Professional Documents
Culture Documents
ABSTRACT
EBITDA is a commonly used performance measure for (i) valuation, (ii) debt contracting, and
(iii) executive compensation. The widespread use of EBITDA by stakeholders may induce man-
agers to focus their attention on EBITDA. Since EBITDA excludes various expenses, managers
who fixate on EBITDA may underweight the excluded expenses when determining their firms’
investments in capital and leverage levels. I find that managers who fixate on EBITDA overinvest
in capital and overlever their firm relative to their industry peers. These results are robust to alter-
native proxies for managers’ focus on EBITDA and alternative specifications. I also find that firms
whose managers focus on EBITDA have weaker operating performance, which is attributed to
higher depreciation expense. My primary proxy for managers’ focus on EBITDA is whether they
choose to disclose EBITDA in annual earnings announcements. I find that the use of EBITDA in
setting executive compensation, the prevalence of EBITDA estimates by analysts, and the use of
EBITDA-based covenants in firms’ debt contracts are all positively associated with the propensity
to disclose EBITDA in earnings announcements. I find weaker evidence of opportunistic motives
explaining EBITDA disclosure. These results are consistent with managers disclosing EBITDA to
portray to investors that it is a metric they seek to maximize. Overall, this study suggests that while
EBITDA is a widely used metric, there is a systematic cost to using this measure—it provides
managers with incentives to overinvest in capital and to acquire excessive debt.
* Accepted by Peter Clarkson. A previous version of this paper was titled “EBITDA Disclosure and Overinvestment in
Capital.” This paper is based on my first-year summer paper in the Columbia Business School doctoral program. I thank
Peter Clarkson, two anonymous reviewers, Dan Amiram, Robert Bloomfield, Brian Cadman, Ted Christensen (discus-
sant), Atif Ellahie, Trevor Harris, Yuan Ji, Chris Jones, Alon Kalay, Sok-Hyon Kang, Zach Kaplan (discussant), Sharon
Katz, Susan Kulp, Fred Lindahl, Nahum Melumad, Beatrice Michaeli, Doron Nissim, Sasha Peng, Steve Penman, James
Potepa, Haim Rosenbaum, Gil Sadka, Andy Schmidt, Steve Stubben, and workshop participants at Columbia University,
The George Washington University, University of Utah, 2012 Trans-Atlantic Doctoral Conference at London Business
School, 2013 EAA annual meeting, and the 2013 AAA annual meeting for their valuable comments. Any errors are
my own.
† Corresponding author.
Contemporary Accounting Research Vol. 36 No. 1 (Spring 2019) pp. 513–546 © CAAA
doi:10.1111/1911-3846.12387
514 Contemporary Accounting Research
principale variable de substitution à la concentration des gestionnaires sur le BAIIA est le choix de
publier ou non le BAIIA dans les annonces des résultats annuels. L’auteur note que l’utilisation du
BAIIA dans la détermination de la rémunération des cadres, la prévalence des estimations du
BAIIA proposées par les analystes et l’utilisation de clauses restrictives basées sur le BAIIA dans
les contrats d’emprunt des sociétés affichent toutes un lien positif avec la propension à publier le
BAIIA dans les annonces de résultats. Il relève des données moins probantes quant aux motiva-
tions opportunistes qui pourraient expliquer la publication du BAIIA. Ces résultats tendent à con-
firmer que les gestionnaires publieraient le BAIIA afin de signaler aux investisseurs leurs efforts
pour maximiser cet indicateur. Dans l’ensemble, l’étude semble indiquer que, même si le BAIIA
est un indicateur largement utilisé, son usage présente un coût systématique : il incite les gestion-
naires à surinvestir dans les immobilisations et à abuser de l’emprunt.
1. Introduction
Setting appropriate performance measures for managers is important for aligning the objectives of
managers with those of shareholders. Many analytical and empirical studies document the adverse
consequences of managers’ focus on inappropriate performance measures. In this study, I exam-
ine the adverse consequences of managers’ focus on EBITDA.
EBITDA is a commonly used performance measure for (i) firm valuation, (ii) debt contract-
ing, and (iii) executive compensation. The use of EBITDA by stakeholders and by compensation
committees in setting executive pay may lead managers to focus on EBITDA when making rou-
tine business decisions. EBITDA excludes various expenses used in determining bottom-line
GAAP earnings. The most significant expense excluded in calculating EBITDA is depreciation.
As a result, EBITDA does not account for the cost of investments in capital, since the expense
resulting from those investments—depreciation—is omitted in calculating EBITDA. To the extent
that managers focus on EBITDA, they may underweight the true cost of investments in capital
and overinvest. Furthermore, managers may overlever the firm to facilitate overinvestment in cap-
ital, since the direct cost of leverage—interest expense—is also excluded in calculating EBITDA. In
this study, I examine whether managers’ fixation on EBITDA is associated with overinvestment in
capital and overleverage.
I find that managers who focus on EBITDA invest in capital between 0.4 percent and 1.5
percent of total assets more than their industry peers.1 I also find that capital investments by firms
that focus on EBITDA are 46 percent higher than capital investments by a matched sample of
firms that do not focus on EBITDA. Additionally, I find that the interest expense (leverage) of
firms whose managers focus on EBITDA is between 0.5 percent and 1.6 percent (2.9 percent and
14.3 percent) higher than their industry peers. I do not find that tax avoidance by firms whose
managers focus on EBITDA is significantly different from their industry peers. Intuitively, man-
agers do not have incentives to overpay taxes since this type of overpayment would not directly
affect EBITDA. Furthermore, managers may use tax savings to invest in capital (Goldman 2016),
in which case, managers who focus on EBITDA may engage in tax avoidance to the same extent
as their industry peers.
I use firms’ disclosure of EBITDA in earnings announcements as a proxy for managers’
focus on EBITDA.2 Managers use earnings announcements to summarize and discuss their firms’
1. Similarly, in untabulated analyses, I find that the net change in intangible assets of firms whose managers focus on
EBITDA is between 1.3 percent and 3.1 percent of total assets higher than their industry peers.
2. I find that the proportion of my sample firms disclosing EBITDA increased from 17 percent in 2003 to 35 percent
in 2011. EBITDA disclosure is relatively persistent; 61 percent of firms in my sample never disclose EBITDA,
while 12 percent of firms in my sample consistently disclose EBITDA. Furthermore, while 15 percent of firms in
my sample switch to disclosing EBITDA, only 3 percent of firms stop disclosing EBITDA after consistently dis-
closing it. Lastly, 9 percent of firms in my sample exhibit a nonpersistent disclosure behavior (i.e., switching back
and forth between disclosing and not disclosing EBITDA).
performance. Therefore, earnings announcements may be a good setting for evaluating which per-
formance measures are important to managers.
I validate my proxy for managers’ focus on EBITDA by examining the determinants of
EBITDA disclosure in earnings announcements. I find evidence consistent with managers disclos-
ing EBITDA when their executive compensation is linked to EBITDA and in response to inves-
tors’ demand for this performance metric. Firms are required to report in their annual proxy
statement the performance metrics used in determining executive compensation (Dechow et al.
1994; Ittner et al. 1997; Lambert and Larcker 1987; Shalev et al. 2013). I find that firms that
include the EBITDA metric in their proxy statement are 14.6 percent more likely to disclose
EBITDA in their annual earnings announcement relative to their industry peers.3 Furthermore, a
1 standard deviation increase in the proportion of analysts estimating EBITDA is associated with
a 2.8 percent increase in the propensity to disclose EBITDA in earnings announcements. I also
find that firms that are subject to EBITDA-based debt covenants are 6.2 percent more likely to
disclose EBITDA. These results suggest that managers’ incentive structure and investors’ use
of EBITDA in valuation are important determinants of EBITDA disclosure in earnings
announcements.
Since my proxy for managers’ focus on EBITDA—disclosure of EBITDA in earnings
announcements—is voluntary by nature, there may be omitted correlated variables that may limit
my inferences. For example, managers may be overlevered and overinvest for various unobserved
reasons and use EBITDA to “window-dress” their performance (reverse causality). I perform vari-
ous robustness tests to mitigate this concern. For example, I employ propensity score matching to
better control for observed characteristics. I also examine firms that change their EBITDA disclo-
sure pattern and employ a firm fixed effects model, which controls for unobserved fixed firm
characteristics, and find consistent results.
Furthermore, in untabulated analyses, I replace my proxy for managers’ focus on EBITDA
with four alternative proxies: (i) an indicator variable that captures whether the firm uses
EBITDA-based executive compensation, (ii) an indicator variable that identifies firms subject to
EBITDA-based debt covenants, (iii) the percentile rank of the proportion of analysts who provide
an EBITDA estimate, and (iv) an aggregate measure of the three proxies. I find a significantly
positive association between each of the four proxies and both overinvestment and overleverage.4
These results are harder to explain with reverse causality (i.e., overinvestment leads investors to
rely more heavily on EBITDA). While these robustness tests provide some assurance on the
validity of my inferences, absent an exogenous shock to managers’ focus on EBITDA, the inter-
pretations of the results are still subject to endogeneity concerns.
I measure overinvestment using Richardson’s (2006) capital investment model.5 Since over-
investment is likely to manifest itself in low profitability, I also use various profitability metrics,
such as return on net operating assets, operating asset turnover, and operating profit margins, as
additional proxies for overinvestment. While overinvestment models are more direct, they may be
measured with error, whereas the profitability metrics are indirect, but better capture the conse-
quences of firms’ investments. I find a negative association between managers’ focus on EBITDA
and firm profitability. Overall, the results in this study are consistent with EBITDA-based incen-
tives and investor demand for EBITDA disclosure leading managers to focus on EBITDA when
making business decisions. Managers’ focus on EBITDA creates incentives for them to overin-
vest in capital and overlever the firm.
3. The correlation between EBITDA disclosure in earnings announcements and the use of EBITDA in executive com-
pensation is 37.6 percent.
4. The results are available from the author upon request.
5. In untabulated tests, I also employ McNichols and Stubben’s (2008) capital investment model. My inferences
remain unchanged.
I do not claim that the use of EBITDA has a net negative effect or that its use should be dis-
couraged. Rather, I highlight possible adverse implications of a managerial focus on EBITDA. It is
still possible that the benefits investors gain from the use of EBITDA in valuation and contracting
outweigh the costs associated with managers’ focus on EBITDA. Particularly, EBITDA is an easy-
to-calculate cash flow proxy, which represents a more comparable performance metric across firms,
and it omits depreciation and amortization expenses, which are subject to managers’ choice.6
This paper contributes to the literature in several ways. First, consistent with Campbell’s
(1979) law,7 it provides evidence on the negative implications of using EBITDA in firm perfor-
mance evaluation. This implication is important given the rise in the use of EBITDA in perfor-
mance evaluation. This paper also provides further evidence on how fixation on specific
performance metrics can lead to suboptimal outcomes.
Second, while there is a growing literature on non-GAAP disclosure behavior,8 EBITDA, a
specific form of non-GAAP earnings, has received little attention, despite its prevalence and the
material differences between EBITDA and other non-GAAP earnings metrics. For example,
Bhattacharya et al. (2007) and Brown et al. (2012) find that non-GAAP disclosure is not persis-
tent and varies from period to period. I find that 91 percent of firms in my sample exhibit a per-
sistent EBITDA disclosure pattern. Furthermore, while some prior research examines the
determinants and usefulness of non-GAAP earnings disclosures,9 I investigate how managers’
focus on EBITDA, proxied by their choice to disclose EBITDA in earnings announcements, is
associated with their investment and leverage choices.
D’Souza et al. (2010) are the only researchers to provide evidence on EBITDA disclosure
behavior. They seek to explain EBITDA disclosure behavior and find that capital-intensive firms
disaggregate earnings into EBITDA and interest, taxes, depreciation, and amortization. In con-
trast, I examine whether management focus on EBITDA leads them to underweight depreciation
and interest expense and consequently overinvest and overlever their firm.
6. Managers’ choice refers to their choice of deprecation method and estimated useful life of PP&E. For example, a
2013 survey by PwC of U.S. wireless companies finds significant variation in the useful life estimates for the same
asset category. Available online at http://www.pwc.se/sv/telekom/assets/north-american-wirelessindustry-survey.pdf.
7. Campbell’s (1979) law states that “The more any quantitative social indicator [or even some qualitative indicators]
is used for social decisionmaking, the more subject it will be to corruption pressures and the more apt it will be to
distort and corrupt the social processes it is intended to monitor.”
8. See, for example, Bhattacharya et al. (2004), Black and Christensen (2009), Bowen et al. (2005), Doyle et al.
(2003), Johnson and Schwartz (2005), and Lougee and Marquardt (2004).
9. See, for example, Black, Black et al. (2017); Black, Christensen et al. (2017); Brown et al. (2017); Christensen et al.
(2017); and Dyreng et al. (2017).
may underweight the real cost of investments in capital and overinvest.10 This observation is con-
sistent with Campbell’s (1979) law, which predicts that “The more any quantitative social indica-
tor is used for social decision making, the more subject it will be to corruption pressures and the
more apt it will be to distort and corrupt the social processes it is intended to monitor.” This rea-
soning leads to my first hypothesis:
Similar to depreciation expense, the direct cost of leverage—interest expense—is also omit-
ted in calculating EBITDA. Consequently, managers who focus on EBITDA may acquire exces-
sive leverage to either increase EBITDA (through investment in capital or other production
inputs, for example, Stulz (1990)), or redistribute it to shareholders (Cohen et al. 2017).
Since EBITDA also excludes taxes, the initial presumption can be that managers’ focus on
EBITDA may lead them to overpay taxes (similar to the previous arguments about overinvest-
ment and overleverage). However, the relation between managers’ focus on EBITDA and tax
avoidance is not clear ex ante. Since EBITDA excludes taxes, it is possible that managers will
not divert resources for tax planning, and, as a result, the firm would overpay taxes relative to its
industry peers. Conversely, managers who focus on EBITDA may engage in tax avoidance and
use the tax savings for activities that increase EBITDA. In this case, managers who focus on
EBITDA may engage in tax avoidance to the same extent as their industry peers. This reasoning
leads to my third hypothesis (stated in null form):
HYPOTHESIS 3. Managers’ focus on EBITDA is not associated with excessive tax avoidance
compared to industry peers.
3. Research design
Identification of managers’ focus on EBITDA
Managers use earnings announcements to summarize and discuss their firms’ performance. There-
fore, earnings announcements may be a good setting for evaluating which performance measures
are important to managers. I use firms’ disclosure of EBITDA in earnings announcements as my
main proxy for managers’ focus on EBITDA.
I use directEDGAR to extract section 2.02 8-K filings that contain earnings announcements.
I then delete filings that are not within a three-day window centered on the COMPUSTAT annual
earnings announcement date. This process purges out 8-K filings that are not annual earnings
announcements. I use directEDGAR to conduct a word search for the words “EBITDA” or
“OIBDA” (operating income before depreciation and amortization). I then create an indicator
10. The consequences of inefficient performance metrics have been studied in other settings. For example, Carter et al.
(2007) find that firms decreased the use of stock-option compensation once they became accountable for these costs
following SFAS 123(R). Bushee (1998) and Graham et al. (2005) suggest that managers’ fixation on earnings can
lead to myopic decisions. This issue has also been raised by Lambert (2001), who notes that “criticism continues to
grow that traditional performance measures motivate dysfunctional behavior by causing managers to pay attention
to the ‘wrong’ things.” These studies as well as mine do not claim that managers are behaving irrationally. Rather,
managers are responding to direct and indirect incentives set forth by the stakeholders in the firm.
variable that is equal to one if the words “EBITDA” or “OIBDA” appear at least once in the
directEDGAR word search, and zero otherwise.
11. I validate the data generated by directEDGAR by following Bentley et al.’s (2017) validation procedure. I choose a
random subsample of 100 observations; 50 of the observations include “EBITDA” or “OIBDA,” and 50 observa-
tions do not. I hand-collect the corresponding executive compensation sections in the proxy statements. I then com-
pare the hand-collected proxy statements to the data generated by directEDGAR. All 100 observations have a
corresponding proxy statement. Two observations that directEDGAR classified as including EBITDA did not
include it in the executive compensation section. The 98 remaining observations were classified correctly by
directEDGAR.
12. Some analysts explicitly forecast EBITDA; 58.92 percent of my sample firm-years that are covered by analysts
have at least one analyst who provides an EBITDA estimate. I obtain the number of analysts who provide EBITDA
estimates from the I/B/E/S summary file (STATSUM_XEPSUS). The file contains the consensus estimates for vari-
ous metrics, which are defined by the variable MEASURE. EBITDA consensus estimates appear in observations
where MEASURE is equal to “EBT.”
13. This approach, to examine whether opportunistic disclosure motives explain actual disclosure, is motivated by
Doyle et al. (2013).
14. The last consensus estimate before the earnings announcement is commonly used in the literature as a proxy for
market expectation and is frequently used in the calculation of earnings surprise (e.g., Amiram et al. 2016; Bagnoli
et al. 1999; Kothari et al. 2009).
EBITDA_EA is my proxy for managers’ focus on EBITDA. I set this variable to one if the
company includes “EBITDA” or “OIBDA” in its annual earnings announcement and zero other-
wise. The remaining variables are described in the Appendix.
where INEW is new investments during year t. The definition of INEW in Richardson (2006)
includes R&D expenditure. However, since I am only interested in expenditures that are
related to EBITDA, I move R&D expenditures to the right-hand side of the model. EBIT-
DA_EA is my proxy for managers’ focus on EBITDA. V/P, Book-to-Market, Age, Size, and
SalesGrowth are proxies for growth opportunities. I add INEW,i,t−1 to control for prior year’s
investments. R&D is research and development expenditure, which is another form of invest-
ment choice that does not directly affect EBITDA. I follow D’Souza et al. (2010) and also
control for firms’ capital intensity (Cap_Int). Financial leverage and low cash reserves may
constrain firms’ ability to invest. I, therefore, add leverage (Leverage), cash (Cash), and cash
flows from operations (CFO) as controls. I use the prior year’s compounded stock returns
(StockReturns) to control for firms’ investment opportunities and ability to raise cash. In the
wake of uncertainty boards of directors may seek to incentivize a risk-averse manager to
invest in capital by tying her compensation to EBITDA. I, therefore, use SalesStd to control
for revenue volatility. YearIndicator is a vector of indicator variables to capture annual fixed
effects. IndustryIndicator is a vector of indicator variables to capture industry fixed effects. I
follow the Fama-French 48-industry classification. I provide detailed explanations on the vari-
ables and their construction in the Appendix. β1 > 0 is consistent with a positive association
between my proxy for managers’ focus on EBITDA and overinvestment.
Testing the association between managers’ focus on EBITDA and overleverage
I follow Blaylock et al. (2017) to test the association between managers’ focus on EBITDA and
overleverage (Hypothesis 2):
I use both leverage and interest expense as dependent variables. EBITDA_EA is my proxy
for managers’ focus on EBITDA. MarginalTaxRate is based on Blouin et al.’s (2010) measure. I
include 1/Assets to control for the denominator in Leverage. Book-to-Market, MarketROA,
StockReturns, and SalesGrowth control for the firm’s growth opportunities. NetPP&E controls for
nondebt tax shields, which may reduce the incentive to increase leverage. Lastly, I include
SalesStd to control for uncertainty, which may affect leverage decisions. I define the variables in
the Appendix. γ 1 > 0 is consistent with a positive association between overleverage and my proxy
for managers’ focus on EBITDA.
Testing the association between managers’ focus on EBITDA and tax avoidance
I follow Dyreng et al. (2010) to test the association between managers’ focus on EBITDA and
tax avoidance (Hypothesis 3):
where the dependent variable, ETR, is the GAAP effective tax rate. EBITDA_EA is my proxy for
managers’ focus on EBITDA. I include a vector of control variables that prior research has docu-
mented to be associated with tax avoidance. For example, Capex is capital expenditures, which
may result in accelerated depreciation and tax planning. ForeignOperations is an indicator vari-
able that captures the existence of foreign operations. NetOperatingLoss is an indicator variable
equal to one if the firm reported a net operating loss in year t, and zero otherwise. Variable defini-
tions are available in the Appendix.
15. I begin the sample in 2003, subsequent to the Sarbanes-Oxley Act and Regulation G.
16. I validate the data generated by directEDGAR by following Bentley et al.’s (2017) procedure to validate machine-
based word searches and classifications. I choose a random sample of 100 observations and hand-collect the corre-
sponding earnings announcements. I then compare the hand-collected earnings announcements to the data generated
by directEDGAR. All 100 observations have a corresponding earnings announcement and all the classifications by
directEDGAR were accurate.
17. Of the 4,319 earnings announcements that include EBITDA, 1,813 refer to “adjusted EBITDA.” Investigating the
other exclusions in firms’ “adjusted EBITDA” is beyond the scope of this study. Inferences remain unaltered when
I remove the 1,813 observations that include “adjusted EBITDA” from the sample.
Descriptive statistics
Panel A of Table 1 reports the proportion of companies that disclose EBITDA in my sample by
industry and year. I categorize industries based on the Fama-French 48-industry classification.
Interestingly, the overall proportion of firms disclosing EBITDA in earnings announcements
increases from 17 percent in 2003 to 35 percent in 2011. The correlation between a time variable
and the proportion of firms disclosing EBITDA is 98 percent (p < 0.01). The industries with the
largest proportion of firms disclosing EBITDA are entertainment (72 percent), telecommunica-
tions (59 percent), and healthcare (47 percent). This evidence is in line with Francis et al.’s
(2003) and D’Souza et al.’s (2010) evidence.
I present my sample firms’ EBITDA disclosure patterns by industry in panel B of Table 1.
I require firms to appear for at least five years in my sample to identify firm-specific disclosure trends.
I then categorize sample firms into five groups. “Non-EBITDA” firms do not disclose EBITDA in
any of the sample years; “EBITDA” firms disclose EBITDA in every year during my sample period;
“Switch-to” firms make a single change from not disclosing EBITDA at all to persistently disclosing
EBITDA; “Switch-from” firms make a single change from consistently disclosing EBITDA to not dis-
closing EBITDA; and finally, “Multiple-switches” firms make multiple changes in their EBITDA dis-
closure pattern. The “non-EBITDA” group consists of 61 percent of my sample. “EBITDA” firms
account for 12 percent of my sample, and “Switch-to” firms account for 15 percent of my sample.
The “EBITDA” and “Switch-to” groups together comprise 27 percent of the sample. It is consistent
with the overall proportion of firms in my sample that disclose EBITDA in annual earnings announce-
ments presented in panel A of Table 1. The “Switch-from” group accounts for only 3 percent of my
sample. This evidence suggests that firms that disclose EBITDA continue to do so persistently. Only
9 percent of firms in my sample appear to disclose EBITDA in a transitory manner.18
Table 2 reports the descriptive statistics for the pooled sample. The average EBITDA_EA is
0.239. Since EBITDA_EA is an indicator variable, it implies that 23.9 percent of earnings
announcements in my sample disclose EBITDA, like the results in panel A of Table 1. The mean
INEW,i,t and INEW,i,t−1 is 0.029, and the median is 0.004, which suggests some right skewness. The
mean EBITDA_Compensation is 0.194. Since EBITDA_Compensation is an indicator variable, it
implies that 19.4 percent of firm-years represented in my sample use EBITDA in setting execu-
tive compensation.19 The mean EBITDA_Analysts is 0.386, which implies that, on average, 38.6
percent of analysts covering a given firm post an EBITDA estimate on I/B/E/S. Furthermore, for
at least 25 percent of firm-years in my sample, there is no EBITDA estimate posted by analysts
on I/B/E/S. Lastly, the mean Relative_Earnings_Surprise is 0.087, which implies that 8.7 percent
of firm-years in my sample meet or beat the analysts’ EBITDA consensus estimate but miss the
analysts’ EPS consensus estimate.
Table 3 reports pairwise correlations (with Pearson correlations above the diagonal and
Spearman correlations below). The Pearson correlation between EBITDA_EA and INEW,i,t is
0.051, consistent with my prediction. Furthermore, Table 3 reveals that firms that disclose
EBITDA in their earnings announcements are younger (correlation of −0.148), invest less in
R&D (correlation of −0.186), and are more levered (correlation of 0.247). The correlation
between EBITDA_EA and Cap_Int is 0.106. This positive correlation is consistent with D’Souza
et al.’s (2010) evidence that capital-intensive firms disclose line items in the income statement
that allow for the calculation of EBITDA. Finally, the correlation between EBITDA_EA and
EBITDA_Compensation is 0.376. This correlation suggests a significantly positive association
between managers’ EBITDA disclosure choice and the use of EBITDA-based incentives.
18. I find that 3 percent of firms disclose EBITDA on a “one-off” basis. Those firms are included in the “Multiple-
switches” category.
19. This proportion is consistent with Bennett et al.’s (2017) evidence. They find that 23 percent of firms in their sam-
ple use EBITDA in setting executive compensation.
Panel A: Proportion of firms that disclose EBITDA in earnings announcements by industry and year
Industry 2003 2004 2005 2006 2007 2008 2009 2010 2011 Industry N Corr
2 Food Products 13% 21% 6% 19% 24% 26% 32% 30% 30% 23% 324 80%
6 Recreational Products 6% 19% 12% 11% 47% 29% 41% 36% 27% 25% 142 67%
7 Entertainment 72% 64% 72% 63% 63% 74% 81% 76% 82% 72% 276 64%
8 Printing and Publishing 40% 47% 40% 53% 50% 42% 36% 44% 108 −15%
14 Chemicals 15% 17% 20% 24% 31% 35% 35% 40% 43% 30% 491 99%
15 Rubber and Plastic Products 13% 6% 21% 21% 20% 25% 38% 36% 38% 23% 152 93%
17 Construction Materials 10% 8% 5% 6% 14% 9% 20% 16% 20% 12% 386 75%
18 Construction 29% 20% 26% 40% 27% 32% 20% 25% 35% 29% 211 15%
19 Steel Works, Etc. 11% 22% 13% 21% 14% 25% 21% 14% 39% 20% 255 57%
21 Machinery 20% 20% 18% 21% 23% 24% 25% 27% 31% 23% 716 93%
22 Electrical Equipment 14% 8% 3% 8% 15% 19% 19% 26% 31% 16% 243 85%
24 Automobiles and Trucks 7% 14% 22% 21% 27% 26% 29% 43% 38% 26% 306 95%
25 Aircraft 0% 0% 6% 7% 14% 13% 20% 9% 102 97%
31 Petroleum and Natural Gas 20% 21% 18% 20% 33% 33% 37% 35% 47% 31% 739 92%
32 Utilities 14% 17% 18% 19% 19% 19% 20% 18% 24% 19% 785 84%
33 Telecommunications 57% 45% 49% 43% 60% 62% 67% 70% 77% 59% 627 83%
34 Personal Services 20% 21% 24% 29% 27% 37% 30% 44% 45% 32% 294 93%
35 Business Services 16% 23% 21% 27% 33% 40% 44% 46% 51% 32% 2,736 99%
36 Computers 10% 8% 11% 9% 13% 16% 20% 22% 26% 15% 886 94%
37 Electronic Equipment 6% 7% 8% 8% 9% 13% 14% 12% 20% 11% 1,580 91%
38 Measuring and Control Equipment 7% 6% 11% 13% 13% 15% 13% 15% 14% 12% 582 85%
39 Business Supplies 26% 26% 20% 23% 28% 33% 37% 38% 44% 31% 243 86%
Panel A: Proportion of firms that disclose EBITDA in earnings announcements by industry and year
Industry 2003 2004 2005 2006 2007 2008 2009 2010 2011 Industry N Corr
41 Transportation 15% 8% 19% 20% 20% 21% 28% 23% 26% 20% 542 84%
42 Wholesale 23% 25% 23% 26% 23% 24% 24% 30% 35% 26% 744 72%
43 Retail 19% 18% 19% 20% 24% 29% 30% 33% 36% 25% 1,190 97%
44 Restaurants, Hotel, Motel 24% 30% 27% 32% 33% 30% 41% 41% 38% 33% 368 86%
Year 17% 18% 18% 20% 23% 26% 29% 30% 35% 24% 18,082 98%
N 1,884 1,822 2,214 2,154 2,170 2,008 1,938 1,836 2,056
Notes: Panel A presents the proportion of firms that include EBITDA in annual earnings announcements by industry and year. “Corr” represents the correlation
between the proportion of firms disclosing EBITDA in earnings announcements and a time variable. Correlations in bold represent significance at the 5 percent level.
Panel B presents the proportion of firms that follow a given EBITDA disclosure pattern, by industry. “Non-EBITDA” represents firms that do not include EBITDA in
earnings announcements throughout the sample period. “EBITDA” represents firms that include EBITDA in earnings announcements consistently throughout the
sample period. “Switch-to” represents firms that change from not disclosing EBITDA at all to disclosing EBITDA consistently. “Switch-from” describes firms that
switch from disclosing EBITDA consistently in earnings announcements to not disclosing EBITDA at all. “Multiple-switches” represent firms that do not have a
persistent EBITDA disclosure pattern within the sample period. The sample period spans from 2003 to 2011. The sample is restricted to firms that appear five years or
more within the sample period.
EBITA and Managers’ Choices 525
TABLE 2
Descriptive statistics
Notes: Table 2 provides descriptive statistics for key variables. Variable definitions are presented in the Appendix.
5. Results
Validating EBITDA disclosure in earnings announcements as a proxy for managers’ focus on
EBITDA
I provide the results from estimating equation (1) in Table 4. The coefficient estimate on EBIT-
DA_Compensation is 0.853 (z-statistic = 8.06). This coefficient estimate implies that the propen-
sity to disclose EBITDA in earnings announcements increases by 14.6 percent for firms that use
EBITDA in setting executive compensation. This result is consistent with firms disclosing
EBITDA to convey to investors that EBITDA is important to them. The coefficient estimate on
EBITDA_Analysts is 0.498 (z-statistic = 2.85). A one standard deviation increase in the proportion
of analysts posting EBITDA estimates on I/B/E/S is associated with a 2.8 percent increase in the
propensity to disclose EBITDA in earnings announcements. This result suggests that firms dis-
close EBITDA in response to investors’ demand.20 The coefficient estimate on EBITDA_Cove-
nant is 0.412 (z-statistic = 3.91). This coefficient estimate implies that the propensity to disclose
EBITDA in earnings announcements increases by 6.2 percent for firms that are subject to
EBITDA-based debt covenants. This result suggests that debtholders’ demand also shapes firms’
EBITDA disclosure choice. The propensity to disclose EBITDA in earnings announcements
increases by 4.7 percent for firms that meet or beat the analysts’ EBITDA consensus estimate but
20. These results are consistent with Bentley et al.’s (2017) evidence that some firms define their non-GAAP earnings
metric following the analysts’ definition. In additional (untabulated) analysis, I find a stronger association between
EBITDA and stock returns for firms that disclose EBITDA (these results are consistent with Kang et al. 2017). I
also find a weaker association between depreciation expense and stock returns for firms that disclose EBITDA.
These results suggest that investors use EBITDA to value firms that disclose this measure.
Correlations
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
(1) EBITDA_EA 0.051 0.065 −0.148 0.068 0.054 −0.186 0.038 0.247 −0.206 0.106
(2) INEW,i,t −0.004 −0.001 −0.062 0.071 0.280 −0.141 −0.104 −0.065 −0.038 0.124
(3) V/P 0.070 0.021 0.148 0.165 0.059 −0.385 0.739 −0.064 −0.335 0.154
(4) Age −0.147 0.033 0.219 0.327 −0.051 −0.173 0.074 0.086 −0.268 0.118
(5) Size 0.081 0.197 0.209 0.336 0.121 −0.370 0.055 0.338 −0.395 0.299
(6) INEW,i,t−1 −0.003 0.467 0.083 0.031 0.223 −0.148 −0.031 0.040 −0.168 0.161
(12) StockReturns 0.006 0.234 −0.187 0.034 0.057 0.014 −0.069 −0.285 −0.019 −0.012 0.032
(13) CFO −0.043 0.307 0.029 0.138 0.258 0.240 −0.197 −0.178 −0.058 −0.117 0.082
(14) SalesStd 0.075 0.199 0.165 0.284 0.863 0.224 −0.341 0.046 0.354 −0.337 0.257
(15) SalesGrowth −0.004 0.231 −0.160 −0.168 −0.017 0.304 0.043 −0.238 −0.076 0.052 −0.063
(16) InterestExpense 0.236 0.017 0.086 0.097 0.390 0.064 −0.274 0.082 0.801 −0.470 0.398
(17) ETR 0.006 0.228 0.245 0.152 0.214 0.207 −0.369 0.098 0.052 −0.281 0.168
(18) EBITDA_Compensation 0.376 −0.016 0.050 −0.091 0.078 −0.006 −0.164 0.026 0.176 −0.125 0.084
(19) EBITDA_Analysts 0.277 0.132 0.121 0.084 0.364 0.115 −0.333 0.042 0.230 −0.282 0.221
(20) EBITDA_Covenant 0.239 0.099 0.160 0.009 0.269 0.107 −0.301 0.066 0.300 −0.373 0.176
(21) Relative_Earnings_Surprise 0.041 −0.003 0.001 −0.014 0.011 −0.023 −0.033 0.030 0.065 −0.036 0.038
(22) EBITDA_EAt−1 0.841 −0.015 0.088 −0.165 0.065 −0.017 −0.247 0.069 0.261 −0.220 0.114
(12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
(1) EBITDA_EA −0.007 0.030 −0.005 0.003 0.133 0.025 0.376 0.294 0.239 0.041 0.841
(2) INEW,i,t 0.177 0.219 0.003 0.117 0.000 0.071 0.015 0.075 0.123 0.021 0.028
(3) V/P −0.160 0.244 0.051 −0.121 −0.128 0.177 0.039 0.111 0.142 −0.002 0.087
(4) Age 0.043 0.162 0.220 −0.207 −0.060 0.092 −0.097 0.044 0.007 −0.013 −0.168
(12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22)
(5) Size 0.052 0.329 0.551 −0.075 0.099 0.129 0.070 0.306 0.240 0.011 0.050
(6) INEW,i,t−1 0.013 0.171 0.022 0.226 0.038 0.055 0.022 0.068 0.131 −0.004 0.044
(7) R&D −0.098 −0.552 −0.136 0.124 0.102 −0.230 −0.138 −0.294 −0.298 −0.008 −0.197
(8) Book-to-Market −0.317 −0.004 0.017 −0.145 −0.009 0.078 0.015 0.051 0.059 0.024 0.064
(9) Leverage −0.057 −0.013 0.129 −0.050 0.476 0.016 0.189 0.200 0.251 0.068 0.259
(10) Cash −0.028 −0.330 −0.165 0.129 −0.044 −0.194 −0.140 −0.299 −0.391 −0.021 −0.223
(11) Cap_Int 0.022 0.115 0.106 −0.042 0.255 0.082 0.084 0.216 0.140 0.044 0.121
(12) StockReturns 0.184 0.014 0.098 −0.094 0.022 −0.022 0.049 0.031 −0.005 −0.013
(13) CFO 0.173 0.111 −0.068 −0.293 0.220 0.024 0.159 0.179 −0.050 0.025
(14) SalesStd 0.069 0.272 −0.019 −0.013 0.031 −0.015 0.128 0.003 0.015 −0.013
(15) SalesGrowth 0.178 0.143 0.077 0.045 −0.049 −0.026 −0.045 −0.052 0.013 −0.008
(16) InterestExpense −0.038 −0.159 0.245 −0.053 −0.071 0.089 0.080 0.054 0.057 0.138
(17) ETR 0.103 0.360 0.219 0.051 −0.011 0.023 0.064 0.117 −0.002 0.024
(18) EBITDA_Compensation −0.011 −0.036 0.066 −0.030 0.161 0.005 0.228 0.170 0.023 0.386
(19) EBITDA_Analysts 0.066 0.146 0.330 −0.017 0.204 0.141 0.214 0.203 0.034 0.284
(20) EBITDA_Covenant 0.038 0.118 0.256 −0.012 0.262 0.162 0.170 0.204 0.032 0.240
(21) Relative_Earnings_Surprise 0.001 −0.070 0.003 0.014 0.075 −0.015 0.023 0.027 0.032 0.023
(22) EBITDA_EAt−1 0.004 −0.055 0.057 −0.014 0.238 0.006 0.386 0.273 0.240 0.023
Notes: Table 3 presents correlations among key variables, with Pearson (Spearman) correlations reported above (below) the diagonal. Variable definitions are presented
in the Appendix. Values in bold are significant at the 5 percent level.
EBITA and Managers’ Choices
527
TABLE 4
Factors associated with EBITDA disclosure in earnings announcements
Notes: This table presents results of logistic estimation of variations of equation (1). Bold type indicates the
variables of interest. Variable definitions are presented in the Appendix. Robust z-statistics are reported in
parentheses. *, **, and *** represent significance levels (two-sided) of 10 percent, 5 percent, and 1 percent,
respectively. Fixed effects are included but not reported.
miss the analysts’ EPS consensus estimate, and is marginally significant, both economically and
statistically. Overall, these results are consistent with managers disclosing EBITDA to convey to
investors that EBITDA is important to them and in response to investors’ demand.21 I find weaker
evidence consistent with managers disclosing EBITDA for opportunistic reasons.
21. Investors’ use of EBITDA and managers’ focus on EBITDA are interrelated. A rational manager who knows that
investors use EBITDA to evaluate her performance will want to maximize this metric.
22. I find a positive association between EBITDA disclosure and financing constraints. Therefore, failure to control for
financing constraints would bias downwards the association between EBITDA disclosure and overinvestment
(Greene 2007, 134).
23. In additional analysis of the subsample of firms that did not change their EBITDA disclosure policy, I aggregate all
time-series observations for each firm into one observation (i.e., each firm in this subsample is represented by one
observation). Inferences under this specification are unchanged. Furthermore, in additional analysis of the subsam-
ple of firms that make a single change to their EBITDA disclosure policy, I limit the sample to the year prior and
following the change to the EBITDA disclosure policy (i.e., each firm is represented by two observations). My
inferences under this specification are unaltered.
TABLE 5
Managers’ focus on EBITDA and overinvestment
TABLE 5 (continued)
Notes: This table presents results of OLS estimation of variations of equation (2). Bold type indicates the
variable of interest. Variable definitions are presented in the Appendix. Robust t-statistics are reported in
parentheses. *, **, and *** represent significance levels (two-sided) of 10 percent, 5 percent, and 1 percent,
respectively. Fixed effects are included but not reported.
(t-statistic = 16.15), and 0.105 (t-statistic = 16.12), respectively. This evidence suggests that man-
agers who focus on EBITDA overlever their firms compared to industry peers.
Panel B of Table 6 reports the results for subgroups based on EBITDA disclosure patterns.
In this panel, I exclude firms that do not have a consistent EBITDA disclosure pattern (i.e., their
disclosure patterns suggest opportunistic behavior). Columns (1) and (4) report the results for
firms that do not change their EBITDA disclosure policy throughout my sample period. Columns
(2) and (5) report time-series results for firms that make a single change to their EBITDA
Notes: This table presents results of OLS estimation of variations of equation (3). Bold type indicates the variable of interest. Variable definitions are presented in the
Appendix. Robust t-statistics are reported in parentheses. *, **, and *** represent significance levels (two-sided) of 10 percent, 5 percent, and 1 percent, respectively.
Fixed effects are included but not reported.
533
TABLE 7
Managers’ focus on EBITDA and tax avoidance
(1) (2)
Sample: Full Full
Dependent variable: ETR ETR
TABLE 7 (continued)
Notes: This table presents results of OLS estimation of variations of equation (4). Bold type indicates the
variable of interest. Variable definitions are presented in the Appendix. Robust t-statistics are reported in
parentheses. *, **, and *** represent significance levels (two-sided) of 10 percent, 5 percent, and 1 percent,
respectively. Fixed effects are included but not reported.
disclosure policy during the sample period. Lastly, columns (3) and (6) report the results for firms
that make no or a single change to their EBITDA disclosure policy. The coefficient estimates on
EBITDA_EA are significantly positive under all specifications, ranging from 0.005 (t-statis-
tic = 3.11) to 0.143 (t-statistic = 10.97). This evidence suggests that managers who focus on
EBITDA discount the true costs of leverage and overlever, since the direct cost of leverage—
interest expense—is excluded from EBITDA.
Testing the association between managers’ focus on EBITDA and tax avoidance
Table 7 reports the results from estimating equation (4), which tests whether managers’ focus on
EBITDA is associated with tax avoidance. The coefficient estimates on EBITDA_EA in columns
(1) and (2) of panel A are −0.004 (t-statistic = −0.58) and −0.010 (t-statistic = −1.37), respec-
tively. These results suggest that firms that disclose EBITDA are no different from their industry
peers with respect to tax avoidance.
Panel B of Table 7 reports the results for subgroups based on EBITDA disclosure patterns.
In this panel, I exclude firms that do not have a consistent EBITDA disclosure pattern (i.e., their
disclosure patterns suggest opportunistic behavior). The EBITDA_EA coefficient estimates are
economically and statistically insignificant in the analyses of all subsamples. These results do not
suggest a significant association between managers’ focus on EBITDA and tax avoidance. Unlike
overinvestment in capital and leverage, the association between managers’ focus on EBITDA and
tax avoidance is not clear ex ante. On the one hand, managers who focus on EBITDA would not
direct resources to tax planning, since taxes are excluded from EBITDA, while tax planning
expenses are not excluded. On the other hand, managers may engage in tax planning as much as
their industry peers and use the tax saving to invest in EBITDA increasing activities. The results
are consistent with the latter argument. Furthermore, these results are inconsistent with opportu-
nistic disclosure motives. If managers were disclosing EBITDA opportunistically, they would
have disclosed EBITDA when their effective tax rate is higher than that of their industry peers;
this is not observed in the data.
6. Additional tests
Managers’ focus on EBITDA and firm performance
I measure overinvestment using Richardson’s (2006) and McNichols and Stubben’s (2008)
capital investment models. Since overinvestment is likely to manifest itself in low profitabil-
ity, I also use several variables to measure the firms’ fundamental performance and its associ-
ation with managers’ focus on EBITDA. While overinvestment models are more direct, they
may be measured with error, whereas the profitability metrics are indirect but better capture
the consequences of firms’ investments. I present the results in Table 8. Due to the skewed
distribution of the profitability variables, I use least absolute deviations regressions (LAD)
and the log of the profitability variables as dependent variables in the OLS regressions. I
include year and industry fixed effects. I categorize industries based on the Fama-French
48-industry classification.
I find that capital intensity is significantly higher for firms whose managers focus on
EBITDA in both the OLS and LAD regressions. ROA is significantly lower for firms that focus
on EBITDA, with a coefficient estimate of −0.233 (t-statistic = −7.65) in the OLS regression and
a coefficient estimate of −0.014 (t-statistic = −10.69) in the LAD regression. Looking further into
the composition of ROA, the results suggest that firms that focus on EBITDA have significantly
lower profit margins and significantly higher asset turnover (ATO). The results for asset turnover
are incomplete since asset turnover includes nonoperating assets.
Return on net operating assets (RNOA) is significantly lower for firms that focus on
EBITDA, with a coefficient estimate of −0.360 (t-statistic = −10.63) in the OLS regression
and −0.030 (t-statistic = −10.97) in the LAD regression. Looking into the components of
RNOA, operating profit margins (OperatingPM) of firms that focus on EBITDA are 0.105
lower (t-statistic = −3.46) in the OLS regression and 0.003 lower (t-statistic = −1.81) in the
LAD regression compared to those of their industry peers. Similarly, operating asset turnover
(OperatingATO) of EBITDA firms is lower in both the OLS regression and the LAD
regression. Furthermore, ROOA is significantly lower for firms that focus on EBITDA, with a coeffi-
cient estimate of −0.222 (t-statistic = −7.84) in the OLS regression and −0.012 (t-statistic = −5.44) in
the LAD regression.
I exclude depreciation and amortization expenses from operating profit margin to test
whether the lower operating profit margin of firms whose managers focus on EBITDA is
Notes: This table reports the difference in key performance metrics between firms that disclose EBITDA in earnings announcements and firms that do not. Variable
definitions are presented in the Appendix. The regressions are estimated using OLS and LAD (least absolute deviations). The dependent variables in the OLS
regressions are logged in order to mitigate the effects of outliers. A vector of industry and year indicators is included but not reported. Robust t-statistics based on
standard errors clustered by firm for the OLS regressions and on bootstrapped standard errors for the LAD regressions are reported in parentheses. *, **, and ***
EBITA and Managers’ Choices
Testing the association between managers’ focus on EBITDA and overinvestment with a
matched sample
Coefficient estimates from equation (2) may be biased because of confounding covariates that dif-
fer between the treatment group (firms that disclose EBITDA) and the control group (firms that
do not disclose EBITDA). I use propensity score matching to ensure that that treatment and con-
trol groups are as comparable as possible on a vector of observable characteristics. First, I esti-
mate a logistic model by industry (based on the Fama-French 48-industry classification):
Variable definitions are available in the Appendix. I use the predicted values from the estima-
tion of equation (5) to create propensity scores. I then use the propensity scores to identify the
matched sample using the nearest neighbor approach (Prawitt et al. 2012). I present the results
from estimating equation (5) using the pooled sample with industry and year fixed effects in col-
umn (1) of Table 4. I find that, within my sample, firms that disclose EBITDA in earnings
announcements are younger, engage less in R&D, are more levered, and hold less cash. Lastly,
firms that disclose EBITDA also have lower operating cash flows and higher growth options.
Panel A of Table 9 provides descriptive statistics for firms that disclose EBITDA in their
annual earnings announcement (column (1)), firms that do not disclose EBITDA in earnings
announcements (column (2)), and a matched sample of firms that do not disclose EBITDA in
earnings announcements (column (3)). Column (4) provides the differences across a vector of var-
iables between firms that disclose EBITDA and the full sample of firms that do not. The results
indicate that those two groups are different across almost all variables. Column (5) provides the
differences across a vector of variables between firms that disclose EBITDA and the matched
sample of firms that do not. The results suggest that the matching process is effective. I find that
only the difference in INEW,i,t is significant (difference = 0.012 and t-statistic = 3.11). These
results provide further evidence that managers who focus on EBITDA overinvest in capital.
I reestimate equation (2) using the matched sample and provide the results in panel B of
Table 9. Inferences are unchanged. Column (1) presents the results from a parsimonious model
with no controls. The coefficient estimate on EBITDA_EA in column (1) is 0.011 (t-statis-
tic = 3.28). I add controls for the firm’s growth opportunities in column (2). The coefficient esti-
mate on EBITDA_EA is 0.010 (t-statistic = 3.41). Lastly, I provide the results from estimating the
full model, which also includes controls for the firm’s financial constraints, in column (3). The
coefficient estimate on EBITDA_EA is 0.011 (t-statistic = 3.77). Overall, the positive association
between my proxy for managers’ focus on EBITDA and overinvestment remains using a matched
sample, which controls for nonlinear differences in observed characteristics.
TABLE 9
Matched sample analysis of managers’ focus on EBITDA and overinvestment
Panel A: Summary statistics before and after the application of propensity score matching
TABLE 9 (continued)
Notes: This table presents summary statistics before and after matching in panel A and results of OLS
estimation of variations of equation (3) for a matched sample in panel B. t-statistics are presented in
parentheses in both panels. Bold type indicates indicate variables of interest. Variable definitions are
presented in the Appendix. *, **, and *** represent significance levels (two-sided) of 10 percent, 5 percent,
and 1 percent, respectively. Fixed effects are included in the analyses presented in panel B but are not
reported.
Discussion on endogeneity
The results of my study suggest a positive association between EBITDA disclosure and overin-
vestment in capital. Figure 1 presents feasible alternative explanations for the observed results.24
The observed results may be attributable to managers’ focus on EBITDA, which leads them to
both disclose EBITDA and overinvest.25 Managers’ focus on EBITDA is possibly influenced by
stakeholders’ use of EBITDA to evaluate the firm. To the extent stakeholders use EBITDA to
evaluate the firm, they can align the manager’s interest with their objectives either directly, by
tying the manager’s compensation to EBITDA, or indirectly by requiring the manager to disclose
EBITDA.26 The positive association between the propensity to disclose EBITDA and both
24. Overleverage does not directly increase EBITDA. Rather, overleverage can be a means to finance overinvestment
in capital, which can increase EBITDA. Therefore, I emphasize overinvestment in capital in the discussion on
endogeneity.
25. The primary construct in my study is managers’ focus on EBITDA. The top dashed arrow in Figure 1 represents
the link between managers’ focus on EBITDA and overinvestment. Since I cannot observe managers’ focus on
EBITDA, I use EBITDA disclosure as a proxy. The dashed arrow in Figure 1 pointing from EBITDA disclosure to
overinvestment represents the link between my proxy and overinvestment.
26. The manager would infer that stakeholders demand for EBITDA to be disclosed because they use it to evaluate the
firm. The manager, whose goal is to accomplish the stakeholders’ objectives, would then also focus on EBITDA.
stakeholders’ demand for EBITDA disclosure and the use of EBITDA in setting executive com-
pensation is consistent with this argument.
There may be other alternative explanations for my results. First, managers may disclose
EBITDA as a result of an industry or firm standard, and not because they fixate on this metric. I
find that the propensity to disclose EBITDA increases with managers’ EBITDA-based compen-
sation and stakeholders’ demand. I also control for the industry and find that EBITDA disclosure
is positively associated with overinvestment. This evidence is inconsistent with managers dis-
closing EBITDA as a result of an industry or firm standard. Second, in periods of uncertainty, a
risk-averse manager may be reluctant to invest in capital. The board of directors may want to
incentivize the manager to invest by linking her compensation to EBITDA since EBITDA does
not penalize investments in capital. I use revenue volatility as a control variable in all the appli-
cable analyses to address this possibility. Third, a powerful manager may elicit EBITDA-based
compensation to extract rents or to support her empire-building objectives. In an untabulated
analysis, I control for CEO power by including an indicator variable that is equal to one if the
CEO is also the chairman of the board and zero otherwise. The coefficient estimate on CEO
power is insignificantly different from zero, and all other inferences are unchanged.
Managers may overinvest and then disclose EBITDA to “window-dress” their performance
(i.e., reverse causality). In untabulated analyses, I replace my proxy for managers’ focus on
EBITDA (EBITDA_EA) in equation (2) with four alternative non-disclosure related proxies: (i) an
indicator variable that captures whether the firm uses EBITDA-based executive compensation,
(ii) an indicator variable that identifies firms subject to EBITDA-based debt covenants, (iii) the
percentile rank of the proportion of analysts who provide an EBITDA estimate, and (iv) an aggre-
gate measure of the three proxies. I find a significantly positive association between each of the
four proxies and both overinvestment and overleverage. These results are harder to explain with
reverse causality (i.e., overinvestment leads investors to rely more heavily on EBITDA). In an
untabulated analysis, I also use a lagged EBITDA disclosure indicator variable, and inferences
remain unaltered.
EBITDA disclosure and tax avoidance is a good setting to evaluate reverse causality. While
managers’ focus on EBITDA is unlikely to induce them to overpay taxes (as such action would
not increase EBITDA significantly), firms that pay high taxes may disclose EBITDA to deempha-
size this. Therefore, a positive association between EBITDA disclosure and higher effective tax
rates is more consistent with window-dressing. I do not find a positive association between
EBITDA disclosure and effective tax rates (Table 7). Lastly, I investigate whether managers’
opportunistic motives to disclose EBITDA explain actual EBITDA disclosure and find little sup-
portive evidence. Specifically, I find that firms that meet or beat analysts’ EBITDA expectations
but miss analysts’ EPS expectations are marginally more likely to disclose EBITDA relative to
their industry peers.
Another concern is that there is an underlying unobservable reason for both EBITDA disclo-
sure and overinvestment. I use propensity score matching to better control for a vector of possibly
confounding observables and find comparable results. I also examine a subset of firms that
change their EBITDA disclosure policy and replace the industry fixed effects with firm fixed
effects. The firm fixed effects allow me to control for unobserved fixed firm characteristics. I find
consistent results under this specification as well. While I conduct multiple robustness tests to
mitigate endogeneity concerns, I acknowledge that these concerns cannot be fully remediated.
7. Conclusion
The use of EBITDA in capital markets and disclosure of EBITDA in earnings announcements is
prevalent. I find that the percentage of firms in my sample that disclose EBITDA increased from
17 percent in 2003 to 35 percent in 2011. Furthermore, while EBITDA disclosure practices vary
by industry, most industries have a nontrivial percentage of firms disclosing EBITDA. This study
suggests that the extensive use of EBITDA may come at a cost.
To the extent that managers that disclose EBITDA also use EBITDA to evaluate business
decisions, they may decide to overinvest in capital, since EBITDA does not capture the cost of
overinvestment. Overinvestment can manifest itself in both higher revenues and higher deprecia-
tion and amortization expenses, resulting in a higher EBITDA because depreciation and amortiza-
tion expenses are excluded in calculating EBITDA, while revenues are not excluded.
Furthermore, managers may obtain excessive leverage since the direct cost of leverage—interest
expense—is also excluded from EBITDA. Consistent with these hypotheses, I find a positive
association between my proxy for managers’ focus on EBITDA and both overinvestment in capi-
tal and overleverage. I conduct the analyses within-industry, and the results are robust to alterna-
tive specifications and propensity score matching. Furthermore, I find that GAAP-based
performance measures are lower for firms whose managers fixate on EBITDA compared to their
industry peers. I validate my proxy for managers’ focus on EBITDA and find that firms’ propen-
sity to disclose EBITDA increases when executive compensation is tied to EBITDA and when
EBITDA is used more extensively by shareholders and debtholders. I only find limited evidence
that firms disclose EBITDA opportunistically.
My analyses contribute to prior research by providing evidence on how the use of EBITDA
may influence managers’ investing and financing decisions. The results provide evidence of the
inferior characteristics of EBITDA, which, given its popularity, may lead managers to overinvest
in capital and obtain excessive leverage, since the adverse effects of overinvestment and over-
leverage are not captured by EBITDA. I do not claim that the use of EBITDA has a net negative
effect or that it should be discouraged. Rather, I highlight possible costs that are associated with
EBITDA.
Appendix
Variable definitions
1/Assets Reciprocal of the market value of assets As defined by Blaylock et al. (2017).1/
(PRCC_F×CSHO + DLC + DLTT)
AdvertisingExpense Annual advertising expense scaled by net As defined by Dyreng et al. (2010). XAD/
sales SALE
Age Age of the firm As defined by Richardson (2006). Log of
the number of years the company has
been listed on CRSP as of the start of the
year
ATO Asset turnover REVTt/ATt−1
Book-to-Market Book-to-market ratio (AT – LT + TXDITC − PSTKL)/
(PRCC_F×CSHO)
Cap_Int Capital intensity PPENT/AT
Cash Cash and short-term investments CHEt/ATt−1
Capex Capital expenditures scaled by gross PP&E As defined by Dyreng et al. (2010). CAPX/
PPEGT
CFO Cash flow from operations minus (OANCFt − DPt)/[(ATt−1 + ATt)/2]
depreciation and amortization
EBITDA_Analysts Proportion of analysts issuing an EBITDA Number of analysts contributing to the
estimate through I/B/E/S EBITDA consensus estimate
immediately preceding the earnings
announcement (obtained from the
I/B/E/S STATSUM_XEPSUS file)
scaled by the number of analysts
contributing to the EPS consensus
estimate immediately preceding the
earnings announcement (obtained from
the I/B/E/S STATSUM_EPSUS file)
EBITDA_Compensation Indicator variable equal to one if EBITDA I use directEDGAR to extract annual proxy
is included in the proxy statement in the statement filings. I then search within the
context of executive compensation, and identified proxy statements for the words
zero otherwise “EBITDA” or “OIBDA” that are in the
same paragraph as either words “award,”
“incentive,” or “target” to identify proxy
statements that include EBITDA-based
compensation
EBITDA_Covenant Indicator variable equal to one if the firm I use DealScan’s “financialcovenant” file
is subject to an EBITDA-based debt from WRDS. I then use DealScan’s
covenant in year t, and zero otherwise variable “CovenantType” to identify
EBITDA-based debt covenants
EBITDA_EA Indicator variable equal to one if EBITDA I use directEDGAR to extract section 2.02
is included in the earnings 8-K filings that contain earnings
announcement, and zero otherwise announcements. I then search within the
identified earnings announcements for
the words “EBITDA” or “OIBDA” to
identify earnings announcements that
include EBITDA
ETR Effective tax rate As defined by Dyreng et al. (2010). TXT/
(PI + SPI)
ForeignOperations Indicator variable equal to one if the firm As defined by Dyreng et al. (2010).
has nonzero, nonmissing income from COMPUSTAT item PIFO
foreign operations, and zero otherwise
GrossPP&E Gross PP&E scaled by total assets As defined by Dyreng et al. (2010).
PPEGT/AT
INEW Investments in new projects Following Richardson (2006).
(CAPXt + AQCt − SPPEt − DPt)/
[(ATt + ATt−1)/2]
IntangibleAssets Intangible assets scaled by total assets As defined by Dyreng et al. (2010).
INTAN/AT
InterestExpense Interest expense scaled by sales (XINT + INTC)/SALE
Appendix (continued)
Leverage Sum of short-term and long-term debt As defined in Richardson (2006). (DLC
deflated by the sum of shareholders’ +DLTT)/(DLC + DLTT + CEQ)
equity and total debt
MarginalTaxRate Marginal tax rate before interest As defined by Blouin et al. (2010). The
variable is available on WRDS
MCAP Market capitalization at the end of the year COMPUSTAT item MKVALT
MarketROA Pre-tax income minus interest expense As defined by Blaylock et al. (2017).
divided by the market value of assets (PI − XINT)/
(PRCC_F×CSHO + DLC + DLTT)
NetOperatingLoss Indicator variable equal to one if the firm As defined by Dyreng et al. (2010).
has nonmissing, nonzero tax loss carry- COMPUSTAT item TLCF
forward, and zero otherwise
NetPP&E Net property, plant, and equipment divided As defined by Blaylock et al. (2017).
by the market value of assets PPENT/
(PRCC_F×CSHO + DLC + DLTT)
NOA Net operating assets As defined by Nissim and Penman (2001).
CEQ + DLC + DLTT + UPSTK – CHE
– IVAO – MIB
OperatingAssets Operating assets As defined by Nissim and Penman (2001).
AT – CHE – IVAO
OperatingATO Operating asset turnover REVT/NOA
OperatingPM Operating profit margin OIADP/REVT
OperatingPM_BD Operating profit margin excluding OIBDP/REVT
depreciation and amortization
PM Profit margin NI/REVT
R&D Research and development expense scaled XRDt/[(ATt + ATt−1)/2]
by average total assets
Relative_Earnings_Surprise Indicator variable equal to one if EBITDA EBITDA (EPS) earnings surprise is equal
earnings surprise is positive while EPS to actual EBITDA (EPS) minus the mean
earnings surprise is negative, and zero consensus EBITDA (EPS) estimate
otherwise immediately preceding the earnings
announcement. Data are extracted from
the I/B/E/S U.S. summary files
RNOA Return on net operating assets OIADPt/NOAt−1
ROA Return on assets NIt/ATt−1
ROE Return on equity NIt/CEQt−1
ROOA Return on operating assets OIADPt/OperatingAssetst−1
SalesGrowth Change in sales in year t−1 (SALEt−1/SALEt−2) – 1
SalesStd Standard deviation of sales in the previous Standard deviation is calculated based on
three years scaled by 1,000 the COMPUSTAT variable SALE
SG&A Selling, general, and administrative As defined by Dyreng et al. (2010). XSGA/
expenses scaled by sales SALE
Size Size of the firm at the beginning of the As defined by Richardson (2006). Log(ATt−1)
year
StockReturns Annual compounded stock returns As defined by Richardson (2006). Change
in the market value of the firm over the
prior year
V/P Measure of growth opportunities As defined by Richardson (2006).
VAIP/MCAP
VAIP Value of the firm As defined by Richardson (2006). (1 − αr)
BV + α(1 + r)X − αrd, where
α = (ω/(1 + r − ω)), r = 12 percent, and
ω = 0.62. ω is the abnormal earnings
persistence parameter from the Ohlson
(1995) framework; BV is the book value
of equity (CEQ); d is annual dividends
(DVC); and X is operating income after
depreciation (OIADP)
References
Amiram, D., E. Owens, and O. Rozenbaum. 2016. Do information releases increase or decrease information
asymmetry? New evidence from analyst forecast announcements. Journal of Accounting and Economics
62 (1): 121–38.
Bagnoli, M., M. Beneish, and S. Watts. 1999. Whisper forecasts of quarterly earnings per share. Journal of
Accounting and Economics 28 (1): 27–50.
Bennett, B., J. C. Bettis, R. Gopalan, and T. T. Milbourn. 2017. Compensation goals and firm performance.
Journal of Financial Economics 124 (2): 307–30.
Bentley, J. W., T. E. Christensen, K. H. Gee, and B. C. Whipple. 2017. Disentangling managers’ and
analysts’ non-GAAP reporting incentives. Working paper, University of Massachusetts Amherst.
Bhattacharya, N., E. Black, T. E. Christensen, and D. Mergenthaler. 2004. Empirical evidence on recent
trends in pro forma reporting. Accounting Horizons 18 (1): 27–43.
Bhattacharya, N., E. Black, and T. E. Christensen. 2007. Who trades on pro forma earnings information?
The Accounting Review 82 (3): 581–619.
Black, D. E., and T. E. Christensen. 2009. US managers’ use of ‘pro forma’ adjustments to meet strategic
earnings targets. Journal of Business Finance and Accounting 36 (3–4): 297–326.
Black, D. E, E. L. Black, T. E. Christensen, and K. H. Gee. 2017. The use of non-GAAP performance mea-
sures in executive compensation contracting and financial reporting. Working paper, Dartmouth
College.
Black, E., T. E. Christensen, T. Joo, and R. Schmardebeck. 2017. The relation between earnings manage-
ment and pro forma reporting. Contemporary Accounting Research 34 (2): 750–82.
Blaylock, B., F. Gaertner, and T. Shevlin. 2017. Book-tax conformity and capital structure. Review of
Accounting Studies 22 (2): 903–32.
Blouin, J., J. Core, and W. Guay. 2010. Have the tax benefits of debt been overestimated? Journal of
Accounting and Economics 98 (2): 195–213.
Bowen, M., K. Davis, and D. Matsumoto. 2005. Emphasis on pro forma versus GAAP earnings in quarterly
press releases: Determinants, SEC intervention, and market reaction. The Accounting Review 80 (4):
1011–38.
Brown, N. C., T. E. Christensen, and W. B. Elliot. 2012. The timing of quarterly ‘Pro Forma’ earnings
announcements. Journal of Business Finance and Accounting 39 (3–4): 315–359.
Brown, N. C., T. E. Christensen, A. Menini, and T. D. Steffen. 2017. Non-GAAP earnings disclosure and
IPO pricing. Working paper, University of Delaware.
Bushee, B. 1998. The influence of institutional investors on myopic R&D investment behavior. The
Accounting Review 73 (3): 305–333.
Campbell, D. T. 1979. Assessing the Impact of Planned Social Change. Evaluation and Program Planning
2 (1): 67–90.
Carter, M. E., L. J. Lynch, and I. Tuna. 2007. The role of accounting in the design of CEO equity compensa-
tion. The Accounting Review 82 (2): 327–57.
Chava, S., and M. Roberts. 2008. How does financing impact investment? The role of debt covenants.
Journal of Finance 63 (5): 2085–121.
Christensen, T. E., H. Pei, S. Pierce, and L. Tan. 2017. Non-GAAP reporting following debt covenant viola-
tions. Working paper, Columbia University.
Cohen, N., S. Katz, S. Mutlu, and G. Sadka. 2017. Does existing debt covenant tightness affect leverage:
Evidence from SFAS 160 during the financial crisis. Working paper, Columbia University.
Dechow, P. M., M. R. Huson, and R. G. Sloan. 1994. The effect of restructuring charges on executives’ cash
compensation. The Accounting Review 69 (1): 138–56.
Doyle, J., R. Lundholm, and M. Soliman. 2003. The predictive value of expenses excluded from pro forma
earnings. Review of Accounting Studies 8 (2–3): 145–74.
Doyle, J., J. Jennings, and M. Soliman. 2013. Do managers define non-GAAP earnings to meet or beat ana-
lyst forecasts? Journal of Accounting and Economics 56 (1): 40–56.
D’Souza, J. D., K. Ramesh, and M. Shen. 2010. Disclosure of GAAP line items in earnings announcements.
Review of Accounting Studies 15 (1): 179–219.
Dyreng, S. D., M. Hanlon, and E. L. Maydew. 2010. The effects of executives on corporate tax avoidance.
The Accounting Review 85 (4): 1163–89.
Dyreng, S. D., R. Vashishtha, and J. Weber. 2017. Direct evidence on the informational properties of
earnings in loan contracts. Journal of Accounting Research 55 (2): 371–406.
Fazzari, S., R. G. Hubbard, and B. C. Petersen. 1988. Financing constraints on corporate investment.
Brookings Papers on Economic Activity 1988 (1): 141–95.
Francis, J., K. Schipper, and L. Vincent. 2003. The relative incremental explanatory power of earnings and
alternative (to earnings) performance measures for returns. Contemporary Accounting Research 20 (1):
121–64.
Goldman, N. 2016. The effect of tax aggressiveness on investment efficiency. Working paper, University of
Arizona.
Graham, J. R., C. R. Harvey, and S. Rajgopal. 2005. The economic implications of corporate financial
reporting. Journal of Accounting and Economics 40 (1–3): 3–73.
Greene, W. H. 2007. Econometric analysis, 6th ed. Upper Saddle River, New Jersey: Pearson Prentice Hall.
Hubbard, R. G. 1998. Capital-market imperfections and investment. Journal of Economic Literature 36 (1):
193–225.
Ittner, C. D., D. F. Larcker, and M. V. Rajan. 1997. The choice of performance measures in annual bonus
contracts. The Accounting Review 72 (2): 231–55.
Jensen, M. 1986. Agency costs and free cash flow, corporate finance and takeovers. American Economic
Review 76 (2): 659–65.
Johnson, W. B., and W. C. Schwartz. 2005. Are investors misled by “pro forma” earnings? Contemporary
Accounting Research 22 (4): 915–63.
Kang, S. H., J. S. Moon, O. Rozenbaum, and Z. Zhu. 2017. The information content and voluntary disclo-
sure effect of EBITDA. Working paper, George Washington University.
Kothari, S. P., S. Shu, and P. Wysocki. 2009. Do managers withhold bad news? Journal of Accounting
Research 47 (1): 241–75.
Lambert, R. A. 2001. Contracting theory and accounting. Journal of Accounting and Economics
32 (1–3): 3–87.
Lambert, R. A., and D. F. Larcker. 1987. An analysis of the use of accounting and market measures of per-
formance in executive compensation contracts. Journal of Accounting Research 25 (Supplement):
85–125.
Lougee, B., and C. Marquardt. 2004. Earnings informativeness and strategic disclosure: An empirical exami-
nation of “pro forma” earnings. The Accounting Review 79 (3): 769–95.
McNichols, M. F., and S. R. Stubben. 2008. Does earnings management affect firms’ investment decisions?
The Accounting Review 83 (6): 1571–603.
Nissim, D., and S. Penman. 2001. Ratio analysis and equity valuation: from research to practice. Review of
Accounting Studies 6 (1): 109–54.
Ohlson, J. 1995. Earnings, book values and dividends in equity valuation. Contemporary Accounting
Research 11(2): 661–87.
Prawitt, D. S., N. Y. Sharp, and D. A. Wood. 2012. Internal audit outsourcing and the risk of misleading or
fraudulent financial reporting: Did Sarbanes-Oxley get it wrong? Contemporary Accounting Research
29 (4): 1109–36.
Richardson, S. 2006. Over-investment and free cash flow. Review of Accounting Studies 11 (2–3): 159–189.
Shalev, R., I. X. Zhang, and Y. Zhang. 2013. CEO compensation and fair value accounting: Evidence from
purchase price allocation. Journal of Accounting Research 51 (4): 819–54.
Smith, M., and B. Stradley. 2010. New research tracks the evolution of annual incentives plans. Towers
Watson Executive compensation bulletin, February 25, 2010.
Stulz, R. 1990. Managerial discretion and optimal financing policies. Journal of Financial Economics 26
(1): 3–27.