Earnings Guidance and Corporate Diversification
DONALD MONK⇤
Rutgers Business School
January 2017
ABSTRACT
Firms with a single business segment have a natural disadvantage compared to thosewith multiple segments with respect to voluntary disclosure policy. Earnings guidanceby a single-segment “focused” firm is more readily allocated to its individual businesssegment and accompanying industry than guidance by a “diversified” firm . In thissense, guidance by focused firms is at a finer level of detail than guidance by diversifiedfirms, potentially creating a competitive disadvantage for focused firms. This paperexamines whether focused firms voluntarily disclose less in order to avoid or mitigatethese e↵ects. Using management earnings guidance data I provide evidence that focusedfirms are less likely to provide an earnings forecast after controlling for typical deter-minants of forecast issuance, including various controls for competitive pressures. Also,firms increasing their corporate diversification tend to increase their disclosures. Testsshowing that disclosure ranking is not related to excess value are inconsistent with thealternative explanation that diversified firms are more likely to disclose in an attemptto mitigate their innate opacity and therefore reap greater benefits of disclosure.
⇤This manuscript is derived from my dissertation, and as such I o↵er very special thanks to my dissertationadvisor Sheri Tice and to my committee members Paul Spindt, Venkat Subramaniam, and John Hund. I appreciatethe feedback I received from participants at the Southern Finance Association Meeting, former colleagues at the SEC,participants at the University of Miami Ph.D. Rookie Camp, and seminar participants at Loyola University of NewOrleans, Kansas University, University of Arkansas, Auburn University, and Rutgers University.
I. Introduction
When a single-segment (“focused”) firm provides earnings guidance, that guidance is easily
assigned to the individual business segment in which the firm operates. On the other hand, when a
multiple-segment (“diversified”) firm provides guidance, it is aggregated across multiple segments
making it more di�cult to assess the competitive position of the firm. For example, Microsoft re-
ported five business segments (plus one Eliminations segment) in its 2008 10-K filing with aggregate
sales of $60.4 billion. Google reported just one business segment with total sales of $21.8 billion on
its 2008 10-K. Of Microsoft’s five segments, four of them operate in the same three-digit Standard
Industrial Company (SIC) code (737) as Google’s single segment. If Microsoft voluntarily discloses
a forecast of an aggregate performance measure such as earnings per share, its competitors would
have to make assumptions as to how those earnings are allocated by segment. If Google voluntarily
discloses such a measure, its competitors are able to assign the forecasted performance with more
precision and adjust their competitive structure accordingly. The example points to a potentially
higher cost of voluntary disclosure for Google due to its focused status.
This study examines whether there is a lower level of voluntary disclosures for focused firms
relative to diversified firms as would be predicted if focused firms have higher proprietary costs of
disclosure. From early analytical models to recent empirical studies, proprietary costs are shown
to a↵ect the threshold level of disclosure. Verrecchia (1983) develops a model in which concerns
of revealing proprietary information rationally limit discretionary disclosure despite its apparent
benefit of a lowering the cost of capital. More recently, Huang, Jennings, and Yu (2016) provide
evidence that a decrease in tari↵s (and the accompanying increase in proprietary costs) are associ-
ated with a decrease in management guidance. Applying this argument to diversification, if focused
firms have higher proprietary costs of disclosure, the threshold level of voluntary disclosure they
maintain will be lower than that of diversified firms. While there is already research using propri-
etary cost di↵erences between diversified and focused firms as a rationale for mandatory disclosure
di↵erences, this paper is distinct in that it examines discretionary disclosures.1
Empirical results confirm that focused firms are less likely than diversified firms to issue an
annual earnings forecast, which is consistent with higher proprietary costs of disclosure for focused
firms than for diversified firms.2 On average over the sample period, less than 16% of focused
1Botosan and Stanford (2005) and Berger and Hann (2007) consider proprietary costs as a motivating factorfor segment disclosures before and after the segment rule change in 1998 with the former showing support for aproprietary cost motivation for segment disclosures and the latter showing support for an agency cost motivation.The new segment rule (Financial Accounting Standards Board, 1997) explicitly mentions proprietary cost concernsfor diversified filers.
2As in Huang, Jennings, and Yu (2016) I focus on annual forecasts rather than quarterly forecasts, though myrationale is slightly di↵erent. Primarily, segment disclosures are only reliable and available on an annual basis, sousing annual forecasts matches the timing of the designation of corporate form. Moreover, the higher proprietary
2
firms issue a forecast in a fiscal year while almost 26% of diversified firms do so. After controlling
for other factors known to a↵ect voluntary disclosure (e.g., growth opportunities, fear of litigation,
earnings volatility, and recent performance), focused firms are still less likely to issue a forecast than
a diversified firm. Also, greater diversification, as measured using the number of business segments
or the dispersion of sales across segments, is associated with a greater likelihood of guidance.
Finally, using a first di↵erence model, I show that firms that diversify tend to contemporaneously
increase their number of disclosures and increase the lead time between their guidance and the
fiscal period end.
I incorporate several tests to rule out other reasons that focused firms may voluntarily disclose
less than diversified firms. First, diversified firms may have higher information asymmetry relative
to focused firms stemming from the more opaque, complicated diversified corporate form. Thus,
diversified firms may have a greater incentive to commit to disclosure in an attempt to mitigate
their inherent, relatively higher information asymmetry. Notwithstanding the results in Thomas
(2002) showing that the assumption that diversified firms have more information asymmetry is
not necessarily true, I include information asymmetry proxies as controls when modeling voluntary
disclosure decisions.
To further examine the alternative explanation that diversified firms provide more disclosure
to o↵set their inherent opacity, I analyze the relationship between earnings guidance and excess
value, a measure that is the relative value of a diversified firm to its imputed value using data
from focused benchmark firms. Bens and Monahan (2004) perform a similar empirical study and
find that diversified firms score lower than focused firms in the AIMR (Association for Investment
Management and Research) disclosure rankings, and their lower ranking is associated with a lower
value for diversified firms relative to focused firms. My study augments Bens and Monahan’s work
by using actual disclosures rather than outsider rankings of disclosure, recent data, new empirical
methods, and a longer sample period.
Overall, the evidence does not support valuation e↵ects of disclosure for diversified firms as an
alternative explanation. Whether using an indicator variable indicating a forecast, a matched sam-
ple of guiding and non-guiding firms, or five di↵erent ranking measures of guidance, the interaction
of guidance measures with a diversified firm status is not a statistically significant determinant of
excess value. The common empirical result that diversified firms have lower excess values (known
as the “diversification discount”) is present, but guidance appears to have no di↵erential impact
on excess value by corporate form.
A second reason that di↵erences in disclosure practices may exist is that focused firms tend to
costs for focused firms described above may only be present in annual forecast due to their longer horizon and theability of competitors to react to competitors’ disclosures before a fiscal period ends.
3
be smaller and younger than diversified firms, and these firm-level characteristics, among others,
are relevant in voluntary disclosure decisions. Here again, I use controls for these variables where
appropriate. Additionally, I use coarsened exact matching to create comparable diversified and
focused firm samples. Following Coller and Yohn (1997), I match firms based on market value
of equity, two-digit SIC code, fiscal year, and primary exchange. Through all of these robustness
checks, focused firms consistently disclose less than their diversified peers.3
Studies of voluntary disclosures generally use proxies for proprietary costs to control for its
e↵ects, and those e↵ects di↵er by the type of proprietary cost proxied (Li, 2010). The link between
proprietary costs and diversification provides another test mechanism. I consider numerous proxies
for proprietary costs, such as the Herfindahl Index, market-to-book ratio, the speed of adjustment
to abnormal profit, and research and development expenses, to compare my results with those of
extant literature. Additionally, I construct two measures of proprietary costs using the distribution
of sales across business segments of the firm: weighted-average Herfindahl Index and weighted-
average market share. Since diversified firms are composed of pieces of di↵erent industries with
potentially di↵erent proprietary costs, these measures are likely to be better indicators of the expo-
sure that a firm has to competitive pressures. Including proprietary cost measures with proxies for
firm diversification status provides additional understanding of the relationships between voluntary
disclosure and corporate form. Despite including various proxies for proprietary costs, the results
consistently show that focused firms are less likely to issue guidance.
One relevant econometric concern is the endogeneity that has been shown in the decision to
diversify. Campa and Kedia (2002) and Villalonga (2004) show that the decision to diversify is
endogenous, and this endogeneity can drastically change results using measures of corporate form.
To ameliorate these concerns I use a two-stage framework that invokes instruments for diversification
in the first stage and then uses a predicted value for diversification in the second stage. Focused
firms are less likely to issue guidance in all models, though the relationship is not statistically
significant in the last model with substantially fewer observations due to data limitations.
These results bring into question commonly used assumptions with respect to corporate diver-
sification in the academic literature and provide evidence that can be useful as regulators consider
changes to standards for segment disclosures.4 Often researchers use the number of segments or
other diversification measures as proxies for greater opacity and information asymmetry. If focused
3Diversified firms may have greater agency costs associated with their access to internal capital markets andmanagement’s desire to build empires rather than act in the best interest of shareholders. Berger and Hann (2007)lend some credence to the claim that agency costs influence segment disclosure decisions, however, it is unclear howa diversified firm could utilize voluntary disclosures to mitigate the higher agency costs associated with its corporateform. The only way to remove those agency costs would be to change from a diversified to a focused form.
4The FASB recently asked for comment on its future agenda, and segment disclosure standards were among thespecific topics that they are considering.
4
firms are indeed less likely to provide guidance as I find, it seems that focused firms have greater
opacity. For regulators, clearly something is creating a di↵erence in guidance practices of firms
based on corporate form, and if there are mandatory disclosures that are working in concert with
market forces to create these di↵erences, any updates to segment disclosure standards should take
this into account.
The remainder of the paper proceeds as follows. Section II reviews relevant literature on vol-
untary disclosure and corporate form and provides further rationale for my study. Section III
continues with a description of the data used in my empirical analysis. In Section IV, I present
the tests and results showing di↵erences in management guidance between diversified and focused
firms, and I address some empirical issues. Section V concludes.
5
II. Literature Review and Motivation
This section details how proprietary costs associated with voluntary disclosure may inhibit
full disclosure, and how such costs have been shown to limit disclosure. With the support of the
literature in conglomerate diversification and voluntary disclosure, I argue that focused firms have
a higher proprietary cost of disclosure and therefore disclose less than diversified firms. Also, I
consider potential alternatives to the proprietary cost hypotheses.
A. Competitive Pressures
Early explanations for non-disclosure relax the assumption in the full disclosure models that the
information can be conveyed with little or no cost. Later models show that the benefits of lowering
information asymmetry and potentially lowering the cost of capital via disclosure could be o↵set
by costs of the disclosure. In the informational setting where a value maximizing manager with
private information chooses whether to reveal his information, models by Verrecchia (1983) and
Bhattacharya and Ritter (1983) yield the full disclosure result for low-cost information, but their
models provide for a threshold level of disclosure when such information production is costly.5
Verrecchia (1983) pinpoints proprietary costs as a mechanism to model the tradeo↵s of disclo-
sure. In his model, firms choose to disclose information based on an expected reaction by traders
to the disclosure or non-disclosure. If the expected detriment is greater than the benefits, the
disclosure should not rationally occur. His model predicts a negative association between product
market competition and disclosure. In the presence of proprietary costs, traders are unable to assess
whether the lack of disclosure is good news or bad news, and the full disclosure premise is no longer
valid.
Other analytical studies addressing proprietary costs make it clear that the type of competition
could be an important factor. For example, Darrough and Stoughton (1990) study a potential en-
trant as the form of competition, and their model predicts that this sort of competition encourages
disclosure, therefore predicting a positive association between threat of entry and disclosure. In the
context of corporate diversification, any increased pressure from threat of entry and an accompa-
nying higher level of disclosure is likely to be borne by focused firms since they tend to be smaller
and depend upon a single industry for income.
5Diamond (1985) provides an explanation for investor demand of such information. A basic premise of much ofaccounting literature and of the full disclosure literature in particular is that managers possess private informationand investors know this fact. In practice, this assumption seems believable, although surely there are cases in whichmanagement knows little or no information. Myers and Majluf (1984) o↵ers a well known example of a financialmodel assuming that agents have superior information. On the other hand, Axelson (2007) develops a security designmodel in which bidders have superior information to management.
6
In the years since the claim by Healy and Palepu (2001) that proprietary costs had received
little attention, a few empirical studies provide support for the impact of proprietary costs on
voluntary disclosures. Bamber and Cheon (1998) find that higher product market competition is
related to a lower probability of a firm o↵ering a forecast in a venue with more “visibility.” This
negative relationship extends to the specificity of the forecast. Moreover, Brockman, Khurana,
and Martin (2008) report a negative relationship between a measure of how far management’s
forecast missed actual earnings-per-share and market-to-book (MB), with MB being their proxy for
proprietary costs (as is also the case in Bamber and Cheon (1998)). Li (2010) performs an analysis
of multiple measures of proprietary costs and provides support for less voluntary disclosure when
product market competition prevails and more disclosure when threat of entry prevails, which is in
agreement with Verrecchia (1983) and Darrough and Stoughton (1990). Finally, Huang, Jennings,
and Yu (2016) show that firms tend to decrease their disclosures when facing decreased tari↵s,
which is associated with an increase in product market competition.
Focused firms that voluntarily disclose private information are revealing a finer level of detail
than diversified firms that reveal aggregate information. For example, providing a forecast of earn-
ings per share for a focused firm will allow competitors to assess how that particular business and
accompanying industry is performing and make adjustments to investment accordingly. Diversified
firms, on the other hand, can provide an earnings forecast for the consolidated firm without reveal-
ing how individual components of the business are performing. Of course, competitors of diversified
firms will be able to use historical or contemporaneous information about the segments of the firm
to apportion aggregate earnings. However, such apportionment is at best equal to the apportion-
ment possible with a focused firm. To the extent that competitors are successful in apportioning
such information, proprietary cost e↵ects will be diminished.6
Overall, firms are operating in a market with both proprietary costs of disclosure and the threat
of entry from peers. For both of these competitive pressures, focused firms are more exposed. The
potential informational advantage for the diversified firm could raise the costs of disclosure for a
focused firm and motivate the focused firm to refrain from providing guidance. On the other hand,
if diversified firms tend to have a greater threat of entry, they could be more likely to disclose.
Since it is uncertain as to which competitive pressure prevails, the following hypothesis is stated in
null form:
Hypothesis 1. Focused firms are equally likely to provide an earnings forecast than diversified
firms.
6Hutton, Miller, and Skinner (2003) show that firms provide supplementary statements concurrently with earningsforecasts approximately two-thirds of the time in their sample; the distribution of statements is almost equal between“good” and “bad” news; and the market only reacts to “good” news forecasts when accompanied by supportingverifiable statements. These results are based on aggregate statements only and do not incorporate the intricacies ofdiversified firm versus focused firm disclosure.
7
Hypothesis 1 considers the relationship between diversification and voluntary disclosure, but
an explicit treatment of the competitive environment will provide further understanding. If the
explanatory variable that measures corporate diversification is simply a noisy proxy for competi-
tive pressures in tests of voluntary disclosure propensity, the inclusion of variables that are more
direct proxies for these measures should decrease the explanatory power of the diversification vari-
able. However, the diversification variable should still capture di↵erences that are not captured in
standard measures.
This logic applies to changes in corporate form as well. A firm that changes from being focused to
diversified will enjoy more obscurity with respect to its filings, in e↵ect lowering its proprietary costs
of voluntary disclosure. If proprietary costs prevail, once the firm is diversified one would expect it
to be more likely to provide voluntary disclosure and to provide more informative disclosures than
it did when it was focused. Likewise, a firm going from diversified to focused will have mandatory
financial statements that detail its now individual business and create greater proprietary costs.
In this case, a focusing firm would be more likely to withhold voluntary disclosures. The following
hypotheses related to changes in corporate form are stated in alternative form:
Hypothesis 2. Firms that change from focused to diversified (diversified to focused) are more
(less) likely to provide earnings guidance and provide more (less) informative voluntary disclosures.
B. Consideration of Alternatives
B.1. Cost of Capital
In the full disclosure model, managers are endowed with private information and investors
know that the manager possesses such information. If this information is disclosed, the information
asymmetry between managers and investors diminishes. Diamond and Verrecchia (1991) show that
this reduction leads to a lower price impact on the firm’s securities that increases demand from
large investors and in turn decreases the cost of capital for the firm. Another line of research that
produces a negative relationship between disclosure and cost of capital centers around estimation
risk. In the models of Coles, Loewenstein, and Suay (1995) and Barry and Brown (1985), firms
that o↵er more information have parameters that are easier to estimate, resulting in lower market
betas and lower expected returns (i.e., lower cost of equity capital). By modeling information as
a noisy indicator of future cash flows, Lambert, Leuz, and Verrecchia (2007) show that increasing
the quality of disclosures creates e↵ects within a CAPM framework that ultimately lead to a lower
cost of capital.
The empirical literature examining the notion of a negative relationship between disclosure
8
and cost of capital o↵ers mixed results. Botosan and Plumlee (2002) find support using analysts’
ratings of disclosure of annual documents, but they find a positive relationship using the ratings of
quarterly reports. Brown and Hillegeist (2007) find more consistent support by showing that annual,
quarterly, and investor relations ratings are negatively related to the probability of informed trade
measure (PIN), which they argue proxies for information asymmetry. Further, Lang and Lundholm
(1996) show that many measures often used to proxy for information asymmetry, such as analyst
coverage and forecast dispersion, accuracy, and variability, are correlated with disclosure in ways
that indicate lower information asymmetry for firms with more disclosure, which is consistent with
a lower cost of capital. Botosan, Plumlee, and Xie (2004) argue that public information could
either be a complement to or a substitute for private information, and when they include public
information, the relationship between cost of equity capital and private information is positive.
In support of the price impact story of Diamond and Verrecchia (1991), Coller and Yohn (1997)
show that information asymmetry as measured by bid-ask spreads is higher for firms providing a
forecast than for non-forecasting firms in the period prior to the forecast, but there is no di↵erence
in spreads after a forecast. Also they show that spreads over the nine days prior to a forecast are
significantly higher than the spreads over the nine days after the management forecast.
A higher likelihood of disclosure for diversified firms could be a result of an increased incentive
to lower their information asymmetry (and their cost of capital) rather than a result of di↵erences
in competitive pressures. The transparency hypothesis o↵ered in Hadlock, Ryngaert, and Thomas
(2001) states that diversified firms have higher information asymmetry due to lower transparency
in the information available about the segments of the firm relative to pure-play firm information.
The empirical evidence on higher information asymmetry in diversified firms generally finds the
opposite, however. Using analysts’ forecasts as a proxy for information asymmetry, Thomas (2002)
shows that diversified firms do not have more information asymmetry than focused firms. He shows
“that greater diversification is associated with smaller forecast errors and less dispersion among
forecasts.” Moreover, he finds that diversified firms have higher earnings response coe�cients (ERC)
indicating that investors impound earnings information into stock prices to a greater extent than for
focused firms. However, the results from Thomas (2002) indicating lower information asymmetry
for diversified firms flip after controlling for return volatility. Using market microstructure measures
of information asymmetry, Clarke, Fee, and Thomas (2004) support the Thomas (2002) findings of
lower information asymmetry for diversified firms.7
Only a few studies o↵er tests related to di↵erences in cost of capital or expected returns between
diversified and focused firms. Hann, Ogneva, and Ozbas (2013) provide evidence that diversified
firms have a lower measured cost of capital relative to their focused peers. Hadlock, Ryngaert, and
7Though not a study including all diversified firms, Krishnaswami and Subramaniam (1999) show that firms thatengage in a spino↵ have higher information asymmetry than a matched control group and gains associated with thespino↵ are related to the decrease in information asymmetry for spino↵ firms.
9
Thomas (2001) show that relative to focused firms, diversified firms su↵er a less negative stock price
reaction to seasoned equity o↵erings than focused firms, which is inconsistent with higher levels
of information asymmetry for diversified firms. The authors attribute their result to lower adverse
selection problems in issuing securities of diversified firms due to lower measurement error from
imperfectly correlated segments than from a bucket of focused segments. If the lower measurement
error for diversified firms that are selling securities is actually due to a commitment to disclosure
above and beyond that of focused firms, increased disclosure could be causing this result. Lamont
and Polk (2001) bring lower cost of capital for diversified firms into question by showing that there
is no di↵erence in returns between diversified and focused firms, but they do find that discounted
diversified firms have higher realized returns than premium diversified firms.8
I address the possibility that diversified firms have a greater incentive to disclose due to dif-
ferences in information asymmetry rather than proprietary cost di↵erences in two ways. First,
in regressions of disclosure on diversification status and proprietary costs, I include variables that
control for information asymmetry. Next, I analyze whether firms that provide voluntary disclosure
have higher valuations relative to their industry peers and whether this result is related to diversi-
fication status. If the latter is true, it is an indication that further analysis is needed to disentangle
the determinants of disclosure and how those determinants a↵ect value.
Bens and Monahan (2004) report that disclosure ranking measured using AIMR ratings, which
is used as an inverse proxy for information asymmetry, is positively associated with excess value,
which is measured as a log ratio of the actual value of a firm to its value imputed from focused
firm rivals, for diversified firms, but the relationship does not exist for focused firms. The authors
attribute the positive association for diversified firms to the increased monitoring that is present
for firms with more revealing disclosure. My empirical structure allows me to update Bens and
Monahan’s work with more recent data, new empirical methods, and with actual disclosures rather
than outsider rankings of disclosure. I test the following hypothesis:
Hypothesis 3. Greater voluntary disclosure is associated with higher excess value.
Moreover, if diversified firms have higher information asymmetry and use voluntary disclosure
to decrease it, there would be a positive interaction e↵ect for diversification status and disclosure
in regressions of excess value. The hypothesis below formalizes this argument:
Hypothesis 4. Higher levels of disclosure positively e↵ect the excess value of diversified firms more
than focused firms.
8Mitton and Vorkink (2008) find that diversified firms have lower skewness in their returns and this is consistentwith investors’ preference for skewness risk and with a discount for diversified firms.
10
B.2. Agency Costs
There is also a strand of literature addressing managers acting in their own interest and adopting
a disclosure policy accordingly. Berger and Hann (2007) provide empirical support for an agency
cost story that managers of diversified firms seek to mask ine�cient behavior among their segments
by aggregating segments with poor performance. Using proxies for disclosure, Aboody and Kasznik
(2000) find evidence that is consistent with firms adapting their voluntary disclosures in favor of
CEO option payo↵s. Brockman, Martin, and Puckett (2008) lend more support to this argument
by showing that firms release information intended to increase management’s stock option payo↵ by
releasing positive disclosure before intended exercise of options and by releasing negative informa-
tion before intended holding of vested options. In a similar agency cost story, insider transactions
are shown to be clustered after voluntary disclosures that result in higher payo↵s for the insiders
in Noe (1999). Bernhardt and Campello (2007) provide evidence that managers “talk down” the
consensus analyst estimate of earnings. While this practice fools investors in that they treat the
changes in analysts’ estimates as unbiased, the earnings “surpise” is not substantial enough to raise
the stock price above its losses from talking down the consensus before the earnings announcement.
Finally, Brockman, Khurana, and Martin (2008) show that managers “talk down” the price of the
firm’s stock using voluntary disclosures prior to repurchasing shares, and the bias in management
forecasts is positively correlated with management’s private incentives.9
Many studies on corporate form point to potential agency costs di↵erences between diversified
and focused firms. At the level of the CEO, Shleifer and Vishny (1989) model an empire building
CEO who overinvests in projects to carve out more rents for herself. Jensen (1986) details another
form of overinvestment borne of greater access to free cash flows in the diversified corporate form.
Rajan, Servaes, and Zingales (2000) develop a model in which incomplete contracting on invest-
ment choice drives self-interested divisional managers to invest in projects that are defensive rather
than those that are most e�cient for the firm. Scharfstein and Stein (2000) show how rent-seeking
managers provide another avenue for a value loss for corporate diversification as managers take on
projects that increase their bargaining power rather than increasing firm value. Lamont (1997),
Lamont and Polk (2002), and Ahn and Denis (2004) provide empirical support for overinvestment
by diversified firms. If managers are behaving in the manner described in these studies, agency
costs will be higher in all cases for diversified firms. As such, they will be considered “lemons” in
the marketplace, and any attempt to mitigate agency costs using disclosure will be moot in equi-
librium. Since the mechanism by which voluntary disclosures could be used to mitigate this aspect
9It could be that the adjustment to disclosure by diversified firms is less than for focused firms because investorsdon’t know enough details to apportion the news to the segments that make up the business. If this is the case, thegood news/bad news studies will have more focused firms in them, and in turn, those samples will be smaller andyounger than excluded firms. Also, dividing the sample based on ”substantial news” (>1% or <-1% move in stockprice) amplifies the aforementioned e↵ect.
11
of di↵erences in corporate form is not evident, I do not include direct agency cost considerations
in my tests, though a number of control variables can be construed as proxies for agency cost.
12
III. Sample and Variable Construction
The primary data that I use to test the hypotheses are derived from the intersection of the First-
Call Company Issued Guidance (CIG) database and segment- and firm-level data from Compustat.
A download of the entire database of forecasts from CIG yields 66,600 annual forecast observations
with announcement years from 1990–2011.10 I restrict the sample to forecasts of earnings per share
on common stock in U.S. dollars that possess an eight-digit CUSIP and a FirstCall code that is
necessary to qualify the specificity of the forecasts. Similar to Anilowski, Feng, and Skinner (2007),
I remove forecasts that are more than 90 days after or more than two years and 90 days before the
subject fiscal period end of the forecast. Finally, I remove a few remaining duplicates in the CIG
and observations in firm fiscal years of 1990–1993 and 2012 as there are very few observations in
those firm fiscal years. The final sample has 57,330 forecasts.
To derive measures of corporate diversification and to weight variables according to segment
distribution, I use the segment-level data from Compustat. SFAS No. 14 created the regulatory
requirement for firms to file segment-level information with implementation and data entries begin-
ning in earnest in the fiscal year of 1978. Restatements of segment or firm information are removed
so the database contains information that was available to investors at the time of filing rather
than adjusted numbers and filings revealed later.
SFAS No. 131 creates the need to make an adjustment to the data on both sides of the rule
change for comparability. The new rule requires firms to report segments based on operating
structure rather than industry composition. As a result, firms reported more segments, but many
of these segments are in the same four-digit SIC code. The procedure I use aggregates sales for
segments in the same 4-digit SIC code thereby making the data after SFAS No. 131 more comparable
to those before it. I also remove segments with sales equal to zero or with missing values, since
many of these are “corporate” segments put in place to allow firms (under the new rule) to allocate
assets and other financial information to the corporate entity rather than business-line segments.11
Finally I merge the forecast and segment data with Compustat firm-level data required to per-
form additional screens for the segment-level data and to calculate other variables used throughout
the study as controls. I remove those firms not reporting segment sales that sum to within 5% of
reported total sales. This firm-level screen is taken from Berger and Ofek (1995) and is in agree-
ment with the empirical diversification literature. Other variables will be described in the sections
below as needed. Short descriptions of all variables are in Appendix A.
10The CIG was discontinued in 2011 after it was bought by Thomson.11Refer to Sanzhar (2006) for more information about the impacts of the new standard with relation to pseudo-
conglomerates. Due to the subjective nature of asset allocation under the new rule, I only use segment sales data inmy analysis.
13
A. Measures of Voluntary Disclosures
Using the FirstCall Company Issued Guidance data described above I create management fore-
cast variables as proxies for voluntary disclosures. I create an indicator variable (Forecast) to
indicate whether management issued a forecast during a fiscal year. Forecast equals one for each
CIG observation that has a matching firm-year observation from Compustat, and it equals zero for
Compustat firm-years that do not have a matching observation in CIG. To examine how often the
firm provides earnings guidance, I calculate the number of forecasts provided by a firm in a fiscal
year, notated by NForecast, including updates to previously issued forecasts but not duplicating
forecasts given on the same day.
To allow for deeper analysis of the disclosure policy of firms, I create variables based on more
than just the sheer number of management forecasts. First, I calculate the number of days between
the guidance announcement date and the fiscal period end (actual announcement date), denoted
by Lead (LeadAct). Note that Lead is negative for those forecasts that are provided after the fiscal
period end but before the actual earnings are announced. Second, I create a variable to denote the
specificity of forecasts, Spec, using the definitions from Baginski, Conrad, and Hassell (1993) and
a numbering scheme that is increasing in specificity as indicated in Appendix A.
The final forecast measure is the ex-post accuracy of the management forecast. Error is cal-
culated as the di↵erence between the management forecast and actual earnings normalized by the
stock price at the end of the most recent quarter, multiplied by 100, and winsorized at the 1% level.
I also use the absolute value of this measure in some tests. Ajinkya, Bhojraj, and Sengupta (2005)
and Brockman, Khurana, and Martin (2008), among others, use a similar measure of management
“bias” in situations of monitoring and repurchasing shares, respectively. In the present context the
measure will be useful in determining if the bias from other research is related to the e↵ects of
proprietary costs and diversification. However, this measure is imperfect because for open-interval
forecasts (e.g., “greater than $0.39 per share”), I simply subtract actual EPS number from the only
EPS value available ($0.39). Also, for range forecasts, I use the mid-point of the range forecast as
management’s forecast following Baginski, Conrad, and Hassell (1993).12
B. Measures of Diversification Status and Value
Using the Compustat segment data I create two measures of diversification. The first and most
commonly used is the diversification indicator variable (Divers) that equals one if a firm reports
12The interpretation of forecast error is also problematic. The primary issue is that it is an ex-post measure of howaccurate management was in predicting actual earnings. Increasing the time between forecast and actual earnings(forecast lead time, which is itself a measure of disclosure) increases the likelihood that confounding factors a↵ectforecast accuracy.
14
multiple segments by four-digit SIC code in a fiscal year. Otherwise, the indicator equals zero. To
provide additional depth to the analysis, I also create entropy (Entropy) as described in Jacquemin
and Berry (1979) as a continuous measure of diversification. The entropy measure of diversification
for firm i is determined at fiscal year t by
Entropyi, t =nX
s=1
Ps, i, t ln1
Ps, i, t
, (1)
where n is the number of four-digit SIC code segments and Ps, i, t is the proportion of sales from
segment s of firm i at t. Entropy equals zero for firms reporting a single business segment, and it is
greater than zero for firms reporting multiple business segments. Importantly, entropy changes as
the distribution of sales across segments changes, even if the number of segments is held constant,
which allows for an analysis of the impact of the degree of diversification on disclosure decisions.
C. Proprietary Costs
Several measures are needed for reliable proxies for the proprietary costs that firms face. As
noted in Li (2010), the type of competition can and does have an impact on voluntary disclosure
equilibrium outcomes. The di↵erence between product market competition and the threat of entry
has been shown to be enough to change the e↵ect of competition on voluntary disclosures. The
variability of proxies for proprietary costs across industries, firms, and segments can be drastically
di↵erent. I separate the measures according to their variability: industry-, firm-, or segment-level.
C.1. Industry-Level Measures
Following Botosan and Stanford (2005) and Harris (1998), for each three-digit firm-level SIC
code I construct the four-firm concentration ratio (Conc4Firm) and the Herfindahl Index (HI).
The former equals the sum of the proportion of annual sales in a three-digit SIC code industry of
the top four producers by sales, whereas the latter is the sum of the squared proportions of sales
coming from all firms in a three-digit SIC code industry. As these measures increase, competition
decreases.
As the last industry-level measure, I use the speed of profit adjustment. Harris (1998) notes that
this measure provides an indicator of the persistence of abnormal profits away from the industry
mean. The value for speed of adjustment, SpeedAdj, is the coe�cient �2j of Eq. 2, which is executed
separately for each industry j. As with Conc4Firm and HI, a higher value for SpeedAdj implies
15
less competition.13
Xijt = �0j + �1j(DnXijt�1) + �2j(DpXijt�1) + "ijt (2)
where
X = di↵erence between the ROA of firm i and the mean ROA of its three-digit
SIC industry j,
Dn = indicator variable of negative X,
Dp = indicator variable of positive X.
C.2. Firm-Level Measures
The equity market-to-book ratio (MB) has been used in the disclosure literature as a measure
of growth opportunities and more loosely as a proxy for proprietary costs. Firms with high growth
opportunities may have a lower incentive to disclose as argued in Bamber and Cheon (1998), but
this relationship could be in the opposite direction if a firm desires to deter entry by signalling that a
particular industry has lower opportunities. Perhaps this ambiguous relationship is demonstrated
in their findings that the lagged value of MB is negatively associated with the level of investor
proactivity of the release venue, but when used as an explanatory variable for forecast specificity
the ratio is no longer significant. Further, Ajinkya, Bhojraj, and Sengupta (2005) include lagged
MB in similar regressions of management forecasts issuance, but in most cases their tests show that
the coe�cient for it is not significantly di↵erent from zero. I calculate MB as the log of the ratio
of the market value of equity at calendar year end to the book value of equity.
Other firm-level variables o↵er more direct proxies for proprietary costs. Research and develop-
ment expense (RD), calculated as the yearly R&D expense over assets, is argued to be positively
related to proprietary costs in Wang (2007). In cases in which R&D expense is missing, I set
the value equal to zero. Also, I include three-digit SIC industry percent rank of profit margin
(PMargin) and market share (MShare) as in Nichols (2009).
C.3. Firm-Level Measures Using Segment-Level Information
Since a diversified firm is composed of multiple segments from potentially multiple industries, I
construct some firm-level variables that are based on segment-level information. For each measure,
I treat the segment as a separate entity within an industry and calculate market share information
13Berger and Hann (2007) use segment abnormal profitability to proxy for management’s desire to withhold segmentinformation from potential entrants. As stated in their paper, such measures for the entire sample of segments aredi�cult to obtain and to verify. Their sample is limited to firms changing corporate form around a rule change. Assuch, they could hand-collect the necessary data more easily.
16
accordingly. By treating each segment as a separate competitor in the industry market, these mea-
sures o↵er a more complete picture of the level of competition that a particular industry participant
is facing. Specifically, I use segment sales and their accompanying industry designation to create a
segment-sales weighted average market share (MShareSeg) and Herfindahl Index (HIwtd). To cal-
culate the latter measure I multiply the proportion of firm sales in a particular three-digit segment
industry by the Herfindahl Index created using sales values from all segments within a three-digit
SIC code industry and then sum over the number of segments in the firm as shown in Eq. 3 and
Eq. 4.
HIsegj =mX
i=1
(siSj
)2 (3)
HIwtdf =nX
k=1
(skSf
) ⇤HIsegj , (4)
where
m = number of segments in three-digit industry j,
s = segment sales,
S = sales from all segments (in industry j or firm f),
n = number of segments in firm f .
Appendix D provides some support for separate consideration of the proprietary cost measures.
Although many of the correlation coe�cients between the measures are significantly di↵erent than
zero, only four have an absolute value greater than 0.5. SpeedAdj, MB, and PMargin have very
little relationship with any of the other measures. Since MB has been used in the disclosure
literature to proxy for other economic e↵ects such as growth opportunities, it will remain in my
analyses. Among the remaining proprietary cost proxies, I include industry-level, firm-level, and
segment-based calculations.
D. Additional Variables
I address two common controls first. Firm size may have a positive or negative association
with disclosure. On one hand, larger firms will have the real resources to produce the information
more easily (Diamond, 1985). On the other hand, more information is generally available publicly
for larger firms, perhaps substituting for some of the information that management would otherwise
release (Brockman, Khurana, and Martin, 2008). Harris (1998) argues that firm size is also a proxy
for the number of segments reported due to filing requirements based on a 10% threshold to list a
segment separately. To control for these possible e↵ects I use the variable Size, measured as the
log of total assets. Brown and Hillegeist (2007) also note the importance of recent performance
17
on a firm’s decision to issue guidance. To capture recent performance I use fiscal year excess firm
return over the CRSP value-weighted index.
Earnings volatility has been used as a measure of the potential for large movements in
management forecasts and susceptibility to litigation. Managers from firms with higher earnings
volatility may have a tougher time forecasting earnings and may be more likely to get the forecast
wrong. Not only is this measure applicable in the study of voluntary disclosures, but also it has been
shown to be an important determinant in studies of corporate diversification. Diversified firms are
shown in Dimitrov and Tice (2006) and Hund, Monk, and Tice (2010) to have significantly lower
volatility in firm performance measures such as ROE, ROA, and EBIT . I calculate earnings
volatility, EarnV ol, as the standard deviation of the previous 12 quarters of earnings before the
period including the forecast winsorized at 1%.
To address information asymmetry, which is one of the primary theoretical determinants of
disclosures, I use a few measures taken from extant literature. First, I use residual stock return
standard deviation, StockV ol, as calculated in Krishnaswami and Subramaniam (1999). StockV ol
is the standard deviation of the market-adjusted daily stock returns over the 36 months preceding
the forecast announcement. I take two other measures of information asymmetry from analyst
information as provided in FirstCall: Analysts and AEDisp. Analysts is the number of analyst
estimates of annual earnings per share preceding the date of the management forecast, and AEDisp
is the standard deviation of all active analyst forecasts as of that same date winsorized at 1%.
There is considerable theoretical and empirical evidence in the disclosure literature showing
that firms disclose good news more readily than bad news.14 I construct an indicator variable
for negative earnings, NegEarn, to control for this e↵ect. However, there is a counterargument to
the preference for good news disclosures. Management’s legal obligation to reveal material private
information can bias their disclosures toward “bad news” as management attempts to prevent suits
after a precipitous fall in stock price as in Baginski, Hassell, and Kimbrough (2002) and Schrand
and Walther (1998).
The legal environment, specifically the probability of litigation surrounding negligent guid-
ance, has been shown to be a factor when issuing guidance, for the frequency of the guidance, and
for its specificity. Congress enacted the Private Securities Litigation Reform Act of 1995 as a means
to address this fear of litigation. Recent results by Rogers and Stocken (2005), Kothari, Shu, and
Wysocki (2009), and Cao, Wasley, and Wu (2007) show that firms are more likely and quicker to re-
veal bad news than good news. I use the negative earnings growth indicator variable (NegEarnG)
from Bamber and Cheon (1998) and Brockman, Khurana, and Martin (2008) to proxy for litigation
exposure. NegEarnG equals 1 if the firm has negative earnings growth over the year, and it equals
14For example, see Dye (1990), Dye and Sridhar (1995), Gennotte and Trueman (1996), and Miller (2002).
18
0 otherwise. Additionally, I include a broader indicator for industries prone to litigation. Using
segment-level data, I calculate LitInd as the proportion of firm total sales coming from segments in
the following four-digit SIC code industries: 2833–2836 and 8731–8734 (biotechnology); 3570–3577
and 7370–7374 (computers); 3600–3674 (electronics); and 5200–5961 (retail).
19
IV. Empirical Tests and Results
In the following section, I merge arguments taken from the Motivation section with the data
described in the previous section to implement empirical tests.
A. Summary Statistics
On average over the sample period, a greater percentage of diversified firms provide voluntary
disclosure than do focused firms. Table I shows descriptive statistics for annual forecasts split into
two panels based on diversification status. The mean number of annual forecasts per firm per
year (NForecast) has increased from about one in 1995 to more than three after 2002, and mean
NForecast is greater for diversified firms in every year after 1997. On average 26% of diversified
firms provide a forecast while only 16% of focused firms do. While diversified firms comprise 28%
of firms not providing annual guidance, they comprise 42% of firms providing an annual forecast
and 46% of total annual forecasts.
The summary statistics in Table II support these di↵erences in disclosure between focused and
diversified firms, and provide support for control variables used in later tests. Both measures of
disclosures (Forecast and NForecast) indicate that focused firms disclose less often and are less
likely to disclose. Almost all of the measures of competitive pressures that have been shown to
a↵ect disclosure are statistically significant in the direction of less disclosure for focused firms: HI
and HIwtd are lower; SpeedAdj, RD, and MB are higher; and NegEarn and NegEarnG are higher.
The only variable not conveying higher competitive pressures for focused firms is profit margin
(PM), which is slightly lower than for diversified firms.
B. Forecast Issuance
I first analyze whether diversified firms are more or less likely than focused firms to issue
a forecast as stated in Hypothesis 1 and whether the e↵ect of diversification changes with the
competitive environment. I test the propensity of providing a management forecast conditioned
on proxies for corporate form and other factors known to a↵ect forecast issuance, such as growth
opportunities, firm size, earnings volatility, and litigation environment (see Rogers and Stocken
(2005) and Matsumoto (2002)). The dependent variable is the indicator variable Forecastt that
equals 1 if a firm provides a forecast in fiscal year t and is 0 otherwise. Due to the binary nature
of the dependent variable, I use a probit model. The tests of forecast issuance take the form:
Pr(Forecastt) = �0 + �1Formt�1 +Xt�1� + "t, (5)
20
where Form is either the multi-segment indicator variable Divers or the entropy measure of di-
versification Entropy, and X is a vector containing control variables.
Table III provides results that are consistent with Hypothesis 1 for various iterations of Equa-
tion 5 using Divers as a measure of diversification. All of the models show that the diversified
corporate form is associated with a greater propensity to issue a forecast. Overall, those mod-
els with ample control variables that contain ample observations perform quite well in explaining
the probability of earnings guidance: Pseudo R2 for Models 2-4 are over 10% and 80% percent is
correctly predicted by the model.
The results in Table III are consistent across models with respect to the control variables. The
coe�cients for size as proxied by ln(Assets) are positive and significant at the 1% level, indicating
that larger firms are associated with higher probability of issuing a forecast, perhaps because size
is a proxy for diversification as in Harris (1998). The negative coe�cients for NegEarn are contrary
to arguments that firms with negative earnings attempt to avoid litigation resulting from poor
performance by being more transparent via disclosures. However, LitInd is positive and significant
in almost all cases, and the inclusion of LitInd makes the interpretation of NegEarn di↵erent with
respect to litigation exposure. Consistent with earlier studies, recent firm performance, as proxied
by ROE, is positive and significant at the 1% level in all models. Some variables have signs consistent
with other studies, but are not significant in many cases. For example, those measures proxying
for di�culty to forecast (stock volatility, earnings volatility, and analyst forecast dispersion) have
the expected negative sign, but are not significant in any model.
Table IV shows that using other proxies for corporate diversification produces very similar
results to those found using Divers. The number of segments (SegN) and greater distribution
of sales across segments (Entropy) are positive and significant at the 1% level in all models. The
results for the control variables are almost identical to the results using Divers as the diversification
indicator.
C. Changes in Forecasting Activity
Using the level in corporate diversification and the level in disclosure practices subjects the
findings to an instant criticism of omitted correlated variables. One way to mitigate this criticism
inasmuch as it is due to time constant factors is to use a changes regression. I employ a method
to investigate changes in disclosure activity regressed on changes in corporate form and changes in
other variables known to a↵ect voluntary disclosure as shown in the following equation:
�Disc = �0 + �1Diversifying + �2Focusing + �3DivDiv +�X� +�". (6)
21
Disc is one of three measures of firm disclosure activityNForecast, Lead, or Error.15 Diversifying
is an variable indicating that a firm went from focused in year t�1 to diversified in year t. Focusing
indicates a firm going from diversified to focused. DivDiv indicates that a firm did not change
form from t� 1 to t and is diversified. The omitted case of FocFoc, in which a firm is focused at
t� 1 and t, will be predicted by the intercept �.
To test Hypothesis 2 I concentrate our attention on �1 and �2 in Equation 6. In the models in
which the change in NForecast or Lead is the dependent variable, a positive and significant coef-
ficient for �1 indicates that firms increasing their corporate diversification are contemporaneously
increasing their disclosure. Since Error is considered an indicator of less disclosure, in models with
Error as the dependent variable, I expect �1 to be negative and significant. The coe�cient �2
represents the e↵ect on those firms that are focusing, and for Hypothesis 2 to hold, I expect the
signs for �2 to be the opposite of �1.
The summary statistics in Table VIII give insight on whether Hypothesis 2 is likely to hold and
evoke some concerns about lack of variation in variables. For the full sample, 7.3% of Diversifying
firms increase their disclosures by starting earnings guidance, and this is a higher rate than other
indicators of changes in corporate form. However, the 3.4% of Focusing firms that stop earnings
guidance is not noticeably di↵erent from the rate of stopping for other changes in corporate form.
Also evident in the statistics is a lack of variation in forecasting changes with approximately 90% of
firms having no change in forecasting status in a given year they are in the full sample. Moreover,
Table VIII shows that very few firms change their corporate form in a particular year. Overall,
for the full sample approximately 95% [=(35574+15997)/54231] keep the same corporate form in
a given year.
To mitigate the first issue of a lack of variation in the dependent variable I use various measures
of disclosure. First, I use the number of forecasts provided during the fiscal year NForeast as that
measure has higher variation. One problem with using this measure is that it equals zero for much
of the sample (See Table I and Section IV.A.) Second, I use the change in average Lead and Error
as separate dependent variables. Of course, these variables have the limitation that they can only
be measured for those firms providing forecasts.
Table IX shows the results from the tests of Hypothesis 2 for the full sample. In three of the four
full sample models, �1 is positive and significant indicating that indeed Diversifying firms tend to
increase the number of disclosures they provide and lengthen the time between the EPS guidance
and the fiscal period end. Also, these coe�cients are greater than those for other indicators of the
change in corporate form. However, it does not appear that Hypothesis 2 holds for Focusing firms.
15A greater value for Error indicates a higher error on average versus industry peers and therefore is an inversemeasure of disclosure.
22
The coe�cient �2 is not significant in any model. Perhaps the greater number of diversifying firms
relative to other changes in form, which have very few observations, allows for better identification
of �1.
I address the lack of variation in the change in corporate form slightly di↵erently. Fig. 1 shows
that in 1998 there was a significant increase in the percentage of diversified firms providing a
forecast. Also, since 1998 the di↵erence between the percentage of diversified firms and focused
firms providing guidance has remained approximately the same. This is exactly the time that
SFAS131 came into e↵ect, and I know from previous studies and results in Appendix E that the
number of diversified firms increased dramatically during the few years as did other measures
of diversified firm transparency like number of segments. Since there was such great change in
corporate form (mainly diversifying) during the years just after the new standard passed, I run the
model from Equation 6 on the sample from 1998–2000.16
In the limited sample focused on potential e↵ects for firms that diversified during the period just
after SFAS 131 implementation, the results are similar to those for the full sample. The models in
Table IX that use�NForecast as the dependent variable show positive and significant �1 coe�cients
as before. The model using �Lead has a �1 coe�cient that is positive, but not significant. Here
again, however, there are very few observations—only 660 observations for the SFAS 131 sample
using �Lead as the dependent variable. Another noticeable di↵erence is that all of the models
testing the SFAS 131 sample show a negative �2, though it is not significant in any model.
Overall, there seems to be weak support for the diversifying firm aspect of Hypothesis 2, but
not the focusing firm aspect. Since using a first di↵erencing model removes a time-constant unob-
servable, any force causing this increase in disclosure for diversified firms must be time-varying. For
example, diversified firms may face competitive pressures outside of those proxied by the controls
listed that change year-to-year according to their particular list of competitors and the proportion
of the business facing each competitor. For focused firms, lack of statistical power notwithstanding,
their list of competitors may be more focused on the proportion of private (non-filing) competitors
they face, since financial statements from focused firms would be more useful to private firms in a
particular business.
16Despite the apparent usefulness of a di↵erence-in-di↵erences model in this situation (Roberts and Whited, 2012,pg. 37), I only use the years immediately following the implementation of SFAS 131. The implementation of Securitiesand Exchange Commission (2000) in late 2000 drastically changed the disclosure dynamic, potentially confoundingany DiD model. Also, there are very few observations for the forecasting data prior to 1998, so establishing pre-eventnorms is problematic.
23
D. Valuation E↵ects as Alternative Explanation
The next tests that I perform are related to the potential benefits of voluntary disclosure that
could be alternative explanations for a higher level of disclosure for diversified firms. Firms that
successfully lower the level of information asymmetry surrounding their firm should enjoy higher
valuations. Moreover, diversified firms that are considered more opaque may benefit more from
such disclosures than their less opaque focused peers. Although all firms would be expected to
gain value if they commit to higher levels of disclosure and disclosure decreases the cost of capital,
diversified firms may benefit even more from disclosure if they have characteristics causing their
cost of capital to be higher relative to focused firms.17
I use the excess value measure to assess valuation di↵erences between diversified and focused
firms. Excess value (EV ) is calculated using a log ratio of reported total capital (market value of
equity plus book value of debt) to the imputed value for the firm. The imputed value is computed
by multiplying the median ratio of total capital to sales for focused firms in a segment’s industry
by the segment’s reported sales and then summing over the number of segments in the firm.18
I test Hypothesis 3 and Hypothesis 4 using the following regression:
EVt = ↵+ �0Diverst + �1Disct + �1DiversXDisct +Xt� + "t, (7)
where Disc is one of many measures of firm disclosure level for fiscal year t: Forecast dummy
or within industry percentile rank of NForecast, Lead, Spec, Error, or |Error|.19 Notably, all
percentile rank disclosure measures can only be calculated for those firms that provide earnings
guidance in our sample. Typical control variables for regressions involving excess value are included
in Xt. All models include year fixed e↵ects and cluster standard errors by year.
The results in Table V support Hypothesis 3 by showing a positive relationship between volun-
tary disclosure and excess value. Every coe�cient for the disclosure measures except the percentile
rank of NForecast is significant and is in the direction of a positive relationship between disclosure
and excess value. Of course, the number of observations drops considerably when the regression
uses the percentile rank disclosure variables since only those firms providing guidance are included
in those tests.17This line of reasoning resulting in a higher cost of capital for diversified firms is exactly the opposite of the
empirical result in Hann, Ogneva, and Ozbas (2013) of a lower cost of capital for diversified firms.18I do not use the asset- or EBIT-multiplier approach for excess value. I forego the former because the allocation
of assets to segments is problematic after the passing of SFAS No. 131, and the latter because EBIT is often missingin the segment data. Appendix B provides greater detail on the formula used to calculate excess value.
19A higher ranking for |Error| indicates a higher error on average versus industry peers and therefore is an inversemeasure of disclosure.
24
As for Hypothesis 4, the results in Table V are mixed. For the models using the indicator
variable Forecast there is support for the hypothesis as the coe�cient for the interaction term
DiversXDisc is positive and significant. This means that diversified firms that provide earnings
forecast are valued higher than their focused peers who provide a forecast. However, none of the
percentile rank disclosure measures is significant and in the correct direction. In fact, two of the
coe�cients are in the opposite direction as hypothesized. For the disclosure measure percentile
rank of Spec, the interaction coe�cient is -0.091 and is significant at the 10% level, which indicates
a negative relationship between higher specificity of the forecasts and excess value for diversified
firms. Likewise, a higher guidance error is positively related to excess value for diversified firms.
The lack of a convincing result for Hypothesis 4 lends some support to the argument that the
higher propensity of providing a forecast for diversified firms found in earlier tests is not necessarily
related to diversified firms disclosing more to benefit from accompanying reductions in their cost
of capital.
E. Empirical Issues and Robustness Tests
E.1. Matched Sample
Although a number of recent academic studies use the FirstCall CIG database for guidance
forecasts, there are some sample selection concerns with the firms covered. Lansford, Lev, and
Tucker (2010) provide an appendix to their work showing that firms providing “soft” guidance
information are less likely to be covered in the CIG. Moreover, Chuk, Matsumoto, and Miller
(2013) provide evidence that firms providing guidance with greater Lead or lower Spec, among
other characteristics, tend to be missing from the CIG database. For this to be a factor in the
results presented here, the omissions from the CIG would have to be systematically related to
diversification status or proprietary costs.
To allay these concerns I change how I determine the sample that did not issue a guidance
forecast. Namely, I use coarsened exact matching to construct the nonforecasting firms from firms
that are matched to those in the FirstCall CIG using a number of criteria. I follow Coller and Yohn
(1997) and match firms on market value of equity, two-digit SIC code, fiscal year, and primary
exchange. I use coarsened exact matching to exactly match on the latter three characteristics and
to match within a range for the market value of equity. As shown in the last column of Table V,
the results using this adapted sample are almost identical to the model using all non-forecasting
firms that is shown in the first column of the table.
25
E.2. Diversification Decision
The decision to diversify has been shown to be a factor in analyzing the e↵ects of diversification
status. Campa and Kedia (2002) and Villalonga (2004) provide evidence that the results of earlier
studies using diversification indicators as exogenous measures are erased or even reversed when
variables correlated with the decision to diversify and the dependent variable in those studies are
included in the empirical framework. In the section below I address this endogeneity first in the
analysis of forecast issuance (Hypothesis 1) and then in the analyses of excess value as an alternative
explanation for my findings (Hypothesis 3 and Hypothesis 4).
For the models examining the factors related to forecast issuance I address endogeneity of
corporate form by fitting a probit model that allows for instrumentation of a continuous endogenous
explanatory variable. Since the implementation of instrumenting a binary endogenous explanatory
variable in a binary response model has some weaknesses, I perform tests using the continuous
variable Entropy rather than Divers as a proxy for diversification.
As instruments for Entropy, I use three measures that have been supported in the literature.
Campa and Kedia (2002) note that there are many reasons why a particular industry may be more
attractive to a particular corporate form. In particular, they mention industry regulation as a
potential factor. I use their measures to capture this potential e↵ect. PSDIV is the fraction of
sales within an industry that come from diversified firms after omitting the sales from the subject
firm. In constructing PSDIV , industry is measured at the two-digit SIC code level in Campa and
Kedia (2002), but I use the three-digit and the four-digit level to allow for comparison with other
measures. Also, I use a sales-weighted average of the industry measures, which a↵ects the values
for multiple-segment firms that operate in multiple industries. These measures are constructed to
be positively associated with industry attractiveness for diversified firms. Following Dimitrov and
Tice (2006), I also include minority interest as shown on the balance sheet (MIB) as an instrument
for the decision to diversify. MIB is an indicator variable that equals one if the firm has non-zero
minority interest on its balance sheet. This indicates that the firm owns a majority of another firm
and therefore has an interest in that firm.
Table VI shows the second stage results of this test. Four of the five models have coe�cients
for Entropy that are positive and significant in agreement with Table IV and in support of focused
firms being less likely to provide a forecast as in Hypothesis 1. The model including controls for
analyst information shows a positive coe�cient for instrumented Entropy, but that coe�cient is
not significantly di↵erent from zero. The lack of results for these particular models weakens my
previous findings, but the last model is suspect since it only has 16,055 observations while the other
models have well over 40,000.
26
For the models examining the potential e↵ects of disclosure on excess value I address endogeneity
of corporate form using a method that much more closely matches those presented in Campa and
Kedia (2002) and Villalonga (2004). Using the aforementioned instruments for diversification, I
employ a two-stage least squares (2SLS) regression model and run the same models used in the
earlier excess value tests shown in Table V. The second stage results for the 2SLS are presented in
Table VII.
After considering the endogeneity of the diversification decision, I reconsider the earlier re-
sults from Table V. The coe�cients on the disclosure measures (the row labeled as Disct) are
mostly insignificant though the models including Error or |Error| remain significant. This weakens
the support for Hypothesis 3. All of the coe�cients on the interaction term (the row labeled as
DiversXDisct) remain insignificant except the one on DiversXError is positive and significant at
the 5% level. Since Error is an inverse measure of the informativeness of disclosure, the positive
coe�cient indicates that less accurate forecasts by diversified firms is correlated with higher excess
value. Overall, these findings show that a lack of support for Hypothesis 4 remains after considering
endogeneity of the diversification decision.
27
V. Conclusion
In disclosing information to the public that is not mandatory, a diversified firm has a choice:
provide segment-level details or provide aggregate information. Focused firms do not enjoy this
option. A focused firm disclosure can be more accurately allocated to a particular business or
industry allowing competitors to react more readily. This situation creates the potential for addi-
tional competitive pressures su↵ered by focused firms that are not incurred by diversified firms. If
the proprietary cost of voluntary disclosures hypothesis posited in Verrecchia (1983) holds, focused
firms could refrain from voluntary disclosures without the fear of incurring a market discount,
resulting in a lower propensity to provide voluntary disclosure for focused firms.
Using voluntary disclosures from the FirstCall Company Issued Guidance database, I show that
focused firms are less likely than diversified firms to issue a forecast. Even after controlling for other
variables that are known to a↵ect the issuance of a forecast (e.g., recent performance, size, analyst
coverage), a focused firm is less likely to issue a forecast than a diversified firm. Most measures of
competitive pressure a↵ect the likelihood of providing guidance in the expected direction. However,
the inclusion of measures of competitive pressure does not remove the significance of the diversifi-
cation status of the firm. Focused firms remain less likely to disclose. Finally, a changes regression
reveals that firms changing from focused to diversified tend to increase disclosure.
I use regressions of excess value on disclosure measures and diversification proxies to test whether
diversified firms are more likely to disclose because they have more to gain from lowering their
increased information asymmetry relative to focused firms. The results show that more informative
disclosures do tend to be positively associated with excess value, but there is limited evidence of a
di↵erential e↵ect for diversified firms. These results generally indicate that diversified firms do not
gain more from disclosure.
This study provides ample indication that further study of the voluntary (and mandatory)
disclosure environment is warranted. Regulation that is written to consider the competitive dis-
advantage of a diversified firm disclosing segment information should also consider the fact that
focused firms are always revealing their “segment” information in full.
28
Appendices
A Variable Descriptions
Variable Definition
Forecast Dummy variable equal to 1 if the firm o↵ered guidance and 0 other-wise
NForecast Number of forecasts provided by a firm per fiscal yearLead Elapsed days from guidance announcement to fiscal period endLeadAct Elapsed days from guidance announcement to the earnings release
dateSpec The specificity of the guidance: 1 is qualitative; 2 is open-ended; 3
is range; and 4 is pointError Di↵erence between the forecast and actual earnings per share, nor-
malized by the most recent end-of-quarter share price, multiplied by100, and winsorized at 1%
|Error| The absolute value of ErrorDivers Dummy variable equal to 1 if firm has multiple business segments
and 0 otherwiseEntropy A measure of firm diversification based on the dispersion of sales
across segmentsConc4Firm Proportion of sales in a three-digit SIC code industry coming from
the top four producers by salesHI Firm-level sales-based Herfindahl Index at the 3-digit SIC code levelSpeedAdj Speed of abnormal profit adjustment as calculated in Eq. 2EarnV ol Standard deviation of 12 quarters of earnings measured at the end
of the fiscal period before the management forecast date winsorizedat 1%
MB Log of the ratio of market value of equity to the book value of equityMShare Firm three-digit SIC code industry sales market share as a percentile
rankPMargin Firm three-digit SIC code industry profit margin (EBIT/Sales) win-
sorized at 1% as a percentile rankRD Research and development yearly expense over total assetsHIwtd Weighted average firm Herfindahl Index using segment sales at the
3-digit SIC code levelMShareSeg Within-firm sales-weighted three-digit segment SIC code industry
sales market share as a percent rank, scaled to 0–100V olatility Standard deviation of monthly market-adjusted returns over the 36
months before the management forecastSize Log of total yearly assetsNegEarn Dummy variable equal to 1 if earnings for a given period are negative
(Continues on the next page.)
29
(Variable descriptions continued)
Variable Definition
NegEarnG Dummy variable equal to 1 if earnings growth (the di↵erence in earn-ings) is negative
ROE Return on equity, calculated as earnings over book equity, winsorizedat 2%
LitInd Dummy equal to 1 if the firm is in an industry that is prone tolitigation: SIC=2833–2836, 8731–8734, 3570–3577, 7370–7374, 3600–3674, and 5200–5961
NumEst Number of analyst with active estimates before the release of themanagement forecast
Dispersion Standard deviation of active estimates before the release of the man-agement forecast winsorized at 1%
RegFD Dummy variable equal to 1 if the management forecast date is afterOctober 23, 2000
SFAS131 Dummy variable equal to 1 if the subject fiscal period end of theforecast is after December 15, 1998
PSDIV For each firm and three-digit SIC code industry, the sales-weightedaverage proportion of sales coming from diversified firms excludingthe subject firm
MinInt Minority interest dummy indicating whether the firm has ...
30
B Excess Value Definition
To calculate excess value (EV ) I use the following formulas (Berger and Ofek, 1995, page 60):
I(V ) =nX
i=1
AIi ⇤ (Indi(V
AI)mf )
EV = ln(V/I(V ))
where
I(V ) = imputed value,V = firm total capital (market value of equity at the end of the calendar year t
plus book value of debt at the end of the firm fiscal year t),AI = accounting item (sales at the end of the firm fiscal year t),Indi(
VAI )mf = ratio of total capital to an accounting item for the median focused firm in
the same industry as segment i,n = the number of segments in segment i’s firm at the end of the firm fiscal year
t.The matched segment median value comes from the finest SIC code level (2-, 3-, or 4-digit) withat least five focused firms.
31
C FirstCall Summary Statistics
Table C1: FirstCall Guidance Statistics by Year
The following table provides summary statistics for annual earnings forecasts from the FirstCall Company IssuedGuidelines database tabulated by the firm fiscal year at the time of the forecast. The data represent forecasts ofearnings per share for U.S. common stock in U.S. dollars. Lead (LeadAct) is the di↵erence in days between theannouncement date and the fiscal period end (actual earnings disclosure date). Spec is the specificity of the forecastcoded numerically as follows: qualitative = 1; open-ended = 2; range = 3; and point = 4. Error is di↵erence betweenthe forecasted and actual earnings per share, normalized by the most recent end-of-quarter share price, multiplied by100, and winsorized at 1%. For more details refer to the variable definitions in Appendix A.
Mean
Year N Lead LeadAct Spec Error
1994 34 193 238 3.4 2.381995 226 175 212 3.4 2.261996 333 196 232 3.5 2.161997 516 183 226 3.4 1.991998 832 208 248 3.5 1.861999 1,074 219 260 3.3 1.632000 1,193 221 262 3.3 2.342001 2,643 225 265 3.2 2.472002 3,646 219 258 3.1 1.662003 4,096 215 257 3.1 1.382004 4,498 217 261 3.1 0.932005 4,500 212 258 3.1 0.872006 4,729 206 251 3.1 0.642007 4,498 204 248 3.1 0.572008 4,442 205 248 3.1 0.942009 3,639 200 242 3.1 0.462010 3,549 188 230 3.1 -0.322011 474 160 199 3.1 -0.32
All Years 44,922 208 251 3.1 1.00
32
D Correlation of Proprietary Cost Measures
Table D1: Correlation Matrix of Proprietary Cost Measures
This table shows the Pearson correlation coe�cients for the various proprietary cost proxies considered. The data span 1994–2011 and are derived fromCompustat firm- and segment-level databases. Variables are described in Appendix A. The lower triangle shows the correlations coe�cients using apairwise method, while the upper triangle shows the coe�cients using a list-wise method. Subscripts indicate the fiscal year of measurement. Starsindicate a statistically significant correlation at the levels of 10% (*), 5% (**), and 1% (***).
Industry-Level Firm-Level Segment-Based
Conc4Firm HI SpeedAdj MB MShare PMargin RD HIwtd MShareSeg
Conc4Firm 1.000 0.842⇤⇤⇤ -0.435⇤⇤⇤ -0.018⇤⇤⇤ -0.020⇤⇤⇤ -0.016⇤⇤⇤ -0.043⇤⇤⇤ 0.042⇤⇤⇤ 0.178⇤⇤⇤
HI 0.842⇤⇤⇤ 1.000 -0.314⇤⇤⇤ -0.012⇤⇤⇤ -0.018⇤⇤⇤ -0.017⇤⇤⇤ -0.065⇤⇤⇤ 0.031⇤⇤⇤ 0.132⇤⇤⇤
SpeedAdj -0.435⇤⇤⇤ -0.319⇤⇤⇤ 1.000 0.017⇤⇤⇤ 0.018⇤⇤⇤ 0.008⇤⇤ 0.205⇤⇤⇤ -0.058⇤⇤⇤ -0.139⇤⇤⇤
MB -0.018⇤⇤⇤ -0.011⇤⇤⇤ 0.017⇤⇤⇤ 1.000 -0.002 -0.000 -0.002 -0.001 -0.003MShare -0.020⇤⇤⇤ -0.020⇤⇤⇤ 0.019⇤⇤⇤ -0.002 1.000 0.754⇤⇤⇤ -0.088⇤⇤⇤ -0.001 0.255⇤⇤⇤
PMargin -0.017⇤⇤⇤ -0.018⇤⇤⇤ 0.008⇤⇤ -0.000 0.750⇤⇤⇤ 1.000 -0.122⇤⇤⇤ -0.001 0.222⇤⇤⇤
RD3 -0.039⇤⇤⇤ -0.063⇤⇤⇤ 0.203⇤⇤⇤ -0.002 -0.087⇤⇤⇤ -0.122⇤⇤⇤ 1.000 -0.024⇤⇤⇤ -0.070⇤⇤⇤
HIwtd 0.060⇤⇤⇤ 0.065⇤⇤⇤ -0.062⇤⇤⇤ -0.001 -0.003 -0.001 -0.024⇤⇤⇤ 1.000 0.036⇤⇤⇤
MShareSeg 0.284⇤⇤⇤ 0.360⇤⇤⇤ -0.160⇤⇤⇤ -0.003 0.282⇤⇤⇤ 0.253⇤⇤⇤ -0.069⇤⇤⇤ 0.063⇤⇤⇤ 1.000
33
E SFAS 131 E↵ects on Change in Corporate Form
34
Table E1: Firms Changing Corporate Form by Year
The following table provides a frequency table with associated percentages of the total for the tabulation of �Diversby fiscal year. �Divers takes four forms based on status change from t� 1 to t: diversified to focused (“Focusing”),focused to focused (“FocFoc”), diversified to diversified (“DivDiv”), and focused to diversified (“Diversifying”).
Focusing FocFoc DivDiv Diversifying TotalYear %Total %Total %Total %Total %Total
1995 69 3,343 968 63 4,4431.6% 75.2% 21.8% 1.4% 100.0%
1996 64 3,446 911 46 4,4671.4% 77.1% 20.4% 1.0% 100.0%
1997 75 3,427 880 56 4,4381.7% 77.2% 19.8% 1.3% 100.0%
1998 64 2,870 784 549 4,2671.5% 67.3% 18.4% 12.9% 100.0%
1999 82 2,459 1,177 309 4,0272.0% 61.1% 29.2% 7.7% 100.0%
2000 84 2,352 1,318 125 3,8792.2% 60.6% 34.0% 3.2% 100.0%
2001 92 2,214 1,230 105 3,6412.5% 60.8% 33.8% 2.9% 100.0%
2002 96 2,293 1,193 81 3,6632.6% 62.6% 32.6% 2.2% 100.0%
2003 84 2,242 1,181 53 3,5602.4% 63.0% 33.2% 1.5% 100.0%
2004 68 2,206 1,148 76 3,4981.9% 63.1% 32.8% 2.2% 100.0%
2005 68 2,152 1,105 51 3,3762.0% 63.7% 32.7% 1.5% 100.0%
2006 68 2,158 1,072 53 3,3512.0% 64.4% 32.0% 1.6% 100.0%
2007 53 2,143 1,024 56 3,2761.6% 65.4% 31.3% 1.7% 100.0%
2008 60 2,052 997 69 3,1781.9% 64.6% 31.4% 2.2% 100.0%
2009 45 2,067 1,000 62 3,1741.4% 65.1% 31.5% 2.0% 100.0%
2010 51 2,002 1,013 43 3,1091.6% 64.4% 32.6% 1.4% 100.0%
2011 50 1,912 962 53 2,9771.7% 64.2% 32.3% 1.8% 100.0%
Total 1,173 41,338 17,963 1,850 62,3241.9% 66.3% 28.8% 3.0% 100.0%
35
References
Aboody, David, and Ron Kasznik, 2000, CEO stock option awards and the timing of corporate voluntarydisclosures, Journal of Accounting and Economics 29, 73–100.
Ahn, Seoungpil, and David J Denis, 2004, Internal Capital Markets and Investment Policy: Evidence fromCorporate Spino↵s, Journal of Financial Economics 71, 489–516.
Ajinkya, Bipin B, Sanjeev Bhojraj, and Partha Sengupta, 2005, The Association between Outside Directors,Institutional Investors and the Properties of Management Earnings Forecasts, Journal of AccountingResearch 43, 343–376.
Anilowski, Carol, Mei Feng, and Douglas J Skinner, 2007, Does earnings guidance a↵ect market returns?The nature and information content of aggregate earnings guidance, Journal of Accounting and Economics44, 36–63.
Axelson, Ulf, 2007, Security Design with Investor Private Information, Journal of Finance 62, 2587–2632.
Baginski, Stephen P, Edward J Conrad, and John M Hassell, 1993, The E↵ects of Management ForecastPrecision on Equity Pricing and on the Assessment of Earnings Uncertainty., The Accounting Review 68,913–927.
Baginski, Stephen P, John M Hassell, and Michael D Kimbrough, 2002, The E↵ect of Legal Environmenton Voluntary Disclosure: Evidence from Management Earnings Forecasts Issued in U.S. and CanadianMarkets., The Accounting Review 77, 25–50.
Bamber, Linda Smith, and Youngsoon Susan Cheon, 1998, Discretionary management earnings forecastdisclosures: Antecedents and outcomes associated with forecast venue and forecast specificity choices,Journal of Accounting Research 36, 167–190.
Barry, Christopher B, and Stephen J Brown, 1985, Di↵erential Information and Security Market Equilibrium,Journal of Financial and Quantitative Analysis 20, 407–422.
Bens, Daniel A, and Steven J Monahan, 2004, Disclosure Quality and the Excess Value of Diversification,Journal of Accounting Research 42, 691–730.
Berger, Philip G, and Rebecca N Hann, 2007, Segment Profitability and the Proprietary and Agency Costsof Disclosure, The Accounting Review 82, 869–906.
Berger, Philip G, and Eli Ofek, 1995, Diversification’s E↵ect on Firm Value, Journal of Financial Economics37, 39–65.
Bernhardt, Dan, and Murillo Campello, 2007, The Dynamics of Earnings Forecast Management, Review ofFinance 11, 287–324.
Bhattacharya, Sudipto, and Jay R Ritter, 1983, Innovation and Communication: Signalling with PartialDisclosure, Review of Economic Studies 50, 331–346.
Botosan, Christine A, and Marlene A Plumlee, 2002, A Re-Examination of Disclosure Level and the ExpectedCost of Equity Capital, Journal of Accounting Research 40, 21–40.
Botosan, Christine A, Marlene A Plumlee, and Yuan Xie, 2004, The Role of Private Information PrecisionIn Determining Cost of Equity Capital, Review of Accounting Studies 9, 000–000.
Botosan, Christine A, and Mary Stanford, 2005, Managers’ Motives to Withhold Segment Disclosures andthe E↵ect of SFAS No. 131 on Analysts’ Information Environment., The Accounting Review 80, 751–771.
36
Brockman, Paul, Inder K Khurana, and Xiumin Martin, 2008, Voluntary Disclosures around Actual ShareRepurchases, Journal of Financial Economics 89, 175–191.
Brockman, Paul, Xiumin Martin, and Andy Puckett, 2008, Voluntary Disclosures and the Exercise of CEOStock Options, Working paper, .
Brown, Stephen, and Stephen A Hillegeist, 2007, How Disclosure Quality A↵ects the level of InformationAsymmetry, Review of Accounting Studies 12, 443–477.
Campa, Jose Manuel, and Simi Kedia, 2002, Explaining the Diversification Discount, Journal of Finance 57,1731–1762.
Cao, Yan, Charles E Wasley, and Joanna S Wu, 2007, Soft-Talk Management Cash Flow Forecasts: Verifia-bility, Credibility and Stock Price E↵ects, Working paper, .
Chuk, Elizabeth, Dawn A Matsumoto, and Gregory S Miller, 2013, Assessing methods of identifying man-agement forecasts: CIG vs. researcher collected, Journal of Accounting and Economics 55, 23–42.
Clarke, Jonathan E, C Edward Fee, and Shawn E Thomas, 2004, Corporate Diversification And AsymmetricInformation: Evidence From Stock Market Trading Characteristics, Journal of Corporate Finance 10,105–129.
Coles, Je↵rey L, Uri Loewenstein, and Jose Suay, 1995, On Equilibrium Pricing under Parameter Uncertainty,Journal of Financial and Quantitative Analysis 30, 347.
Coller, Maribeth, and Teri Lombardi Yohn, 1997, Management Forecasts and Information Asymmetry: AnExamination of Bid-Ask Spreads, Journal of Accounting Research 35, 181–191.
Darrough, Masako N, and Neal M Stoughton, 1990, Financial Disclosure Policy In An Entry Game, Journalof Accounting and Economics 12, 219–243.
Diamond, Douglas W, 1985, Optimal Release of Information By Firms, Journal of Finance 40, 1071–1094.
Diamond, Douglas W, and Robert E Verrecchia, 1991, Disclosure, Liquidity, and the Cost of Capital, Journalof Finance 46, 1325–1359.
Dimitrov, Valentin, and Sheri Tice, 2006, Corporate Diversification and Credit Constraints: Real E↵ectsacross the Business Cycle, Review of Financial Studies 19, 1465–1498.
Dye, Ronald A, 1990, Mandatory Versus Voluntary Disclosures: The Cases of Financial and Real External-ities, The Accounting Review 65, 1–24.
Dye, Ronald A, and Sri S Sridhar, 1995, Industry-Wide Disclosure Dynamics, Journal of Accounting Research33, 157–174.
Financial Accounting Standards Board, 1976, Financial Reporting for Segments of a Business, Statement ofFinancial Accounting Standards.
Financial Accounting Standards Board, 1997, Disclosures about Segments of an Enterprise and RelatedInformation, Statement of Financial Accounting Standards.
Gennotte, Gerard, and Brett Trueman, 1996, The Strategic Timing Of Corporate Disclosures, Review ofFinancial Studies 9, 665–690.
Hadlock, Charles J, Michael Ryngaert, and Shawn E Thomas, 2001, Corporate Structure and Equity O↵er-ings: Are There Benefits to Diversification?, Journal of Business 74, 613–635.
37
Hann, Rebecca N, Maria Ogneva, and Oguzhan Ozbas, 2013, Corporate Diversification and the Cost ofCapital, Journal of Finance 68, 1961–1999.
Harris, Mary S, 1998, The Association between Competition and Managers’ Business Segment ReportingDecisions, Journal of Accounting Research 36, 111–128.
Healy, Paul M, and Krishna G Palepu, 2001, Information Asymmetry, Corporate Disclosure, and the CapitalMarkets: A Review of the Empirical Disclosure Literature, Journal of Accounting and Economics 31, 405–440.
Huang, Ying, Ross Jennings, and Yong Yu, 2016, Product Market Competition and Managerial Disclosureof Earnings Forecasts: Evidence from Import Tari↵ Rate Reductions, The Accounting Review.
Hund, John, Donald Monk, and Sheri Tice, 2010, Uncertainty about Average Profitability and the Diversi-fication Discount, Journal of Financial Economics 96, 463–484.
Hutton, Amy P., Gregory S Miller, and Douglas J Skinner, 2003, The Role of Supplementary Statementswith Management Earnings Forecasts, Journal of Accounting Research 41, 867.
Jacquemin, Alexis P, and Charles H Berry, 1979, Entropy Measure of Diversification and Corporate Growth,Journal of Industrial Economics 27, 359–369.
Jensen, Michael C, 1986, Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers, AmericanEconomic Review 76, 323–329.
Kothari, S.P., Susan Shu, and Peter D. Wysocki, 2009, Do Managers Withhold Bad News?, Journal ofAccounting Research forthcoming, 000–000.
Krishnaswami, Sudha, and Venkat Subramaniam, 1999, Information Asymmetry, Valuation, and the Corpo-rate Spin-o↵ Decision, Journal of Financial Economics 53, 73–112.
Lambert, Richard, Christian Leuz, and Robert E Verrecchia, 2007, Accounting Information, Disclosure, andthe Cost of Capital, Journal of Accounting Research 45, 385–420.
Lamont, Owen A, 1997, Cash Flow and Investment: Evidence from Internal Capital Markets, Journal ofFinance 52, 83–109.
Lamont, Owen A, and Christopher Polk, 2001, The Diversification Discount: Cash Flows Versus Returns,Journal of Finance 56, 1693–1722.
Lamont, Owen A, and Christopher Polk, 2002, Does Diversification Destroy Value? Evidence from theIndustry Shocks, Journal of Financial Economics 63, 51–77.
Lang, Mark H, and Russell Lundholm, 1996, Corporate Disclosure Policy and Analyst Behavior, The Ac-counting Review 71, 467–492.
Lansford, Benjamin, Baruch I Lev, and Jenny Tucker, 2010, Causes and Consequences of DisaggregatingEarnings Guidance, Working paper, .
Li, Xi, 2010, The impacts of product market competition on the quantity and quality of voluntary disclosures,Review of Accounting Studies 15, 663–711.
Matsumoto, Dawn A, 2002, Management’s Incentives to Avoid Negative Earnings Surprises, The AccountingReview 77, 483–514.
38
Miller, Gregory S, 2002, Earnings Performance and Discretionary Disclosure, Journal of Accounting Research40, 173–204.
Mitton, Todd, and Keith Vorkink, 2008, Why Do FirmsWith Diversification Discounts Have Higher ExpectedReturns?, Journal of Financial and Quantitative Analysis 45, 1367–1390.
Myers, S C, and N S Majluf, 1984, Corporate financing and investment decisions when firms have informationthat investors do not have, Journal of Financial Economics.
Nichols, D. Craig, 2009, Proprietary Costs and Other Determinants of Nonfinancial Disclosures, Workingpaper, .
Noe, Christopher F., 1999, Voluntary Disclosures and Insider Transactions, Journal of Accounting and Eco-nomics 27, 305–326.
Rajan, Raghuram, Henri Servaes, and Luigi Zingales, 2000, The Cost of Diversity: The DiversificationDiscount and Ine�cient Investment, Journal of Finance 55, 35–80.
Roberts, Michael R, and Toni MWhited, 2012, Endogeneity in Empirical Corporate Finance, Working paper,.
Rogers, Jonathan L., and Phillip C. Stocken, 2005, Credibility of Management Forecasts, The AccountingReview 80, 1233–1260.
Sanzhar, Sergey V, 2006, Discounted But Not Diversified: Organizational Structure and ConglomerateDiscount, Working paper, .
Scharfstein, David S, and Jeremy Stein, 2000, The Dark Side of Internal Capital Markets: Divisional Rent-seeking and Ine�cient Investment, Journal of Finance 55, 2537–2564.
Schrand, Catherine M, and Beverly R Walther, 1998, Discretionary Reporting of Earnings Components,Working paper, .
Securities and Exchange Commission, 2000, Selective Disclosure and Insider Trading, Final Rulehttp://www.sec.gov/rules/final/33-7881.htm.
Shleifer, Andrei, and Robert W Vishny, 1989, Management Entrenchment, Journal of Financial Economics25, 123–139.
Thomas, Shawn E, 2002, Firm Diversification and Asymmetric Information: Evidence from Analysts’ Fore-casts and Earnings Announcements, Journal of Financial Economics 64, 373–396.
Verrecchia, Robert E, 1983, Discretionary Disclosure, Journal of Accounting and Economics 5, 179–194.
Villalonga, Belen, 2004, Does Diversification Cause the “Diversification Discount”?, Financial Management33, 5–27.
Wang, Isabel Yanyan, 2007, Private Earnings Guidance and Its Implications for Disclosure Regulation, TheAccounting Review 82, 1299–1332.
39
Tables
Table I: Descriptive Statistics—Forecasts and Diversification
The following table provides summary statistics for annual company issued earnings per share forecasts for the sample meeting screening requirementstabulated by fiscal year at the time of the forecast. Panel A provides results for single-segment (“focused”) firms, and Panel B provides results formultiple-segment (“diversified”) firms. The data represent forecasts of earnings per share for U.S. common stock in U.S. dollars. “Nonforecasting”indicates that a firm is not present in the FirstCall Company Issued Guidance database, but is present in Compustat firm and segment files. NForecast
is the number of forecasts per firm per fiscal year.
Panel A: Focused Firms Panel B: Diversified Firms
Nonforecasting Forecasting Count Mean Nonforecasting Forecasting Count MeanYear Firms Firms % Forecasts NForecast Firms Firms % Forecasts NForecast
1995 3,914 123 3 149 1.2 1,068 42 4 55 1.31996 3,895 210 5 264 1.3 996 64 6 81 1.31997 3,954 312 7 422 1.4 960 80 8 117 1.51998 3,092 337 10 490 1.5 1,250 223 15 428 1.91999 2,535 385 13 614 1.6 1,341 332 20 667 2.02000 2,482 369 13 636 1.7 1,227 357 23 691 1.92001 2,103 587 22 1,275 2.2 998 455 31 1,204 2.62002 1,973 587 23 1,700 2.9 914 450 33 1,548 3.42003 1,916 608 24 2,009 3.3 873 448 34 1,659 3.72004 1,829 647 26 2,289 3.5 864 447 34 1,810 4.02005 1,955 572 23 2,158 3.8 824 406 33 1,788 4.42006 1,923 599 24 2,298 3.8 798 445 36 1,978 4.42007 1,889 576 23 2,217 3.8 769 404 34 1,844 4.62008 1,909 499 21 2,202 4.4 761 390 34 1,864 4.82009 1,852 422 19 1,804 4.3 814 334 29 1,620 4.92010 1,804 417 19 1,849 4.4 809 320 28 1,621 5.12011 1,811 391 18 1,152 2.9 746 326 30 1,022 3.1
All Years 40,836 7,641 16 23,528 3.1 16,012 5,523 26 19,997 3.6
40
Table II: Forecasting Status Summary Statistics
This table presents statistics for variables of interest by forecasting status for the firm-level sample for the fiscal years1995–2011. A firm is considered “Focused” (“Diversified”) in a particular fiscal year if it has only one (more thanone) business/operating segment. Variables are described in Appendix A. The “Di↵erence” column indicates thedi↵erence between focused and diversified firm means, and asterisks indicate if the di↵erence is significant at the 10%(*), 5% (**), or 1% (***) level.
Focused DiversifiedMean Mean Di↵erence Observations
Forecast 0.158 0.256 -0.099*** 70,012NForecast 0.822 1.429 -0.608*** 70,012HI 0.158 0.202 -0.044*** 70,012SpeedAdj 0.734 0.640 0.094*** 70,011PM 0.476 0.567 -0.091*** 69,508RD 2.928 0.087 2.840** 70,012HIwtd 0.122 0.151 -0.029** 70,012MB 6.339 6.114 0.225 69,882Earnings Vol. 0.041 0.027 0.015*** 58,075Stock Vol. 0.155 0.125 0.029*** 70,012ln(Assets) 5.285 6.677 -1.392*** 70,011NegEarn 0.342 0.204 0.138*** 70,012NegEarnG 0.409 0.390 0.019*** 70,012exannret 0.081 0.071 0.011 70,012LitInd 0.337 0.172 0.166*** 70,012Analysts 5.360 6.620 -1.261*** 36,830AE Disp. 0.248 0.106 0.142 27,860
41
Table III: Forecast Issuance—Diversification Dummy
This table contains the coe�cients from a probit regression where the binary outcome is whether or not a firm issueda management forecast in a given fiscal year. Data for management forecasts are derived from the FirstCall CompanyIssued Guidance database for the time period 1995–2011. Each column heading indicates the model. Variablesare described in Appendix A and “L.” prepended to the variable name indicates that it is lagged one year. Theparentheses contain standard errors adjusted for year clustering. Asterisks indicate if the coe�cient is significant atthe 10% (*), 5% (**), or 1% (***) level.
1 2 3 4 5
L.Divers 0.374*** 0.183*** 0.183*** 0.218*** 0.290***(0.015) (0.013) (0.013) (0.014) (0.022)
L.ln(Assets) 0.149*** 0.150*** 0.177*** 0.098***(0.012) (0.012) (0.014) (0.012)
L.MB -0.000 -0.000* -0.000** -0.000(0.000) (0.000) (0.000) (0.000)
L.Stock Vol. -0.266 -0.250 -0.156 -0.264(0.424) (0.428) (0.410) (0.683)
L.ROE 0.347*** 0.318*** 0.345*** 0.341***(0.028) (0.028) (0.037) (0.059)
L.NegEarn -0.318*** -0.313*** -0.312*** -0.413***(0.041) (0.041) (0.040) (0.054)
L.NegEarnG -0.041 -0.040 -0.031 0.044(0.028) (0.028) (0.032) (0.043)
L.LitInd 0.368*** 0.369*** 0.377*** 0.384***(0.027) (0.027) (0.028) (0.044)
L.SpeedAdj 0.068* 0.030 0.098**(0.027) (0.038) (0.037)
L.RD -0.036* -0.031 -0.053(0.018) (0.019) (0.033)
L.Earnings Vol. -0.182 -0.396(0.251) (0.449)
L.Analysts 0.016**(0.006)
L.AE Disp. -0.010(0.021)
Constant -0.971*** -1.815*** -1.863*** -1.993*** -1.113***(0.078) (0.171) (0.186) (0.196) (0.209)
N 54,231 54,135 54,134 45,171 18,257Pseudo R
2 0.015 0.104 0.105 0.124 0.085% Correct 80 80 80 79 65
42
Table IV: Forecast Issuance—Diversification Measures
This table contains the coe�cients from a probit regression where the binary outcome is whether or not a firm issueda management forecast in a given fiscal year. Data for management forecasts are derived from the FirstCall CompanyIssued Guidance database for the time period 1995–2011. Each column heading indicates the model. Variablesare described in Appendix A, and “L.” prepended to the variable name indicates that it is lagged one year. Theparentheses contain standard errors adjusted for year clustering. Asterisks indicate if the coe�cient is significant atthe 10% (*), 5% (**), or 1% (***) level.
1 2 3 4 5 6
L.SegN 0.156*** 0.036*** 0.122***(0.007) (0.008) (0.013)
L.Entropy 0.455*** 0.172*** 0.367***(0.023) (0.030) (0.044)
L.ln(Assets) 0.155*** 0.099*** 0.152*** 0.098***(0.013) (0.013) (0.013) (0.014)
L.MB -0.000 -0.000 -0.000 -0.000(0.000) (0.000) (0.000) (0.000)
L.Stock Vol. -0.249 -0.252 -0.241 -0.243(0.433) (0.683) (0.429) (0.680)
L.ROE 0.315*** 0.344*** 0.318*** 0.346***(0.028) (0.059) (0.028) (0.059)
L.NegEarn -0.317*** -0.415*** -0.315*** -0.414***(0.042) (0.055) (0.042) (0.055)
L.NegEarnG -0.039 0.046 -0.040 0.046(0.028) (0.042) (0.028) (0.043)
L.LitInd 0.354*** 0.375*** 0.361*** 0.378***(0.028) (0.046) (0.028) (0.046)
L.SpeedAdj 0.050 0.088* 0.057* 0.088*(0.028) (0.038) (0.027) (0.039)
L.RD -0.037* -0.054 -0.037* -0.054(0.018) (0.034) (0.018) (0.034)
L.Earnings Vol. -0.423 -0.407(0.447) (0.445)
L.Analysts 0.015** 0.015**(0.006) (0.006)
L.AE Disp. -0.011 -0.011(0.020) (0.021)
Constant -1.088*** -1.869*** -1.192*** -0.938*** -1.841*** -1.083***(0.080) (0.187) (0.212) (0.077) (0.189) (0.218)
N 54,231 54,134 18,257 54,231 54,134 18,257Pseudo R
2 0.012 0.102 0.083 0.013 0.104 0.084% Correct 80 80 64 80 80 64
43
Table V: Excess Value on Diversification and Disclosure Ranking
This table presents results from an ordinary least squares regression with the dependent variable of firm excess valueas described in Appendix B. Each column heading indicates the disclosure measure used in the various models, andthe associated coe�cient for each disclosure measure is labeled Disct. Forecast is an indicator variable equal toone if a firm provides annual EPS guidance during a fiscal year. The disclosure measures used in columns 2–6 arethe percentile rank within three-digit SIC code industry of the yearly disclosure measure in the column heading.The last results column is the same as the first except a matched sample of forecasting and non-forecasting firms isused. Matches are determined using coarsened exact matching (CEM) on year, two-digit SIC code industry, primaryexchange, and market value of equity. Variables are described in Appendix A. Year fixed e↵ects are included ineach model. Data for management forecasts are derived from the FirstCall Company Issued Guidance database overthe period 1995–2011. Standard errors are clustered by year, and resulting t-statistics are presented in parentheses.Asterisks indicate if the coe�cient is significant at the 10% (*), 5% (**), or 1% (***) level.
Percentile Rank CEM
Forecast NForecast Lead Spec Error |Error| Forecast
Divers -0.301*** -0.248*** -0.202*** -0.183*** -0.246*** -0.235*** -0.221***(-32.78) (-12.40) (-10.78) (-8.79) (-11.37) (-11.62) (-18.33)
Disct 0.070** -0.015 0.116*** 0.105* -0.182*** -0.260*** 0.105***(3.22) (-0.43) (4.09) (2.80) (-5.07) (-7.14) (4.70)
DiversXDisct 0.175*** 0.075 -0.035 -0.091* 0.081* 0.053 0.072***(18.62) (1.79) (-0.89) (-2.24) (2.20) (1.18) (4.73)
ln(Assets) -0.020* 0.057*** 0.056*** 0.058*** 0.049*** 0.045*** -0.011(-2.85) (8.35) (8.24) (8.28) (6.73) (6.05) (-2.08)
Capx/Sales -0.001 0.021 0.021 0.020 0.154* 0.142 0.001(-0.39) (0.61) (0.63) (0.59) (2.18) (2.01) (0.97)
Profit Margin -0.001** -0.027*** -0.027*** -0.027*** 0.030 0.025 -0.001(-3.09) (-4.54) (-4.48) (-4.54) (1.12) (0.91) (-1.94)
Constant 0.147** -0.287*** -0.335*** -0.338*** -0.182*** -0.123* 0.035(3.75) (-6.05) (-6.62) (-7.63) (-4.14) (-2.51) (1.09)
N 66,645 13,023 13,023 13,022 11,469 11,469 54,626Adj. R2 0.039 0.047 0.048 0.048 0.047 0.053 0.028
44
Table VI: Forecast Issuance with Instrumented Entropy
This table presents results after consideration of the endogeneity of the level of diversification of the firm, Entropy.The coe�cients from a probit regression where the binary outcome is whether or not a firm issued a managementforecast in a given fiscal year using instrumentation for Entropy are presented below. Data for management forecastsare derived from the FirstCall Company Issued Guidance database for the time period 1995–2011. Each columnheading indicates the model. Variables are described in Appendix A, and “L.” prepended to the variable nameindicates that it is lagged one year. The parentheses contain standard errors adjusted for firm clustering. Asterisksindicate if the coe�cient is significant at the 10% (*), 5% (**), or 1% (***) level.
1 2 3 4 5
L.Entropy 0.994*** 1.059*** 1.083*** 1.426*** 0.621(0.199) (0.232) (0.243) (0.272) (0.530)
L.ln(Assets) 0.098*** 0.098*** 0.112*** 0.072(0.009) (0.010) (0.011) (0.045)
L.MB -0.000 -0.000 -0.000*** -0.000(0.000) (0.000) (0.000) (0.000)
L.Stock Vol. -0.235 -0.212 -0.092 -0.164(0.379) (0.383) (0.338) (0.654)
L.ROE 0.319*** 0.291*** 0.301*** 0.299***(0.025) (0.025) (0.034) (0.061)
L.NegEarn -0.296*** -0.292*** -0.274*** -0.411***(0.053) (0.054) (0.056) (0.073)
L.NegEarnG -0.060* -0.059* -0.050 0.031(0.026) (0.026) (0.029) (0.040)
L.LitInd 0.420*** 0.409*** 0.436*** 0.389***(0.023) (0.023) (0.020) (0.041)
L.SpeedAdj 0.144*** 0.141** 0.090*(0.043) (0.053) (0.039)
L.RD -0.035* -0.030 -0.053(0.017) (0.017) (0.034)
L.Earnings Vol. -0.322 -0.576(0.279) (0.573)
L.Analysts 0.018(0.010)
L.AE Disp. -0.307***(0.076)
Constant -1.056*** -1.653*** -1.754*** -1.859*** -0.985***(0.090) (0.126) (0.146) (0.138) (0.112)
N 50,170 50,078 50,077 41,484 16,055% Correct 81 81 81 80 64
45
Table VII: Excess Value on Endogenous Diversification Dummy
This table presents results after consideration of the endogeneity of the diversification status of the firm, Divers.Using instruments for Divers and DiversXDisct in the first stage of a two-stage least squares approach, the resultsbelow show the coe�cients from the second stage regression of instrumented diversification status and disclosuremeasures on excess value. Each column heading indicates the disclosure measure used in the various models, andthe associated coe�cient for each disclosure measure is labeled Disct. Forecast is an indicator variable equal to oneif a firm provides annual EPS guidance during a fiscal year. The disclosure measures used in columns 2–6 are thepercentile rank within three-digit SIC code industry of the yearly disclosure measure in the column heading. Data formanagement forecasts are derived from the FirstCall Company Issued Guidance database for the time period 1995–2011. Variables are described in Appendix A. The parentheses contain standard errors adjusted for firm clustering.Asterisks indicate if the coe�cient is significant at the 10% (*), 5% (**), or 1% (***) level.
Percentile Rank
Forecast NForecast Lead Spec Error |Error|
Divers -0.390** -0.762** -0.664** -0.625** -0.990*** -0.807***(-3.08) (-3.25) (-2.71) (-2.68) (-3.81) (-3.66)
Disct 0.088 -0.192 -0.011 -0.007 -0.563*** -0.461***(1.43) (-1.26) (-0.08) (-0.04) (-3.79) (-3.33)
DiversXDisct 0.107 0.487 0.220 0.171 0.996** 0.554(0.66) (1.37) (0.69) (0.43) (2.65) (1.59)
ln(Assets) -0.020* 0.087*** 0.087*** 0.086*** 0.080*** 0.077***(-2.27) (5.25) (5.24) (5.20) (4.64) (4.48)
Capx/Sales -0.001 0.015 0.015 0.015 0.115 0.098(-0.36) (0.51) (0.49) (0.49) (1.62) (1.34)
Profit Margin -0.001** -0.028*** -0.029*** -0.029*** 0.014 0.007(-2.81) (-5.33) (-5.43) (-5.43) (0.46) (0.24)
Constant 0.175*** -0.271*** -0.339*** -0.341*** -0.079 -0.099(5.25) (-3.31) (-4.61) (-3.91) (-0.94) (-1.16)
N 61,853 11,423 11,423 11,422 9,963 9,963
46
Table VIII: Descriptive Statistics—Changes in Diversification and Changes in Forecasting
The following table provides a frequency table with associated percentages of the total for the two-way tabulationof �Divers and �Forecast. �Divers takes four forms based on status change from t � 1 to t: diversified to fo-cused (“Focusing”), focused to focused (“FocFoc”), diversified to diversified (“DivDiv”), and focused to diversified(“Diversifying”). �Forecast is divided into firms that “stop” forecasting (�Forecast = �1), firms with no change(�Forecast = 0), and firms that “start” forecasting (�Forecast = 1).
�Forecast (full sample) �Forecast (1998-2000)
Stop No Change Start Total Stop No Change Start Total
Focusing 35 942 45 1,022 11 180 13 204%Total 3.4% 92.2% 4.4% 100.0% 5.4% 88.2% 6.4% 100.0%
FocFoc 1,134 32,921 1,519 35,574 285 6,238 427 6,950%Total 3.2% 92.5% 4.3% 100.0% 4.1% 89.8% 6.1% 100.0%
DivDiv 576 14,642 779 15,997 148 2,539 269 2,956%Total 3.6% 91.5% 4.9% 100.0% 5.0% 85.9% 9.1% 100.0%
Diversifying 61 1,457 120 1,638 41 765 76 882%Total 3.7% 89.0% 7.3% 100.0% 4.7% 86.7% 8.6% 100.0%
Total 1,806 49,962 2,463 54,231 485 9,722 785 10,992%Total 3.3% 92.1% 4.5% 100.0% 4.4% 88.5% 7.1% 100.0%
47
Table IX: Changes in Forecasting Activity on Changes in Corporate Form
This table reports coe�cients from a regression of the change in various measures of forecasting activity on the change in corporate form. The columntitles indicate the dependent variable and the sample tested. The full sample covers the period 1995–2011 and the SFAS131 sample covers the years1998–2000. The first three control variables are indicators of the change in corporate form: focused to diversified (“Diversifying”), diversified to focused(“Focusing”), and diversified to diversified (“DivDiv”). Variables are described in Appendix A and “D.” prepended to the variable name indicates thatit is the change from t� 1 to t. The parentheses contain standard errors, and asterisks indicate if the coe�cient is significant at the 10% (*), 5% (**),or 1% (***) level.
Dep. Var. �NForecast �NForecast �NForecast �NForecast �Lead �Lead �Error �ErrorSample Full SFAS131 Full SFAS131 Full SFAS131 Full SFAS131
Diversifying 0.116** 0.128** 0.094* 0.115** 24.168** 39.224 -0.255 -0.866(0.043) (0.044) (0.043) (0.044) (7.353) (23.799) (0.210) (0.441)
Focusing 0.035 -0.054 0.048 -0.039 -0.136 -13.244 0.004 -0.484(0.054) (0.087) (0.054) (0.088) (7.582) (48.888) (0.204) (0.821)
DivDiv 0.011 0.077** 0.013 0.081** -0.632 -15.497 -0.257*** -0.644**(0.016) (0.027) (0.016) (0.027) (2.152) (13.020) (0.060) (0.240)
D.ln(Assets) 0.199*** 0.124*** 20.355*** 18.882 -0.398** -1.031*(0.023) (0.032) (4.757) (22.984) (0.133) (0.422)
D.MB 0.000 -0.000 -0.012 -1.079 -0.002 -0.005(0.000) (0.000) (0.069) (0.961) (0.002) (0.020)
D.Stock Vol. -0.039 0.220** -1.747 34.036 2.079*** 3.595(0.063) (0.080) (17.701) (96.965) (0.505) (2.043)
D.ROE 0.018 0.031 -7.040 -1.632 -1.755*** -1.433**(0.020) (0.029) (4.913) (23.933) (0.146) (0.484)
D.NegEarn -0.038 0.040 14.455*** 53.907** 1.192*** 0.890*(0.021) (0.033) (3.759) (20.687) (0.105) (0.404)
D.NegEarnG 0.058*** 0.071*** -1.184 -6.981 0.282*** 0.311(0.012) (0.019) (1.703) (9.667) (0.047) (0.172)
D.LitInd -0.130 -0.203 32.177 1.965 1.225* 0.691(0.101) (0.141) (18.580) (77.250) (0.511) (1.415)
Constant 0.052*** 0.088*** 0.035*** 0.066*** 0.699 8.595 0.171*** 0.827***(0.009) (0.015) (0.009) (0.016) (1.560) (10.349) (0.043) (0.194)
Adj. R2 0.000 0.001 0.002 0.005 0.006 0.014 0.106 0.103Observations 54,231 10,992 54,097 10,943 8,387 660 7,431 497
48
Figure
1:Percentag
eof
FirmsProvidingaForecast
This
figu
represents
thepercentage
offirm
sfrom
each
corp
orateform
catego
rythat
provideaforecast
within
agivenfiscal
year.
49