
0.1893
−0.1783
−0.2165
−0.2423
1
0.1732
0.4389
0.4023
0.4984
1
−0.1472
0.1726
0.1468
1
0.2372
0.2868
1
0.3943
1
The sample consists of 102,024 firm-month observations for the period 2001–2006. Bold numbers in Panel
B are significant at p < 0.05. RMRF is excess return on the value-weighted market portfolio. Excess returns
equal the value-weighted return on the portfolios less the risk free rate. SMB is the value-weighted sizemimicking portfolio return. HML is the value-weighted book-to-market-mimicking portfolio return. UMD is
the value-weighted momentum-mimicking portfolio return. LIQ is the value-weighted liquidity-mimicking
portfolio return. HILOQlt Seg is the value-weighted Qlt Seg-mimicking portfolio return. HILOEarnings Qlt is
the value-weighted Earnings Qlt-mimicking portfolio return.
(five, including liquidity) factor model, as firms providing poor segment information
quality have greater average excess returns unexplained.26
In the second column of Table 11, we include an additional factor (HERF) to
consider the level of competition. To create HERF we take a long position on the two
quintiles with the lowest Herf index and a short position on the two quintiles with the
highest Herf index. We find that HERF is not a risk factor (p = 0.318) and even when
we take into account the competition level, the effect of segment disclosure on cost of
equity capital is negative and significant (−0.1694 and p = 0.000).
In Table 12, Panel A shows the average factor loadings and average R2 of timeseries regressions of monthly contemporaneous portfolio excess stock returns (stock
return minus the risk-free rate) on the three Fama–French factors, the momentum and
liquidity factors, the Qlt Seg factor and the Earnings Qlt factor. We report the results
of averaging multivariate betas from 25 value-weighted portfolios sorted by Size and
B/M. We find that the sensitivity of portfolios’ returns to segment information quality
26 The positive coefficient of RMRF, which is connected to some extent to the CAPM beta, may be at odds
with the lower beta that we expect from a fixed portfolio in which there is an increase in voluntary segment
information (see Table 5). However, we should note that the voluntary segment information hedge portfolio
– long (short) – in high (low) voluntary segment information quality – is composed of different stocks that
are rebalanced each month. Hence the hedge portfolio is not fixed but changes every month.
C
2014 John Wiley & Sons Ltd
33
SEGMENT DISCLOSURE AND COST OF CAPITAL
Table 11
Time-Series Regressions of HEDGE Qlt Seg on the Fama–French, Momentum and
Liquidity Factors
RMRF
SMB
HML
UMD
LIQ
Coef. (p-value)
Coef. (p-value)
0.0219
(0.000)
−0.0103
(0.000)
0.0318
(0.000)
0.0593
(0.001)
0.0347
(0.083)
0.0252
(0.000)
−0.0114
(0.000)
0.0324
(0.000)
0.0602
(0.000)
0.0385
(0.062)
0.0029
(0.318)
−0.1724
(0.000)
0. 1529
−0.1694
(0.000)
0. 1531
HERF
Cons
R2
The sample consists of 102,024 firm-month observations for the period 2001–2006. RMRF is excess return
on the value-weighted market portfolio. Excess returns equal the value-weighted return on the portfolios
less the risk free rate. SMB is the value-weighted size-mimicking portfolio return. HML is the valueweighted book-to-market-mimicking portfolio return. UMD is the value-weighted momentum-mimicking
portfolio return. LIQ is the value-weighted liquidity-mimicking portfolio return. HERF is the value-weighted
Herf-mimicking portfolio return. HEDGE Qlt Seg is the hedge value-weighted segment information quality
portfolio, buying top decile segment quality information and selling bottom decile segment quality
information.
is negative (−0.0086 with p = 0.004 if we do not include the earnings quality factor,
and −0.0083 with p = 0.005 if we do). The earnings quality factor is also significantly
negative, but with a larger standard error (p = 0.096). Our results are robust to the
inclusion of competition (HERF) as an additional factor in the Fama–French model
(see last column of Panel A), and to the use of the CAPM (reported in Table 12, Panel
B) instead of the augmented Fama–French model.
As a final test to study whether segment reporting quality is a priced factor, in
Table 13 we present the coefficients from 72 monthly cross-sectional regressions
of excess value-weighted portfolio returns on portfolio factor loadings (the slope
coefficients from the regressions in Table 12). We find that the risk premium linked
to segment information quality is negative (−0.3943; p = 0.004 in column 3, Panel A),
that is, firms providing better quality segment information have lower excess realized
returns. However, earnings quality bears no significant risk premium (−0.0952; p =
0.119). This indicates that earnings quality, as we measure it, is weakly associated with
excess realized returns. Our results regarding earnings quality are in line with the
evidence in Core et al. (2008). Overall, our findings indicate that firms with better
quality segment information enjoy a lower cost of equity capital. As in the first stage
reported in Table 12, results are robust to the use of the CAPM (Panel B).
Finally, we include the HERF factor to consider the level of competition. As in
Table 12, we find that competition is not a risk factor (p = 0.324 in column 4, Panel
A) and the sign of the coefficients barely changes. Overall, even when we include the
C
2014 John Wiley & Sons Ltd
34
BLANCO, GARCIA LARA AND TRIBO
Table 12
Average Factor Loadings across Fama–French 25 Size-B/M Portfolios
Panel A: 5 Factor
Model Coef. (p-value)
SMB
HML
UMD
LIQ
0.4983
(0.001)
0.2934
(0.004)
0.0396
(0.006)
0.0289
(0.103)
0.4992
(0.001)
0.2875
(0.004)
0.0383
(0.004)
0.0283
(0.104)
0.5002
(0.001)
0.2865
(0.004)
0.0398
(0.005)
0.0346
(0.093)
1.0002
(0.000)
1.0004
(0.000)
−0.0086
(0.004)
0.0238
(0.000)
0.9014
2.38
<0.01
0.0302
(0.000)
0.9085
3.61
<0.01
1.0004
(0.000)
−0.0083
(0.005)
−0.0009
(0.096)
0.0295
(0.000)
0.9136
4.93
<0.01
HERF
RMRF
HILO Qlt Seg
HILO Earnings Qlt
Cons
R2
GRS-stat
GRS p-value
Panel B: CAPM
Coef. (p-value)
0.5001
(0.001)
0.2866
(0.005)
0.0399
(0.005)
0.0328
(0.095)
0.0243
(0.293)
1.0001
(0.000)
−0.0078
(0.006)
−0.0006
(0.106)
0.0294
(0.000)
0.9139
5.02
<0.01
1.1679
(0.000)
1.1686
(0.000)
−0.0106
(0.005)
0.0983
(0.000)
0.7832
4.89
<0.01
0.0981
(0.000)
0.7856
4.92
<0.01
1.1698
(0.000)
−0.0105
(0.004)
−0.0009
(0.085)
0.0978
(0.000)
0.7893
4.94
<0.01
This table presents average factor loadings and average R2 of time-series regressions of monthly contemporaneous portfolio excess stock returns (stock return minus the risk-free rate) on the four Fama–French
factors, the Liquidity factor, the Qlt Seg factor and the Earnings Qlt factor. The sample consists of 102,024
firm-month observations for the period 2001–2006. We form 25 portfolios sorting stocks into quintiles based
on Size-B/M. SMB is the value-weighted size-mimicking portfolio return. HML is the value-weighted bookto-market-mimicking portfolio return. UMD is the value-weighted momentum-mimicking portfolio return.
LIQ is the value-weighted liquidity-mimicking portfolio return. HERF is the value-weighted Herf-mimicking
portfolio return. RMRF is excess return on the value-weighted market portfolio. Excess returns equal the
value-weighted return on the portfolios less the risk free rate. HILOQlt Seg is the value-weighted Qlt Segmimicking portfolio return. HILOEarnings Qlt is the value-weighted Earnings Qlt-mimicking portfolio return.
The reported p-values are calculated from the portfolio-specific standard errors of the average parameters
(25 coefficients on each variable). The GRS statistic is the Gibbons et al. (2005) test on whether the
estimated intercepts are jointly zero.
competition level, results are consistent with segment reporting quality being a priced
factor and with investors rewarding firms that report better segment disclosure with a
lower cost of equity capital.27
As an additional battery of robustness tests, we also run asset pricing tests using as
test assets a set of 25 portfolios from the intersection of the quintiles of B/M with the
quintiles of Qlt Seg. We run also the test using as test assets a set of 25 portfolios from
the intersection of the quintiles of Size with the quintiles of Qlt Seg. Finally, we repeat
the tests using as a test asset a set of 100 portfolios of Qlt Seg. The results are robust
across these different test asset specifications. Overall, the findings of the asset pricing
tests show that better segment reporting quality is associated with a lower cost of equity
capital.
27 Results are similar when we use other competition proxies.
C
2014 John Wiley & Sons Ltd
35
SEGMENT DISCLOSURE AND COST OF CAPITAL
Table 13
Monthly Regressions of Returns of 25 Size-B/M Portfolios on Factor Loadings
Panel A: FF 5
Factor Model Coef. (p-value)
SMB
HML
UMD
LIQ
0.2532
(0.131)
0.3995
(0.000)
0.0321
(0.004)
0.0219
(0.113)
0.2532
(0.134)
0.4123
(0.000)
0.0392
(0.003)
0.0278
(0.109)
0.2624
(0.130)
0.4213
(0.000)
0.0392
(0.001)
0.0283
(0.106)
0.3123
(0.101)
0.3120
(0.104)
−0.3967
(0.004)
0.7934
(0.000)
0.5482
0.8025
(0.000)
0.5570
0.3190
(0.103)
−0.3943
(0.004)
−0.0952
(0.119)
0.8238
(0.000)
0.5608
HERF
RMRF
HILO Qlt Seg
HILO Earnings Qlt
Cons
R2
0.2619
(0.125)
0.4226
(0.000)
0.0393
(0.001)
0.0286
(0.106)
0.0980
(0.324)
0.3053
(0.101)
−0.3840
(0.003)
−0.0956
(0.121)
0.8268
(0.000)
0.5616
Panel B: CAPM
Coef. (p-value)
0.4328
(0.005)
0.4216
(0.008)
−0.4892
(0.001)
0.9318
(0.000)
0.4567
0.8993
(0.000)
0.4783
0.4209
(0.009)
−0.4896
(0.000)
−0.1025
(0.120)
0.8849
(0.000)
0.4786
This table presents average coefficients from 72 monthly cross-sectional regressions of excess value-weighted
portfolio returns on portfolio factor loadings (the slope coefficients from the regressions in Table 12, Panel
A). The sample consists of 102,024 firm-month observations for the period 2001–2006. SMB is the valueweighted size-mimicking portfolio return. HML is the value-weighted book-to-market-mimicking portfolio
return. UMD is the value-weighted momentum-mimicking portfolio return. LIQ is the value-weighted
liquidity-mimicking portfolio return. HERF is the value-weighted Herf-mimicking portfolio return. RMRF
is excess return on the value-weighted market portfolio. Excess returns equal the value-weighted return on
the portfolios less the risk free rate. HILOQlt Seg is the value-weighted Qlt Seg-mimicking portfolio return.
HILOEarnings Qlt is the value-weighted Earnings Qlt-mimicking portfolio return. The reported p-values are
calculated from the standard errors of the average monthly parameter estimates following the Fama and
MacBeth (1973) procedure.
(v) Separate Analysis of Business and Geographic Segment Information
We replicate the four sets of tests (forecast errors, covariance of returns, ex-ante cost of
capital tests and asset pricing tests), splitting our main proxy for segment disclosure
into two separate proxies for business segment disclosures and geographic segment
disclosures. In both cases, results are qualitatively similar to what we obtain when we
use the aggregate segment disclosure score (Qlt Seg). The results regarding geographic
segment disclosures are especially relevant given the debate on the reduction of
geographic segment information requirements introduced with the passage of SFAS
131.
(vi) Additional Robustness Tests (Alternative Measure of Disclosure Quality, and
Excluding Firms with Losses)
We replicate our main tests using a different measure of segment disclosure quality.
In particular, we create a measure that captures whether firms are disclosing segments
that differ substantially in growth. To do so, we calculate the growth in sales for each
C
2014 John Wiley & Sons Ltd
36
BLANCO, GARCIA LARA AND TRIBO
segment, and then, for each firm, we calculate the standard deviation of the growth
rate of the different reported segments. The idea behind this measure, which is in line
with the empirical work by Ettredge et al. (2006), is that segment disclosure is of high
quality whenever the standard deviation is high. A high standard deviation implies
that the firm is providing information about segments that are actually different,
and therefore that the information provided is really useful for investors. Using this
alternative measure of disclosure quality we obtain results qualitatively similar to those
reported in the paper. Finally, we also replicate the main tests in the paper excluding
firms with losses, and inferences do not change.
5. SUMMARY AND CONCLUSIONS
In this paper we analyze the relation between segment disclosure and cost of capital.
We find a negative relation between better segment reporting and (a) analysts’ forecast
errors, (b) the covariance between the firm’s returns and returns of the rest of firms
operating in the same industry, and (c) ex-ante estimates of cost of equity capital.
These results suggest that improved segment reporting reduces cost of capital. We
also run formal asset pricing tests and results show that segment reporting quality
is a priced risk factor in the CAPM and in the Fama–French three-factor model
(augmented with momentum and liquidity factors), and that it contributes to reducing
cost of equity capital beyond the reduction in cost of equity capital (if any) driven by
high quality earnings alone. Our results are robust to the use of aggregate segment
disclosure scores, and to scores based on business or geographic segment disclosures
only.
Our findings raise the question as to why, given that improved segment disclosure
leads to lower cost of capital, not all firms increase their segment disclosure levels.
As shown in prior research (Hayes and Lundholm, 1996), the firm will have to
reach equilibrium between enhancing segment disclosure to improve the information environment of the firm and keeping key information away from competitors
(Yosha, 1995). In fact, competition largely affects the disclosure of voluntary segment
information (Stanford Harris, 1998). Our results show that competition moderates
the relation between segment disclosure and cost of equity capital. We find that
the cost of equity capital benefits of improving segment disclosure quality are
reduced for firms subject to more competitive pressures. An additional explanation
why managers would not disclose even if this entails a larger cost of capital is
that they behave strategically, trying to maximize their own wealth (e.g., following
empire-building strategies) instead of firm value. The findings in Hope and Thomas
(2008) that segment disclosure constrains managerial empire-building behavior (in
particular, investments in foreign operations that do not maximize firm value) are
consistent with this explanation. Finally, another alternative explanation is that Chief
Financial Officers do not believe that improved segment disclosure leads to reduced
cost of capital. However, the results in Graham et al. (2005) do not support this
argument.
Lastly, we also contribute to the debate, started after the passage of SFAS 131, on
whether it is advisable to reduce the amount of geographic segment information that
firms are obliged to present. We find, consistent with previous evidence (Hope and
Thomas, 2008), that permitting firms to reduce the amount of geographic segment
information may have undesirable economic consequences.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
37
APPENDIX 1
Procedure for Elaborating the Segment Disclosure Score Qtt seg
To elaborate an index for the quantity of voluntary segment disclosure (Qtt Seg),
we do as follows. For every reported business/geographic segment in each firm, we
analyze whether the segment is reported on a compulsory or voluntary basis. For
the compulsory segments, we distinguish between the items reported compulsorily as
required by SFAS 131, and the items reported on a voluntary basis. Next, we create the
business/geographic segment score (Qtt Seg Bus)/(Qtt Seg Geo) by adding 1 point for
every voluntarily disclosed item in every mandatory segment, and 1 point for every item
in the voluntarily disclosed segments. Finally, we create the overall index of quantity of
voluntary segment disclosure (Qtt Seg) by adding the business and geographic segment
scores.
Distinguishing between Mandatory and Voluntary Segment Information
A.1. Identifying Business Segments Reported Mandatorily. To identify which business segments are reported mandatorily, we analyze whether the business segments reported
by the firm meet the quantitative thresholds, according to paragraph 18 of SFAS 131:
(a) “Its reported revenue, including both sales to external customers and intersegment sales or transfers, is 10 percent or more of the combined revenue, internal
and external, of all operating segments.
(b) The absolute amount of its reported profit or loss is 10 percent or more of
the greater, in absolute amount, of (1) the combined reported profit of all
operating segments that did not report a loss or (2) the combined reported
loss of all operating segments that did report a loss.
(c) Its assets are 10 percent or more of the combined assets of all operating
segments.”
If any given reported segment meets any of these thresholds, we consider the
segment as reported mandatorily.28 If the firm reports segments that do not meet
any of the thresholds in paragraph 18, to analyze whether these additional business
segments are compulsorily or voluntarily reported, we study the requirements of
paragraph 20 of SFAS 131: “If total of external revenue reported by operating segments
constitutes less than 75 percent of total consolidated revenue, additional operating
segments shall be identified as reportable segments (even if they do not meet the
criteria in paragraph 18) until at least 75 percent of total consolidated revenue is
included in reportable segments.” Given this, we sum the revenues of the segments
that meet any of the quantitative thresholds according to paragraph 18. If they
account for 75 percent of consolidated revenue, we consider all the other reported
segments as reported voluntarily. If they do not account for 75 percent of consolidated
28 Given the requirements in paragraph 18.b., it could be the case that we classify a segment as reported
on a voluntary basis when it is in fact reported on a mandatory basis. However, this would bias the results
against our hypotheses given that we predict that only voluntary segment disclosure would produce effects
on a firm’s cost of capital. Hence, an overestimation of voluntary disclosure would, if anything, reduce the
significance of our findings.
C
2014 John Wiley & Sons Ltd
38
BLANCO, GARCIA LARA AND TRIBO
revenue, we consider additional segments as compulsorily reported until all of the
segments considered as compulsorily reported account for 75 percent of consolidated
revenue. All other reported segments are considered as reported voluntarily. We also
consider segments reported mandatorily on a given year those segments considered
as reportable (according to paragraphs 18 and 20) in the previous year, and include
those in the calculation of the 75 percent.
Finally, we take into account that in paragraph 24, SFAS 131 recommends to include
only a maximum of 10 reportable segments, even if doing so breaches the limits
established in paragraphs 18 and 20. Consequently, we consider that only 10 segments
are reported mandatorily. Any segment in addition to these 10 will be considered as
voluntary. For firms reporting more than 10 segments, we classify as compulsory the
10 segments with the largest revenues.
A.2. Identifying Mandatory Items for Mandatory Business Segments. Compustat provides a
maximum of 29 items of information for each business segment. From these 29 items,
we analyze whether they are reported by the firm on a compulsory or voluntary
basis. If they are described as compulsory by paragraphs 26–31 of SFAS 131, then we
consider them as compulsory. We consider them voluntary otherwise. The items that
we consider mandatory out of the 29 total items are the following:
1. Business segment name (as the general information required in paragraph 26)
2. Identifiable assets per segment.
3. Depreciation, depletion and amortization per segment.
4. Equity in earnings per segment.
5. Operating profit per segment.
6. Sales to principal customer per segment.
7. Sales of principal product per segment.
8. Customer name per segment.
9. Investment at equity per segment.
The remaining 20 items available in Compustat, which we consider voluntarily
reported, are the following:
1. Business segment availability code.
2. Business segment ID.
3. Capital expenditure per business segment.
4. Capital expenditure note per business segment.
5. Employees per business segment.
6. Employees per business segment note.
7. Equity in earnings per segment note.
8. Foreign governments per segment.
9. Operating profit note per segment.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
39
10. Order backlog per segment.
11. Principal product name per segment.
12. Principle product SIC per segment.
13. R&D – company sponsored per segment.
14. R&D – company sponsored per segment note.
15. R&D – customer sponsored per segment.
16. Sales to domestic government per segment.
17. Sales to foreign government per segment.
18. Sales net per segment.
19. Sales net per segment note.
20. SIC codes per business segment (primary and secondary).
A.3. Voluntary Business Segment Information. We consider as voluntary business segment
information the other items for mandatorily reported segments, and all items for
segments reported voluntarily.
B. Mandatory vs Voluntary Geographic Segment Information
SFAS 131, paragraph 38, states “An enterprise shall report the following geographic
information unless it is impracticable to do so:
(a) Revenues from external customers (1) attributed to the enterprise’s country
of domicile and (2) attributed to all foreign countries in total from which the
enterprise derives revenues. If revenues from external customers attributed to
an individual foreign country are material, those revenues shall be disclosed
separately. An enterprise shall disclose the basis for attributing revenues from
external customers to individual countries.
(b) Long-lived assets other than financial instruments, long-term customer relationships of a financial institution, mortgage and other servicing rights,
deferred policy acquisition costs, and deferred tax assets (1) located in the
enterprise’s country of domicile and (2) located in all foreign countries in total
in which the enterprise holds assets. If assets in an individual foreign country
are material, those assets shall be disclosed separately.”
Given the above, we consider as reportable geographic segments all geographic segments available in Compustat (11 items of information for each geographic segment),
and consider the following as mandatory items for each reportable segment:
1. Net sales per segment (as required in paragraph 38-a).
2. Identifiable assets per segment (as required in paragraph 38-b).
C
2014 John Wiley & Sons Ltd
40
BLANCO, GARCIA LARA AND TRIBO
We consider as voluntary geographic segment information the other available items
for every reportable geographic segment.
APPENDIX 2
Description of the Estimation of the Implied Cost of Equity Capital Measures rPEG and rMPEG
To calculate both proxies, and following Easton (2004), we assume that the market
expects zero growth in abnormal earnings beyond the forecast horizon. Under this
assumption, the dividend discount model can be solved for expected returns directly.
The PEG ratio proposed by Easton (2004), as calculated in Botosan and Plumlee
(2005), is as follows:
rP EG =
ep s 5 − ep s 4
P0
(A1)
where epst is earnings per share in year t. We use five-year long-term growth rates from
I/B/E/S to calculate eps4 and eps5 . P0 is the market price of a firm’s stock. To calculate
rPEG , Botosan and Plumlee (2005) use earnings per share forecasts in years 4 and 5
because for rPEG to be meaningful it requires positive changes in forecasted earnings,
and changes between years 4 and 5 are more likely to be positive than changes in
near-term forecasts.
We calculate the MPEG ratio, also proposed by Easton (2004), as follows:
rMP EG
ep s 2 − ep s 1
= A + A2 +
;
P0
A=
dp s 1
2P0
(A2)
where dps is dividends per share. While rPEG assumes that the market expects no
dividends in year t+1, rMPEG relaxes this assumption.
rOJN and rGM
Both models are a special case of the abnormal earnings growth valuation model, in
which Ohlson and Juettner-Nauroth (2005) impose some assumptions in the market’s
expectations of near-term earnings, abnormal earnings, and the rates of short- and
long-term growth in abnormal earnings. Following Botosan et al. (2011), to calculate
short-term earnings growth rates we use analysts’ forecasts, and to calculate the
infinite growth in abnormal earnings we use rf less 3 percent (γ ). The only difference
between rOJN and rGM is that rOJN is estimated with short- and long-term growth in
earnings, while rGM is only estimated with short-term growth in earnings.
The proxy proposed by Ohlson and Juettner-Nauroth (2005) is as follows:
r OJ M = A +
A=
A2 +
ep s 1
×
P0
(ep s 3 − ep s 2 ) /ep s 2 + (ep s 5 − ep s 4 ) /ep s 4
− (γ − 1) ;
2
(γ − 1) + (dp s 1 /P0 )
2
(A3)
C
2014 John Wiley & Sons Ltd
41
SEGMENT DISCLOSURE AND COST OF CAPITAL
and the proxy proposed by Gode and Mohanram (2003) is the following:
r GM = A +
A2 +
ep s 1
×
P0
(ep s 2 − ep s 1 )
− (γ − 1) ;
ep s 1
A=
(γ − 1) + (dp s 1 /P0 )
.
2
(A4)
REFERENCES
Allayannis, G. and J. Ihrig (2001), ‘Exposure and Markups’, Review of Financial Studies, Vol. 14,
No. 3, pp. 805–35.
Armitage, S. and C. Marston (2008), ‘Corporate Disclosure, Cost of Capital and Reputation:
Evidence from Finance Directors’, The British Accounting Review, Vol. 40, No. 4, pp. 314–36.
Association for Investment, Management and Research (AIMR) (1993), ‘Financial Reporting in
the 1990s and Beyond’, Report prepared by Peter Knutson, Charlottesville, VA.
Balakrishnan, R., T. Stanford Harris and P. K. Sen (1990), ‘The Predictive Ability of Geographic
Segment Disclosures’, Journal of Accounting Research, Vol. 28, No. 2, pp. 305–25.
Baldwin, B. (1984), ‘Segment Earnings Disclosure and the Ability of Security Analysts to Forecast
Earnings Per Share’, The Accounting Review, Vol. 59, No. 3, pp. 376–89.
Bens, D. A. and S. J. Monahan (2004), ‘Disclosure Quality and the Excess Value of Diversification’, Journal of Accounting Research, Vol. 42, No. 4, pp. 691–730.
Berger, P. and R. Hann (2003), ‘The Impact of SFAS No. 131 on Information and Monitoring’,
Journal of Accounting Research, Vol. 41, No. 2, pp. 163–223.
——— and ——— (2007), ‘Segment Profitability and the Proprietary and Agency Costs of
Disclosure’, The Accounting Review, Vol. 82, No. 4, pp. 869–906.
Black, F., M. C. Jensen and M. Scholes (1972), ‘The Capital Asset Pricing Model: Some Empirical
Findings’, in M. C. Jensen (ed.), Studies in the Theory of Capital Markets. (Westport, CT:
Praeger).
Blanco, B., J. M. Garcia Lara and J. A. Tribo (2014), ‘Geographic segment disclosure and firm
investment efficiency’, Working paper (Melbourne: University of Melbourne).
Board, O. (2009), ‘Competition and Disclosure’, The Journal of Industrial Economics, Vol. 57, No.
1, pp. 197–213.
Botosan, C. A. (1997), ‘Disclosure Level and the Cost of Equity Capital’, The Accounting Review,
Vol. 72, pp. 323–49.
——— and M. A. Plumlee (2002), ‘A Re-Examination of Disclosure Level and the Expected Cost
of Equity Capital’, Journal of Accounting Research, Vol. 40, pp. 21–40.
——— and ——— (2005), ‘Assessing Alternative Proxies for the Expected Risk Premium’, The
Accounting Review, Vol. 80, No. 1, pp. 21–53.
———, ——— and H. Wen (2011), ‘The Relation between Expected Returns, Realized Returns,
and Firm Risk Characteristics’, Contemporary Accounting Research, Vol. 28, No. 4, pp. 1085–
122.
——— and M. Stanford (2005), ‘Managers’ Motives to Withhold Segment Disclosures and the
Effect of SFAS No. 131 on Analysts’ Information Environment’, The Accounting Review, Vol.
80, No. 3, pp. 751–71.
Bresnahan, T. F. and P. C. Reiss (1991), ‘Entry and Competition in Concentrated Markets’,
Journal of Political Economy, Vol. 99, No. 5, pp. 977–1009.
Bugeja, M., R. Czernkowski and D. Moran (2014), ‘The Impact of the Management Approach
on Segment Reporting’, Paper presented at the Journal of Business Finance & Accounting
Capital Markets Conference 2014.
Burks, J. J., C. Cuny, J. Gerakos and J. Granja (2013), ‘Competition and Voluntary Disclosure:
Evidence from Deregulation in the Banking Industry’, Chicago Booth Research Paper No.
12–29 (Chicago, IL: University of Chicago Booth School of Business).
Buzby, S. L. (1975), ‘Company Size, Listed Versus Unlisted Stocks, and the Extent of Financial
Disclosure’, Journal of Accounting Research, Vol. 13, No. 1, pp. 16–37.
C
2014 John Wiley & Sons Ltd
42
BLANCO, GARCIA LARA AND TRIBO
Carhart, M. (1997), ‘On the Persistence of Mutual Fund Performance’, The Journal of Finance,
Vol. 52, No. 1, pp. 57–82.
Caves, R. E. (1982), Multinational Enterprises and Economic Analysis (Cambridge: Cambridge
University Press).
Chen, P. and G. Zhang (2003), ‘Heterogeneous Investment Opportunities in Multiple-Segment
Firms and the Incremental Value Relevance of Segment Accounting Data’, The Accounting
Review, Vol. 78, No. 2, pp. 397–428.
Chen, S., DeFond, M. L. and Park, C. W. (2002), ‘Voluntary Disclosure of Balance Sheet
Information in Quarterly Earnings Announcements’, Journal of Accounting and Economics,
Vol. 33, pp. 229–51
Cheynel, E. (2013), ‘A Theory of Voluntary Disclosure and Cost of Capital’, Review of Accounting
Studies, Vol. 18, No. 4, pp. 987–1020.
Claus, J. and J. Thomas (2001), ‘Equity Premia as Low as Three Percent? Evidence from Analysts’
Earnings Forecasts for Domestic and International Stock Markets’, The Journal of Finance,
Vol. 56, No. 5, pp. 1629–66.
Clinch, G. and R. Verrecchia (2013), ‘Voluntary Disclosure and the Cost of Capital’, Working
Paper.
Collins, D. (1975), ‘SEC Product Line Reporting and Market Efficiency’, Journal of Financial
Economics, Vol. 2, No. 2, pp. 125–64.
——— (1976), ‘Predicting Earnings with Subentity Data: Some Further Evidence’, Journal of
Accounting Research, Vol. 14, No. 1, pp. 163–77.
——— and S. Henning (2004), ‘Write-Down Timeliness, Line-of-Business Disclosures and
Investors’ Interpretations of Segment Divestiture Announcements’, Journal of Business
Finance & Accounting, Vol. 31, Nos. 9&10, pp. 1261–99.
Core, J. E., W. R. Guay and R. Verdi (2008), ‘Is Accruals Quality a Priced Risk Factor?’, Journal of
Accounting and Economics, Vol. 46, No. 1, pp. 2–22.
De Andres, P., G. de la Fuente and P. Velasco (2014), ‘Growth Opportunities and the Effect of
Corporate Diversification on Value’, The Spanish Review of Financial Economics, Vol. 12, No.
2, pp. 72–81.
Dechow, P. and I. Dichev (2002), ‘The Quality of Accruals and Earnings: The Role of Accrual
Estimation Errors’, The Accounting Review, Vol. 77, Supplement, pp. 35–59.
———, W. Ge and C. Schrand (2010), ‘Understanding Earnings Quality: A Review of the
Proxies, their Determinants and their Consequences’, Journal of Accounting and Economics,
Vol. 50, Nos. 2&3, pp. 344–401.
———, R. Sloan and A. Sweeney (1995), ‘Detecting Earnings Management’, The Accounting
Review, Vol. 70, No. 2, pp. 193–225.
Dhaliwal, D. S., S. X. Huang, I. K. Khurana and R. Pereira (2014) ‘Product Market Competition
and Conditional Conservatism’, Review of Accounting Studies, Vol. 19, No. 4, pp. 1309–45.
———, O. Z. Li, A. Tsang and Y. G. Yang (2011), ‘Voluntary Non-Financial Disclosure and the
Cost of Equity Capital: The Initiation of Corporate Social Responsibility Reporting’, The
Accounting Review, Vol. 86, No. 1, pp. 59–100.
———, B. H. Spicer and D. Vickrey (1979), ‘The Quality of Disclosure and the Cost of Capital’,
Journal of Business Finance & Accounting, Vol. 6, No. 2, pp. 245–66.
Diamond, D. and R. Verrecchia (1991), ‘Disclosure, Liquidity, and the Cost of Capital’, The
Journal of Finance, Vol. 46, No. 4, pp. 1325–59.
Easley, D. and M. O’Hara (2004), ‘Information and the Cost of Capital’, The Journal of Finance,
Vol. 59, No. 4, pp. 1553–83.
Easton, P. D. (2004), ‘PE ratios, PEG Ratios, and Estimating the Implied Expected Rate of Return
on Equity Capital’, The Accounting Review, Vol. 79, No. 1, pp. 73–95.
——— and S. Monahan (2005), ‘An Evaluation of Accounting-Based Measures of Expected
Returns’, The Accounting Review, Vol. 80, No. 2, pp. 501–38.
Emmanuel, C. and R. H. Pick (1980), ‘The Predictive Ability of UK Segment Reports’, Journal of
Business Finance & Accounting, Vol. 7, No. 2, pp. 201–18.
Ernst and Young (2004), ‘International GAAP 2005. Generally Accepted Accounting Practice
under International Financial Reporting Standards’, The financial reporting group of Ernst
and Young, Global edition 1, LexisNexis.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
43
Escribano, A., A. Fosfuri and J. A. Tribo (2009), ‘Managing External Knowledge Flows: The
Moderating Role of Absorptive Capacity’, Research Policy, Vol. 38, No. 1, pp. 96–105.
Ettredge, M. L., S. Y. Kwon, D. B. Smith and M. S. Stone (2006), ‘The Effect of SFAS No. 131
on the Cross-Segment Variability of Profits Reported by Multiple Segment Firms’, Review of
Accounting Studies, Vol. 11, No. 1, pp. 91–117.
———, ———, ——— and P. Zarowin (2005), ‘The Impact of SFAS no. 131 Business Segment
Data on the Market’s Ability to Anticipate Future Earnings’, The Accounting Review, Vol. 80,
No. 3, pp. 773–804.
Fama, E. and K. French (1992), ‘The Cross-Section of Expected Stock Returns’, Journal of
Finance, Vol. 47, No. 2, pp. 427–65.
——— and ——— (1993), ‘Common Risk Factors in the Returns on Bonds and Stocks’, Journal
of Financial Economics, Vol. 33, No. 1, pp. 3–56.
——— and J. MacBeth (1973), ‘Risk, Return, and Equilibrium: Empirical Tests’, Journal of
Political Economy, Vol. 81, No. 3, pp. 607–36.
Foster, G. (1975), ‘Security Price Revaluation Implications of Sub-Earnings Disclosure’, Journal
of Accounting Research, Vol. 13, No. 2, pp. 283–92.
Francis, J., R. LaFond, P. Olsson and K. Schipper (2004), ‘Costs of Equity and Earnings
Attributes’, The Accounting Review, Vol. 79, No. 4, pp. 967–1010.
———, D. J. Nanda and P. Olsson (2008), ‘Voluntary Disclosure, Earnings Quality, and Cost of
Capital’, Journal of Accounting Research, Vol. 46, No. 1, pp. 53–99.
Franko, L. G. (1989), ‘Global Corporate Competition: Who’s Winning, Who’s Losing, and the
R&D Factor as one Reason Why’, Strategic Management Journal, Vol. 10, No. 5, pp. 449–74.
Garcia Lara, J. M., B. Garcia Osma and F. Penalva (2011), ‘Conditional Conservatism and Cost
of Capital’, Review of Accounting Studies, Vol. 16, No. 2, pp. 247–71.
Gaspar, J. M. and M. Massa (2006), ‘Idiosyncratic Volatility and Product Market Competition’,
The Journal of Business, Vol. 79, No. 6, pp. 3125–52.
Gebhardt, W. R., C. M. C. Lee and B. Swaminathan (2001), ‘Toward an Implied Cost of Capital’,
Journal of Accounting Research, Vol. 39, No. 1, pp. 135–76.
Gibbons, M. R., S. A. Ross and J. Shanken (1989), ‘A Test of the Efficiency of a Given Portfolio’,
Econometrica, Vol. 57, No. 5, pp. 1121–52.
Gietzmann, M. and J. Ireland (2005), ‘Cost of Capital, Strategic Disclosures and Accounting
Choice’, Journal of Business Finance & Accounting, Vol. 32, Nos. 3&4, pp. 599–634.
——— and A. Ostaszewski (2014), ‘Why Managers with Low Forecast Precision Select High
Disclosure Intensity: an Equilibrium Analysis’, Review of Quantitative Finance and Accounting,
Vol. 43, No. 1, pp. 121–53.
Gode, D. and P. Mohanram (2003), ‘Inferring the Cost of Capital using the Ohlson-Juettner
Model’, Review of Accounting Studies, Vol. 8, No. 3, pp. 399–431.
Gong, G., H. Louis and A. X. Sun (2008), ‘Earnings Management, Lawsuits, and Stock-for-Stock
Acquirers’ Market Performance’, Journal of Accounting and Economics, Vol. 46, No. 1, pp.
62–77.
Graham, J. R., C. R. Harvey and S. Rajgopal (2005), ‘The Economic Implications of Corporate
Financial Reporting’, Journal of Accounting and Economics, Vol. 40, Nos. 1–3, pp. 3–73.
Greenstein, M. M. and H. Sami (1994), ‘The Impact of the SEC’s Segment Disclosure
Requirement on the Bid-Ask Spread’, The Accounting Review, Vol. 69, No. 1, pp. 179–99.
Hail, L. and C. Leuz (2006), ‘International Differences in the Cost of Equity Capital: Do Legal
Institutions and Securities Regulation Matter?’, Journal of Accounting Research, Vol. 44, No. 3,
pp. 485–531.
Hann, R., M. Ogneva and O. Ozbas (2013), ‘Corporate Diversification and the Cost of Capital’,
The Journal of Finance, Vol. 68, No. 5, pp. 1961–99.
Harvey, C. R. (1991), ‘The World Price of Covariance Risk’, Journal of Finance, Vol. 46, No. 1, pp.
111–57.
Hayes, R. and R. Lundholm (1996), ‘Segment Reporting to the Capital Market in the Presence
of a Competitor’, Journal of Accounting Research, Vol. 34, No. 2, pp. 261–79.
Hope, O. K. (2003), ‘Disclosure Practices, Enforcement of Accounting Standards, and Analysts’
Forecast Accuracy: An International Study’, Journal of Accounting Research, Vol. 41, No. 2, pp.
235–72.
C
2014 John Wiley & Sons Ltd
44
BLANCO, GARCIA LARA AND TRIBO
———, T. Kang, W. B. Thomas and F. Vasvari (2008), ‘Pricing and Mispricing Effects of SFAS
131’. Journal of Business Finance & Accounting, Vol. 35, Nos. 3&4, pp. 281–306.
——— and W. B. Thomas (2008), ‘Managerial Empire Building and Firm Disclosure’, Journal of
Accounting Research, Vol. 46, No. 3, pp. 591–626.
Hughes, J.S., J. Liu and J. Liu (2007), ‘Information Asymmetry, Diversification, and Cost of
Capital’, The Accounting Review, Vol. 82, No. 3, pp. 705–30.
Jensen, M. C. and W. H. Meckling (1976), ‘Theory of the Firm: Managerial Behaviour, Agency
Costs and Ownership Structure’, Journal of Financial Economics, Vol. 3, No. 4, pp. 305–60.
Johnstone, D. J. (2013), ‘Information, Uncertainty and the Cost of Capital in a Mean-Variance
Efficient Market’, Working Paper (Sydney: University of Sydney).
Jones, J. J. (1991), ‘Earnings Management during Import Relief Investigations’, Journal of
Accounting Research, Vol. 29, No. 2, pp. 193–228.
Kim, W. C., P. Hwang and W. P. Burgers (1989), ‘Global Diversification, Strategy and Corporate
Profit Performance’, Strategic Management Journal, Vol. 10, No. 1, pp. 45–57.
———, ——— and ——— (1993), ‘Multinationals’ Diversification and the Risk-Return Tradeoff’, Strategic Management Journal, Vol. 14, No. 4, pp. 275–86.
Kimbrough, M. D. and H. Louis (2011), ‘Voluntary Disclosure to Influence Investor Reactions
to Merger Announcements: An Examination of Conference Calls’, The Accounting Review,
Vol. 86, No. 2, pp. 637–67.
Kinney, W. (1971), ‘Predicting Earnings: Entity vs. Sub-Entity Data’, Journal of Accounting Research,
Vol. 9, No. 1, pp. 127–36.
Lambert, R., C. Leuz and R. E. Verrecchia (2007), ‘Accounting Information, Disclosure, and the
Cost of Capital’, Journal of Accounting Research, Vol. 45, No. 2, pp. 385–420.
———, ——— and ——— (2012), ‘Information Asymmetry, Information Precision and the
Cost of Capital’, Review of Finance, Vol. 16, No. 1, pp. 1–29.
Lang, M. H. (1991), ‘Time-Varying Stock Price Response to Earnings Induced by Uncertainty
about the Time Series Process of Earnings’, Journal of Accounting Research, Vol. 29, No. 2, pp.
229–57.
Leung, E. and A. Verriest (2014), ‘The Impact of IFRS 8 on Geographical Segment Information’, Paper presented at the Journal of Business Finance & Accounting Capital Markets
Conference 2014.
Leuz, C. (2004), ‘Proprietary Versus Non-Proprietary Disclosures: Evidence from Germany’, in
C. Leuz, D. Pfaff and A. Hopwood (eds.), The Economics and Politics of Accounting (Oxford:
Oxford University Press), pp. 164–97.
——— and R. E. Verrecchia (2000), ‘The Economic Consequences of Increased Disclosure’,
Journal of Accounting Research, Vol. 38, Supplement, pp. 91–124.
Li, X. (2010). ‘The Impacts of Product Market Competition on the Quantity and Quality of
Voluntary Disclosures’, Review of Accounting Studies, Vol. 15, No. 3, pp. 663–711.
Lim, T. (2001), ‘Rationality and Analysts’ Forecast Bias’, The Journal of Finance, Vol. 56, No. 1, pp.
369–85.
Liu, W. and M. Du (2014), ‘Liquidity Risk and Asset Pricing: Evidence from Daily Data, 1926–
2009’, Paper presented at the Journal of Business Finance & Accounting Capital Markets
Conference 2014.
Lobo, G. J., S. S. Kwon and G. A. Ndubizu (1998), ‘The Impact of SFAS No. 14 Segment
Information on Price Variability and Earnings Forecast Accuracy’, Journal of Business Finance
& Accounting, Vol. 25, Nos. 7&8, pp. 969–85.
Lubatkin, M. and R. C. Rogers (1989), ‘Diversification, Systematic Risk, and Shareholder Return:
A Capital Market Extension of Rumelt’s 1974 Study’, The Academy of Management Journal, Vol.
32, No. 2, pp. 454–65.
Markarian, G. and J. Santalo (2014), ‘Product Market Competition, Information and Earnings
Management’, Journal of Business Finance & Accounting, Vol. 41, Nos. 5&6, pp. 572–99.
McInnis, J. M. (2010), ‘Earnings Smoothness, Average Returns, and Implied Cost of Equity
Capital’, The Accounting Review, Vol. 85, No. 1, pp. 315–42.
McNichols, M. (2002), ‘Discussion of the Quality of Accruals and Earnings: the Role of Accrual
Estimation Errors’, The Accounting Review, Vol. 77, Supplement, pp. 61–9.
Modigliani, F. and M. H. Miller (1958), ‘The Cost of Capital, Corporation Finance and the
Theory of Investment’, American Economic Review, Vol. 48, No. 3, pp. 261–97.
C
2014 John Wiley & Sons Ltd
SEGMENT DISCLOSURE AND COST OF CAPITAL
45
Nagar, V., D. Nanda and P. Wysocki (2003), ‘Discretionary Disclosure and Stock-Based Incentives’, Journal of Accounting and Economics, Vol. 34, Nos. 1–3, pp. 283–309.
Ohlson, J. and B. Juettner-Nauroth (2005), ‘Expected EPS and EPS Growth as Determinants of
Value’, Review of Accounting Studies, Vol. 10, Nos. 2&3, pp. 349–65.
Palepu, K. G., P. M. Healy and V. L. Bernard (2004), Business Analysis and Valuation: Using
Financial Statements, Third Edition (Mason, OH: Thomson South-Western).
Pastor, L., M. Sinha and B. Swaminathan (2008), ‘Estimating the Intertemporal Risk–Return
Tradeoff using the Implied Cost of Capital’, The Journal of Finance, Vol. 63, No. 6, pp. 2859–
97.
Prodhan, B. K. and M. C. Harris (1989), ‘Systematic Risk and the Discretionary Disclosure of
Geographical Segments: an Empirical Investigation of US Multinationals’, Journal of Business
Finance & Accounting, Vol. 16, No. 4, pp. 467–85.
Raffournier, B. (1995), ‘The Determinants of Voluntary Financial Disclosure by Swiss Listed
Companies’, European Accounting Review, Vol. 4, No. 2, pp. 261–80.
Roberts, P. W. and G. R. Dowling (2002), ‘Corporate Reputation and Sustained Superior
Financial Performance’, Strategic Management Journal, Vol. 23 No. 12, pp. 1077–93.
Saini, J. S. and D. Herrmann (2013), ‘Cost of Equity Capital, Information Asymmetry, and
Segment Disclosure’, Advances in Quantitative Analysis of Finance and Accounting, Vol. 11,
pp. 143–73.
Silhan, P. (1983), ‘The Effects of Segmenting Quarterly Sales and Margins on Extrapolative
Forecasts of Conglomerate Earnings: Extension and Replication’, Journal of Accounting
Research, Vol. 21, No. 1, pp. 341–7.
Stanford Harris, M. (1998), ‘The Association between Competition and Managers’ Business
Segment Reporting Decisions’, Journal of Accounting Research, Vol. 36, No. 1, pp. 111–28.
Stickel, S. E. (1990), ‘Predicting Individual Analyst Earnings Forecasts’, Journal of Accounting
Research, Vol. 28, No. 2, pp. 409–17.
Subrahmanyam, M. G. and S. B. Thomadakis (1980), ‘Systematic Risk and the Theory of the
Firm’, The Quarterly Journal of Economics, Vol. 94, No. 3, pp. 437–51.
Swaminathan, S. (1991), ‘The Impact of SEC Mandated Segment Data on Price Variability and
Divergence of Beliefs’, The Accounting Review, Vol. 66, No. 1, pp. 23–41.
Tang, Y. (2014), ‘Information Disclosure and Price Discovery’, Journal of Financial Markets, Vol.
19, pp. 39–61.
Tse, S. (1989), ‘Attributes of Industry, Industry Segment, and Firm-Specific Information in
Security Valuation’, Contemporary Accounting Research, Vol. 5, No. 2, pp. 592–614.
Valta, P. (2012), ‘Competition and the Cost of Debt’, Journal of Financial Economics, Vol. 105, No.
3, pp. 661–82.
Vives, X. (2008), ‘Innovation and Competitive Pressure’, The Journal of Industrial Economics, Vol.
56, No. 3, pp. 419–69.
Yoo, J. and I. Semenenko (2012), ‘Segment Information Disclosure and the Cost of Equity
Capital’, Journal of Accounting, Business and Management, Vol. 19, No. 1, pp. 103–23.
Yosha, Y. (1995), ‘Information Disclosure Cost and the Choice of Financing Source’, Journal of
Financial Intermediation, Vol. 4, No. 1, pp. 3–20.
C
2014 John Wiley & Sons Ltd