Strategic Revenue Recognition to Achieve Earnings

Benchmarks



Marcus L. Caylor
University oI South Carolina

Moore School oI Business
1705 College Street
Columbia, SC 29208

Email: marcus.caylormoore.sc.edu
OIIice: 803-777-6081
Fax: 803-777-0712

This study is based in part on my doctoral dissertation at Georgia State University. I am grateIul Ior
helpIul comments and suggestions Irom my dissertation committee: Larry Brown (chair), Lynn Hannan,
Jayant Kale, and Siva Nathan. This paper has also beneIited Irom the helpIul comments and suggestions oI
Ashiq Ali, Tony Chen, Bill Cready, Robert Freeman, Artur Hugon, Scott Jackson, Ross Jennings, Steve
Kachelmeier, Bill Kinney, Krishna Kumar, Yen Lee, Andrew Leone, Tom Lopez, Arianna Pinello, Suresh
Radhakrishnan, Galen Sevcik, Scott Vandevelde, Rich White and workshop participants at the 2006
American Accounting Association Annual Meetings, Georgia State University, the University oI South
Carolina, the University oI Texas at Austin and the University oI Texas at Dallas. I am grateIul to
Thomson Financial/I/B/E/S Ior providing data on analysts' earnings Iorecasts.
1


Strategic Revenue Recognition to Achieve Earnings
Benchmarks

SYNOPSIS: I examine whether managers use discretion in the two accounts related to
revenue recognition, accounts receivable and deIerred revenue, to avoid three common
earnings benchmarks. I Iind that managers use discretion in both accounts to avoid
negative earnings surprises. I Iind that neither oI these accounts is used to avoid losses or
earnings decreases. For a common sample oI Iirms with both deIerred revenue and
accounts receivable, I show that managers preIer to exercise discretion in deIerred
revenue vis-a-vis accounts receivable. I provide a reason Ior why managers might preIer
to manage a deIerral rather than an accrual: lower costs to manage (i.e., no Iuture cash
consequences). My results suggest that iI given the choice, managers preIer to use
accounts that incur the lowest costs to the Iirm.



Keywords: Revenue recognition, earnings surprises, earnings management, accounts
receivable, deferred revenue.

Data availability: All data are available from public databases identified in the paper.

















2
Strategic Revenue Recognition to Achieve Earnings
Benchmarks

1. Introduction
Revenue recognition is one oI the most important Iinancial reporting issues Iacing
Iirms today, usually representing the largest line item on income statements. It is a
timely issue to examine as evidenced by the Revenue Recognition joint project currently
undertaken by The Financial Accounting Standards Board (FASB) and the International
Accounting Standards Board (IASB).
1
Depending on the nature oI a Iirm`s business,
there is an accrual and a deIerral that relates to the amount oI revenue recognized in an
accounting period: accounts receivable and deIerred revenue, respectively.
2
I examine
both accounts to see iI discretion in revenue recognition is used to achieve three common
earnings benchmarks common in the literature: avoid losses, avoid earnings decreases,
and avoid negative earnings surprises (Brown and Caylor 2005; Burgstahler and Dichev
1997; Degeorge, Patel and Zeckhauser 2001). Both oI these accounts are alternative
ways Ior recognizing revenue so it represents a unique opportunity to examine how
managers choose between two rather diIIerent types oI earnings management.
3
I examine
whether Iirms with both accounts available to them express a preIerence Ior discretion in
the accrual or deIerral account.

1
See http://Iasb.org/project/revenuerecognition.shtml Ior more details.
2
DeIerred revenue goes by several other names including advances Irom customers, unearned revenue and
revenue received in advance. Surprisingly, little research has examined the deIerred revenue account. The
only study that directly examines this account is Bauman (2000). Using a sample oI 22 Iirms Irom the
publishing industry, he Iinds that sales increases in the current year do not persist into the Iuture unless
accompanied by increases in deIerred revenue, suggesting that deIerred revenues are a leading indicator oI
Iuture earnings.
3
A common theme oI prior studies is to examine speciIic accruals and reserves related to expense
recognition (e.g., Frank and Rego (2006)).
3
I provide an explanation Ior why discretion in the deIerral (i.e., deIerred revenue)
would be preIerred based on accounting intuition. First, gross accounts receivable is
managed primarily through real activities, such as easing credit policies. Management oI
deIerred revenue, on the other hand, represents a situation where cash has already been
received. Thus, management oI deIerred revenue relates more to manipulation oI
accounting estimates. Managing gross accounts receivable should be more costly to
Iirms as it relates to accelerating sales where cash has not been collected yet. Thus, it has
Iuture cash consequences whereas deIerred revenue does not.
I construct a model Ior the normal change in short-term deIerred revenue to
determine abnormal changes in short-term deIerred revenue.
4
I derive a similar model Ior
the normal change in gross accounts receivable to determine abnormal changes in gross
accounts receivable. I use gross accounts receivable in lieu oI net accounts receivable
because abnormal changes in net accounts receivable could reIlect changes in the
allowance Ior bad debt and I am solely interested in discretion in revenue recognition
(i.e., the revenue account).
5
I create a pre-managed distribution oI earnings by removing
the discretionary component related to the account in question (Dhaliwal, Gleason and
Mills 2004; Frank and Rego 2006), and then test whether abnormal changes in each oI
these revenue accounts are higher than would be expected Ior Iirms with pre-managed
earnings that just miss an earnings benchmark. Next, I examine Iirms with both accounts
to see iI managers preIer one account as a means Ior discretion in revenue recognition.

4
I use the short-term deIerred revenue component and ignore the long-term component because the long-
term component oI deIerred revenue does not reIlect revenue that should have been recognized during the
current period.
5
The manipulation oI the allowance Ior bad debt involves the manipulation oI accounting estimates related
to receivables that are not expected to be collected.
4
My results indicate that both deIerred revenue and accounts receivable are
managed in an attempt to avoid negative earnings surprises.
6
I Iind that neither is
managed to avoid losses or earnings decreases. In addition, I provide evidence that Iirms
preIer to exercise discretion in deIerred revenue relative to accounts receivable to avoid
negative earnings surprises when given the choice.
My study makes several contributions to the literature. First, I provide the Iirst
descriptive evidence on deIerred revenue, showing that many high technology industries
have it on their balance sheets. Second, I provide the Iirst comprehensive analysis oI
revenue manipulation in relation to all three earnings benchmarks and show that
discretion is used to avoid negative earnings surprises but not the other two benchmarks.
Third, my study is the Iirst to examine a common sample oI Iirms with a deIerral and an
accrual available that are used Ior the same purpose (i.e., increase revenue) but diIIer in
terms oI costliness to see whether managers preIer to manage one type oI account. I
provide evidence that deIerred revenue is the preIerred earnings management account,
indicating that when given the choice, managers preIer to exercise discretion in a
deIerral, suggesting that they preIer to minimize real costs to the Iirm. Fourth, I derive a
discretionary model oI deIerred revenue Ior potential use in Iuture research on
discretionary deIerrals.
The remainder oI my paper is organized as Iollows. The second section reviews
the relevant literature and develops hypotheses. Section three introduces the research

6
In a contemporaneous study, Stubben (2007) also Iinds that accounts receivable are used to exceed the
analyst benchmark. However, his study does not examine deIerred revenue nor does it examine whether a
preIerence exists.

5
design. Section Iour provides results oI my study, and section Iive contains implications
oI my study.
2. Hypotheses Development
Burgstahler and Dichev (1997) and Degeorge et al (2001) introduce three
earnings benchmarks that Iirms attempt to achieve. In two recent studies, researchers
Iind no evidence that discretion is used in gross accounts receivable to avoid losses and
earnings decreases (Marquardt and Weidman 2004; Roychowdhury 2006).
7
Recent
evidence also suggests that the analyst benchmark is the most important benchmark
sought by managers (Dechow, Richardson and Tuna 2003; Brown and Caylor 2005)
indicating that Iirms may be more likely to manage earnings to achieve this benchmark
than the others. Firms can recognize more revenue to meet or just beat earnings
benchmarks via discretion in accounts receivable, deIerred revenue or a combination oI
both. I hypothesize that managers will use discretion in revenue recognition to avoid
negative earnings surprises but not Ior the other two benchmarks given the above
evidence.
However, I may not Iind results Ior the analyst benchmark Ior three reasons.
First, I examine a post-SAB 101 environment where accounting regulations on revenue
recognition are speciIically written to prevent aggressive recognition (see SAB 101).
8

However, Rountree (2006) Iinds that deIerred revenue was more likely to be targeted by
SAB 101, suggesting that deIerred revenue may be more likely to be aIIected in

7
Marquardt and Weidman (2004) Iind that managers do not exercise discretion in gross accounts
receivable to avoid an earnings decrease. Roychowdhury (2006) Iinds no signiIicant evidence that gross
accounts receivable are managed to avoid a loss (see Iootnote 25 oI his study).
8
In 1999, the Securities Exchange Commission (SEC) issued StaII Accounting Bulletin (SAB) No. 101,
which states that revenue can be recognized only when the Iollowing Iour criteria are met: 1) persuasive
evidence oI an arrangement exists, 2) delivery has occurred or services have been rendered, 3) the seller's
price to the buyer is Iixed or determinable, and 4) collectibility is reasonably assured.
6
comparison to accounts receivable. Second, managers use other (non-revenue-based)
accruals Ior avoiding negative surprises (Moehrle 2002; Frank and Rego 2006; among
others) potentially mitigating the eIIect I expect to Iind Ior revenue-based accruals.
Third, managers have another mechanism, expectations management, to achieve the
analyst benchmark also reducing the power oI my tests. In addition, it is unclear whether
the results Iound by Marquardt and Weidman (2004) and Roychowdhury (2006) extend
to deIerred revenue given the lower real costs that discretion imposes on the Iirm.
I hypothesize that managers do use discretion in revenue recognition to avoid
negative earnings surprises but not losses or earnings decreases. More Iormally, my Iirst
two hypotheses are:
H
1
: Firms with pre-managed earnings that just miss the analyst (loss or earnings
decrease) benchmark do (do not) experience an abnormal increase in gross
accounts receivable.
H
2
: Firms with pre-managed earnings that just miss the analyst (loss or earnings
decrease) benchmark do (do not) experience an abnormal decrease in short-term
deIerred revenue.
Major diIIerences exist between the manipulation oI accounts receivable and
short-term deIerred revenue. First, with deIerred revenue, cash has already been received
and a journal entry has been made. Three oI the Iour criteria Irom SAB 101 Ior revenue
recognition have usually been met with the recording oI deIerred revenue (i.e., persuasive
evidence oI an arrangement exists, the seller's price to the buyer is Iixed or determinable,
and collectibility is reasonably assured). Thus, discretion arises in deIerred revenue as to
when delivery has occurred or services have been rendered.
7
Second, with deIerred revenue, managers are less able to manipulate through real
activities. For gross accounts receivable, managers accelerate the recognition oI revenue
through real activities manipulation, such as providing Iavorable credit terms, easing
creditworthiness restrictions, and speeding up the shipment oI goods.
9
In contrast to
gross accounts receivable, managers accelerate recognition oI deIerred revenue by
increasing estimates oI services provided.
10
This causes manipulation oI deIerred
revenue to be relatively less costly vis-a-vis accounts receivable in terms oI its Iuture
cash consequences. For example, providing Iavorable credit terms to speed up the
recognition oI a receivable has Iuture cash consequences, as a higher percentage oI
customers may deIault payment in the Iuture. In addition, managing gross accounts
receivable has long-term reputation eIIects with customers, even iI it has no direct Iuture
cash consequences. For instance, a large supplier may push unwanted merchandise on
small retailers to speed up recognition oI a receivable.
I expect managers to preIer discretion in deIerred revenue relative to accounts
receivable as a result oI the real costs imposed by accounts receivable. However, a
preIerence may not exist iI managers are not concerned with costs imposed on their Iirm.
My third hypothesis is:
H
3
: Firms with deIerred revenue will use less discretion in gross accounts
receivable to avoid negative earnings surprises.


9
Gross accounts receivable can be managed through subjective estimates oI how much revenue has been
earned, but only apply to a very Iew industries where long-term construction projects exist.
10
While managers can use real activities manipulation to determine when a credit sale is recorded, with
deIerred revenue the use oI real activities manipulation is less likely. For instance, it is unlikely that
managers would withhold services to customers to reduce the amount oI revenue recognized. DeIerred
revenue could be manipulated through altering contract terms, however, because a customer has to agree to
the new terms, such an action is less likely to occur.
8
3. Sample Selection and Research Design
3.1 Sample Selection
I obtain annual earnings, short-term deIerred revenue, accounts receivable, sales,
total assets, cash Ilow Irom operations and other Iinancial statement accounts Irom the
2005 Annual Compustat File. I obtain annual analyst earnings Iorecast data and reported
annual earnings Ior computing earnings surprises Irom the split-unadjusted I/B/E/S Detail
File. To be consistent with prior literature on earnings management (e.g., Burgstahler
and Dichev 1997), I exclude utilities and Iinancial Iirms (i.e., SIC codes between 4400
and 5000 and SIC codes between 6000 and 6500). I also exclude any Iirms related to
public administration (i.e., SIC codes oI 9000 or higher). Fiscal years beIore 2001 are
not used because deIerred revenue data coverage in Compustat begins in Iiscal year 2000.
3.2 Modeling Normal Changes in Gross Accounts Receivable
To derive a model Ior the expected amount oI gross accounts receivable in time t,
I make a couple oI simpliIying assumptions. I assume that gross accounts receivable is
some proportion oI current period`s sales, as accounts receivable are included in this
period`s sales.
11
I also assume that gross accounts receivable are some proportion oI next
period`s cash Ilow Irom operations, since accounts receivable will turn over in the next
period. This implies that changes in gross accounts receivable should be positively
related to contemporaneous changes in sales and Iuture changes in cash Ilow Irom
operations.
12
I include both oI these variables in my model to capture any non-
discretionary component that is not captured by the other. Thus, iI gross accounts

11
This assumption is adopted Irom Dechow, Kothari, and Watts (1998).
12
I Iind that changes in gross accounts receivable are signiIicantly and positively correlated with both
changes in current period`s sales and changes in Iuture period`s cash Ilow Irom operations using both
Pearson and Spearman correlations.
9
receivable are a greater proportion oI this period`s sales or next period`s cash Ilow Irom
operations than expected, more accounts receivable may have been recorded than
expected.
13
Based on these assumptions, I estimate abnormal changes in gross accounts
receivable by running linear regressions by industry (2-digit SIC code) and Iiscal year
using all available Iirms with the requisite data:
14

^Gross A/R
t
/ A
t-1
o
0
o
1
*(1/ A
t-1
)
1
*(^S
t
/ A
t-1
)
2
*(^CFO
t1
/ A
t-1
) c
t
(1)
where:

A Gross A/R
t
change in gross accounts receivable during year t (change in Compustat
Annual Data Item 2 plus Compustat Annual Data Item 67),
AS
t
change in sales during year t (change in Compustat Annual Data Item 12),
ACFO
t1
change in cash Ilow Irom operations during year t1 (change in Compustat
Annual Data Item 308), and
A
t-1
beginning oI the year total assets (Compustat Annual Data Item 6).
15

In addition to the scaled intercept term Iound in prior discretionary accrual
studies, I also include a constant term based on Kothari, Leone and Wasley (2005) who
Iind that it results in better-speciIied, more symmetric discretionary models.
16
I compute
the abnormal change in gross accounts receivable Ior the current period as the diIIerence
between the actual change in gross accounts receivable and the predicted (or expected)
change obtained Irom these industry-year regressions. An abnormal increase in gross

13
The opposite implies that less accounts receivable may have been recorded than expected.
14
I estimate at the 2-digit level to be consistent with prior literature. I also winsorize all variables entering
both oI my discretionary models at the extreme (1st and 99th) percentiles oI their respective distributions to
be consistent with prior literature. In addition, I require at least eight industry-year observations to estimate
the model.
15
I choose a cross-sectional industry discretionary model in lieu oI a Iirm-speciIic model due to data
restrictions (i.e., a small time-series oI data). Bartov, Gul and Tsui (2000) provide evidence that suggests a
cross-sectional discretionary model is superior to a time-series model.
16
Kothari, Leone, and Wasley (2005) include both a scaled and unscaled intercept term in their
discretionary accrual models.
10
accounts receivable occurs when the actual change exceeds the predicted 'normal
change. Abnormally low growth in gross accounts receivable occurs when the actual
change is less than the predicted 'normal change.
3.3 Modeling Normal Changes in Deferred Revenue
To derive a model Ior the expected amount oI short-term deIerred revenue, I make
a similar set oI assumptions to those made Ior accounts receivable modiIied Ior the
opposite behavior oI deIerrals relative to accruals. More speciIically, I assume that short-
term deIerred revenue is a proportion oI next period`s sales, since deIerred revenue is to
be recognized in the next period. I also assume that short-term deIerred revenue is a
proportion oI the current period`s cash Ilow Irom operations, because deIerred revenue in
the current period is reIlected in the current period`s cash Ilow Irom operations. This
implies that changes in short-term deIerred revenue should be positively related to
contemporaneous changes in cash Ilow Irom operations and Iuture changes in sales.
17

Thus, iI short-term deIerred revenue is a greater proportion oI either the current period`s
cash Ilow Irom operations or next period`s sales than expected, more short-term deIerred
revenue remains than expected. Based on these assumptions, I estimate abnormal
changes in deIerred revenue by running linear regressions by industry (2-digit SIC code)
and Iiscal year using all available Iirms with the requisite data:
18

^Def Rev
t
/ A
t-1
o
0
o
1
*(1/ A
t-1
)

1
*(^S
t1
/ A
t-1
)
2
*(^CFO
t
/ A
t-1
) c
t
(2)
where:


17
I Iind that changes in short-term deIerred revenue are signiIicantly and positively correlated with changes
in Iuture sales and changes in current period`s cash Ilow Irom operations using both Pearson and Spearman
correlations.
18
I require at least eight industry-year observations to estimate this model similar to the constraint Ior the
gross accounts receivable model.
11
ADeI Rev
t
change in short-term deIerred revenue during year t (change in Compustat
Annual Data Item 356),
AS
t1
change in sales during year t1 (change in Compustat Annual Data Item 12),
ACFO
t
change in cash Ilow Irom operations during year t (change in Compustat Annual
Data Item 308), and
A
t-1
beginning oI the year total assets (Compustat Annual Data Item 6).
I compute the abnormal change in deIerred revenue Ior the current period as the
diIIerence between the actual change in deIerred revenue and the predicted change Irom
these industry-year regressions. An abnormal increase in deIerred revenue occurs when
the actual change exceeds the predicted change. Similarly, an abnormal decrease in
deIerred revenue occurs when the actual change Ialls short oI the predicted change. My
measure is a proxy Ior the change in deIerred revenue relative to what the expected or
'normal value should be given changes in cash Ilow Irom operations and Iuture sales.
19

I use the short-term component and ignore the long-term component because the latter
does not reIlect revenue that should be recognized during the current period.
3.4 Models to Test Hypotheses
I use pre-managed earnings in lieu oI post-managed earnings because this better
reIlects ex-ante behavior. Pre-managed earnings are obtained by removing the
discretionary component oI earnings (Dhaliwal, Gleason, and Mills 2004; Frank and

19
In an attempt to provide some evidence on my discretionary deIerred revenue model relative to prior
discretionary models, I Iormed deciles based on the level oI discretionary accruals Ior all Iirms in the
Compustat universe Ior the same time period using a modiIied-Jones model and examined the mean
abnormal change in deIerred revenue within these deciles. An interesting Ieature oI this comparison is that
deIerred revenue is likely to be only a small proportion oI aggregate accruals on average and discretionary
deIerred revenue will move in an opposite direction to aggregate discretionary accruals, thus any negative
relation between the two will provide comIort that my model is eIIectively picking up discretionary
behavior. Untabulated analyses reveal that Ior every decile the sign oI mean abnormal changes in deIerred
revenue is opposite to the sign oI mean discretionary accruals.
12
Rego 2006; among others). To compute pre-managed earnings Ior gross accounts
receivable, I subtract the abnormal change in gross accounts receivable Irom reported
earnings. To compute pre-managed earnings Ior deIerred revenue, I add the abnormal
change in deIerred revenue to reported earnings. I convert the abnormal change oI the
revenue account to an undeIlated amount by multiplying by lagged total assets and then
scaling by common shares outstanding used to calculate EPS (Compustat Annual Data
Item #54) in order to adjust I/B/E/S reported earnings per share. I deIine pre-managed
earnings in year t as pre-managed annual earnings (Compustat Data Item 18) in year t. I
deIine a pre-managed earnings change in year t as pre-managed annual earnings
(Compustat Data Item 18) in year t minus annual earnings in year t-1. I deIine a pre-
managed earnings surprise in year t as pre-managed earnings in year t minus the
consensus analyst Iorecast oI earnings in year t.
20

To test my Iirst two hypotheses, I examine how abnormal changes in gross
accounts receivable (deIerred revenue) are related to instances where a Iirm`s pre-
managed earnings just misses the analyst`s Iorecast. I estimate the regression:
Abnormal^Gross A/R
t

(or Abnormal^Def Rev
t
)



0

1
* PRE-
MANAGEDJUSTMISS
t

2
* PRE-MANAGEDMEETJUSTBEAT
t


1
*SIZE
t-1

2
*BM
t-1
c
t


(3)

where AbnormalA Gross A/R (AbnormalADeI Rev ) is the abnormal change in gross
accounts receivable (deIerred revenue). For the loss and earnings decrease benchmarks,
PRE-MANAGEDJUSTMISS is deIined as an indicator variable equal to 1 iI a Iirm

20
I calculate the consensus annual earnings Iorecast based on the median oI the last individual earnings
Iorecasts made by all analysts in the 90-day period preceding the end oI the Iiscal year. This has the
advantage over using I/B/E/S summary Iorecasts because it avoids the stale Iorecast problem.
13
reports a pre-managed loss (earnings decrease) in year t oI no more than 0.5 (0.25) oI
the end oI the prior Iiscal year`s stock price.
21
PRE-MANAGEDMEETJUSTBEAT is
deIined as an indicator variable equal to 1 iI a Iirm reports a pre-managed non-negative
earnings (earnings decrease) in year t oI less than 0.5 (0.25) oI the end oI the prior
Iiscal year`s stock price. For the analyst benchmark, PRE-MANAGEDJUSTMISS is
deIined as an indicator variable equal to 1 iI a Iirm reports a pre-managed negative
earnings surprise in year t oI no more than 0.2 oI the end oI the prior Iiscal year`s stock
price. PRE-MANAGEDMEETJUSTBEAT is deIined as an indicator variable equal to
1 iI a Iirm reports a pre-managed non-negative earnings surprise in year t oI less than
0.2 oI the end oI the prior Iiscal year`s stock price.
22

PRE-MANAGEDMEETJUSTBEAT is used as a natural reIerence group since
these Iirms should have no motives to manage revenue given they were able to achieve
the benchmark beIore discretion in revenue is considered. I expect to Iind a signiIicant
and positive (negative) coeIIicient on PRE-MANAGEDJUSTMISS Ior accounts
receivable (deIerred revenue). An F-test is conducted between PRE-
MANAGEDJUSTMISS and PRE-MANAGEDMEETJUSTBEAT when PRE-
MANAGEDJUSTMISS is signiIicantly diIIerent Irom zero and PRE-
MANAGEDMEETJUSTBEAT is oI the same sign. To control Ior systematic
diIIerences in abnormal changes in gross accounts receivable, I include SIZE, the natural

21
These intervals are consistent with Burgstahler and Dichev (1997).
22
Any deIinition oI small miss or small beat is arbitrary. My choice is based on prior research that has
examined the avoidance oI negative earnings surprises. I use a 0.2 interval width Ior earnings surprises
consistent with Burgstahler and Eames (2006). An additional advantage oI this choice is that it represents
the best trade-oII between the smallest interval width and the most observations to make reliable statistical
inIerences. However, my results are qualitatively similar using other interval widths (e.g., 0.3).
14
logarithm oI a Iirm`s beginning oI the year market value oI equity. To control Ior growth
opportunities, I include BM, the book-to-book ratio.
23

The sample related to testing my Iirst hypothesis pertaining to accounts receivable
is 4,562 Iirm-year observations Ior Iiscal years 2001-2003. My sample to test my second
hypothesis pertaining to deIerred revenue is 1,378 Iirm-year observations Ior Iiscal years
2001-2003. Fiscal years beIore 2001 are not used because deIerred revenue data
coverage in Compustat begins in Iiscal year 2000. Fiscal year 2004 is not included in my
Iinal samples because I require cash Ilow Irom operations one year ahead in order to
compute the abnormal change in accounts receivable and I require sales one year ahead in
order to compute the abnormal change in deIerred revenue.
To test my third hypothesis, I require Iirms to have both accounts receivable and
deIerred revenue. This reduces my sample to 962 Iirm-year observations. I then re-
estimate model 3 Ior this common sample. To the extent that one account aIIects the
other, I include the abnormal change oI the other revenue account as an additional control
Ior this analysis. II my third hypothesis is correct, I should Iind signiIicance on PRE-
MANAGEDJUSTMISS Ior deIerred revenue but not Ior gross accounts receivable.
4. Results
4.1 Descriptive Evidence
Little is known about the deIerred revenue account. I begin by providing some
industry-speciIic evidence oI this account. For Iiscal year 2004, I Iind that 30.4 oI
Iirms reported non-zero short-term deIerred revenue.
24
Table 1 provides descriptive
inIormation Ior the 48 Fama-French (FF) industry groups Ior Iiscal years 2001-2004,

23
I winsorize this ratio at the top and bottom 1 oI its distribution.
24
Firms with missing assets were excluded. I do not exclude utilities and Iinancial Iirms Ior purposes oI
providing descriptive evidence in Table 1. A similar proportion is Iound Ior earlier years in my sample.
15
ranked in ascending order by percentage oI Iirms in the industry with non-zero short-term
deIerred revenue (Fama and French 1997). All 48 FF industry groups have some Iirms
with short-term deIerred revenue on their balance sheets. The top ten industry groups in
terms oI percent oI Iirms with short-term deIerred revenue were Printing and Publishing
(66.67), Computers (55.56), Business Services (51.14), Measuring and Control
Equipment (45.38), Telecommunications (45.11), Pharmaceutical Products (39.81),
Personal Services (37.22), Electronic Equipment (35.65), DeIense (33.33), and
Medical Equipment (33.17). With the exceptions oI Printing and Publishing and
Personal Services, the rest are high technology sectors that relate to medical or
computer/electronic technology. The bottom ten industry groups in terms oI percent oI
Iirms with short-term deIerred revenue were Banking (1.37), Textiles (4.17), Utilities
(5.01), Shipbuilding, Railroad Equip. (5.26), Precious Metals (5.39), Agriculture
(6.45), AircraIt (6.84), Food Products (7.06), Steel Works, Etc. (7.25), and
Rubber and Plastic Products (7.30).
---------------------------
Insert Table 1 here
----------------------------
Table 2 reports the mean coeIIicient estimates Irom estimating the models Ior the
normal change in gross accounts receivable and the normal change in deIerred revenue.
T-statistics are computed by dividing the mean oI the distribution across all industry-year
observations Ior each oI the variables in the model by the standard error oI this
distribution.

16
---------------------------
Insert Table 2 here
----------------------------
The model Ior gross accounts receivable has an adjusted R-square oI almost 37.
As expected, there is a signiIicant and positive relationship between changes in gross
accounts receivable and changes in current sales (coeIIicient 0.0979; t-statistic 14.27)
and Iuture cash Ilow Irom operations (coeIIicient 0.0573; t-statistic 2.92). The model
Ior deIerred revenue has an adjusted R-square oI almost 30. Also, as expected, there is
a signiIicant and positive relationship between changes in short-term deIerred revenue
and changes in Iuture sales (coeIIicient 0.0234; t-statistic 4.45) and current cash Ilow
Irom operations (coeIIicient 0.0365; t-statistic 1.94).
Table 3 provides descriptive statistics Ior all Iirms with either accounts receivable
or deIerred revenue Ior Iiscal years 2001-2003. Gross accounts receivable has a mean oI
nearly 291 million dollars and a median oI nearly 17.69 million dollars. The mean
change in gross accounts receivable is approximately 0.1 oI beginning assets. DeIerred
revenue has a mean oI nearly 40 million dollars and a median oI nearly 2.14 million
dollars. The mean change in deIerred revenue is approximately 0.9 oI beginning assets.
By deIinition, the mean abnormal change in short-term deIerred revenue and the mean
abnormal change in gross accounts receivable are zero since I include a constant term in
my discretionary models.
---------------------------
Insert Table 3 here
----------------------------
17
4.2 Earnings Benchmark Results
Table 4 reports the results oI OLS regressions examining my Iirst hypothesis
related to abnormal changes in gross accounts receivable. Consistent with my Iirst
hypothesis, I Iind that abnormal changes in gross accounts receivable are more positive
than normal Ior PRE-MANAGEDJUSTMISS Ior the analyst benchmark (coeIIicient
0.0017; t-statistic 1.65).
25
I Iind a negative and signiIicant coeIIicient on PRE-
MANAGEDMEETJUSTBEAT. As expected, I Iind no evidence Ior the other two
benchmarks. Although I Iind a signiIicant and positive coeIIicient on
PREMANAGEDJUSTMISS Ior the loss avoidance benchmark, an F-test reveals that it
is not signiIicantly diIIerent Irom that oI PREMANAGEDMEETJUSTBEAT (F-test
0.06).
---------------------------
Insert Table 4 here
----------------------------
Table 5 provides results related to my second hypothesis. The evidence is similar
to that reported in Table 4 Ior gross accounts receivable. More speciIically, there is a
negative and signiIicant coeIIicient on PRE-MANAGEDJUSTMISS Ior the analyst
benchmark (-0.0061; t-statistic -4.14) Ior abnormal changes in deIerred revenue. I Iind
an insigniIicant coeIIicient on PRE-MANAGEDMEETJUSTBEAT. I do not Iind
evidence Ior the other two benchmarks.
26


25
All coeIIicient estimates are presented in decimal Iorm. For instance, this coeIIicient translates into an
abnormal increase in gross accounts receivable that was 0.17 oI beginning total assets.
26
To the extent that my proxy Ior discretionary revenue recognition contains measurement error, a
correlation may be induced between pre-managed earnings and the abnormal change in revenue account
(Leone and Rock 2002). However, it is unclear why such a relation would exist Ior only the PRE-
MANAGEDJUSTMISS interval in relation to the other pre-managed intervals. Nonetheless, I perIorm
18
To provide some evidence on the prevalence oI revenue recognition to avoid
negative earnings surprises, I also conduct an analysis similar to that oI Frank and Rego
(2006). I examine the proportion oI Iirms that use discretion in revenue accounts to cross
over the analyst threshold. I Iind that 75.8 (72.7) oI Iirms that had pre-managed
earnings just missing analyst Iorecasts were able to use discretion in gross accounts
receivable (short-term deIerred revenue) to meet or beat the benchmark.
---------------------------
Insert Table 5 here
----------------------------
4.3 Do Managers Express a Preference for Revenue Management?
Table 6 provides results pertinent to my third hypothesis. Consistent with my
expectations, I Iail to Iind signiIicant evidence that Iirms with deIerred revenue use
discretion in gross accounts receivable (coeIIicient 0.0016; t-statistic 0.71). I
continue to Iind a signiIicant and negative coeIIicient on abnormal changes in deIerred
revenue (coeIIicient -0.0058; t-statistic -3.14). An alternative explanation could be
that these Iirms have a smaller stock oI receivables than deIerred revenue so these Iirms
would not Iind discretion in accounts receivable to be as economically Ieasible.
However, I Iind that these Iirms actually have a much higher mean stock oI receivables
than deIerred revenue by a Iactor oI 10 indicating that such an alternative explanation is
not plausible.


two additional analyses. In the Iirst, I regress the abnormal change in revenue account on pre-managed
earnings and Iind insigniIicant coeIIicients Ior both revenue measures. I also include PRE-
MANAGEDEARNINGS, the magnitude oI pre-managed earnings, in the regressions reported in tables 4-
5 to control Ior this correlation iI it exists. I obtain qualitatively similar results.
19
---------------------------
Insert Table 6 here
----------------------------
5. Conclusions and Implications
I examine whether managers use accounting discretion in two accounts related to
revenue recognition, short-term deIerred revenue and gross accounts receivable, to avoid
missing three common earnings benchmarks. My evidence suggests that managers
accelerate the recognition oI revenue using both accounts when pre-managed earnings
miss the analyst benchmark by a small amount. I Iind no signiIicant evidence Ior the
other two benchmarks. Using a common sample, I Iind that managers preIer to exercise
discretion in deIerred revenue as opposed to accounts receivable to avoid negative
earnings surprises. I oIIer an explanation why deIerred revenue would be the preIerred
account, the lower costs oI discretion associated with deIerred revenue. However, iI
managers do not have a choice they will choose a mechanism that does have Iuture
consequences in order to avoid negative surprises. Finally, I introduce a discretionary
model Ior deIerred revenue that Iuture researchers can use when studying deIerrals.
My results have implications Ior policy makers and auditors. First, although not
the Iocus oI my study, my results suggest that discretion in revenue recognition continues
post-SAB 101. Standards setters should evaluate whether policy can be created to reduce
this discretion. The Revenue Project undertaken by the FASB and IASB with its
emphasis on an asset and liability model over an earnings process model could represent
a step in the right direction.
27
Standards setters should also examine my evidence
regarding the preIerence Ior discretion in deIerred revenue. While a common allegation

27
http://Iasb.org/boardhandouts/11-20-07RREducSession.pdI
20
is that managers are only short-term Iocused at the expense oI long-term value creation,
my results suggest that managers preIer the revenue recognition mechanism that has the
least long-term consequences. My results suggest that Iuture policy should consider
trade-oIIs between alternate revenue recognition mechanisms and consider that some
managerial discretion can be more harmIul in the long-term than others. Finally, my
results provide evidence that suggests that auditors should be cognizant oI a potential
preIerence Ior discretion in revenue recognition.
















21
REFERENCES
Bauman, M. 2000. 'The Unearned Revenue Liability and Firm Value: Evidence Irom the
Publishing Industry. Working paper, University oI Illinois at Chicago.
Brown, L., and M. Caylor. 2005. 'A Temporal Analysis oI Quarterly Earnings
Thresholds: Propensities and Valuation Consequences. The Accounting Review 80
(April): 423-440.
Burgstahler, D., and I. Dichev. 1997. 'Earnings Management to Avoid Earnings

Decreases and Losses. Journal of Accounting and Economics 24 (December): 99-126.

Burgstahler, D., and M. Eames. 2006. 'Management oI Earnings and Analysts' Forecasts

to Achieve Zero and Small Positive Earnings Surprises. Journal of Business Finance &

Accounting 33 (June/July): 633-652.

Dechow, P., S. Kothari, and R. Watts. 1998. 'The Relation between Earnings and Cash
Flows. Journal of Accounting and Economics 25 (May): 133-168.
Dechow, P., S. Richardson, and I. Tuna. 2003. 'Why are Earnings Kinky? An
Examination oI the Earnings Management Explanation. Review of Accounting Studies 8
(June): 355-384.
Degeorge, F., J. Patel, and R. Zeckhauser. 1999. 'Earnings Management to Exceed
Thresholds. Journal of Business 72 (January): 1-33.
Dhaliwal, D., C. Gleason, and L. Mills. 2004. 'Last-chance Earnings Management: Using
the Tax Expense to Meet Analysts` Forecasts. Contemporarv Accounting Research 21
(Summer): 431-459.
Fama, E., and K. French. 1997. 'Industry Costs oI Equity. Journal of Financial
Economics 43 (February): 153-193.
22
Frank, M., and S. Rego. 2006. 'Do Managers Use the Valuation Allowance Account to
Manage Earnings around Certain Earnings Targets? Journal of the American Taxation
Association 28 (Spring): 43-65.
Kothari, S., A. Leone, and C. Wasley. 2005. 'PerIormance Matched Discretionary

Accrual Measures. Journal of Accounting & Economics 39 (February): 163-197.

Leone, A., and S. Rock. 2002. 'Empirical Tests oI Budget Ratcheting and its EIIect on

Managers` Discretionary Accrual Choices. Journal of Accounting & Economics

33 (February): 43-67.

Marquardt, C., and C. Wiedman. 2004. 'How are Earnings Managed? An Examination oI
SpeciIic Accruals. Contemporarv Accounting Research 21 (Summer): 461-491.
Moehrle, S. 2002. 'Do Firms Use Restructuring Charge Reversals to Meet Earnings
Targets? The Accounting Review 77 (April): 397-413.
Newey, W., and K. West. 1987. 'A Simple, Positive Semi-deIinite, Heteroskedasticity
and Autocorrelation Consistent Covariance Matrix. Econometrica 55 (May): 703-708.
Rountree, B. 2006. 'Mandatory Accounting Changes and Firms` Financial Reporting

Environments. Working paper, Rice University.

Roychowdhury, S. 2006. 'Earnings Management through Real Activities Manipulation.
Journal of Accounting and Economics 42 (December): 335-370.
Securities and Exchange Commission (SEC), 1999. Revenue Recognition in Financial
Statements. Staff Accounting Bulletin No. 101. Washington, D.C.: Government Printing
OIIice.
Stubben, S. 2007. 'Do Firms Use Discretionary Revenues to Meet Revenue and Earnings
Targets? Working paper, University oI North Carolina.
23
TABLE 1 Short-Term Deferred Revenue by Fama-French Industry Group

FF Industry N oI Firms Mean Median
Banking 52 1.37 0.89 0.01
Textiles 4 4.17 0.01 0.01
Utilities 71 5.01 1.28 0.75
Shipbuilding, Railroad Equip. 3 5.26 0.01 0.01
Precious Metals 18 5.39 1.57 0.49
Agriculture 6 6.45 8.46 10.90
AircraIt 8 6.84 13.66 2.15
Food Products 29 7.06 0.95 0.01
Steel Works, Etc. 28 7.25 1.81 0.44
Rubber and Plastic Products 20 7.30 3.93 0.89
Fabricated Products 7 8.05 6.55 2.41
Construction 26 8.36 2.09 0.45
Apparel 35 9.72 1.41 0.65
Alcoholic Beverages 11 9.73 0.65 0.01
Shipping Containers 9 10.11 7.04 0.08
Candy and Soda 8 10.13 1.28 0.09
Construction Materials 52 11.21 0.66 0.26
Miscellaneous 72 12.37 3.83 0.89
Automobiles and Trucks 59 13.02 5.47 0.69
Petroleum and Natural Gas 158 13.45 2.08 0.29
Consumer Goods 60 13.51 5.88 1.90
Business Supplies 41 13.71 0.87 0.71
Wholesale 148 14.04 4.08 1.43
Nonmetallic Mining 32 14.55 0.69 0.01
Trading 250 14.59 3.19 0.12
Chemicals 86 14.73 2.53 0.35
Healthcare 72 15.48 3.84 1.48
Real Estate 50 15.77 0.66 0.22
Recreational Products 38 15.97 5.57 2.47
Electrical Equipment 77 18.69 3.35 1.02
Transportation 153 19.52 2.97 0.79
Tobacco Products 9 20.00 0.01 0.01
Insurance 194 20.25 2.55 0.01
Machinery 200 22.10 4.74 1.53
Retail 310 23.27 2.31 1.30
Coal 11 23.40 0.42 0.42
Restaurants, Hotel, Motel 141 24.96 2.67 1.57
Entertainment 159 29.01 4.57 1.24
Medical Equipment 342 33.17 5.70 2.15
DeIense 16 33.33 2.69 1.22
Electronic Equipment 668 35.65 3.73 1.68
Personal Services 118 37.22 18.47 9.16
Pharmaceutical Products 783 39.81 4.87 1.38
Telecommunications 627 45.11 3.37 1.26
Measuring and Control Equip 285 45.38 43.54 1.43
Business Services 2320 51.14 13.16 6.38
Computers 715 55.56 7.93 3.91
Printing and Publishing 156 66.67 6.59 2.80
24
Note. This table reports the Fama-French industry name, total number oI non-missing
and non-zero observations Ior the ratio oI short-term deIerred revenue-to-total assets
(Compustat Annual Data Item 356 divided by Compustat Annual Data Item 6),
percentage oI Iirms in that industry with non-missing and non-zero short-term deIerred
revenue, as well as the mean and median oI the ratio oI short-term deIerred revenue-to-
total assets. I deIine industries consistent with Fama and French (1997). I multiply ratios
by 100 to convert to percentages Ior expositional purposes.







































25
TABLE 2
Model Parameters for Normal Change Models

AGross A/R
t
/ A
t-1
u
0
u
1
*(1 / A
t-1
)
1
*(AS
t
/ A
t-1
)
2
*(ACFO
t1
/ A
t-1
) c
t

ADeI Rev
t
/ A
t-1
u
0
u
1
*(1 / A
t-1
)
1
*(AS
t1
/ A
t-1
)
2
*(ACFO
t
/ A
t-1
) c
t


Independent Variables Expected Sign Dependent Variable:
AGross A/R
t
/ A
t-1

Dependent Variable:
ADeI Rev
t
/ A
t-1

Intercept ? -0.0027**
(-2.03)
0.0025
(1.20)

1 / A
t-1
? -0.0004
(-0.01)
0.0162
(0.39)

AS
t
/ A
t-1
0.0979***
(14.27)
N/A

AS
t1
/ A
t-1

N/A
0.0234***
(4.45)

ACFO
t
/ A
t-1
N/A 0.0365*
(1.94)

ACFO
t1
/ A
t-1
0.0573***
(2.92)
N/A

Adjusted R-square 36.6 29.2

***, **, and * denote statistical signiIicance at the 1, 5 and 10 two-tailed levels, respectively.

Note. This table provides parameter estimates Ior the normal change models oI gross accounts receivable
and short-term deIerred revenue. I require at least eight non-missing observations within an industry-year
Ior estimation. To be consistent with prior literature on earnings management (e.g., Burgstahler and
Dichev 1997), I exclude utilities and Iinancial Iirms (i.e., SIC codes between 4400 and 5000 and SIC codes
between 6000 and 6500). I also exclude any Iirms related to public administration (i.e., SIC codes oI 9000
or higher). I winsorize all variables that enter the models at the top and bottom 1 oI their respective
distributions. The coeIIicient estimates are based on means oI industry-years and t-statistics are based on
the standard error oI those means. The coeIIicient estimates Ior the abnormal change in gross accounts
receivable model is based on 46 industries and 130 industry-years over 2001-2003, and the coeIIicient
estimates Ior the abnormal change in deIerred revenue is based on 22 industries and 39 industry-years over
2001-2003. I also report the associated mean oI the adjusted R
2
s across these industry-years. The
dependent variables are AGross A/R
t
, deIined as the change in gross accounts receivable (change in
Compustat Annual Data Item 2 plus Compustat Annual Data Item 67), and ADeI Rev
t
, deIined as the
change in short-term deIerred revenue (change in Compustat Annual Data Item 356). The independent
variables include a constant term, an intercept scaled by lagged total assets, 1/A
t-1
(Compustat Annual Data
Item 6), change in sales Ior year t, AS
t
(change in Compustat Annual Data Item 12), change in sales in year
t1, AS
t1
, change in cash Ilow Irom operations during year t, ACFO
t
(change in Compustat Annual Data
Item 308), and change in cash Ilow Irom operations during year t1, ACFO
t1
.


26
TABLE 3
Descriptive Statistics


Panel A: Firms with Accounts Receivable





Panel B: Firms with Deferred Revenue



Note. This table provides descriptive statistics. All variables are scaled by lagged total
assets, except Ior the raw values oI gross accounts receivable and deIerred revenue, book-
to-market ratio, and log oI size. AGross A/R is deIined as the change in gross accounts
receivable (change in Compustat Annual Data Item 2 plus Compustat Annual Data Item
67). AbnormalAGross A/R is the abnormal change in gross accounts receivable deIined
using the model in the text. ADeIerred revenue is deIined as the change in short-term
deIerred revenue (change in Compustat Annual Data Item 356). AbnormalADeI Rev is
the abnormal change in short-term deIerred revenue deIined using the model developed
in the text. Log (MVE) is the natural logarithm oI a Iirm`s size using beginning oI the
year market value oI equity (Compustat Annual Data Item 25 Compustat Annual Data
Item 199). Book-to-market is the beginning oI the year book-to-market ratio
((Compustat Annual Data Item 60 Compustat Annual Data Item 74) / (Compustat
Annual Data Item 25 Compustat Annual Data Item 199)).






Mean Std. Dev. 25 Median 75
Gross A/R (in $ mil) 290.787 3121.742 3.351 17.693 84.603
AGross A/R
t
0.001 0.085 -0.027 0.000 0.023
AbnormalAGross A/R
t
0.000 0.063 -0.023 -0.001 0.021
Log (MVE)
t
4.650 2.522 2.874 4.665 6.394
Book-to-Market
t
0.609 1.945 0.244 0.548 1.055
Mean Std. Dev. 25 Median 75
DeIerred Revenue (in $ mil) 39.324 284.213 0.269 2.135 11.837
ADeI Rev
t
0.009 0.053 -0.003 0.001 0.014
AbnormalADeI Rev
t
0.000 0.048 -0.015 -0.002 0.008
log (MVE)
t
4.856 2.362 3.283 4.958 6.389
Book-to-Market
t
0.498 0.956 0.192 0.431 0.814
27
TABLE 4
Abnormal Changes in Gross Accounts Receivable to Avoid Missing Earnings
Benchmarks


AbnormalAGross A/R
t


0

1
*PRE-MANAGEDJUSTMISS
t

2
*PRE-
MANAGEDMEETJUSTBEAT
t
*CONTROLS

c
t








***, ** and * denote statistical signiIicance at the 1, 5 and 10 two-tailed levels,
respectively (except Ior F-tests which are based on one-tailed signiIicance levels).

Note. This table provides regression results Ior my Iirst hypothesis. The primary
independent variable is PRE-MANAGEDJUSTMISS corresponding to the range in
which a Iirm just misses an earnings benchmark using a distribution based on pre-
managed earnings. PRE-MANAGEDMEETJUSTBEAT

is included as a natural
reIerence group, in which I conduct an F-test between this coeIIicient and that oI the
primary variable. The control variables are SIZE, deIined as the natural logarithm oI a
Iirm`s size using beginning oI the year market value oI equity and BM, deIined as the
beginning oI the year book-to-market ratio. T-statistics are reported in parentheses under
the coeIIicient estimates based on the Newey-West standard error correction Ior
autocorrelation and heteroskedasticity (Newey and West 1987). CoeIIicient estimates are
reported in decimal Iorm. For ease oI exposition, I exclude coeIIicient estimates Ior the
intercept and control variables.











Avoid Loss Avoid Earnings
Decrease
Avoid Negative
Earnings Surprise

1
0.0075***
(2.69)
0.0015
(0.65)
0.0017*
(1.65)

2
0.0064*
(1.89)
0.0021
(0.87)
-0.0055***
(-5.62)

1

2
0.06 N/A 76.01***
28
TABLE 5
Abnormal Changes in Deferred Revenue to Avoid Missing Earnings Benchmarks

AbnormalADeI Rev
t


0

1
*PRE-MANAGEDJUSTMISS
t

2
*PRE-
MANAGEDMEETJUSTBEAT
t
*CONTROLS

c
t





***, ** and * denote statistical signiIicance at the 1, 5 and 10 two-tailed levels,
respectively (except Ior F-tests which are based on one-tailed signiIicance levels).

Note. This table provides regression results Ior my second hypothesis. The primary
independent variable is PRE-MANAGEDJUSTMISS corresponding to the range in
which a Iirm just misses an earnings benchmark using a distribution based on pre-
managed earnings. PRE-MANAGEDMEETJUSTBEAT

is included as a natural
reIerence group, in which I conduct an F-test between this coeIIicient and that oI the
primary variable. The control variables are SIZE, deIined as the natural logarithm oI a
Iirm`s size using beginning oI the year market value oI equity and BM, deIined as the
beginning oI the year book-to-market ratio. T-statistics are reported in parentheses under
the coeIIicient estimates based on the Newey-West standard error correction Ior
autocorrelation and heteroskedasticity (Newey and West 1987). CoeIIicient estimates are
reported in decimal Iorm. For ease oI exposition, I exclude coeIIicient estimates Ior the
intercept and control variables.










Avoid Loss Avoid Earnings
Decrease
Avoid Negative
Earnings Surprise

1
-0.0023
(-0.59)
-0.0083
(-1.54)
-0.0061***
(-4.14)

2
-0.0089*
(-1.66)
0.0010
(0.37)
-0.0019
(-1.13)

1

2
N/A N/A 10.68***
29
TABLE 6
Abnormal Changes in Gross Accounts Receivable (Deferred Revenue) to Avoid
Negative Earnings Surprises Using a Common Sample

AbnormalAGross A/R
t
(AbnormalADeI Rev
t

)
0

1
*PRE-MANAGEDJUSTMISS
t


2
*PRE-MANAGEDMEETJUSTBEAT
t
*CONTROLS c
t



***, ** and * denote statistical signiIicance at the 1, 5 and 10 two-tailed levels,
respectively (except Ior F-tests which are based on one-tailed signiIicance levels).

Note. This table provides regression results Ior my third hypothesis using a common
sample with accounts receivable and short-term deIerred revenue. The primary
independent variable is PRE-MANAGEDJUSTMISS corresponding to the range in
which a Iirm just misses analysts` Iorecasts using a distribution based on pre-managed
earnings. PRE-MANAGEDMEETJUSTBEAT is included as a natural reIerence group,
in which I conduct an F-test between this coeIIicient and that oI the primary variable.
The control variables are SIZE, deIined as the natural logarithm oI a Iirm`s size using
beginning oI the year market value oI equity and BM, deIined as the beginning oI the
year book-to-market ratio. I include the abnormal change oI the other account as an
additional control. T-statistics are reported in parentheses under the coeIIicient estimates
based on the Newey-West standard error correction Ior autocorrelation and
heteroskedasticity (Newey and West 1987). CoeIIicient estimates are reported in decimal
Iorm. For ease oI exposition, I exclude coeIIicient estimates Ior the intercept and control
variables.






.



AbnormalAGross A/R AbnormalADeI Rev

1
0.0016
(0.71)
-0.0058***
(-3.14)

2
-0.0021
(-0.96)
-0.0011
(-0.50)



F-test:

1

2
N/A 7.75***