TSpace Research Repository - tspace.library.utoronto.ca
Accounting Adjustments and the Valuation of Financial Statement Note Information in 10-K
Filings
Gus De Franco, M. H. Franco Wong and Yibin Zhou
Version Post-print/Accepted Manuscript
Citation (published version)
Wong, M. H. Franco, Gus De Franco and Yibin Zhou. Accounting Adjustments and the Valuation of Financial Statement Note Information in 10-K Filings. The Accounting Review. (2011), Vol. 86, No.5, pp.1577-1604. DOI: 10.2308/accr-10094
How to cite TSpace items Always cite the published version, so the author(s) will receive recognition through services that track
citation counts, e.g. Scopus. If you need to cite the page number of the author manuscript from TSpace because you cannot access the published version, then cite the TSpace version in addition to the published
version using the permanent URI (handle) found on the record page.
This article was made openly accessible by U of T Faculty. Please tell us how this access benefits you. Your story matters.
Accounting Adjustments and the Valuation of Financial Statement Note
Information in 10-K Filings
Gus De Franco
Rotman School of Management
University of Toronto
105 St. George Street
Toronto, Canada, M5S 3E6
(416) 978-3101
M.H. Franco Wong
Rotman School of Management
University of Toronto
105 St. George Street
Toronto, Canada, M5S 3E6
(416) 946-0729
Yibin Zhou
School of Management
University of Texas at Dallas
800 W. Campbell SM41
Richardson, TX 75083
(972) 883-2738
March 10, 2010
We appreciate the helpful comments of Ole-Kristian Hope, Hai Lu, Chia Chun Hsieh, Artur Hugon, Michael
Mikhail, Visarut Sribunnak, and workshop participants at Arizona State University, Chulalongkorn University, the
University of Toronto, and the 2009 AAA Annual Meeting. We thank Andrea Blackman, Philip Robinson, and
John Sullivan of Moody’s Investors Service for their assistances and thorough feedback. Richard Gao, Yang Han,
Yu Yeung (Kenny) Ho, Philip Rabenok, and Konstantin Shestopaloff provided excellent research support. We
thank I/B/E/S Inc. for the analyst earnings forecast data, available through the Institutional Brokers Estimate
System. The data has been provided as part of their academic program to encourage earnings expectation research. We gratefully acknowledge the financial support of the Social Sciences and Humanities Research Council of
Canada, the Rotman School of Management at the University of Toronto, and the School of Management at the
University of Texas at Dallas. All errors remain our responsibility.
Accounting Adjustments and the Valuation of Financial Statement Note
Information in 10-K Filings
Abstract
We examine the valuation of financial statement note information at the time of 10-K
filings. We conjecture that financial statement users explore the information in these notes to
compute accounting adjustments to correct the imperfections in financial statements. We find
that stock returns around 10-K filings are positively related to accounting adjustments, calculated
from note information. We further find that the likelihood of equity analysts updating their target
price estimates at the 10-K dates is increasing in the magnitude of the adjustments. Those
analysts who do update their target prices at this time revise their estimates consistent with the
sign and magnitude of the adjustments. These findings are consistent with financial statement
users using financial statement note information to make accounting adjustments, thereby
incorporating this information into stock prices.
Keywords: Accounting adjustments; financial statement note information; 10-K filings; equity
analysts; target price estimates
JEL Classifications: G14; G29; M40; M41; M44
1
Accounting Adjustments and the Valuation of Financial Statement Note
Information in 10-K Filings
1. Introduction
This study examines the equity market’s and equity analysts’ responses to financial
statement note information at the time of 10-K filings. Financial statement note information is
an integrated part of the financial statements. The amount of information disclosed in the notes
has increased over time because business transactions conducted by companies have become
more complex and user demand for more supplementary information has risen. The note
information first becomes available to the public when companies file their 10-K reports. Prior
studies examining the information content of 10-K filings document mixed evidence (e.g.,
Foster et al. 1978; Stice1991; Easton and Zmijewski 1993; Qi et al. 2000; Griffin 2003;
Asthana et al. 2004). However, these researchers use unsigned market reaction measures and
their tests might not be as powerful had they been able to isolate the market response to a
specific set of information in the 10-K reports. On the other hand, numerous studies have
documented the value relevance of individual accounting item disclosed in the notes to the
financial statements using a long-window research design (see, e.g., Landsman 1986; Barth,
Beaver, and Landsman 1992; Bowman 1980; Ely 1995; Aboody, Barth, and Kasznik 2004;
Landsman, Peasnell, and Shakespeare 2008). In this study, we draw on these two strands of
literature by investigating the market reaction to a comprehensive set of note information
around its release in the 10-K filings.
We conjecture that detailed note information is impounded into stock prices around 10-K
filings, because financial statement users explore note information when conducting accounting
analysis and computing accounting adjustments. Financial statements are not perfect and,
2
hence, financial statement users conduct rigorous accounting analysis to improve the
usefulness of the financial statements for assessing the economic performance and health of the
company in question.1 A key product of accounting analysis is accounting adjustments, which
are calculated using information disclosed in financial statement notes. For example, note
information can be used to capitalize the assets and liabilities associated with operating leases.
These adjustments correct missing or misclassified expense and revenue items, and thus make
the adjusted income statement numbers more informative of the economic performance of a
company.2
To shed further light on our conjecture that note information is incorporated into stock
prices because financial statement users employ it to conduct accounting analysis and compute
accounting adjustments, we examine the role played by equity analysts in this process. Equity
analysts are sophisticated financial statement users, who routinely study the financial
statements and associated notes of the companies they cover. We hypothesize that equity
analysts use the note information to assess and calculate the accounting adjustments needed to
mitigate the shortcomings in the financial statements. When the amount of the adjustments
required is different from the analysts’ prior expectations, it will affect their assessments of the
1 Financial statements are not perfect because they are prepared under Generally Accepted Accounting Principles
(GAAP), which have shortcomings, including not accurately reflecting the economics of the underlying
transactions (gaps) and providing too much flexibility. One concern with gaps in GAAP is that managers can
take advantage of the gaps by strategically structuring a transaction to obtain a management-preferred
accounting treatment. Entering into operating leases and granting at-the-money employee stock options (before
the SFAS 123R period) are two examples. One concern with flexibility in GAAP is that management can
choose alternatives that benefit them to the detriment of other stakeholders. 2 Accounting analysis and adjustments are discussed ubiquitously in financial reporting and financial statement
analysis courses and textbooks. See, e.g., Palepu, Healy, and Bernard (2004), Penman (2006), Revsine, Collins,
and Johnson (2004), Stickney, Brown, and Wahlen (2007), White, Sondhi, and Fried (2002), Wild,
Subramanyam, and Halsey (2006), and Koller et al. (2005). As a specific example, Palepu et al. (2004, 3-1)
state: “The purpose of accounting analysis is to evaluate the degree to which a firm’s accounting captures its underlying business reality. By identifying places where there is accounting flexibility, and by evaluating the
appropriateness of the firm’s accounting policies and estimates, analysts can assess the degree of distortion on a
firm’s accounting numbers. Another important skill is adjusting a firm’s accounting numbers…to ‘undo’ any
accounting distortions....” The importance of adjusted financial statement information is also stressed in the
valuation of companies because accounting forecasts are a critical input for various valuation models.
3
company’s future performance and valuation. Given the new information, analysts revise their
target price estimates, thereby impounding financial statement note information into stock
prices.
We test our conjecture using a large sample of firms with adjusted income statement data
from Moody’s Corporate Financial Metrics (MFM). MFM systematically makes standard
adjustments to eight accounting items, as well as non-standard adjustments to highly
judgmental accounts (Moody’s 2006). The methodology used by MFM is similar to the
treatments found in many financial reporting and financial statement analysis books. Hence,
financial statement users can potentially construct a similar set of accounting adjustments,
making the MFM numbers a reasonable proxy for the accounting adjustments that financial
statement users would make, given the note information.
Our sample includes a panel of non-financial firms, from 2002 to 2007, with non-
missing MFM adjusted income statement data, the dates of the 10-K filing from the SEC, and
analysts’ reports from Investext.3 The results of our empirical analysis support our conjecture.
First, stock returns around 10-K filings are positively related to the changes in accounting
adjustments. This is consistent with equity market participants finding the accounting
adjustments and, hence, the note information used to compute them, useful in determining firm
value. Second, analysts are more likely to revise their target price estimates shortly after 10-K
date for companies that require a large unexpected accounting adjustment (in either direction).
Third, those analysts who do update their target prices in a short window around the 10-K date,
3 Although the MFM data are available mainly for firms with Moody’s rated debts, discussions with Moody’s staff
indicates that the clients of MFM include equity institutional investors. This suggests that the MFM adjustments
appeal to more than debt investors and analysts. The fact that the MFM data are geared toward debt market
participants might be an advantage for us in the sense that our results are less likely attributed to a mechanical
relation between our variables of interest (stock returns and equity analysts’ actions) and the MFM-based
accounting adjustments.
4
revise their target price estimates consistent with the sign and magnitude of the change in
accounting adjustments. The last two pieces of evidence lend support for our hypothesized
mechanism by which detailed financial statement note information is impounded into security
prices.
Our study contributes to four strands of literature. First, our results contribute to the 10-K
literature. While early studies such as Foster et al. (1978), Stice (1991), and Easton and
Zmijewski (1993) find little evidence of a 10-K market reaction, later studies such as Qi et al.
(2000), Griffin (2003), and Asthana et al. (2004) document a significant market response to 10-
K reports during the EDGAR era. However, these studies are silent as to what specific piece of
information in the 10-K reports investors are reacting to. These studies use unsigned measures
such as return volatility, absolute value of returns, or trading volume around the 10-K filing
date. Balsam, Bartov, and Marquardt (2002) find that the releases of the full set of financial
statements at 10-Q filings allow investors to estimate discretionary accruals and respond to it
accordingly at the 10-Q filing dates. Our study is similar to Balsam et al. (2002) in that we
identify specific information in the 10-K report, namely notes to the financial statements, and
use it to predict ex ante how the market would react. Hence, we contribute to this literature by
showing that both the stock market and equity analysts react to the financial statement note
information in a predictable way at the time of 10-K filings.4
Second, we add to the value relevance literature, which shows that financial statement note
disclosures are reflected in equity prices (see, e.g., Landsman 1986; Barth, Beaver, and
Landsman 1992; Bowman 1980; Ely 1995; Aboody, Barth, and Kasznik 2004; and Landsman,
Peasnell, and Shakespeare 2008). In particular, we find evidence consistent with equity
4 Li (2008), Brown and Tucker (2009) and Feldman et al. (2009) examine the information content in the10-K’s
MD&A section using textual analysis.
5
analysts using detailed note information to make accounting adjustments, which induce them to
revise their target price estimates that in turn impact stock prices. By tying the price
implication of the note information to a subset of the sophisticated financial statement users,
we help rule out the possibility that the results reported in prior studies are partly attributed to
spurious correlation. We also conduct our empirical analysis using a short-window event-study
design, which allows us to tightly link the availability of note information to the market’s
reaction and analysts’ actions, thereby mitigating the possibility of correlated omitted variables
problem that affects prior studies using a long-window research design (see, Barth, Beaver and
Landsman 2001 and Holthausen and Watts 2001 for further discussions).
Third, our results provide additional support for the role of analysts as information
intermediaries as opposed to information gatherers (see, e.g., Lang and Lundholm 1996).
Analysts process the note information in company 10-K reports and transmit it to the capital
market in the form of target price revisions. Fairfield and Whisenant (2001) show that the
Center for Financial Research and Analysis (CFRA) successfully identifies companies with
aggressive accounting and document that companies “caught” by CFRA experience a drop in
operating performance and exhibit 10% negative abnormal returns in the following year. CFRA
applies its custom analysis on a small set of firms, which tend to have highly unusual activities
or very aggressive accounting for a specific accounting item. In contrast, our study uses MFM
data, which apply a comprehensive and systematic set of adjustments to a large sample of
firms.5
5 Other studies also use Moody’s MFM data. Kraft (2008) and Batta, Ganguly, and Rosett (2009) examine the
relation between Moody’s adjustments and bond ratings and yields. Ge, Imhoff, and Lee (2008) use Moody’s
data to investigate investors’ assessment of equity risk attributed to operating leases and find that it is not
reflected in stock price for companies not covered by MFM.
6
Fourth, we supplement prior studies examining other types of adjusted earnings: Pro
forma earnings (e.g., Bradshaw and Sloan 2002; Bhattacharya et al. 2003; Lundholm et al.
2003) and economic value added (e.g., Biddle, Bowen, and Wallace 1997). Pro-forma earnings
are GAAP earnings excluding certain costs that the company deems to be transitory in nature,
such as one-time losses, asset write-downs, and restructuring charges. MFM adjustments
include not only these types of adjustments but also adjustments to other items and, hence, they
are broader than pro-forma earnings. While the MFM adjustments are systemically applied, pro
forma earnings are incomparable across companies, because the latter are prepared by the
companies themselves.6 Economic value added (EVA
®) is residual income computed using
accounting numbers adjusted by Stern Stewart & Co. Biddle et al. (1997) find that the
accounting adjustment component of EVA has an insignificant association with market-
adjusted returns, suggesting that EVA accounting adjustments are of little value. While there is
some overlap in the adjustments made by Stern Stewart & Co. and Moody’s MFM, such as
how operating leases are handled, Stern Stewart’s adjustments are not identical to MFM and it
does not apply the same set of adjustments systematically to all companies. Our study also
differs from this literature as we conduct our tests using a short-window event study design.
We also investigate the actions of equity analysts in the incorporation of financial statement
note information into stock price.
The next section describes our data and provides summary statistics on Moody’s
adjustments. Section 3 investigates the market response to accounting adjustments around the
release of financial statements at 10-K filing dates. Section 4 examines the role of equity
6 The SEC finds pro-forma earnings potentially misleading if not properly disclosed. In January 2003, the SEC
enacted Regulation G that became effective March 28, 2003 regulating companies’ presentation of pro forma
earnings, including the requirement that companies to reconcile pro forma numbers to GAAP earnings.
7
analysts in the incorporation of the accounting adjustments and, hence, financial statement note
information into stock prices. Section 5 concludes.
2. Data
2.1. Sample selection
The initial sample includes all companies in Moody’s Corporate Financial Metrics (MFM)
database, which is our source of the adjusted financial statement data. We download all annual
financial statement data, including GAAP reported and MFM adjusted balance sheets and
income statements for the time period from 2002 until 2007, for U.S. non-financial companies.
This process results in 8,327 firm-year observations for 1,564 U.S. firms. We match the MFM
data with the CRSP-COMPUSTAT merged database (CCM) at the firm level by equity ticker
symbol and firm name. We hand-collect target price estimates from analysts’ reports
downloaded from Investext; we match our sample firms to Investext by name and ticker
symbol. Our requirement of stock returns limits the analysis to firms with publicly traded
equity. By using some financial statement data from CCM, we avoid losing one year of sample
observations. For example, in certain tests we use lagged assets as a scaler. After this step, our
sample contains 4,944 firm-year observations for 958 U.S. firms.
Our empirical tests focus on a short window centered on the 10-K filing date. We obtain
the filing dates from the Securities and Exchange Commission. While it is possible that
Moody’s analysts have obtained 10-K information privately before the 10-K is publicly filed,
MFM does not publicly issue their adjusted data for at least seven to ten business days
following the 10-K filing date. Hence for our primary tests that focus on the 10-K event
8
window, equity analysts and other market participants have access to 10-K information but do
not have access to MFM adjusted numbers.
2.2. Accounting adjustments
Our source of accounting adjustment data is Moody’s Corporate Financial Metrics (MFM)
database.7 MFM systematically adjusts the financial statements of non-financial corporations to
improve consistency in its rating practice among analysts. The adjustments are reviewed and
approved by its team of analysts and accounting specialists. Appendix A and Moody’s (2006)
provide a detailed description of these adjustments and the underlying methodology.
MFM systematically makes standard adjustments, if applicable, to underfunded defined
benefit pensions, operating leases, capitalized interest, employee stock compensation, hybrid
securities, securitizations, unusual and non-recurring items, and inventory on a LIFO cost
basis. Underfunded defined benefit pensions, operating leases, hybrid securities, and
securitizations lead to understated liabilities. MFM adds these debt-like liabilities to the
balance sheet and corrects the classification of related expenses on the income statement.
Adjustments for capitalized interest and employee stock compensation add missing interest
expense and compensation expense, respectively, back to the income statement. Adjustments
for unusual and non-recurring items include moving transitory components to below net
income before unusual items. MFM also makes non-standard adjustments to other accounts to
make the financial statements more reflective of the company’s underlying economics. These
7 While there are some other data providers of accounting adjustments, such as Capital IQ and Bloomberg, their
adjustment data suffer a number of limitations. For example, Capital IQ, which has more adjustment information
than Bloomberg, does not have any adjustment information on pensions, capitalized interests, hybrids, and
makes no non-standard adjustments like Moody’s does. In addition, when Capital IQ provides some adjustment
information, for example on operating leases, Capital IQ fails to fully articulate how this adjustment information
affects the complete set of financial statements, like Moody’s does.
9
involve judgmental items such as asset valuation allowances, impairments of assets, and
contingent liabilities. Table 1 summarizes each of these adjustments and their effects on the
income statement.
We believe these adjustments correct missing or misclassified expense and revenue items
on the income statement, making the adjusted earnings subtotal numbers more useful. In
particular, adjusted earnings before interest and taxes (EBIT) better reflects the operating
performance of the company, while the adjusted items between EBIT and net income before
unusual items better reflects the financing and tax strategies of the company. Finally, the
adjusted items below net income before unusual items will better reflect the transitory
component of earnings.8
Table 2, panel A provides descriptive statistics about the effect of each adjustment on net
income before unusual items. Consider adjustments to underfunded defined benefit pensions.
The mean adjustment for all firm-year observations is -0.22% of (lagged total) assets,
consistent with the frequency and magnitude of negative adjustments outweighing that of
positive adjustments. The table shows that negative income adjustments occur for 47% of
observations while 21% of adjustments affect income positively. For the 5% of observations
with the most negative adjustments, income is adjusted downward by 1.35% or more of assets.
In contrast, the 95% percentile of adjustments is 0.17% of assets.
The standard deviation column indicates that some adjustments vary considerably more
than others. Most of the variation in the Total adjustments, the sum of the individual
adjustments, is driven by variation in unusual and non-recurring items, followed by non-
standard adjustments, stock compensation and underfunded pensions. The mean Total
8 Financial statement analysis and adjustments also have the potential to improve accounting comparability across
companies. In this study, we do not examine the effects of adjustments on comparability, using for example
measures as in De Franco, Kothari, and Verdi (2009).
10
adjustment to income is -0.05% of assets, consistent with most Total adjustments being
negative (i.e., 68% of observations).
Panel B of table 2 provides the same descriptive statistics for the effect of each adjustment
on EBIT. The prevalence of positive adjustments to EBIT is much higher than the
corresponding adjustments to net income before unusual items reported in panel A. For
example, the mean Total adjustment is 1.05% of assets, compared with that of -0.05% on net
income before unusual items. Of the Total adjustments, 66.5% result in an increase to EBIT,
compared to only 30.5% of adjustments (from Panel A) leading to an increase in net income
before unusual items. This pattern is consistent with many of the adjustments decreasing
expenses before interest and taxes (hence, EBIT increases) and increasing expenses below
EBIT (hence, net income before unusual items decreases).
2.3. Statistical properties of adjusted earnings
Financial statement users exert effort to conduct analysis of and make adjustments to the
financial statements because they believe that the adjusted earnings numbers have the ability to
better reflect the underlying economics of the company than the reported GAAP numbers. In
this subsection, we compare the statistical properties of MFM adjusted earnings with those of
GAAP earnings to provide support for such an assumption.
We examine seven earnings properties: accrual quality, persistence, predictability,
smoothness, value relevance, timeliness, and conservatism. For each earnings property measure
and industry, we estimate one measure using MFM adjusted data and one using reported
GAAP data. We then produce one matched-pair difference per industry for each measure. We
form 32 industry groups using Moody’s industrial classification, with each industry group
11
containing between 7 and 94 companies. Based on the 32 industry matched-pair differences,
we conduct a parametric t-statistic and a non-parametric binomial test. Appendix B describes
the estimation of these properties in detail.
Table 3 presents summary statistics and test results. The average number of firm-year
observations per industry varies by earnings property because the calculation of some
properties requires additional data.9 The first row shows that the mean and median Accrual
Quality for MFM adjusted data is greater than that for reported GAAP data. The mean
difference of 0.003 is about a 7.1% (=0.003/0.042) “improvement” over mean GAAP Accrual
Quality and is significantly different than zero. Of the 32 matched-pair differences, 84% (i.e.,
27) are positive. A binomial test indicates that this percentage is significantly different than
50%. The other clear difference is Predictability. The mean and median Predictability for
Moody’s is greater than that for reported GAAP data. The mean matched-pair difference of
0.207 represents a statistically significant 14.5% (=0.207/1.425) improvement over the mean
reported GAAP value. A binomial test also confirms the statistical significance of this result.
As for the other five earnings property measures, the t-statistics provide no evidence of a
statistical difference between measures estimated using MFM and reported GAAP data. The
binomial tests, however, indicate that more often than not Smoothness and Timeliness is greater
for MFM adjusted data than for reported GAAP data. A binomial test also indicates that
MFM’s Conservatism is lower than GAAP Conservatism.
In sum, the results reported in table 4 indicate that the MFM adjusted earnings exhibit
9 The underlying sample for this analysis is the same one used to calculate the Table 2 descriptive statistics,
subject to the following additional requirements. In the case of the accrual quality tests, besides current-year information, we also require lead and lag operating cash flows to estimate the model, which hence removes the
first and last year of observations per firm. Similarly, for the persistence and predictability measures we require
lagged EPS to run the AR(1) time-series model, and for the value relevance measure, we require change in
earnings. For the market-based properties (value relevance, timeliness and conservatism) we also require return
data from CRSP.
12
higher accrual quality (as in Dechow and Dichev 2002) and more predictability (as in Lipe
1990) and, to a lesser degree, are more smooth and timely than the GAAP reported earnings.
3. Stock Market Reaction around 10-K Filings
We investigate stock market reaction to the accounting adjustments around the release of
the 10-K reports to test the market valuation of the financial statement note information. The
10-K filing event is critical because the information needed to calculate the adjustments (e.g.,
the interest on the projected benefit obligation, actual gains or losses on pension assets, the
interest and depreciation components of operating lease expense, the amount of capitalized
interest, or employee stock option costs) is included in the notes to financial statements filed
with the 10-K report and is typically not available before that time. We test whether the
adjustments are associated with the market reaction around the 10-K filing using the following
regression equations:
CARit = β0 + β1 ∆AdjNIit + β2 AdjUNUSUALit + β3 LagAdjUNUSUALit
+ β4 EarnSurpriseit + β5 Book-Marketit + β6 Sizeit + ε (1)
CARit = β0 + β1 ∆AdjEBITit + β2 ∆AdjOTHit
+ β3 AdjUNUSUALit + β4 LagAdjUNUSUALit
+ β5 EarnSurpriseit + β6 Book-Marketit + β7 Sizeit + ε (2)
CARit is the cumulative abnormal return obtained from CRSP from one day before to five days
after the year-t 10-K filing date. Abnormal returns are computed as firm raw returns less the
CRSP size-matched decile index returns. The 10-K filing date is obtained directly from the
SEC’s EDGAR website.
∆AdjNI is Moody’s adjustments to firm i’s net income before unusual items for year t less
Moody’s adjustments for year t-1, scaled by lagged total assets. By subtracting last year’s
adjustments, we are implicitly using last year’s adjustments as analysts’ expected amount. The
13
admittedly coarseness of, and hence the noise in, our measure of expectations should weaken
the power of our tests.10
The second equation decomposes the net income (before unusual
items) adjustments into those that affect EBIT and those that affect the difference between
EBIT and net income (before unusual items). ∆AdjEBIT is Moody’s adjustments to firm i’s
EBIT for year t less Moody’s adjustments for year t-1, scaled by beginning total assets, and
∆AdjOTH is ∆AdjNI less ∆AdjEBIT. EBIT is not only an important fundamental for debt
analysis, but is also an important accounting metric used in equity valuation.
Adjustments to net income (or its components) exclude the adjustments to unusual and
non-recurring items. Instead we augment the specification with the variable AdjUNUSUAL,
which is Moody’s unusual and non-recurring adjustments to firm i’s net income for year t. For
these adjustments, the predictions are not clear. By definition these items are not expected to
persist in the future, so we do not use last year’s amounts as an expectation. Hence, we use the
level of this variable rather than the change. We expect the coefficient on this value to be
significantly smaller compared to that on adjustments to EBIT or net income before unusual
items, which we do expect to persist. However, prior research shows that these types of items
do persist, at least in part (e.g., restructuring charges). If so, the contemporaneous value in
levels may contain significant noise. To help alleviate any possible noise we include the
previous year’s value LagAdjUNUSUAL in the model. If there is persistence in these
adjustments then we would expect this variable to load in a direction opposite to the coefficient
on its contemporaneous counterpart, AdjUNUSUAL.
Using earnings announcement dates from I/B/E/S, we find that about 26% of our
observations have earnings released within the 10-K filing window. To control for the
10 In subsequent sensitivity analysis, we control for the possibility that some of this information has already
appeared in 10-Q reports.
14
confounding effect of the earnings announcement, we include in the model the variable
EarnSurprise, which equals firm i’s earnings surprise if earnings is announced inside the 10-K
filing window, and zero otherwise. We compute earnings surprise as actual earnings per share
(EPS) for period t minus the mean of analysts’ EPS forecast, scaled by stock price measured at
the end of year t-1. Other firm controls include Book-Market, measured as the ratio of the book
value of equity to the market value of equity for year t, and Size, measured as the logarithm of
the market value of equity for year t. We also control for year fixed effects and cluster the
standard errors at the firm level.
Table 4, Panel A presents descriptive statistics for the variables used in these tests. The
sample for these tests consists of 3,069 firm-year observations, which is smaller than that used
in the tests reported in Tables 2 and 3 because of data requirement for the explanatory
variables. The mean seven-day cumulative abnormal return (CAR) from one day before to five
days after 10-K filing is 0.2%, with a standard deviation of 4.4%. The average change in the
adjustment to net income before unusual items is 0.3% of lagged total assets. The book-to-
market ratio of the average firm is 0.436.
Panel B of Table 4 summarizes the estimation of Equations 1 and 2. Column (1) shows
that the estimated coefficient on ∆AdjNI is 0.075 and statistically different from zero,
suggesting that positive adjustments to net income are associated with positive stock returns
around the 10-K filings. (Significance levels are based on one-tailed tests where there is a
prediction for the sign of the coefficient and based on two-tailed tests otherwise for all tests in
this study.) Similarly, Column (2) finds that the estimated coefficients on ∆AdjEBIT and
∆AdjOTH, the two components of ∆AdjNI, are 0.096 and 0.060, respectively. This result
indicates that the market reacts more to the adjustments to EBIT than to the adjustments to the
15
line items below EBIT. Furthermore, AdjUNUSUAL exhibits a statistically negative association
with stock return around the 10-K filing date. Taken together, these results are consistent with
stock market participants using 10-K note information to analyze accounting issues and to
compute the appropriate adjustments to the financial statements. As a result, the detailed note
information is incorporated into stock price because it is being used by financial statement
users to compute accounting adjustments.
Some of the note information needed to analyze the accounting issues or make
adjustments appears in the 10-Q quarterly reports and equity market participants can
potentially take advantage of these interim data to extrapolate the annual figures. As a result,
the market’s expectation of the adjustments for the current year will not be based on last year’s
numbers and we will measure unexpected adjustments with errors. To allow for such a
possibility, we add to the regressions the corresponding changes in the adjustment terms
cumulated over the first three quarters of the year (∆Adj3Q_DUM, ∆AdjNI3Q, ∆AdjEBIT3Q,
and ∆AdjOTH3Q). Table 4, Columns (3) and (4) show that, even after allowing for the update
of market’s expectation using quarterly note information, the results on all the annual
adjustment terms are similar to those reported in columns (1) and (2), and hence corroborate
our original inferences. In particular, the market values ∆AdjNI, ∆AdjEBIT, and ∆AdjOTH
positively, but not the quarterly update terms.
Taken together, the findings reported in table 4 are consistent with financial statement
users employing the 10-K note information to analyze accounting issues and make accounting
adjustment, thereby impounding the note information into stock prices. These results add to the
literature on the information content of 10-K filings by showing that notes to the financial
statements, a subset of the information in the 10-K report, is informative. Further, these results
16
add to the value relevance literature by showing that note disclosures are also priced in a short-
window around their releases.
4. Analysts’ Target Price Revisions around 10-K Filings
The results from the last section are consistent with the accounting adjustments and the
corresponding financial statement note information being informative. In this section, we
examine the potential mechanism by which this note information is incorporated into stock
prices. By doing so, we hope to better understand the underlying link between the note
information and stock prices and to rule out the possibility that the documented results are
spurious.
Specifically, we investigate the actions taken by equity analysts in responding to the
release of note information in the 10-K filings. Equity analysts are known to scrutinize the
financial statements and the associated notes of the companies they follow. If analysts use the
note information to compute the necessary accounting adjustments, any news in accounting
adjustments will change analysts’ assessments of the company being analyzed.11
In Section 4.1, we examine whether analysts are more likely to revise their target price
estimates for companies that require a large unexpected accounting adjustment to their GAAP
earnings. In Section 4.2, we test whether the direction of the target price revision is
proportional to the direction of the unexpected accounting adjustment, conditional on analysts
making an accounting adjustment. Unlike the tests in section 3, which are conducted using
firm-year data, the tests in this section are conducted at the analyst-firm-year level.
11 For example, retail-industry analysts often capitalize operating leases (e.g., Lejuez et al. 2006). Some large
brokerage firms have separate analysts who engage exclusively in accounting analysis. These analysts often
monitor the views and regulations of the SEC or the FASB (e.g., Zion and Carcache 2005) or focus on specific
accounting issues as they become more topical about a large cross-section of companies (e.g., Levinson 2006).
17
We use target price as our main variable of interest, because it represents the sum of the
discounted future cash flows and, hence, it gives us a comprehensive measure of the
importance of the adjustment information. Accounting adjustments are meant to generate a
better measure of economic earnings and they have implications for both short-term and long-
term expected cash flows. For example, an unexpectedly higher employee stock compensation
adjustment (in the pre-SFAS 123R period) can be due to an increase in the usage of stock-
based compensation, which will have an immediate effect on analysts’ forecasts of near-term
cash flows, and could via its incentive effect, have an incremental effect on long-term expected
cash flows. Revisions to target price reflect the changes in the sum of the discounted expected
future cash flows and, hence, it provides the desired measure that captures the effect of the
accounting adjustments on both short- and long-term cash flows.12
4.1. Analysts’ decision to revise their target prices around 10-K filings
We first test whether analysts are more likely to revise their target prices shortly after the
release of the 10-K reports if the magnitude of the unexpected adjustments is larger. If note
information is being used by analysts to conduct accounting adjustments, we expect that the
likelihood of analysts revising their target price estimates to increase with the absolute
magnitude of the unexpected adjustment. We test this prediction using the following logistic
models:
TgtPriceRevDumitj = β0 + β1 |∆AdjNIit|
+ β2 | AdjUNUSUALit| + β3 | LagAdjUNUSUALit |
+ β4 | EarnSurpriseitj| + β5 Book-Marketit + β6 Sizeit + ε (3)
12 An alternative is to use analysts’ earnings forecast. While earnings forecasts are more widely available than
target price estimates, the former reflect only changes in near-term GAAP-based earnings expectations. By their
nature, many accounting adjustments should not affect analysts’ assessment of future GAAP earnings.
18
TgtPriceRevDumitj = β0 + β1 |∆AdjEBITit| + β2 |∆AdjOTHit|
+ β3 | AdjUNUSUALit| + β4 | LagAdjUNUSUALit |
+ β5 | EarnSurpriseitj| + β6 Book-Marketit + β7 Sizeit + ε (4)
TgtPriceRevDumitj is an indicator variable that equals one if analyst j issues a report that
contains a target price revision for firm i one day before to five days after the year-t 10-K filing
date, zero otherwise.13
The explanatory variables are defined in section 3, under equation 2.
We take the absolute value of adjustment variables (∆AdjNI, ∆AdjEBIT, ∆AdjOTH,
AdjUNUSUAL, and LagAdjUNUSUAL) because we predict that larger adjustment changes in
either direction will cause the analyst to update their target prices. Different from the firm-
level market reaction tests in section 3, EarnSurpriseitj is specific to each analyst j.
EarnSurpriseitj equals firm i’s actual earnings per share (EPS) for period t minus analyst j’s
forecast of EPS, scaled by stock price measured at the end of year t-1 if earnings is announced
inside the 10-K filing window, and zero otherwise. We also control for year fixed effects and
cluster the standard errors at the firm level.
The sample for these tests includes all analysts who actively follow the company. In
particular, to be included in this analysis, an analyst must issue at least one report for firm i
within the six-month period ending on the day of firm i’s earnings announcements for year t
and at least one report for the same firm within the 12-month period beginning on the day of
the earnings announcement. All reports, the source of our target prices, are collected from
Investext. A total of 24,133 observations at the analyst-firm-year level meet this criterion. The
number of firms in the sample decreases from the previous analysis because of the coverage
13 To be classified as an observation in which TgtPriceRevDum = 1, a report must be issued in the 10-K window
and the report must have a target price revision. We have replicated our analysis when we only require that a
report be issued in the 10-K window (regardless of whether there is a target price revision) in order for the
observation to be classified as TgtPriceRevDum = 1. Untabulated results and inferences are similar.
19
requirement, which is generally associated with the selection of larger firms with more
actively-traded securities.
Of our analyst-firm-year observations, 1,742 or 7.2% have target-price revisions around
the 10-K filings (i.e., TgtPriceRevDum = 1). Hence, the majority of analysts do not react to the
10-K information. This low percentage might be due to at least three reasons. First, it is
possible that some of this information has been pre-empted by quarterly reports or analysts’
private research. Second, analysts might update their target price estimates after our seven-day
event window. Third, analysts might deem the required incremental adjustments immaterial (or
low signal-to-noise ratio) to alter their prior expectation. To the extent that not enough analysts
respond to the 10-K filings, it will make it difficult for us to find support for our empirical
prediction.
Table 5, Panel A presents descriptive statistics, partitioned by the dependent variable
(whether an analyst issued a target price revision around the 10-K filings), for the variables
used in these tests. Consider the first set in which a target price was revised (i.e.,
TgtPriceRevDum = 1). The mean absolute value of the annual change in adjustments to net
income (i.e., |∆AdjNI|) and EBIT (|∆AdjEBIT|) are both 0.9% of total lagged assets. The
corresponding figures for the set in which a target price was not revised (i.e., TgtPriceRevDum
= 0) are only 0.6%. Moreover, the standard deviations of these two variables are greater for
companies with than for those without a revised target price.
Panel B of Table 5 summarizes the results of estimating Equations 3 and 4. Consistent
with our prediction, column (1) indicates that the coefficient on |∆AdjNI| is positive and
significantly different than zero. This finding supports our prediction that the likelihood of
analysts revising their target price estimates increases with the magnitude of the adjustment to
20
net income before unusual items. Column (2) shows that the effect of adjustments to net
income is driven by adjustments to EBIT and not by the non-EBIT adjustments. The coefficient
on |∆AdjEBIT| is positive and significantly different than zero, while the |∆AdjOTH| coefficient
is insignificant. In other words, adjustments to the operating components of earnings (EBIT)
affect the likelihood that analysts revise their target price estimates.
There is no evidence that adjustments to unusual items in the current or lagged period
affect target price revision occurrence. In terms of control variables, the estimated coefficient
on the magnitude of earnings surprises for those firms with earnings announcements around
10-K filing is significantly positive, which is consistent with our expectation. The probability
of analysts revising target prices at the time of the 10-K filing is increasing in firm size,
possibly because the economic significance of 10-K information is greater for larger firms.
To allow for the possibility that analysts use the note information in 10-Q reports to update
their expectation for this year’s adjustments for current year, we add to the regressions the
corresponding absolute changes in the adjustment terms over the first three quarters of the year.
Columns (3) and (4) show that, even after allowing for the update of analysts’ expectation
using quarterly note information, the results on our key variables |∆AdjNI|, |∆AdjEBIT|, and
|∆AdjOTH| are similar to those reported in columns (1) and (2), and hence corroborate our
original inferences. Both |AdjUNUSUAL| and |LagAdjUNUSUAL| exhibit an insignificant effect
on the probability of analysts issuing a target price revision.
Overall, the table 5 results are consistent with analysts using the detailed note information
provided by the 10-K reports to analyze accounting issues when the potential benefits are
greater, i.e., when the unexpected adjustments to the financial statements are greater in
magnitude.
21
4.2. Analysts’ target price revisions around 10-K filings
Having shown that the likelihood of analysts issuing a target price revision is increasing
with the absolute magnitude of the accounting adjustment, we next turn to examine the effect
of the adjustments on target price revisions for those analysts who issue a revision around the
10-K filing date. We use the following linear equations, which include a set of independent
variables that are similar in spirit to equations 3 and 4. Since the dependent variable is the
signed value of analysts’ target price revisions and, hence, the accounting adjustment variables
are also based on their signed values. The regression models are:
TgtPriceRevitj = β0 + β1 ∆AdjNIit + β2 AdjUNUSUALit + β3 LagAdjUNUSUALit
+ β4 EarnSurpriseitj + β5 Book-Marketit + β6 Sizeit + ε (5)
TgtPriceRevitj = β0 + β1 ∆AdjEBITit + β2 ∆AdjOTHit + β3 AdjUNUSUALit
+ β4 LagAdjUNUSUALit + β5 EarnSurpriseitj
+ β6 Book-Marketit + β7 Sizeit + ε (6)
TgtPriceRev is analyst j’s target price forecast for firm i issued one day before to five days after
the year-t 10-K filing date minus analyst j’s prior target price forecast for firm i, scaled by the
prior target price forecast.
Since we estimate Equations 5 and 6 on a subsample of analysts who issue a target price
revision within seven days of the 10-K filings (i.e., TGTPriceRevDum=1), ordinary least
squares estimates are subject to sample selection bias. We correct for this bias using the
Heckman (1979) two-stage method, which assumes that the error terms in Equations 3 and 5
are jointly normal (the same for those in Equations 4 and 6). We include the Inverse Mills’
ratio, computed from the Logistic estimates of Equations 3 and 4, into Equations 5 and 6,
respectively, as an additional explanatory variable.
22
Descriptive statistics for the variables used in these tests are presented in Panel A of Table
6. The mean target price revision is 0.034. The average annual change in the adjustment to net
income before unusual items scaled by lagged total assets, ∆AdjNI, is 0.001, while the averages
for its two components, ∆AdjEBIT and ∆AdjOTH, are -0.001 and 0.002, respectively. By
construction, the statistics for Book-Market and Size are identical to those reported for the
“TgtPriceRevDum = 1” subsample in Panel A of Table 5.
Panel B of Table 6 reports the estimation results. Column (1) shows that ∆AdjNI exhibits a
positive association with analysts’ target price revisions. Hence, an unexpected positive
(negative) adjustment to net income before unusual items induces analysts to revise their target
prices upward (downward). Column (2) indicates that the unexpected adjustment to the EBIT
component of net income has the expected positive effect on target revisions; the coefficient on
∆AdjEBIT is positive and statistically significant. However, the other component ∆AdjOTH
exhibits an insignificant association with the change in target price estimates.
As in section 4.1, we also augment the two regressions with the cumulative changes in the
adjustment terms over the first three quarters of the year, thereby allowing for the possibility
that analysts update their expectation from prior year’s annual adjustments using quarterly note
information. The results reported in columns (3) and (4) are qualitatively unchanged.
Overall, these results provide further support for the notion that analysts who update their
target prices at the 10-K date are using the note information to analyze firms’ accounting
issues. In addition to our main variables of interest, we find that the earnings surprise is not
related to the target price revisions for those observations when earnings announcements fall
within the 10-K filing window.
23
4.3. Market reaction to accounting adjustments and analysts’ revisions around 10-K filings
Given the findings that equity analysts also response to accounting adjustments around 10-
K filings with an update in their target price estimates, we examine the stock market reaction to
accounting adjustments that is incremental to the reaction to analysts’ revisions. In particular,
we re-estimate Equations 1 and 2 (results reported in table 4) augmented with the mean target
price revision (TgtPriceRev) and mean earnings forecast revision (EarnFcstRev) for each firm
and year. TgtPriceRev and EarnFcstRev equal to the value of the revision, if at least one
analyst revises his/her target price estimate and earnings forecast, respectively, over the 10-K
filing window; and zero otherwise. This test is conducted at the firm-year level and, hence, the
sample is identical to the one used in Table 4.
Table 7 reports that, across all four model specifications, both analysts’ target price and
earnings forecast revisions have a statistically positive effect on the market reactions around
10-K filings. However, analysts’ revisions do not subsume the explanatory power of the
accounting adjustment variables. Specifically, the values and statistically significance of the
coefficients on the adjustment terms are about the same as those reported in table 4. (The only
exception is that the estimated coefficients on AdjUNUSUAL are no longer distinguishable
from zero in columns 1 and 3). This finding suggests that other financial statement users, such
as institutional investors and money managers, are also exploring the note information in their
calculation of accounting adjustments. As a result, the accounting adjustment terms exhibit
incremental explanatory power over and above those of analysts’ target price and earnings
forecasts revisions.
As documented in sections 4.1 and 4.2, ∆AdjEBIT but not ∆AdjOTH affects the likelihood
of an analyst issuing a target price revision and the amount of the target price revision,
24
respectively. This suggests that equity analysts find adjustment to EBIT more informative in
their assessment of future firm performance and valuation than adjustment to other component
of net income before unusual items. This could explain why the market values ∆AdjEBIT more
than it values ∆AdjOTH, as shown in both Tables 4 and 7.
5. Conclusion
We examine the valuation of financial statement note information around 10-K filings (the
point at which this information first becomes available to the public). We conjecture that
financial statement users explore the information in these notes to compute accounting
adjustments to correct the imperfections in financial statements. We use a comprehensive set of
financial statement adjustments implemented by Moody’s to proxy for the accounting
adjustments that financial statement users could make. We find that stock returns around 10-K
filings are positively related to the accounting adjustments, calculated from note information.
We further find that the likelihood of equity analysts updating their target price estimates at the
10-K date is increasing in the magnitude of the adjustments. Those analysts who do update
their target prices at this time revise their estimates consistent with the sign and magnitude of
the adjustments. These findings are consistent with financial statement users using financial
statement note information to make accounting adjustments, thereby incorporating this
information into stock prices.
Prior studies show that (a) 10-K filings are informative without isolating the specific
information in the 10-K reports that is valued by investors, (b) financial statement note
information is reflected in equity prices using a long-window research design, and (c) equity
analysts can play an important role in this process. We contribute to these strands of literature
25
by examining the valuation of accounting adjustments and, hence the financial statement note
information used to compute them, in a short-window event-study design as well as by
showing the role played by equity analysts in the incorporation of this detailed note
information into stock prices. Our short-window research design allows us to more tightly link
the 10-K accounting information to the market and analysts’ reactions and rule out the
possibility that our results are driven by correlated omitted variables.
26
References
Aboody, D., M.E. Barth, and R. Kasznik. “SFAS No. 123 Stock-based employee compensation
and equity market values.” The Accounting Review 79 (2004): 251–275.
Ali. A., and P. Zarowin. “The role of earnings levels in annual earnings-returns studies.”
Journal of Accounting Research 30 (1992): 286-296.
Asthana, S., S. Balsam, and S. Sankaraguruswamy. “Differential response of small versus large
investors to 10-K filings on EDGAR.” The Accounting Review 79 (2004): 571-589.
Ball R., S.P. Kothari, and A. Robin. “The effect of International Institutional factors on
properties of accounting earnings.” Journal of Accounting and Economics 29 (2000): 1-51.
Balsam, S., E. Bartov, and C. Marquardt. “Accruals management, investor sophistication, and
equity valuation: Evidence from 10-Q filings.” Journal of Accounting Research 40 (2002),
987-1012.
Barth, M.E., W. Beaver, and W. Landsman. “The market valuation implications of net periodic
pension cost components.” Journal of Accounting and Economics 15 (1992): 27-62.
Barth, M.E., W. Beaver, and W. Landsman. “The relevance of the value relevance literature for
financial accounting standard setting: another view.” Journal of Accounting and
Economics 31 (2001): 77-104.
Basu, S. “The conservatism principle and the asymmetric timeliness of earnings.” Journal of
Accounting and Economics 24 (1997): 3-37.
Batta, G., A.R. Ganguly, and J.G. Rosett. “The valuation implications of financial statement
recasting for credit risk assessment.” Working paper, Claremont McKenna College, 2009.
Bhattacharya, N., Black, E., Christensen, T., Larson, C., 2003. Assessing the relative
informativeness and permanence of pro forma earnings and GAAP operating earnings.
Journal of Accounting and Economics 36 (1–3), 285-319.
Biddle, G.C., R.M. Bowen, and J.S. Wallace. “Does EVA Beat Earnings? Evidence on
Associations with Stock Returns and Firm Values.” Journal of Accounting and Economics
24 (1997): 301-336.
Bowman, R.G. “The debt equivalence of leases: An empirical investigation.” The Accounting
Review 55 (1980), 237-253.
Bradshaw, M.T., Sloan, R.G., 2002. GAAP versus the street: an empirical assessment of two
alternative definitions of earnings. Journal of Accounting Research 40, 41–66.
Brown, L.D., 2001. A temporal analysis of earnings surprises: profits versus losses. Journal of
Accounting Research 39, 221–241.
Brown, L.D., Sivakumar, K., 2003. Comparing the quality of three earnings measures. Review
of Accounting Studies 8, 561–572.
Bushman, R., Q. Chen, E. Engel, and A. Smith. “Financial accounting information,
organizational complexity and corporate governance systems.” Journal of Accounting and
Economics 37 (2004): 139-201.
Collins, D., E. Maydew, and I. Weiss. “Changes in the value-relevance of earnings and book
values over the past forty years.” Journal of Accounting and Economics 24 (1997): 39-67.
27
Dechow, P., and I. Dichev. “The quality of accruals and earnings: The role of accrual
estimation errors.” The Accounting Review 77 (2002): 35-59.
De Franco, G., S.P. Kothari, and R. Verdi. “The Benefits of Financial Statement
Comparability. Working paper MIT, University of Toronto, 2009.
Easton, P.D. and M.E. Zmijewski. “SEC Form 10K/10Q reports and annual reports to
shareholders: Reporting lags and squared market model prediction errors.” Journal of
Accounting Research 31 (1993): 113-129.
Ely, K.M. “Operating lease accounting and the market’s assessment of equity risk.” Journal of
Accounting Research 33 (1995): 397-415.
Fairfield, P.M. and J.S. Whisenant. “Using fundamental analysis to assess earnings quality:
Evidence from the Center for Financial Research and Analysis.” Journal of Accounting,
Auditing, and Finance 16(2001): 273-295.
Foster, T.W., III and D.W. Vickrey. “The incremental information content of the 10-K.” The
Accounting Review 58 (1978): 921-934.
Francis, J., and K. Schipper. “Have financial statements lost their relevance?” Journal of
Accounting Research 37 (1999): 319-352.
Francis, J., R. LaFond, P.M. Olsson, and K. Schipper. “Cost of equity and earnings attributes.”
The Accounting Review 79 (2004): 967-1010.
Ge, W., G. Imoff, and L.F. Lee. “Is recognition for operating lease necessary?” Working paper,
University of Washington and University of Michigan, 2008.
Givoly, D., and C. Hayn. “The changing time-series properties of earnings, cash flows and
accruals: Has financial reporting become more conservative?” Journal of Accounting and
Economics 29 (2000): 287-320.
Griffin, P. A. “Got information? Investor response to Form 10-K and Form 10-Q EDGAR
filings.” Review of Accounting Studies 8 (2003): 433-460.
Heckman, J. “Sample selection bias as a specification error.” Econometrica 47 (1979): 153-
161.
Koller, T., M. Goedhart, and D. Wessels. “Valuation measuring and managing the value of
companies.” 4th edition. John Wiley & Sons, Inc., 2005.
Kraft, P. “Rating agency adjustments to GAAP financial statements and their effect on ratings
and bond yields.” Working paper, University of Chicago, 2008.
Landsman, W. “An empirical investigation of pension fund property rights.” The Accounting
Review 61 (1986): 662-691.
Landsman, W., K. Peasnell, and C. Shakespeare. “Are securitization transactions really off
balance sheet?” The Accounting Review, forthcoming 2008.
Lang, M.H. and R.J. Lundholm. “Corporate disclosure policy and analyst behavior.” The
Accounting Review 71 (1996): 467-492.
Lejuez, P., G. Balter, M. Exstein, E. Kelly, and O. Saad. “The potential impact of capitalizing
leases.” Credit Suisse Equity Research April 18, 2006.
28
Lev, B. “Some economic determinants of the time-series properties of earnings.” Journal of
Accounting and Economics 5 (1983): 31-38.
Levi, S., and B. Segal. “The Impact of Debt-Equity Reporting Classifications on Firms’
Decision to Issue Hybrid Securities.” Working paper, University of California and
INSEAD, 2005.
Leuz, C., D. Nanda, and P. Wysocki. “Earnings management and investor protection: An
international comparison.” Journal of Financial Economics 69 (2003): 505-527.
Levinson, M. “Pensions in the deep freeze.” J.P. Morgan Securities Inc. North America
Corporate Research January 31, 2008.
Lipe, R. “The relation between stock returns and accounting earnings given alternative
information.” The Accounting Review 65 (1990): 49-71.
Lundholm, R., J. Doyle, and M. Soliman. “The predictive value of expenses excluded from
‘pro forma’ earnings.” Review of Accounting Studies 8 (2003): 145-174.
Moody’s, 2006, Moody’s approach to global standard adjustments in the analysis of financial
statements for non-financial corporations – Part I: Standardized adjustments to enable
global consistency for US and Canadian GAAP Issuers, February 2006.
Palepu, K.G., P. Healy, and V. Bernard. “Business Analysis & Valuation: Using Financial
Statements.” 3rd
edition, Thomson South-Western, 2004.
Penman, S. “Financial Statement Analysis and Security Valuation.” 3rd
edition, McGraw-Hill
College, 2006.
Pope, P., and M. Walker. “International differences in the timeliness, conservatism and
classification of earnings.” Journal of Accounting Research 37 (1999): 53-99.
Qi, D., W. Wu, and I. Haw. “The incremental information content of SEC 10-K reports filed
under the EDGAR system.” Journal of Accounting, Auditing and Finance (2000): 25-46.
Revsine, L., D. Collins, B. Johnson, and F. Mittelstaedt. “Financial Reporting and Analysis.”
4th
edition, McGraw-Hill/Irwin.
Stice, E.K. “The market reaction to 10-K and 10-Q filings and to subsequent Wall Street
Journal earnings announcements.” The Accounting Review 66 (1991): 42-55.
Stickney, C. P., P. Brown, and J. M. Wahlen. “Financial Reporting, Financial Statement
Analysis, and Valuation: A Strategic Perspective.” 6th edition, South-Western College
Publishing, 2006.
White, G.I., A.C. Sondhi, and D. Fried. “The analysis and use of financial statements.” 3rd
edition, John Wiley & Sons Inc, 2002.
Wild, J.J., K.R. Subramanyam, and R.F. Halsey. “Financial statement Analysis.” 9th edition,
McGraw-Hill Higher Education, 2006.
Zion, D., and B. Carcache. “Here it comes: FASB could add pension accounting to agenda this
fall.” Credit Suisse First Boston Equity Research September 22, 2005.
29
APPENDIX A
Description of Moody’s Accounting Adjustments
In this Appendix, we describe each of Moody’s standard and non-standard adjustments,
their purposes, and their effect on net income before unusual items and EBIT. Moody’s (2006)
provides more details on its methodology. The first six adjustments take a more standardized
form and are summarized in Table 1. Two other adjustments reflected in our analysis include
Unusual & Non-Recurring Items and Non-Standard Adjustments, which depending on their
nature can affect any account.
a) Underfunded defined benefit pensions. Under current GAAP, pension liabilities are
not reflective of a company’s obligation to its pension trust, and pension expenses are subject
to smoothing and the recognition of actuarial gains on pension assets. Moody’s recognizes as
debt the underfunded pension amount, estimated by the difference between the projected
benefit obligation (PBO) and the fair value of pension assets. No adjustments are made to over
funded pension amounts. Moody’s defines pension expense as the year’s service cost, plus
Moody’s imputed interest on the PBO, minus actual earnings on plan assets (up to the amount
of interest cost imputed to restrict interest cost to be non positive). Hence, Moody’s eliminates
the amortization of prior service cost and actuarial gains and losses in the computation of
pension expense. Instead, it recognizes actual gains or losses on pension assets in other non-
recurring expenses (gains) below EBIT. This adjustment eliminates smoothing. By capping the
amount of gains allowed up to the amount of interest cost, it avoids reporting net periodic
pension income.
After Moody’s adjustment, cost of goods sold, SG&A and operating expenses increases14
by the service cost minus GAAP pension costs. Interest expense increases by Moody’s imputed
interest cost on the PBO. Other non-recurring expenses or gains increase (decrease) by actual
losses (gains) on pension assets, with gains allowed up to the amount of interest cost. Tax
expense decreases by the tax effect of the additional pension costs. Unusual and non-recurring
items decrease by the after-tax effect of the additional pension costs. Depending on the amount
of GAAP service cost and pension costs, EBIT and net income before unusual items can
increase or decrease.
14 Unless specified otherwise, Moody’s adjusts the cost of goods sold, SG&A and operating expense accounts on a
proportional basis.
30
b) Operating leases. Unlike capital leases, the obligations associated with operating leases
are not required to be recognized as liabilities under current GAAP. Companies can structure
the lease transactions to ensure it is an operating lease and, hence, operating leases represent a
common form of off-balance sheet financing. Moody’s treats operating leases as capital leases
to better reflect the underlying economics of the transactions. It reclassifies rental expense from
COGS, SG&A and operating expenses to interest and depreciation expenses.
After Moody’s adjustment, COGS, SG&A and operating expenses decrease by the interest
and depreciation components of rental expense, and interest and depreciation expenses increase
by one-third and two-third of rental expense, respectively. As a result, EBIT increases by the
interest component of rental expense. These adjustments offset each other, leaving net income
before unusual items unchanged. In other words, this adjustment shifts income from below to
above EBIT.
c) Hybrid securities. Under current GAAP, hybrid securities are classified and accounted
for as pure debt, pure equity, or minority interest. This could be misleading because hybrid
securities typically exhibit attributes of both debt and equity. Moody’s reclassifies hybrid
securities to debt and equity on a pro-rata basis according to the debt-equity continuum it
assigns to the securities, which is based on Moody’s own research. Moody’s adjusts interest
expense and preferred dividends accordingly.
After Moody’s adjustment, on one hand, preferred dividends decreases by the interest
expense for the assigned debt portion of hybrid securities classified as all equity and increases
by the preferred dividends for the assigned equity portion of hybrid securities classified as pure
debt or minority interest. On the other hand, interest expense increases by the interest expense
for the calculated debt portion of hybrid securities classified as all equity or minority interest
and decreases by the preferred dividends for the calculated equity portion of hybrid securities
classified as pure debt. As a result, both EBIT and net income before unusual items are
unaffected.
d) Securitizations. Under current GAAP, companies report as sales the transfer of assets
to securitization trusts. However, Moody’s find that companies usually retain significant risks
related to the assets transferred for nearly all of the securitizations they review. Moody’s
reclassifies securitization transactions that do not fully transfer risk as collateralized
borrowings.
31
After Moody’s adjustment, other expense decreases by the inputed interest on the amount
of additional collateralized borrowings, while interest expense increases by the imputed interest
on the amount of additional collateralized borrowings. As a result, EBIT increase or remains
unchanged (dependent upon where the company has recorded the securitization expense), but
net income before unusual items is unaffected.
e) Capitalized interest. Under current GAAP, interest costs attributed to the financing of
self-constructed assets are capitalized. This not only overstates assets and net income, but it
also mixes up financing transactions with operating activities. Moody’s expenses the interest
capitalized during the current period and adjusts related balance sheet items accordingly.
After Moody’s adjustment, interest expense increases by the amount of capitalized interest
during the current period. Tax expense decreases by the tax effect of the additional interest
expense. Unusual and non-recurring items decrease by the after-tax effect of the additional
interest expense. As a result, EBIT is not affected, while net income before unusual items
decreases.
f) Employee stock compensation. Before June 30, 2005, GAAP did not require the
expensing of employee stock options (ESO) and most firms chose not to voluntarily expense
these costs. This leads to the understatement of operating expenses for many firms, especially
when compared to their counterparts that are light users of employee stock options. Moody’s
expenses the cost of employee stock options.
After Moody’s adjustment, SG&A expense increases by the ESO cost and tax expense
decreases by the tax effect of the ESO expense. Unusual and non-recurring items decrease by
the after-tax effect of the additional ESO expense. As a result, both EBIT and net income
before unusual items decrease.
g) Unusual and non-recurring items. Under current GAAP, certain unusual and non-
recurring items are reported under operating income or EBIT. However, these unusually large
or infrequent transactions are transitory in nature and, therefore, have little implications for
future earnings. Moody’s reclassifies the effects of these transactions, net of the related tax
effect, to a special income statement line item (Unusual & Non-Recurring Items – Adjustments
after tax) that is below net income before unusual items.
Depending on the line items being adjusted, EBIT can be affected, but net income before
unusual items will be affected by construction. These items include atypical large transactions
32
that are unlikely to recur, and gains or losses from the infrequent sales of non-operating assets.
For example, the selling of real estate by a company that rarely sells real estate and tax benefits
of deductible goodwill whose depreciable life is ending.
h) Non-standard adjustments. Besides the standard adjustments, Moody’ also make non-
standard adjustments on a case-by-case basis. These adjustments have the same objective as
standard adjustments, which is to better reflect economic reality by using estimates or
assumptions that are more prudent. Unlike the standard adjustments, however, these
adjustments are at the discretion of the Moody’s analyst. Non-standard adjustments usually
relate to highly judgmental areas such as asset valuation allowances, impairments of assets, and
contingent liabilities.
i) Inventory on a LIFO cost basis. Current U.S. GAAP allows companies to choose the
LIFO cost flow assumption for inventory accounting. In an inflationary environment, LIFO
inventory cost understates the value of inventory. Moody’s adjusts the reported inventory
balance on the LIFO basis using the FIFO cost assumption. However, it makes no adjustment
to the income statement, because it believes that LIFO cost of goods sold better reflects current
cost. As a result, this adjustment has no effect on the income statement, and given our earnings
focus in this study, this adjustment is hence excluded from our analysis.
33
APPENDIX B
Estimation of Earnings Properties
In this appendix, we describe the underlying construct for each earnings property and
provide a brief description of how it is estimated. We adopt the convention that larger values of
the attributes indicate higher levels of the construct. Each earnings property is estimated by
industry group with firm and year fixed effects included. All regression variables are
winsorized at the 1% and 99% level by industry.
a) Accrual Quality. We follow Dechow and Dichev (2002) to create a measure that
captures the degree to which earnings map into cash. We regress total current accruals on
lagged, current, and future cash flows from operation (CFO), with all variables scaled by
lagged assets. (This and all other earnings property regressions are estimated with intercepts).
We take CFO from the statement of cash flows as opposed to using the balance sheet method
as in Dechow and Dichev. Accrual Quality equals the standard deviation of the estimated
residuals from this regression multiplied by -1. Larger values indicate higher accrual quality.
b) Persistence and Predictability. Persistence captures the degree to which current year’s
earnings continues into the future. Following previous research, such as Lev (1983) and Ali
and Zarowin (1992), we regress current year’s earnings per share (EPS) on last year’s EPS (i.e.,
an autoregressive AR(1) time-series model). Persistence is the coefficient on last year’s EPS.
Larger values indicate more persistence. These regressions also provide our measure of
Predictability, which represents earnings ability to predict itself. Consistent with Lipe (1990),
Predictability is the square root of the error variance from this regression, multiplied by -1.
Larger values indicate more predictable earnings.
c) Smoothness. This measure captures the degree of variance in earnings relative to that of
cash flows. Following Leuz, Nanda, and Wysocki (2003), Smoothness is the ratio of the
standard deviation of net income before extraordinary items (scaled by lagged assets) to the
standard deviation of CFO (scaled by lagged assets), multiplied by -1. This attribute calculation
is an exception to our industry level approach. It is computed at the firm level, and then
aggregated to form an industry measure. Larger values indicate smoother earnings.
d) Value Relevance. This measure captures the ability of variation in earnings to explain
variation in returns. Following Francis and Schipper (1999), Collins, Maydew, and Weiss
(1997), and Bushman, Chen, Engel, and Smith (2004), we regress returns on the level and
34
changes in earnings. We use 12-month returns from nine months before to three months after
the end of fiscal year t; Francis et al. (2004) use 15-month returns ending three months after the
end of fiscal year t. Value Relevance is the R-squared from this regression. Larger values
indicate more value-relevant earnings.
e) Timeliness and Conservatism. Timeliness captures the ability of earnings to reflect the
news in returns, while Conservatism captures the differential ability of earnings to reflect the
negative versus positive news in returns. We regress earnings on three variables: an indicator
variable equal to one if returns are negative, returns, and an interaction of the indicator variable
and returns. From this regression, we obtain estimated coefficients of α1, β1, and β2 on these
three variables, respectively. As in Ball, Kothari, and Robin (2000) and Bushman et al. (2004),
Timeliness is the adjusted R-square from this regression. Larger values indicate more timely
earnings. Francis et al. follow prior research by Basu (1997), Pope and Walker (1999), and
Givoly and Hayn (2000) and use (β1+ β2)/β1 as their conservatism measure. However, because
many of our β1 estimates are negative, we simply use the β2 coefficient as our measure of
Conservatism. Larger values indicate more conservative earnings.
35
TABLE 1
The Effect of Six Moody’s Standard Adjustments on Income Statement Line Items
This table summarizes the effect of six Moody’s standard adjustments on the income statement. Two other
adjustments Unusual & Non-Recurring Items and Non-Standard Adjustments do not follow a standard form and
can affect any account. See the text and Appendix A for a detailed discussion of these adjustments.
Underfunded
Defined Benefit
Pensions
Operating
Leases Hybrid Securities Securitizations
Capitalized
Interest
Employee
Stock
Compensation
Cost of Goods Sold /
SG&A Expenses /
Operating Expenses
Increases
proportionally by
service cost minus
GAAP pension
costs.
Decreases
proportional-
ly by interest
component of
rent expense.
SG&A
increases by
amount of pre-
tax ESO
expense.
Depreciation Increases by
depreciation
component of
rent expense.
Other Income/
(Other Expense)
Decreases by
inputed
interest on
amount of
additional
collateralized
borrowings.
Adjusted EBIT Increases or
Decreases
Increases No effect Increases or
No effect
No effect Decreases
Interest Expense Increases by
interest cost on the
PBO.
Increases by
interest
component of
rent expense.
Increases by interest
expense for debt portion
of hybrid securities
classified as pure equity
or minority interest.
Decreases by preferred
dividends for equity
portion of hybrid
securities classified as
pure debt.
Increases by
inputed
interest on
amount of
additional
collateralized
borrowings.
Increases by
amount of
capitalized
interest during
current period.
Other Loss or Expense
(Gain or Income )
Increase by actual
losses pension
assets. Decrease by
actual gains on
pension assets up to
amount of interest
cost.
Decreases by interest
expense for debt portion
of hybrid securities
classified as pure equity.
Increases by preferred
dividends for equity
portion of hybrid
securities classified as
pure debt or minority
interest.
Taxes Expense Decreases by tax
effect of additional
pension costs.
Decreases by
tax effect of
additional
interest
expense.
Decreases by
tax effect of
additional ESO
expense.
Adjusted Net Income
After-tax Before
Unusual Items
Increases or
Decreases
No effect No effect No effect Decreases Decreases
Moody’s Unusual &
Non-Recurring Items -
Adjustments After-tax
Decreases by after-
tax effect of
additional pension
costs.
Decreases by
after-tax effect
of additional
interest
expense.
Decreases by
after-tax effect
of additional
ESO expense.
GAAP Net Income No effect No effect No effect No effect No effect No effect
36
TABLE 2
Descriptive Statistics on Moody’s Accounting Adjustments
This table provides descriptive statistics about our sample of 4,944 firm-year observations for a panel of 958 U.S.
firms, covering the period 2002 through 2007. Panels A and B report summary statistics on the effect of eight
adjustments to net income before unusual items and earnings before interest and taxes (EBIT), respectively, scaled
by beginning total assets. Moody’s also adjusts LIFO inventory balance using the FIFO cost assumption, but
makes no adjustment to the income statement.
Panel A: Adjustments to net income before unusual items (percent of lagged total assets)
Percentiles
Mean Std Dev % < 0 % = 0 % > 0 5% 50% 95%
Underfunded defined benefit pensions -0.22% 0.97% 47.3% 32.2% 20.5% -1.35% 0.00% 0.17%
Operating leases 0.00% 0.00% 0.0% 100.0% 0.0% 0.00% 0.00% 0.00%
Hybrid securities 0.00% 0.00% 0.0% 100.0% 0.0% 0.00% 0.00% 0.00%
Securitizations 0.00% 0.00% 0.0% 100.0% 0.0% 0.00% 0.00% 0.00%
Capitalized interest -0.04% 0.21% 25.3% 74.7% 0.0% -0.24% 0.00% 0.00%
Employee stock compensation -0.30% 1.00% 42.5% 56.7% 0.8% -1.42% 0.00% 0.00%
Unusual and non-recurring items 0.49% 3.48% 18.9% 54.6% 26.5% -0.46% 0.00% 2.86%
Non-standard adjustment 0.03% 2.56% 10.1% 86.7% 3.3% 0.00% 0.00% 0.00%
Total adjustment -0.05% 4.49% 67.9% 1.6% 30.5% -2.79% -0.13% 2.51%
Panel B: Adjustments to EBIT (percent of lagged total assets)
Percentiles
Mean Std Dev % < 0 % = 0 % > 0 5% 50% 95%
Underfunded defined benefit pensions 0.04% 0.53% 28.0% 32.8% 39.2% -0.53% 0.00% 0.61%
Operating leases 0.71% 1.15% 7.1% 1.8% 91.1% 0.04% 0.36% 2.66%
Hybrid securities 0.00% 0.01% 7.1% 92.6% 0.3% 0.00% 0.00% 0.00%
Securitizations 0.02% 0.12% 7.1% 86.2% 6.7% 0.00% 0.00% 0.05%
Capitalized interest 0.00% 0.00% 7.1% 92.9% 0.0% 0.00% 0.00% 0.00%
Employee stock compensation -0.44% 1.39% 42.5% 56.7% 0.8% -2.06% 0.00% 0.00%
Unusual and non-recurring items 0.69% 4.32% 19.7% 51.8% 28.4% -0.81% 0.00% 4.35%
Non-standard adjustment 0.04% 1.60% 12.6% 80.6% 6.8% -0.09% 0.00% 0.16%
Total adjustment 1.05% 4.74% 33.0% 0.5% 66.5% -2.25% 0.40% 6.66%
37
TABLE 3
Properties of Reported GAAP and Adjusted Earnings
This table provides a comparison of the earnings properties calculated using Moody’s adjusted data with the earnings properties calculated using reported GAAP
data. The underlying sample includes 4,944 firm-year observations for a panel of 958 U.S. firms, covering the period 2002 through 2007. Thirty-two industry
groups are formed, according to Moody’s industrial classification, with each industry group containing between 7 and 94 companies. Earnings properties are
estimated by industry group with firm and year fixed effects, using reported GAAP earnings and adjusted earnings, respectively. All regression variables are
winsorized at the 1% and 99% level by industry. Appendix B discusses the estimation of these measures in details. The table reports the average number of observations per industry, which depends on each property’s data requirements. Mean and median values for each earnings property as well as the matched-pair
difference across the 32 industry groups are provided. A t-statistic tests whether the mean industry difference is different than zero. The table also provides the
number and percentage of industry differences that are greater than zero. A binomial test indicates whether this percentage is significantly different than 50%.
***, **, and * denote statistical significance at the 1%, 5%, and 10% levels (two-sided), respectively. Accrual Quality is the standard deviation of the estimated
residuals from a regression of current accruals on lagged, current, and future cash flows from operations (CFO), multiplied by -1. Persistence is the coefficient on
previous year’s earnings from a regression of current year’s earnings on previous earnings (i.e., an AR(1) model). Predictability is the square root of the error
variance from this regression, multiplied by -1. Smoothness is the ratio of the standard deviation of net income to the standard deviation of CFO, multiplied by -1.
Value Relevance is the R-squared from a regression of returns on the level and changes in annual earnings. Timeliness is the adjusted R-square from a regression
of earnings on three variables: an indicator variable equal to one if returns are negative, returns, and an interaction of the indicator variable and returns.
Conservatism is the coefficient on the interaction of the indicator variable and returns from this regression.
Average
Number of
Firm-Year
Observations
Reported GAAP
Earnings Adjusted Earnings
32 Industry Matched-Pair Differences
(Adjusted less Reported GAAP)
Variable per Industry Mean Median Mean Median Mean t-statistic Median Obs. > 0 % > 0
Accrual Quality 107 -0.042 -0.039 -0.039 -0.038 0.003 3.87*** 0.002 27 84.4***
Persistence 126 0.078 0.071 0.092 0.162 0.014 0.31 0.014 19 59.4
Predictability 126 -1.425 -1.076 -1.218 -0.900 0.207 4.01*** 0.113 27 84.4***
Smoothness 155 -1.580 -1.206 -3.509 -1.113 -1.929 -1.13 0.063 21 65.6*
Value Relevance 126 0.345 0.333 0.371 0.311 0.026 1.15 0.004 17 53.1
Timeliness 153 0.152 0.133 0.172 0.150 0.020 1.23 0.009 21 65.6*
Conservatism 153 0.387 0.230 0.350 0.236 -0.036 -1.19 -0.023 10 31.2**
38
TABLE 4
Stock Market Response to Accounting Adjustments around 10-K Filings
This table examines the effect of accounting adjustments on stock returns around the 10-K filing window. Panel A
reports summary statistics on firm-year level data. Panel B presents the multivariate test results. We regress event-
window cumulative abnormal stock returns (CAR) on annual changes in adjustments to net income and its
components, plus controls. Intercept and year fixed effects are included for each model but not tabulated. We
estimate each model as a panel and cluster the standard errors at the firm level. Coefficient t-statistics are in parentheses. Significance levels are based on one-tailed tests where there is a prediction for the sign of the
coefficient and based on two-tailed tests otherwise. ***, **, and * denote statistical significance at the 1%, 5%, and
10% levels, respectively. CAR is the cumulative abnormal return from one day before to five days after the year-t
10-K filing date, where abnormal returns is computed as firm raw returns less the CRSP size-matched decile index
returns. ∆AdjNI is Moody’s adjustments to firm i’s net income before unusual items for year t less Moody’s
adjustments for year t-1, scaled by lagged total assets. ∆AdjEBIT is Moody’s adjustments to firm i’s EBIT for year t
less Moody’s adjustments for year t-1, scaled by beginning total assets. ∆AdjOTH is ∆AdjNI less ∆AdjEBIT.
AdjUNUSUAL is Moody’s unusual and non-recurring adjustments to firm i’s net income for year t.
LagAdjUNUSUAL is AdjUNUSUAL for year t-1. ∆Adj3Q_DUM is an indicator variables that equals one if Moody’s
provides quarterly adjusted income statement for the first three quarters of year t and year t-1; zero otherwise.
∆AdjNI3Q is Moody’s adjustments to firm i’s net income before unusual items for the first three quarters of year t
less Moody’s adjustments for the first three quarters of year t-1, scaled by lagged total assets. ∆AdjEBIT3Q is Moody’s adjustments to firm i’s EBIT for the first three quarters of year t less Moody’s adjustments for the first
three quarters of year t-1, scaled by beginning total assets. ∆AdjOTH3Q is ∆AdjNI3Q less ∆AdjEBIT3Q.
EarnSurprise is firm i’s earnings surprise if earnings is announced inside the 10-K filing window, and zero
otherwise. Earnings surprise is the actual earnings per share (EPS) for period t minus mean analysts’ forecast of
EPS, scaled by stock price measured at the end of year t-1. Book-Market is the ratio of the book value of equity to
the market value of equity for year t. Size is the logarithm of the market value of equity for year t.
Panel A: Descriptive statistics
Percentile
Variable No. of Obs. Mean Std. Dev. 5% 50% 95%
CAR 3,069 0.002 0.044 -0.058 0.001 0.066
∆AdjNI 3,069 0.003 0.052 -0.005 0.000 0.019
∆AdjEBIT 3,069 0.001 0.039 -0.008 0.001 0.015
∆AdjOTH 3,069 0.001 0.045 -0.008 -0.000 0.018
AdjUNUSUAL 3,069 0.004 0.027 -0.005 0.000 0.024
EarnSurprise 3,069 0.000 0.019 -0.006 0.000 0.007
Book-Market 3,069 0.436 0.401 0.060 0.415 0.939
Size 3,069 8.242 1.426 6.138 8.102 10.788
(Table continues on next page)
39
TABLE 4 (Continued)
Panel B: Multivariate Analysis
Explanatory
Variable
Predicted
Sign (1) (2) (3) (4)
∆AdjNI + 0.075 0.076
(3.44)*** (3.49)***
∆AdjEBIT + 0.096 0.098
(3.57)*** (3.77)***
∆AdjOTH + 0.060 0.061
(4.87)*** (4.86)***
AdjUNUSUAL ? -0.127 -0.127 -0.127 -0.127
(-2.88)*** (-2.89)*** (-2.88)*** (-2.89)***
LagAdjUNUSUAL ? 0.014 0.013 0.014 0.014
(0.45) (0.44) (0.47) (0.45)
∆Adj3Q_DUM 0.003 0.003
(1.01) (0.99)
∆AdjNI3Q -0.136
(-1.45)
∆AdjEBIT3Q -0.138
(-1.34)
∆AdjOTH3Q -0.114
(-0.58)
EarnSurprise + 0.369 0.372 0.368 0.369
(2.67)*** (2.69)*** (2.65)*** (2.64)***
Book-Market ? 0.002 0.002 0.002 0.002
(0.46) (0.45) (0.44) (0.43)
Size ? -0.001 -0.001 -0.001 -0.001
(-1.51) (-1.53) (-1.57) (-1.58)
Year fixed effects Yes Yes Yes Yes
No. of obs. 3,069 3,069 3,069 3,069
Adj. R2 (%) 2.1 2.2 2.2 2.3
40
TABLE 5
Probability of Analysts’ Target Price Revisions around 10-K Filings
This table examines the relation between the probability of analysts revising their target price around the 10-K filing
window and the magnitude of the accounting adjustments. Panel A reports summary statistics on analyst-firm-year
level data. In panel B, we estimate a logit model in which the dependent variable is whether an analyst updates his or
her target price during the 10-K window. Independent variables include annual changes in adjustments to net income
and its components, plus controls. Intercept and year fixed effects are included for each model but not tabulated. We estimate each model as a panel and cluster the standard errors at the firm level. Coefficient standard errors are in
parentheses. Significance levels are based on one-tailed tests where there is a prediction for the sign of the
coefficient and based on two-tailed tests otherwise. ***, **, and * denote statistical significance at the 1%, 5%, and
10% levels, respectively. TgtPriceRevDum is an indicator variable that equals one if analyst j issues a report that
contains a target price revision for firm i one day before to five days after the year-t 10-K filing date, zero otherwise.
∆AdjNI is Moody’s adjustments to firm i’s net income before unusual items for year t less Moody’s adjustments for
year t-1, scaled by lagged total assets. ∆AdjEBIT is Moody’s adjustments to firm i’s EBIT for year t less Moody’s
adjustments for year t-1, scaled by beginning total assets. ∆AdjOTH is ∆AdjNI less ∆AdjEBIT. AdjUNUSUAL is
Moody’s unusual and non-recurring adjustments to firm i’s net income for year t. LagAdjUNUSUAL is
AdjUNUSUAL for year t-1. ∆Adj3Q_DUM is an indicator variables that equals one if Moody’s provides quarterly
adjusted income statement for the first three quarters of year t and year t-1; zero otherwise. ∆AdjNI3Q is Moody’s
adjustments to firm i’s net income before unusual items for the first three quarters of year t less Moody’s adjustments for the first three quarters of year t-1, scaled by lagged total assets. ∆AdjEBIT3Q is Moody’s
adjustments to firm i’s EBIT for the first three quarters of year t less Moody’s adjustments for the first three quarters
of year t-1, scaled by beginning total assets. ∆AdjOTH3Q is ∆AdjNI3Q less ∆AdjEBIT3Q. EarnSurprise is firm i’s
earnings surprise if earnings is announced inside the 10-K filing window, and zero otherwise. Earnings surprise is
the actual earnings per share (EPS) for period t minus analyst j’s forecast of EPS, scaled by stock price measured at
the end of year t-1. Book-Market is the ratio of the book value of equity to the market value of equity for year t.
Size is the logarithm of the market value of equity for year t. | • | indicates that we take the absolute value of the
variable.
Panel A: Descriptive statistics (N = 24,133 analyst-firm-year-level obs.)
TgtPriceRevDum = 1 (N=1,742) TgtPriceRevDum = 0 (N=22,391)
Percentile Percentile
Variable Mean Std. Dev. 5% 50% 95% Mean Std. Dev. 5% 50% 95%
|∆AdjNI| 0.009 0.071 0.000 0.002 0.026 0.006 0.037 0.000 0.002 0.020
|∆AdjEBIT| 0.009 0.072 0.000 0.003 0.022 0.006 0.028 0.000 0.002 0.019
|∆AdjOTH| 0.006 0.013 0.000 0.002 0.025 0.006 0.030 0.000 0.001 0.021
|AdjUNUSUAL| 0.009 0.032 0.000 0.000 0.048 0.007 0.025 0.000 0.000 0.032
|EarnSurprise| 0.002 0.005 0.000 0.000 0.011 0.000 0.003 0.000 0.000 0.001
Book-Market 0.396 0.270 0.042 0.361 0.852 0.403 0.250 0.070 0.370 0.843
Size 8.913 1.377 6.757 8.868 11.345 8.862 1.407 6.623 8.824 11.310
(Table continues on next page)
41
TABLE 5 (Continued)
Panel B: Multivariate Analysis
Explanatory Predicted
Variable Sign (1) (2) (3) (4)
|∆AdjNI| + 0.701 0.740
(1.52)* (1.68)**
|∆AdjEBIT| + 1.177 1.194
(6.66)*** (7.46)***
|∆AdjOTH| + -0.667 -0.296
(-0.77) (-0.42)
|AdjUNUSUAL| +/0 -0.033 -0.041 -0.146 -0.131
(-0.03) (-0.03) (-0.11) (-0.10)
|LagAdjUNUSUAL| –/0 -0.335 -0.353 -0.254 -0.304
(-0.29) (-0.30) (-0.22) (-0.27)
∆Adj3Q_DUM 0.493** 0.465**
(2.23) (2.14)
|∆AdjNI3Q| -35.977**
(-2.09)
|∆AdjEBIT3Q| -15.450
(-1.42)
|∆AdjOTH3Q| -1.597
(-0.13)
|EarnSurprise| + 92.562 92.567 96.537 95.741
(8.78)*** (8.78)*** (8.76)*** (8.80)***
Book-Market ? -0.030 -0.020 -0.036 -0.023
(-0.14) (-0.09) (-0.17) (-0.11)
Size ? 0.081 0.083 0.078 0.077
(2.33)** (2.36)** (2.24)** (2.17)**
Year fixed effects Yes Yes Yes Yes
No. of obs. 24,133 24,133 24,133 24,133
Pseudo R2 (%) 1.5 1.5 1.6 1.6
% Concordant 63.3 63.2 64.1 64.0
42
TABLE 6
Analysts’ Target Price Revisions around 10-K Filings
This table examines the effect of accounting adjustments on analysts’ target price revisions around the 10-K filing
window. Panel A reports summary statistics on analyst-firm-year level data. Panel B presents the multivariate test
results. In panel B, we regress signed target price revisions (TgtPriceRev) on annual changes in adjustments to net
income and its components, plus controls. Intercept and year fixed effects are included for each model but not
tabulated. We estimate each model as a panel and cluster the standard errors at the firm level. Coefficient t-statistics are in parentheses. Significance levels are based on one-tailed tests where there is a prediction for the sign of the
coefficient and based on two-tailed tests otherwise. ***, **, and * denote statistical significance at the 1%, 5%, and
10% levels, respectively. TgtPriceRev is analyst j’s target price forecast for firm i issued one day before to five days
after the year-t 10-K filing date minus analyst j’s prior target price forecast for firm i, scaled by the prior target price
forecast. ∆AdjNI is Moody’s adjustments to firm i’s net income before unusual items for year t less Moody’s
adjustments for year t-1, scaled by lagged total assets. ∆AdjEBIT is Moody’s adjustments to firm i’s EBIT for year t
less Moody’s adjustments for year t-1, scaled by beginning total assets. ∆AdjOTH is ∆AdjNI less ∆AdjEBIT.
AdjUNUSUAL is Moody’s unusual and non-recurring adjustments to firm i’s net income for year t.
LagAdjUNUSUAL is AdjUNUSUAL for year t-1. ∆Adj3Q_DUM is an indicator variables that equals one if Moody’s
provides quarterly adjusted income statement for the first three quarters of year t and year t-1; zero otherwise.
∆AdjNI3Q is Moody’s adjustments to firm i’s net income before unusual items for the first three quarters of year t
less Moody’s adjustments for the first three quarters of year t-1, scaled by lagged total assets. ∆AdjEBIT3Q is Moody’s adjustments to firm i’s EBIT for the first three quarters of year t less Moody’s adjustments for the first
three quarters of year t-1, scaled by beginning total assets. ∆AdjOTH3Q is ∆AdjNI3Q less ∆AdjEBIT3Q.
EarnSurprise is firm i’s earnings surprise if earnings is announced inside the 10-K filing window, and zero
otherwise. Earnings surprise is the actual earnings per share (EPS) for period t minus analyst j’s forecast of EPS,
scaled by stock price measured at the end of year t-1. Book-Market is the ratio of the book value of equity to the
market value of equity for year t. Size is the logarithm of the market value of equity for year t. The Inverse Mills
ratio is computed using the corresponding estimates from table 5.
Panel A: Descriptive statistics
Percentile
Variable No. of Obs. Mean Std. Dev. 5% 50% 95%
TgtPriceRev 1,742 0.034 0.2656 -0.208 0.000 0.278
∆AdjNI 1,742 0.001 0.071 -0.005 0.000 0.023
∆AdjEBIT 1,742 -0.001 0.073 -0.009 0.001 0.017
∆AdjOTH 1,742 0.002 0.014 -0.008 -0.000 0.022
AdjUNUSUAL 1,742 0.005 0.033 -0.006 0.000 0.039
EarnSurprise 1,742 -0.000 0.005 -0.004 0.000 0.007
Book-Market 1,742 0.396 0.270 0.042 0.361 0.852
Size 1,742 8.913 1.377 6.757 8.868 11.345
(Table continues on next page)
43
TABLE 6 (Continued)
Panel B: Multivariate Analysis
Explanatory Predicted
Variable Sign (1) (2) (3) (4)
∆AdjNI + 0.063 0.065
(2.18)** (2.21)**
∆AdjEBIT + 0.066 0.068
(1.85)** (2.01)**
∆AdjOTH + 0.548 0.248
(1.20) (0.64)
AdjUNUSUAL +/0 0.177 0.185 0.182 0.190
(0.54) (0.56) (0.55) (0.57)
LagAdjUNUSUAL –/0 0.052 0.050 0.056 0.046
(0.40) (0.38) (0.43) (0.35)
∆Adj3Q_DUM 0.006 0.006
(0.33) (0.37)
∆AdjNI3Q -1.312
(-1.18)
∆AdjEBIT3Q -0.552
(-0.40)
∆AdjOTH3Q 1.555
(0.59)
EarnSurprise + 0.543 0.554 0.553 0.665
(0.53) (0.54) (0.52) (0.63)
Book-Market ? -0.016 -0.015 -0.016 -0.016
(-0.55) (-0.53) (-0.55) (-0.54)
Size ? 0.005 0.005 0.005 0.005
(1.24) (1.22) (1.27) (1.18)
Inverse Mills’ ratio 0.002 0.003 0.001 -0.004
(0.07) (0.08) (0.03) (-0.12)
Year fixed effects Yes Yes Yes Yes
No. of obs. 1,742 1,742 1,742 1,742
Adj. R2 (%) 1.4 1.5 1.5 1.6
44
TABLE 7
Stock Market Response to Accounting Adjustments and Analysts’ Revisions around 10-K
Filings
This table examines the effect of accounting adjustments on stock returns around the 10-K filing window. We
regress event-window cumulative abnormal stock returns (CAR) on annual changes in adjustments to net income and
its components, plus controls. In addition to the control variables reported in table 4, we also include analysts’ target
price revisions and earnings forecast revisions. Intercept and year fixed effects are included for each model but not
tabulated. We estimate each model as a panel and cluster the standard errors at the firm level. Coefficient t-statistics
are in parentheses. Significance levels are based on one-tailed tests where there is a prediction for the sign of the
coefficient and based on two-tailed tests otherwise. ***, **, and * denote statistical significance at the 1%, 5%, and
10% levels, respectively. CAR is the cumulative abnormal return from one day before to five days after the year-t
10-K filing date, where abnormal returns is computed as firm raw returns less the CRSP size-matched decile index
returns. ∆AdjNI is Moody’s adjustments to firm i’s net income before unusual items for year t less Moody’s
adjustments for year t-1, scaled by lagged total assets. ∆AdjEBIT is Moody’s adjustments to firm i’s EBIT for year t less Moody’s adjustments for year t-1, scaled by beginning total assets. ∆AdjOTH is ∆AdjNI less ∆AdjEBIT.
AdjUNUSUAL is Moody’s unusual and non-recurring adjustments to firm i’s net income for year t.
LagAdjUNUSUAL is AdjUNUSUAL for year t-1. TgtPriceRev is analyst j’s target price forecast for firm i issued one
day before to five days after the year-t 10-K filing date minus analyst j’s prior target price forecast for firm i, scaled
by the prior target price forecast. EarnFcstRev is analyst j’s forecast of year t+1 earnings issued one day before to
five days after the year-t 10-K filing date minus analyst j’s prior forecast of corresponding earnings, scaled by the
standard deviation of all forecasts made within 180 days before the 10_K filing date. ∆Adj3Q_DUM is an indicator
variables that equals one if Moody’s provides quarterly adjusted income statement for the first three quarters of year
t and year t-1; zero otherwise. ∆AdjNI3Q is Moody’s adjustments to firm i’s net income before unusual items for the
first three quarters of year t less Moody’s adjustments for the first three quarters of year t-1, scaled by lagged total
assets. ∆AdjEBIT3Q is Moody’s adjustments to firm i’s EBIT for the first three quarters of year t less Moody’s adjustments for the first three quarters of year t-1, scaled by beginning total assets. ∆AdjOTH3Q is ∆AdjNI3Q less
∆AdjEBIT3Q. EarnSurprise is firm i’s earnings surprise if earnings is announced inside the 10-K filing window, and
zero otherwise. Earnings surprise is the actual earnings per share (EPS) for period t minus analyst j’s forecast of
EPS, scaled by stock price measured at the end of year t-1. A firm-year measure of EarnSurprise, TgtPriceRev and
EarnFcstRev is created by taking the firm-year mean of the analyst-firm-year values if there is more than one value,
and equal to zero if there is no value. Book-Market is the ratio of the book value of equity to the market value of
equity for year t. Size is the logarithm of the market value of equity for year t.
(Table continues on next page)
45
TABLE 7 (Continued)
Explanatory
Variable
Predicted
Sign (1) (2) (3) (4)
∆AdjNI + 0.062 0.063
(3.69)*** (3.77)***
∆AdjEBIT + 0.077 0.079
(3.24)*** (3.42)***
∆AdjOTH + 0.052 0.053
(4.71)*** (4.67)***
AdjUNUSUAL ? -0.120 -0.120 -0.120 -0.120
(-2.69) (-2.70)*** (-2.69) (-2.70)***
LagAdjUNUSUAL ? 0.017 0.017 0.017 0.017
(0.56) (0.56) (0.58) (0.57)
TgtPriceRev + 0.026 0.026 0.026 0.026
(2.62)*** (2.62)*** (2.61)*** (2.62)***
EarnFcstRev + 0.010 0.010 0.010 0.010
(5.71)*** (5.70)*** (5.70)*** (5.69)***
Adj3q_dum 0.003 0.003
(0.95) (0.93)
adj3q_d -0.140
(-1.55)
adjebit3q_d -0.143
(-1.46)
adjoth3q_d -0.137
(-0.74)
EarnSurprise + 0.327 0.329 0.325 0.327
(2.47)** (2.49)** (2.45)*** (2.45)***
Book-Market ? 0.002 0.002 0.002 0.002
(0.69) (0.68) (0.67) (0.65)
Size ? -0.001 -0.001 -0.001 -0.001
(-1.62) (-1.63) (-1.67)* (-1.68)*
Year fixed effects Yes Yes Yes Yes
No. of obs. 3,069 3,069 3,069 3,069
Adj. R2 (%) 7.3 7.3 7.3 7.4