BUSINESS CYCLES AND EARNINGS MANAGEMENT

By
Richard Kelly Rebagliati
B Econ., University o f Northern British Columbia. 2004

THESIS SUBMITTED IN PARTIAL FULFILLMENT OF
THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE
IN
BUSINESS ADMINISTRATION

UNIVERSITY OF NORTHERN BRITISH COLUMBIA
OCTOBER 2012

©Richard Kelly Rebagliati, 2012

1+1

Library and Archives
Canada

Bibliotheque et
Archives Canada

Published Heritage
Branch

Direction du
Patrimoine de I'edition

395 Wellington Street
Ottawa ON K1A0N4
Canada

395, rue Wellington
Ottawa ON K1A 0N4
Canada
Your file Votre reference
ISBN:

978-0-494-94102-7

Our file Notre reference
ISBN:

978-0-494-94102-7

NOTICE:

AVIS:

The author has granted a non­
exclusive license allowing Library and
Archives Canada to reproduce,
publish, archive, preserve, conserve,
communicate to the public by
telecommunication or on the Internet,
loan, distrbute and sell theses
worldwide, for commercial or non­
commercial purposes, in microform,
paper, electronic and/or any other
formats.

L'auteur a accorde une licence non exclusive
permettant a la Bibliotheque et Archives
Canada de reproduire, publier, archiver,
sauvegarder, conserver, transmettre au public
par telecommunication ou par I'lnternet, preter,
distribuer et vendre des theses partout dans le
monde, a des fins commerciales ou autres, sur
support microforme, papier, electronique et/ou
autres formats.

The author retains copyright
ownership and moral rights in this
thesis. Neither the thesis nor
substantial extracts from it may be
printed or otherwise reproduced
without the author's permission.

L'auteur conserve la propriete du droit d'auteur
et des droits moraux qui protege cette these. Ni
la these ni des extraits substantiels de celle-ci
ne doivent etre imprimes ou autrement
reproduits sans son autorisation.

In compliance with the Canadian
Privacy Act some supporting forms
may have been removed from this
thesis.

Conform em ent a la loi canadienne sur la
protection de la vie privee, quelques
formulaires secondaires ont ete enleves de
cette these.

W hile these forms may be included
in the document page count, their
removal does not represent any loss
of content from the thesis.

Bien que ces formulaires aient inclus dans
la pagination, il n'y aura aucun contenu
manquant.

Canada

APPROVAL

ABSTRACT

This study investigates the impact of business cycles on earnings
management in the United States. Using a large cohort of firms in the United
States from the S&P 1500 index and the period of 2000-2010, we employ
estimates based on a pooled least squares model, a fixed effects model, and a
random effects model. Our findings show that firm discretionary accruals increase
during expansionary economic periods and decrease during contractionary
periods. We also find that the Sarbanes-Oxley Act has had no effect on mitigating
discretionary accruals. Our primary contribution to the existing literature is a
thorough econometric analysis of discretionary accruals and their relationship to
economic cycles and the Sarbanes-Oxley Act using a large and comprehensive
data set.

TABLE OF CONTENTS
Approval

ii

Abstract

iii

Table of Contents

iv

List of Tables

vi

Acknowledgement

vii

Chapter I

Introduction

I

Chapter II

Review of Literature and Hypothesis
Development

6

Section 2.1
Section 2.2
Section 2.3
Section 2.4

Chapter III
Section 3.1
Section 3.2

Chapter IV
Section 4.1
Section 4.2
Section 4.3
Section 4.4

Chapter V

Definitions of Earnings Management
Accrual Determination Models
Empirical Studies in Earnings Management and
Hypothesis Development
Earnings Management, Reversals, and Recent
Developments

6
8

Database and Methodology

36

Database
Methodology

36
39

Empirical Results

41

Descriptive Statistics
Correlation Matrix
Empirical Results
Summary of Results

41
42
44
48

Conclusions

50

Bibliography

17
32

52

iv

V

LIST OF TABLES

Table 2.1

Discretionary Accrual Proxies

16

Table 3.1

Description of Variables Used in the Study

38

Table 4.1

Descriptive Statistics of Financial Parameters o f US
Non-Financial Corporate Sector Firms 2000-2010
(Balanced)

42

Table 4.2

Correlation Matrix of Variables, Balanced Panel

43

Table 4.3

Determinants of Discretionary Accruals in the
United States, 2000-2010

47

vi

ACKNOWLEDGEMENT

Thanks and love to my father, who guided my interest to the field of business
when I had no idea what I wanted to do, and my mother who has given me
unwavering support and confidence my entire life.

CHAPTER I
INTRODUCTION
Earnings management refers to the willful attempts by managers o f a firm
to manipulate their earnings to meet pre-determined targets. ‘Earnings’ refers in
its simplest form to the profits of a company, and earnings management refers to
the practice of cooking books and creating juiced-up accounts. Investors and
analysts look to earnings to determine the attractiveness of a particular stock. The
management of profits is sometimes used interchangeably with income
smoothing1. Motivations for this management of earnings vary, but common
reasons are (a) to meet targets of profitability set by analysts or the market, (b) to
convey information about future earnings, (c) to signal the market as a low risk
firm, and (d) to appropriate executive compensations, like bonuses, stock options
etc.
Earnings management has become a topic of increased interest for
financial regulators. The numerous recent accounting scandals (like Enron,
WorldCom, Parmlat, Waste Management, Tyco, Satyam, Olympus, etc.) has cast
doubts about truthfulness o f the financial statements o f firms and eroded investor
confidence and adversely affected market sentiments. Industry regulators,
auditors, analysts and company stakeholders (whether individual or institutional
investors) all have a substantial interest in transparent and accurate earnings

1 Income smoothing is a form o f earnings management and is generally defined as the smoothing
of reported earnings over time (Ronen and Yaari, 2008, p. 317)

1

information and can greatly benefit from greater accuracy in financial reporting
by firms.
Healy and Wahlen (1999) state ‘...earnings management occurs when
managers use judgment in financial reporting and in structuring transactions to
alter financial reports to either mislead some stakeholders about the underlying
economic performance o f the company or to influence contractual outcomes that
depend on reported accounting numbers’. Most o f the empirical literature on
earnings management has centered on how firms keep two sets of accounts (one
internally) with one to be publicly issued, whether for required reporting or for an
initial public offer (Dechow et al., 1995; Healy and Wahlen 1999; Graham et al.,
2005; Ball and Shivakumar, 2008).
Prior literature has classified earnings management into two broad
categories: real earnings management (i.e. affecting cash flows) and accruals
management (through changes in accounting policies and calculations).
Roychowdhury (2006) states that operational decisions such as acceleration o f
sales, alterations in shipment schedules, delaying of research and development,
and delaying o f maintenance expenditures are all real earnings management
methods available to managers. The amount of managed earnings is the difference
between reported earnings and true earnings. The most common way of detecting
accrual-based earnings management is through company financial statements.
Accounting adjustments known as accruals are the difference between reported
earnings and operating cash flows. Accruals consist of a discretionary portion
which is often manipulated by managers and a non-discretionary portion which is
2

dictated by business conditions. There exists considerable difficulty to accurately
separate reported accruals into their managed (discretionary) and unmanaged
(non-discretionary) components. Researchers use empirical models to decompose
total accruals into non-discretionary and discretionary accruals. Discretionary
accruals are then used as proxy for earnings management. The most widely used
discretionary accrual models are the Jones and ‘modified’ Jones models which
used the variables o f ‘firm revenues’, ‘gross property, plant, and equipment’, and
‘total assets’, to break down the total accruals values into non-discretionary and
discretionary components.
Earnings management is broadly classified into three types: white, gray,
and black (Ronen and Yaari, 2008). White refers to earnings management where
reports are made more transparent to emphasize private information about fixture
cash flows. Gray earnings management refers to choosing a particular accounting
treatment that is opportunistic or economically efficient. Black earnings
management refers to willful tricks and misrepresentation, or purposefully
decreasing the transparency of financial reports (Ronen and Yaari, 2008).
An understanding of earnings management practices helps public
authorities (like governments and regulators) improve functioning o f financial
markets, reduce asymmetry of information, reduce cost of capital, protect small
and minority shareholder’s interests, promote financial stability, and lead to
efficient allocation of capital. A regulation like the Sarbanes-Oxley Act in the
United States in 2002 is a classic example of an effort to promote more accurate
financial reporting, standards, and accountability to company issued financial

3

statements (Cohen et al., 2007). These interventions allow auditors to have a more
consistent and precise framework for evaluating the financial statements of firms.
In turn, both financial analysts and shareholders benefit from not only more
accurate financial information, but more consistent financial reporting by firms.
This applies across industries, allowing the best possible conclusions to be drawn.
Much research has been done in the past in attempts to determine levels of
earnings management that firms indulge in by studying their financial statements.
Various models have been developed to detect earnings management by studying
different models of accruals. Most models in the area of earnings management
relate to its prevalence over time or at the time o f IPO issue (Teoh, Welch, and
Wong 1998; Teoh, Wong, & Rao 1998), seasoned equity offerings (Rangan,
1998), and mergers and acquisitions (Erickson and Wang, 1999). Very little
attempt has been made to examine whether earnings management has diminished
after the introduction of regulations like the Sarbanes-Oxley Act2 and whether
discretionary accmals vary over the course of business cycles3. The present study
contributes to the literature by examining the determinants o f accruals in the US
corporate sector for the period 1980-2010 by examining its behavior over
different phases o f business cycles using a large cohort of firms (1125 firms)
which could provide robust results.

2 Studies of interest in firm earnings management behavioral changes before and after the
introduction of the Sarbanes-Oxley Act include Graham et al., 2005; Cohen, Dey, & Lys 2007;
and Cohen & Zarowin 2010.
3 Although much more limited, the best work on the relationship of accruals and business cycles
was examined by Teoh, Welch, & Wong 1998; Teoh, Wong, & Rao 1998; Hirshleifer et al., 2009;
and Kang et al., 2010. Additionally, important work on stock prices and economic cycles has been
done by Braun & Larrain 2005; Wei 2009; Covas & Den Haan 2011; and Naes et al., 2011.

4

This study is organized into five chapters: Chapter II reviews the literature
on the subject and sets the hypotheses for empirical investigation. Chapter III
discusses the database and methodology o f the study. Chapter IV presents the
empirical results, and Chapter V summarizes the concluding observations.

5

CHAPTER II
REVIEW OF LITERATURE AND
HYPOTHESIS DEVELOPMENT

This chapter briefly reviews the literature on the subject of earnings
management. This chapter is organized as follows: Section 2.1 briefly reviews the
definition of and literature on earnings management, Section 2.2 discusses the
different discretionary accruals models, and Section 2.3 summarizes the results of
the empirical studies related to accrual-based earnings management, and develops
hypothesis for empirical investigation.

2.1 Definition o f Earnings Management
In addition to white, gray, and black definitions of earnings management
discussed in Chapter 1, earnings management can further be classified into two
forms: (a) real earnings management and (b) accrual-based earnings management.
Real earnings management refers to ‘changes in the timing or structure of an
operating, investing, or financial activity to affect earnings’ (Edelstein et al.,
2009). From a practical point of view, this can involve changes in the timing of
product shipments, strategically timed pricing discounts, or sales of long-term
assets. All of these actions represent ‘real’ alterations in company operations with
the motivation and objective of altering a company’s quarterly or annual financial
data. A commonly studied form of real earnings management is the opportunistic
6

reduction in R&D expenditure to reduce reported expenses (Rowchowdhury,
2006, p.338). In addition, there is anecdotal evidence of managers engaging in
providing limited time discounts to increase sales and building up excess
inventory to lower reported cost of goods sold (ibid., p.338). Additionally, Bens et
al., (2002, 2003) report that managers repurchase stock to avoid eamings-pershare dilution arising from exercising employee stock options or stock option
grants.
Accrual-based earnings management is a more subtle and sophisticated
method of accomplishing the same task. Through a company’s accrual accounts
(accounts receivable, accounts payable, provisioning, etc.) management has the
ability to manipulate their earnings to meet pre-determined targets. Accruals as
defined in accounting are accounts on a balance sheet representing liabilities or
non-cash assets. Because of leeway provided by accounting standards and
practices, management has the ability to increase or decrease income by creating
these accruals (Li et al., 2009). Discretionary accruals can be considered changes
in the value of accruals that are based on inventory write down, alterations o f debt
valuations, provisioning, etc. Because values in these categories have a certain
level of subjectivity, management has the ability to alter these numbers to achieve
pre-determined goals. Isolating the discretionary and non-discretionary portions
of an accrual account is the most important factor in developing a good earnings
management detection model.

7

2.2 Accrual Determination Models
Many models have been developed by researchers for the estimation of
non-discretionary

and

discretionary

accruals

components

from

financial

statements of firms. The difficulty in isolating the non-discretionary and
discretionary portions from total accruals by investigators (auditors, analysts,
investors, and researchers) makes it an ideal mechanism for firms looking to
engage in earnings management. One of the earliest discretionary accrual models
is the Healy model (1985) which is discussed below:
(a) The Healy Model (1985)
The earliest discretionary accrual model was developed by Healy (1985).
In this model, earnings management could be detected by looking at the
deviations in the accruals from the normal (mean) level of past accruals:

A r, n

ALKt -

(ACAt -& CLt - b C a s h t + b S T D t-D e p t )
-

Where:
ACRt = total working capital accruals (total accruals).
ACAt = change in current assets.
ACLt = change in current liabilities.
ACasht = change in cash and cash equivalents.
ASTDt = change in debt included in current liabilities.
Dept = depreciation and amortization expense.
At-i = assets in the previous period.

8

Non-discretionary accruals are given as:

NDAt =

(2.2)

Where:
NDAt = non-discretionary accruals
TAt = total accruals scaled by lagged total assets
t = subscript for year included in the estimation period
T= subscript indicating a year in the event period
T= a year subscript for years included in the estimation period
The result of TA - NDA then gives the value for discretionary accruals.

(b) The DeAngelo Model (1986)
The subsequent model by DeAngelo (1986) assumed that first order
differences in accruals have an expected value of zero.
Therefore:
NDAt = TAt_t

(2.3)

However, it is unlikely that accruals are constant over time, or dependent
on the previous year in such a one dimensional way.
(c) The Industry Model (1991)
The industry model (Dechow and Sloan, 1991) is a further refined attempt
to isolate discretionary accruals. Instead o f directly isolating non-discretionary
accruals to obtain discretionary accruals, the model assumes that ‘variation in the

9

determinants of non-discretionary accruals is common across firms in the same
industry’ (ibid., 1991). The Industry model for nondiscretionary accruals is:
NDAt = Yi + Yzm ediariiiTAf)

(2.4)

Where:
Yi and y 2

=

median] (TAt) =

firms 1 and 2
the median value of total accruals scaled by lagged assets
for all non-sample firms in the same 2-digit SIC code

(d) The Jones Model (1991)

The model by Jones (1991) used the variables of ‘firm revenues’, ‘gross
property, plant, and equipment’, and ‘total assets’, to break down the total
accruals values into non-discretionary and discretionary components. The original
Jones model is given as:

e-‘(£)+*®+*‘(£r)+'»'
Where:
TAi,= total accruals in year t for firm i

AREVit = revenues in year t minus revenues in year t-1 for firm i
PPEjt = gross property, plant, and equipment in year t for firm i
Ait-i = total assets in year t-1 for firm i
£jt = error term in year t for firm i
10

( 2 -5 >

Non-discretionary accruals are calculated as:

NDAt = a x

) + a 2(AREVt) + a 3(PPEt)

(2.6)

The result of TA - NDA then gives the value for discretionary accruals.
Although this model did give some predictability, it has subsequently been
improved on and modified, most notably by Dechow et al., (1995) and Kothari et
al., (2005).
(e) The ‘Modified’Jones Model (1995)
A major limitation of the Jones model lies in its inability to capture the
impact o f sales-based manipulation since changes in sales are assumed to give rise
to non-discretionary accruals. Dechow, Sloan, and Sweeney (1995) proposed a
modification to the standard Jones model. The ‘modified’ Jones model is identical
to the standard Jones model, with the exception that the changes in ‘debtors’
(AREC) is subtracted from AREV at the second stage. In effect, the ‘modified’
Jones model assumes that all changes in credit sales in the event period result
from earnings management. Dechow et al., use this ‘modified’ Jones model to
detect earnings management among firms and to test the results o f this model in
comparison to results from the DeAngelo, Healy, Jones, and Industry models o f
discretionary accrual calculation. Their ‘modified’ Jones model is designed to
‘eliminate conjectured tendency of the Jones model to measure discretionary
accruals with error when discretion is exercised over revenues’ (Dechow et al.,
11

1995). The formula for non-discretionary accruals in the ‘modified’ Jones model
is as follows:

NDAt = a x

+ a 2(AREVt - ARECt) + <x3(PPEt)

(2.7)

Where:
NDA, = non-discretionary accruals in year t
At_i = total assets in year t-1
AREVt = net revenues in year t less net revenues in year t-1 scaled by total assets
at t-1
ARECt = net receivables in year t less net receivables in year t-1 scaled by total
assets at t-1
PPEt = gross property, plant, and equipment in year t

The above formula (equation 2.7) ‘implicitly assumes that all changes in
credit sales in the event period result from earnings management’ (Dechow et al.,
1995). As in the earlier versions non-discretionary accruals are subtracted from
the total accruals value to derive the discretionary accruals value. This formula
has become the most widely used in empirical literature and gives the best
predictability in discretionary accruals based earnings management detection
(Kothari et al., 2005).
(j) The Kothari ‘Modified’Jones Model (2005)
A significant contribution to the work on derivation of discretionary
accruals was done by Kothari, Leone, and Wasley (2005). Kothari et al., (2005)
12

created a ‘performance-matched’ discretionary accruals formula which involved
derivation of a control sample of firms that are assumed to have a ‘mean’ level o f
earnings management. Against this benchmark, individual firms were compared
to derive ‘abnormal’ earnings management, with managing earnings at a rate
higher or lower than the control sample (Kothari et al., 2005, p. 165). The authors
find that the ‘modified’ Jones model continues to be the one with the greatest
ability to detect earnings management, but they augment the existing ‘modified’
Jones model with the addition of the variables of current return on assets (ROAt)
and past return on assets (ROAt-i)- The rationale for this addition to the equation
is that ‘earnings deflated by assets equals return on assets, which measures
performance, and prior research analyzing long-run abnormal stock return
performance and abnormal operating performance finds matching on ROA is
better specified and more powerfully tests compared to other matching variables’
(Kothari et al., 2005, p. 169). The authors also found that for their regressions the
firm ‘matching’ technique used by Teoh et al.,(1998) provides a better result
compared with the industry matching technique. Kothari et al. argue believe that
firms of similar size and industry are likely to have similar non-discretionary
accrual amounts, and argue that discretionary accruals for any firm arise from
three causes:
1)

Accruals related to the ‘treatment’ event (eg. A seasoned equity offering).

2)

Accruals related from incentives (eg. Employee bonuses, meeting
analyst’s expectations).

3)

Accruals correlated with performance.
13

Kothari et al., argue that the firm being tested and the control (matching)
firm are likely to have similar values for the second and third causes, as they are
of similar size and industry. Therefore, Kothari et al., are able to isolate the
earnings management that is directly correlated to the ‘treatment’ event (Kothari
et al., 2005, p. 171). The authors create a new version of the ‘modified’ Jones
equation using the ROA for periods t and t-1:

A L E S ^ + a 3(PPElt) + u ^ R O A ^ o r it-^)

(2 .8)

Where:
TAit = total accruals in year t for firm i
Ajt-i = total assets in year t-1 for firm i
ASALESit = change in sales in year t for firm i
PPEit = gross property, plant, and equipment in year t for firm i
ROAjt(orit-i) = return on assets in time t or t-1 for firm i
Vit = the residual

Kothari et al., (2005) found in their empirical investigationsthat

the

‘matching’ technique yielded better results, and that their alteration o f the
‘modified’ Jones model yielded lower chance of misspecification than the original
‘modified’ Jones model. The design of the ‘modified’ Jones model, assumed that
all credit sales represent accrual manipulation, although this is unlikely to be the
case all the time (Kothari et al., 2005, p. 186). The authors conclude that the Jones
14

and ‘modified’ Jones models suffer from ‘severe misspecification in stratified
random samples’ (Kothari et al., 2005, p. 186). When there are negative
discretionary accruals, there is an over-rejection of the null hypothesis, and when
there are positive discretionary accruals, there is an under-rejection of the null
hypothesis. They admit that their modification of the formula does have its own
misspecification problems, but that it is a viable alternative to the existing reliance
on the ‘modified’ Jones formula for earnings management detection.
In summary, the accrual-based earnings management models discussed
above have relied on a number of firm-specific variables in the attempt to
estimate accurate accruals and non-discretionary accruals. Table 2.1 presents a
summary o f the various accrual models used in the earnings management
literature. In a survey of five commonly used models of discretionary accruals,
Dechow, Sloan, and Sweeney (1995) conclude that the ‘modified’ Jones model
works best. Moreover, the results of all the five models are fairly similar.
Therefore, in our empirical investigation (presented in Chapter IV), we use the
‘modified’ Jones model (equation 2.7) for estimating discretionary accruals.

15

Table 2.1 Discretionary Accrual Proxies
Panel A: Aggregate accrual proxy
Authors
Healy (1985)

Discretionary accrual proxy
Total accruals

DeAngelo (1986)

Change in total accruals

Jones(1991)

Residual fiom regression o f total accruals on change
in sales and property, plant and equipment

Modified Jones Model from Dechow et al., (1995)

Residual from regression o f total accruals on change
in sales and property, plant and equipment, where
revenue is adjusted for change in receivables in the
event period

Kang and Sivaramakrishnan (1995)
Residual from a regression o f noncash current assets
less liabilities on lagged levels o f these balances,
adjusted for increases in revenues, expenses and
plant and equipment
Panel B: Specific accrual models
Authors
McNichols and Wilson (1988)

Discretionary accrual proxy
Residual provision for bad debt, estimated as the
residual from a regression o f the provision for bad
debts on the allowance beginning balance, and
current and future write-offs

Petroni (1992)

Claim loss reserve estimation error, measured as the
five year development of loss reserves o f property
casualty insurers

Beaver and Engel (1996)

Residual allowance for loan losses, estimated as the
residual from a regression o f the allowance for loan
losses on net charge-offs, loan outstanding,
nonperforming assets and one year ahead change in
nonperforming assets

Beneish (1997)
Days in receivables index, gross margin index, asset
quality index, depreciation index, selling general and
administrative expense index, total accruals to total
assets index

Beaver and McNichols (1998)

Serial correlation of one year development o f loss
reserves o f property casualty insurers
Panel C: Frequency distribution approach
Authors
Burgstahler and Dichev (1997)

Test for earnings management
Test whether the frequency o f annual earnings
realizations in the region above (below) zero earnings
and last year’s earnings is greater (less) than
expected

Degeorge et al. (1999)
Test whether the frequency o f quarterly earnings
realizations in the region above (below) zero
earnings, last quarter’s earnings and analysts’
forecasts is greater (less) than expected
Myers and Skinner (1999)
Test whether the number o f consecutive earnings
increase is greater than expected absent earnings
management

Source: McNichols (2000), p. 317.

16

2.3 Empirical Studies o f Earnings Management

The creation and refinement of the above mentioned models to isolate the
discretionary portion of firm total accruals data has led to a substantial amount of
empirical research. Application o f these models to different aspects o f finance has
been researched with the objective of identifying firms who actively engage in
accruals based earnings management. The empirical literature can be broadly
classified as (a) earnings management and stock price, (b) earnings management
and seasoned equity offerings, (c) earnings management and initial public
offerings, (d) earnings management and cumulative abnormal returns, (e) earnings
management and mergers, and (f) earnings management and banks.
(a) Earnings Management and Stock Price
As the ‘modified’ Jones formula for discretionary accruals determination
became widespread, K.R. Subramanyam’s (1996) study attempted published a
paper to determine if discretionary accrual values of firms impacted firm’s stock
prices. Using the ‘modified’ Jones formula, Subramanyam’s study of US firms
using the CRSP database o f 2808 firms over a 20 firm-years period found that
discretionary accruals were an important variable in future predictability of stock
prices across many industries, and the inclusion of discretionary accruals
‘improved the ability

of firm

income to

explain future

profitability’

(Subramanyam, 1996, p.273). With regard to earnings management, the author
also found that there was ‘evidence of pervasive income smoothing that improves
the persistence and predictability of earnings’ (Subramanyam, 1996, p.273).
17

Subramanyam’s (1996) paper was one of the first to show that earnings
management had an active role in the modification of discretionary accrual values
for the purposes of income smoothing.
(b) Earnings Management and Seasoned Equity Offerings
A study by Srinivasan Rangan (1998) tested the discretionary accrual
values of firms (using the ‘modified’ Jones model) during the period surrounding
a seasoned equity offering (non-IPO). Specifically, discretionary accrual values
were tested in the years t-1, t, and t+1 surrounding an equity offering by an
established firm. Using quarterly data for the US, Rangan (1998) found that there
was reliable predictability in the stock price o f firms. Using data for 230
established firms over a four year period (1987-1990), Rangan (1998) found that
‘a one-standard-deviation increase in earnings management (discretionary
accruals) during the year before the equity offering resulted with a decline in
market-adjusted returns in the year following the seasoned equity offering of
about 10%’ (Rangan, 1998, p. 102). Rangan (1998) also found that earnings
management was most common in the quarter in which the equity offering was
announced and the quarter following the announcement.
(c) Earnings Management and IPO ’s (Initial Public Offerings)
Empirical studies on earnings management have generally found it to be
more prevalent in the event of initial public offerings (IPO’s). Altering the
financials of a firm and therefore manipulating the expectations and consensus of
stakeholders and analysts in the lead up to an IPO can have significant positive
18

effects on the stock price at the time of offering. There has been substantial
research showing the existence of earnings management (both accrual-based and
real) in the lead-up to IPO’s (Teoh, Welch, & Wong 1998; Teoh, Wong, & Rao
1999; DuCharme et al., 2001).
Teoh, Welch, and Wong (1998) further discuss evidence o f discretionary
accruals management and income smoothing for companies during IPO’s. The
authors found that companies who had high earnings management (highest
quartile) at the time of the IPO, had in the third year after the IPO (t+3) an
average stock price of 20% less than companies in the lowest quartile of earnings
management. Companies with high earnings management at the time of the IPO
also had 20% less seasoned equity offerings in the 5-year period following the
IPO (Teoh, Welch, & Wong 1998, p. 1935). In their study they used data for
1,974 IPO’s from the 1980-1984 periods, and a further 3,197 IPO’s for the 19851992 periods, using data from the CRSP database (ibid., p. 1942). With
alternative measures of abnormal returns, benchmarks, cumulation periods,
sample partitions, and regression test specifications (cross-sectional, time-series,
Fama-MacBeth) the authors found that ‘discretionary current accruals reliably
predict post-IPO returns’ (ibid., p. 1949).
A subsequent paper published by Teoh, Wong, and Rao (1998) compared
‘abnormal’ accruals of firms in the year o f their IPO. ‘Abnormal’ accruals were
calculated as accruals above the benchmark for firms of similar industry and size.
Because there is an incentive for firms to seek a boost in earnings before their
initial public offering, the authors found that financial statements of the firms
19

showed ‘unusually high positive accruals’ (Teoh, Wong, & Rao 1998, p. 176).
The authors also found that ‘firms with unusually high accruals in the IPO year
consistently had low post-IPO earnings, and that high IPO accrual levels predicted
low post-EPO earnings compared to industry benchmarks’ (ibid., p. 176). In their
sample, the authors used 1,682 IPO’s between 1980 and 1990. For companies to
qualify, they had to have an IPO stock price o f <$1.00, gross proceeds of
>$1,000,000, only common stock offered, and the offering handled by an
investment bank. Data was gathered from the Compustat database. In the
calculation of abnormal accruals, the authors compared the firm accruals across
industry benchmarks, and used the ‘modified’ Jones method of accruals
calculation. One interesting technique that the authors used in their paper was an
alternative system of capturing abnormal accruals. Because IPO firms are likely to
have extreme performance compared to the overall industry, many of the IPO
firm financial values will be outliers compared to the industry. To properly
capture abnormal accruals o f EPO firms, the authors matched each IPO firm with a
firm in the same industry and of the same size, but which was not having an IPO.
The authors state that this ‘matching’ technique is beneficial as ‘systematic errors
in the Jones model abnormal accruals for similar performing firms are eliminated’
(ibid., p. 183). The authors do note, however, that accruals information can be
underestimated, as there still may be motivation for the non-IPO-issuing matching
firm to engage in earnings management themselves for reasons of their own. In
their results, the authors found that there were inferior returns for IPO firms in the
years following the IPO. Compared both to industry benchmarks and the

20

‘matching firm’, the IPO firms underperformed in years t through t+6 of the IPO.
It is of interest that the underperformance was worse for the IPO firms when using
the ‘matching firm’ technique. In their summary, the authors claim that ‘abnormal
current accruals has the greatest consistent explanatory power among all the
proxies, perhaps because it is the component most easily subject to successful
managerial manipulation’ (ibid., p. 195). In general, evidence suggests that firms
in the lead up to their IPO have significant negative abnormal cash flows and
manipulate accmals to inflate reported earnings (Bao et al., 2012). In addition, it
has been shown that decisions to manipulate earnings in the lead-up to an IPO are
positively related to IPO proceeds, and negatively related to analyst reputation
ranking (Bao et al., 2012).
(d) Earnings Management and CAR’s (Cumulative Abnormal Returnsj
In a recent paper Hirshleifer, Hou, and Teoh (2009) examined whether
accmals contained information about the discount rate, or whether firms managed
earnings in response to market under or overvaluation. The authors used the
CRSP value-weighted market index for their data, over the period 1965-2005.
Similar to Kang et al., (2010), they found that firms with high accmals but low
cash flows were consistently overvalued, and suffered from low future cumulative
abnormal returns (CAR’s). Similarly, they found that firms with low accrual
levels but high cash flows were consistently undervalued by the markets, and
enjoyed high future CAR’s. They felt that the cash flows at the firm level should
be dissected into cash and accmal components to give the best picture of the
actual firm’s status (Hirshleifer et al., 2009, p. 390). The authors also found that at
21

the aggregate level, a one-standard-deviation increase in accruals in time t led to a
7% increase in the stock price in time t+1. They also found that high aggregate
cash flow levels negatively affected stock prices in the aggregate. Like Kang et
al., (2010), the authors found that the ‘lean against the wind’ hypothesis was also
a valid explanation of their findings. If firms become undervalued, they will be
especially eager to report higher earnings by increasing accruals relative to their
cash flows (Hirshleifer et al., 2009, p. 405). However, the authors note that some
explanation must be made as to why firms are prone to this ‘leaning’ effect more
often during aggregate (industry or market) undervaluation rather than simply
firm-specific undervaluation.
Related to Hirshleifer et al., (2009), one of the most telling papers related
to accruals-based earnings management detection is ‘Predicting stock market
returns with aggregate discretionary accruals’ by Kang, Liu, and Qi (2010).
Published shortly after Hirshleifer et al., (2009), the authors make more direct
conclusions than the Hirshleifer et al., paper. They find that on the aggregate,
discretionary accruals contain little information about overall firm business
conditions compared to normal non-discretionary accruals, but ‘aggregate
(industry or market) accrual levels reflect aggregate fluctuations in earnings
management, thereby favoring the behavioral explanation that managers time
aggregate equity markets to report earnings’ (Kang et al., 2010, p. 815). The
authors begin with the premise that a change in accmals in the aggregate
represents either a change in the discount rate, or the fact that firms are managing
earnings in response to market undervaluation. They found that aggregate
22

accruals can positively predict aggregate stock returns. They also found using the
‘modified’ Jones formula that the forecasting power was entirely driven by
discretionary accruals (as opposed to total or non-discretionary accruals). Nondiscretionary accruals provided no predictive power whatsoever,

while

discretionary accruals provided very robust results. The authors did add that there
is a misspecification problem that exists as the ‘modified’ Jones accrual formula
fails to take into account business cycles. They also noted that non-discretionary
accrual levels correlated with the rate of GDP growth. Additionally, discretionary
accruals tended to have no correlation with any other macroeconomic variables.
The authors were able to completely rule out the argument that discretionary
accrual amounts were based on changes in the discount rate. They limited the
causes of changes in discretionary accrual amounts to be based on manager’s
decisions to ‘lean against the wind’ in the form o f managing earnings based on
market timing. Managers also responded to decreases in equity market firm
valuations by the adjusting up of current period accruals, and vice versa during
times of increased firm valuations.
Kang et al., used three different regressions in deriving their results. The
first was the standard ‘modified’ Jones model with all firms included. The second
was the same, but with the deletion of firms experiencing ‘extreme events’. In the
third regression they used the Kothari version of the ‘modified’ Jones formula.
They used the CRSP database to obtain data for 2,450 US firms over the period
1965-2004. Any firm with less than ten data points was omitted.

23

Interestingly, the authors speculated that in the face of reputation damage
or litigation, managers will manage earnings based on the aggregate market level
rather than their own firm’s individual stock price. In addition, they found that
aggregate level discretionary accruals showed a stronger ability to predict firmlevel returns than firm-level discretionary accrual values did. Predictability also
increased in power when the target firm was o f a larger size. The authors
speculated this was because the managers of very large firms have ‘more at stake’
(Kang et al., 2010, p. 820).
(e) Earnings Management and Mergers
The study Erickson and Wang (1999) examined earnings management by
acquiring firms when using their own stock during a merger. In these mergers,
stock of the acquiring firm is used as payment. There is an agreed upon price
between the acquiring and target firms, and that price is paid by the equity (stock)
o f the acquiring firm. Logic follows that if the acquiring firm can increase the
price of their stock by some means (including earnings management) they will be
able to obtain the target firm for a lower ‘real’ price (lower acquiring-firm number
o f shares) than if acquiring-firm stock was valued at a lower price without
earnings management. The authors also believe that this artificial stock price
inflation is in the interest of the existing acquiring-firm shareholders because
‘existing shareholders prefer a higher price to minimize the likelihood of earnings
dilution, and secondly a stock issue dilutes voting power and control of existing
shareholders’ (Erickson & Wang, 1999, p. 150).

24

To analyze the hypothesis, the authors looked at 55 acquiring firms who
used stock for a merger between 1985 and 1990. They used the Kothari version of
the ‘modified’ Jones model for their analysis, but scaled all variables by total
assets o f the firm:

i lr = e°(jk) + * ( ^ f ) +

....+ &<?.+

# 7*85 + .......+ # 1 2 *9 0 + £it

(2-9)

Where:
TTACjt = Total accmals of firm i in time t
ASTjt = Total assets of firm i in time t

AREVjt = Change in revenue for firm i in time t
PPEjt = Gross property, plant, and equipment for firm i in time t
q , _ Q k = Quarterly variable taking a value of 1 for quarters 1-4 o f the fiscal year,

and 0 otherwise
Y85 - Y90 = A year indicator variable taking the value of 1 for 1985-1990 and 0
otherwise

Regressions by Erickson and Wang showed that there was consistent
earnings management on the part of the acquiring firm, and this earnings
management was ‘based on an increasing function of the economic benefits at
stake in the merger by relative deal size’ (Erickson & Wang, 1999, p. 151). The
authors also studied competing firms who had completed mergers of similar scale
during the same time frame, but used only cash to acquire the target firm. They

25

found in that case that there was ‘no evidence of pre-merger earnings
management by these firms’ (ibid., p. 151).
Over twenty years of articles related to earnings management using
discretionary accruals has led to an almost complete consensus on the use of the
‘modified’ Jones version when attempting to calculate discretionary accruals.
That being the case, there has been some criticism as to the ‘modified’ Jones
models ability to accurately separate discretionary accruals

from non-

discretionary accruals. There is also criticism as to whether discretionary accruals
data is the most important or reliable means o f detecting earnings management.
An article by Fields, Lys, and Vincent (2001) examines the expansion in
the 1990’s of many scholarly accounting techniques and critiques their relevance,
their actual advancement of accounting literature, and their applicability and
predictability in the real world. Although the paper looks at many techniques used
for firm analysis, they devote a section of the paper to the current state of affairs
in discretionary accrual analysis. The authors cite an article published by Kang
and Sivaramakrishnan who use an ‘instrumental variables’ approach to
discretionary accrual detection, and show that their model performs better than the
benchmark ‘modified’ Jones model (Fields et al., 2001, p.289). The authors
believed that three approaches were open to future scholarly papers regarding
detection of earnings management:
1)

The continued use of discretionary accruals in earnings management
detection

26

2)

The development of more powerful techniques, like those developed by
Kang & Sivaramakrishnan

3)

To ‘return to the basics’ and use accounting expertise to directly measure
accounting choice via financial statements

(Fields et al., 2001, p.290)

The study by Kang and Sivaramakrishnan proposes an alternate method of
discretional accrual calculation, as they feel that the existing ‘modified’ Jones
model causes ‘simultaneity, errors-in-variables, or omitted variable problems, any
of which leads to reduced statistical power and erroneous inferences regarding
earnings management’ (Kang & Sivaramakrishnan, 1995, p. 354).
The authors create a model based on instrumental variables (IV) method,
and measure the results against the ‘modified’ Jones model, and also measure the
amount of Type I and II errors versus ‘modified’ Jones. Using the GMM
regression technique (generalized method of moments) they find that their model
is superior to the ‘modified’ Jones model in discretionary accrual detection, and
also has less Type I errors. In the conclusion of the paper, the authors state that
there is great opportunity to further develop the model by creating more specific
variables than those used in their paper depending on what future scholars wish to
capture with their analysis. In addition, the GMM technique allows lagged or
double lagged variables to be added to the model without any need to change the
model itself. Although this model appears to show improvement in discretionary
accmals detection, there seems to be no real expansion of this model in other
scholarly work.
27

(f) Earnings Management and Banks
Like all other firms, earnings management in the form o f income
smoothing improves the risk perception o f a bank to its investors, analysts, and
regulators. It maintains a steady compensation to managers regardless o f their
actual competency. In addition, bank failures, declining earnings, deposit flights
to mutual funds, erosion o f reserves, hostile takeovers, tightened regulations, and
pressures from boards of directors significantly increase the pressures on banks to
smooth their incomes (Bhat, 1996, p.505). Unlike firms from other industries,
banks earn a large proportion o f their income from loans. Therefore, a significant
portion o f risk regarding earnings comes from loans. During the 1990’s in the
United States there was significant evidence that banks were under-reporting or
over-reporting their loan book sizes, and ‘maintaining significant amounts of
unsupported reserves ... not clearly linked to likely losses’ (Liu & Ryan, 2006).
The difference in reported book values of loans and actual book values, as well as
the value of loans that are ‘written o ff and the actual value that are ‘written o ff
can be determined by managers. Likewise, the size of reserves that they must
keep on hand to cover these ‘written o ff loans can likewise be determined by
management’s discretion. Prior literature on management of provision for loan
losses states that in good times banks have an incentive to decrease income, and
increase income in bad times. In addition, provisions can be managed for the
purposes of managerial compensation (Liu & Ryan, 2006, p.424). Accrual-based
earnings management is likely as prevalent in this industry as in any other, but

28

because o f the significantly different accounting reporting standards in the
financial industry (Teoh & Wong, 2002, p. 873) the inclusion o f the banking
sector has not been included in the scope o f this paper.
2.4 Hypothesis Development
Dechow et al., (1995) and Kasznik (1999) both describe how discretionary
accruals estimated from the Jones base model or the ‘modified’ Jones model
variation are both positively related to return on assets. This positive correlation
has been accredited to the misspecification inherent in these two models, with the
assumption that there is no relationship between earnings management and firm
performance or growth (Jevons Lee et al., 2006, p. 306). Additionally, managers
choose the level of reported earnings to maximize their utility, which is an
increasing function of the firm’s market value. This leads to our first hypothesis:

Hi: Earnings management and return on assets (RoA) are positively related.
Earnings management is related to firm characteristics like size, leverage,
etc. It also makes practical sense that highly leveraged firms would have a greater
impetus to meet or exceed their industry peers as well as analyst predictions when
it comes to earnings reporting. Any substantial drop in their share price would
have serious negative effects on their debt-to-equity ratio, as it would increase in
the event of a lowering o f their stock price. Consequentially, an increase in the
debt-to-equity ratio could erode investor confidence in the firm’s ability to
manage payments on the higher leverage ratio. Therefore, it is hypothesized that:

29

H2: Earnings management and firm leverage are positively related.
Large firms are more closely monitored by investors (especially
institutional investors) and analysts. There are incentives to step-up earnings
reports. Specifically, managers of large companies have greater amounts of
compensation at risk, as well as personal prestige and reputation. Because o f this
they have a greater incentive to indulge in earnings management than managers of
smaller firms (Kim et al., 2003), especially for the purposes o f avoiding reports o f
earnings decreases and potentially putting their compensation at risk. This leads to
the following hypothesis:

H3 : Earnings management and firm size (TA) are positively related.
The Sarbanes-Oxley Act is a federal law that set enhanced standards for
all US public company boards, management, and public accounting firms (Kieso
et al., 2005). With the passage of this act, management must now individually
certify the accuracy of financial information and penalties for fraudulent financial
activity are much more severe. The act also increased the independence o f outside
auditors who review corporate financial statements, and increased the oversight of
boards of directors. Prior to the passage of the Sarbanes-Oxley Act (2002)
managers were able to manage earnings with a lower chance of detection.
Compared to industry peers, earnings shortfalls were rare and when they did
occur the market interpretation of the earnings shortfall was substantial. Cohen et
al., (2007) present findings that after the passage of the Sarbanes-Oxley Act the

30

amount of accrual-based earnings management was attenuated. This leads to the
following hypothesis:

H4 : The amount of accrual-based earnings management will be smaller after
the implementation of heightened financial regulation (SOX) compared to
before the passage of the increased financial regulation.

Utilizing the train of thought of Barberis et al., (1998) and Veronesi
(1999) during expansionary economic phases investors will become highly
confident that the market is in a good state. Under such circumstances, firmspecific bad news causes firm stock prices to fall, since bad news causes poor
investor sentiment that the firm is in a good state in relation to the industry as a
whole. Additionally, as uncertainty in the true state of the economy increases,
risk-averse investors ask for a higher expected return (greater firm earnings). In
times o f economic uncertainty a further asymmetry is caused in investor response
to bad news. When investors believe that the economy is in a bad state, additional
bad news will have minimal impact on an investor’s firm-specific sentiment, as
they cannot separate the firm specific event from overall macroeconomic
uncertainty. There is therefore a motivation for firms to maintain their earnings at
levels similar to their industry peers during good times, and report lower-thanactual amounts during bad times. This leads to hypothesis five:

H5 : Positive earnings management and business cycles are positively related.

31

2.4 Earnings Management, Reversals and Recent Developments

The most recent development in earnings management literature is related
to accrual reversals. The study by Baber, Kang, and Li (2011) examine accrual
‘reversals’ which is based on the argument that any positive or negative
discretionary accruals that are created for the purposes of earnings management
must be reversed in future financial reporting, as they must be ‘paid back’. With
this assumption Baber et al., (2011) examine not only absolute accrual values as
in the ‘modified’ Jones model, but also examine changes in discretionary accrual
values between different time periods. The differences in these values are referred
to as ‘net’ discretionary accruals, and Baber et al., (2011) propose that these are
the true indicators for determining companies’ levels of earnings management.
With the addition of these net discretionary accruals to the ‘modified’ Jones
model, the authors have created a test that has more robust results than traditional
regression modeling (cross section and time series), and also has little
specification problem that has been inherent to discretionary accruals modeling
empirical discourse.
In the study of these reversals, Baber et al., point out that there is an
inverse association between the reversal speed and the probability o f meeting or
beating analyst forecasts (Baber et al., 2011, p.l 191). The reversal speed refers to
the length o f time that a discretionary accrual can exist ‘on the books’ before it
must be paid back. The authors also note that since the introduction of SarbanesOxley regulation there has been a reduction in discretionary accruals-based

32

earnings management (ibid., p. 1190). The framing o f reversals is given as the
following:

Xt = X*t + (d ?ew - ^ v e rse d )

(2 . 10)

Where:
Xt = reported earnings
Xt* = unmanged earnings
dtnew= new discretionary accruals
^reversed _ prevjous discretionary accruals that reverse in time t
(dtnew- dtrcversed) therefore gives the ‘net’ discretionary accrual for t

The authors state the proxies for reversals are the residual autocorrelation
coefficients. Given discretionary accruals for a firm, the order of autocorrelation
o f residuals that is the smallest is equal to the period of full reversal (ibid.,
p. 1195). The isolation of net non-current discretionary accruals is given through
the following formula:

ACC ——— — /?ot + P i t ^

ASALES,

(2 . 11)

Where:

ACC -N Ct = the difference in working capital accounts (ACC_WCt) and total
accruals (ACCt>
ASALESjt= change in sales in year t
PPEit = gross property, plant, and equipment in year t
et =

net non-current discretionary accruals

33

Faster reversal speed of prior discretionary accruals imposes greater
constraints

on

management’s

ability

to undertake subsequent

earnings

management. In addition, a given amount of net discretionary accrual
overstatement imposes ‘different degrees of constraint on subsequent earnings
management depending on the reversal speed of discretionary accruals’ (ibid.,
p. 1209).
The most recent study integrating net discretionary accrual and reversals is
by Dechow, Hutton, Kim, and Sloan (2012). Working under similar equations to
Baber et al., (equation 2.10), the authors describe in detail why the addition of
reversals in current discretionary accmals calculation improves the current
formula. Specifically, they find that reversals ‘increase test power by 40% and
mitigate misspecification problems from correlated omitted variables’ (Dechow et
al., 2012, p.2).
In their testing, the authors create a sample from 1950-2009, for a total of
209,530 firm years. They omit financial firms, as discretionary accmals
calculation is based on working capital, and this variable is of less meaning to
financial firms (ibid., p. 15). In their tests, they find that if the researcher models
reversals when they do not actually exist, the test power decreases. However, if
they are right about the timing of reversals 50% of the time, test power increases,
and if their timing is as right for reversals as it is for the timing of earnings
management, predictive power is increased by >50% compared to traditional ttests. In addition, they find that the best results come from modeling reversals as
occurring in t+1 and t+2 periods (ibid., p.26).
34

To overcome misspecification, the authors find the correlated omitted
variable bias ‘is overcome by reversals as long as the omitted variables don’t
reverse in the same period as the discretionary accruals’ (ibid., p.30). They find
that modeling reversals in any period >t+2 causes over-correction. As opposed to
Kothari et al., the authors also find that firm performance matching does not work
well, as omitted variables cannot be known.
The examination of earnings management to date has been extensive.
Developments in modeling discretionary accrual isolation (using various models)
have evolved, and studies involving the application of these models are
substantial. Most of the studies are based on the ‘modified’ Jones model and these
have shown that there exists substantial earnings management, especially in the
United States. The majority of studies focus on discretionary accruals and their
relationship to initial public offerings, seasoned stock offerings, firm returns,
cumulative abnormal returns, mergers, and the effectiveness o f regulations (like
the Sarbanes-Oxley Act).

35

CHAPTER III
DATA AND METHODOLOGY
This chapter discusses the database and methodology used in the empirical
investigation. This chapter is divided into two sections. Section 3.1 discusses the
database used in the study and Section 3.2 discusses the methodology used in the
study.
3.1 Database
In order to capture the overall picture o f the US corporate sector, we
started the investigation with all firms in the S&P 1500 in 2010. Data was
obtained from the CRSP (The Center for Research in Security Prices) database.
Using the S&P 1500 index, all relevant financial variables required to calculate
discretionary accruals values using the ‘modified’ Jones formula were obtained
for the panel 2000-2010. Using SIC (Standard Industry Classification) codes, the
6000 series o f companies were removed from the panel, as their financial
dynamics are quite different from non-financial firms (Dechow et al. 2012). The
residual panel data set contained 1125 of the original S&P 1500 companies for the
periods 2000-2010.
Regressions were completed using a balanced panel, which was composed
of data from the 2000-2010 periods. The list of variables used in both the
‘modified’ Jones model of discretionary accrual calculation and the empirical
investigation are provided in Table 3.1
36

To derive values for the business cycle dummy variable data was obtained
from the National Bureau of Economic Research (NBER). Quarterly economic
data was obtained for the time period 2000-1010 for the United States. Any years
in this period with one or more quarters of economic contraction were given a
value of zero. Years having economic expansion in all four quarters were given a
value of one.

37

Table 3.1 Description of Variables Used in the Study
Notation
DA

Definition
Discretionary accruals

NDA

Non-discretionary
accruals

AREV

Net revenues

AREC

Net receivables

PPE
SOX

Property, plant, and
equipment
Dummy Variable 1

BUS_CYC

Dummy Variable 2

ROA

Return on assets

TOTAL ASSETS

Total assets

LEV

Leverage

38

Description
Discretionary accrual
values as determined by
‘modified’ Jones
Non-discretionary
accruals as determined by
‘modified’ Jones
Net revenues in year t
less net revenues in year
t-1 scaled by total assets
att-1
Net receivables in year t
less net receivables in
year t-1 scaled by total
assets at t-1
Gross property, plant, and
equipment in year t
Dummy variable with the
value o f zero before the
introduction of the
Sarbanes-Oxley Act
(1980-2001) and a value
of one after (2002-2010)
Dummy variable with the
value o f zero in periods
of economic peak to
trough, and one in periods
of trough to peak
Earnings before interest
and taxes divided by total
assets o f the firm
Total assets o f the firm
Debt to equity ratio of the
firm

3.2 Methodology
Equation 2.7 was estimated using a panel regression framework. From this
regression non-discretionary values were obtained for the S&P 1500 using the
‘modified’ Jones model:

+ a 2(AREVt - ARECt) + a 3(PPEt)

These values were then subtracted from given total accrual values for the
matching year, and discretionary accrual amounts were obtained for each firm.
With these discretionary accrual values, an equation was estimated of the
following form:
DAit = p 0 + piROAit + (S2TOTAL_ASSETSit + p3LEVit + p4S 0 X it +
p sECON_CYCit + eit
Where:
P i ->

■> 0>/?3 > 0 ,/?4 < 0,/?5 > 0

The details of data and their description are reported in Table 3.1. In terms
of empirical methodological frameworks, we present estimates based on pooled
least squares, fixed effects model, and random effects model. It should be noted
that each model comes with its own shortcomings.
Fixed effects estimation assumes that the firm-specific effects are
uncorrelated with the explanatory variation of any individual variable from all
past, current, and future time periods. Assuming that the changes in the firm39

specific portion of these variables is constant over time, the fixed effects model
will attribute changes in the dependent variable to influences other than these
‘fixed’ components (Stock & Watson, 2011, p. 372). Unfortunately, this type of
regression comes with the inherent problem that it is unlikely that all the
unobserved variation that affects the dependent variable is static over time. From
a practical perspective in this study, the unbalanced panel data set contains
observations over a 30 year time period, and it is impossible that all o f the
unobserved variation in the regression had no effect on the dependent variable
(discretionary accruals).
The random effects model also attempts to eliminate a portion of variation
from the model, but in this case assumes that the individual firm variables are
constant over time, all of the variation is attributed to changes over time. Because
time contingent variation is important for this form of regression, the constant is
excluded as it exhibits no change over time periods. Although this model can also
be helpful in identifying the portion of change that is purely a function of changes
in time (and perhaps the phase of the macroeconomic cycle), it will also not be
able to identify all o f the variation alone. It is also worth noting that both the fixed
and random effects models are inferior if the panel data set contains many outliers
(extreme values) (Stock & Watson, 2011, p. 361).

40

CHAPTER IV
EMPIRICAL RESULTS
This chapter discusses the empirical results of the panel regressions of
determinants of discretionary accruals of 1125 US firms for the period 2000-2010.
This chapter is organized as follows: Section 4.1 discusses the descriptive
statistics of the variables used in the empirical investigation. Section 4.2 provides
the correlation matrix of these variables. Section 4.3 discusses the empirical
results of the pooled least squares, random effects, and fixed effects regression
models. Section 4.4 summarizes the conclusions of this chapter.
4.1 Descriptive Statistics
Table 4.1 reports descriptive statistics of the variables used in the
empirical investigation. The average firm size for this period was US$7.25 billion
with the largest firm having a value of US$ 797.76 billion and the smallest having
a value of US$ 2.62 million. The median size was US$ 1.41 billion which
indicates that there is a substantial variation in the data. The average debt-toequity ratio for this period was 50%, with the highest leveraged firm having a
ratio of 354% and the lowest leveraged ratio being -76%. The mean discretionary
accruals value (as a percentage o f total accruals) is 2% of total accruals per year,
with the highest discretionary accruals percentage being 16.37% of total accruals,
and the lowest being -10.82%. Return on assets, as measured by earnings before
interest and taxes divided by total assets, had a mean value o f 9%. The highest
41

return on assets was 488%, and the lowest was -569%. The economy of the
United States was in a state of expansion for 64% of the periods included in the
panel, as determined by BUS_CYC, a dummy variable with a value of 0 during
any year with at least one quarter of recessionary activity, and 1 otherwise.

Table 4.1: Descriptive Statistics of Financial Parameters of US Non-Financial
Corporate Sector 2000-2010 (Balanced).
This table presents the variables fo r the analysis o f measuring the relationship o f
discretionary accruals to a number o f firm-specific, regulatory, and business
cycle based variables. The variable DA is defined as discretionary accrual
amounts as a percentage o f total assets calculated using the ‘modified’ Jones
model. SOX refers to the introduction o f the Sarbanes-Oxley Act. BUS CYC states
whether the economy is in an expansionary or contractionary phase. ROA is
defined as firm earnings before taxes and interest divided by total assets. TOTAL
ASSETS refers to firm total assets in millions. LEV is defined as firm debt-toequity ratio in percentage (See Table 3.1 fo r descriptions). Note: Data relates to
1125 firms fo r the period 2000-2010.

Mean
Median
S. Dev
Skewness
Minimum
Maximum

TOTAL
DA(%
LEV (%)
SOX BUS
ROA
CYC
o f total
ASSETS
(%)
(in billions)
accruals)
45.44
0.02
0.83
0.67
9.01
7.26
1.00
1.41
-0.02
1.00
9.35
47.92
0.29
0.37
0.47
18.59
22.89
27.55
-0.71
48.35
9.15 -1.79
-6.73
16.08
-10.82
0.00
0.00 -569.62
0.0026
0
16.37
1.00
1.00 487.47
797.77
354.27

4.2 Correlation Matrix
None of the variables in the correlation matrix are highly correlated, either
positively or negatively. This shows that there is very little chance o f
multicollinearity occurring in panel regression. In the next section, we present the
results of the empirical investigation of equation 2.1.

42

Table 4.2 Correlation Matrix of Variables, Balanced Panel
This table presents the correlation o f the variables used in the analysis o f the
relationship o f discretionary accruals to a number o f firm-specific, regulatory,
and business cycle based variables. The variable DA is defined as discretionary
accrual amounts as a percentage o f total assets calculated using the ‘modified’
Jones model. SOX refers to the introduction o f the Sarbanes-Oxley Act. BUS CYC
states whether the economy is in an expansionary or contractionary phase. ROA
is defined as firm earnings before taxes and interest divided by total assets.
TOTAL ASSETS refers to firm total assets in millions. LEV is defined as firm
debt-to-equity ratio in percentage (See Table 3.1 fo r descriptions). Note: Data
relates to 1125 firm s fo r the period 2000-2010.
DA

SOX

ECON

ROA

CYC
DA

1

SOX

0.10

TOTAL

LEV

ASSETS

A

A

REC

REV

PPE

1

(0.00)
ECON

0.18

0.13

CYC

(0.00)

(0.00)

ROA

0.09

0.03

0.07

(0.00)

(0.00)

(0.00)

TOTAL

-0.01

0.03

-0.01

-0.01

ASSETS

(0.29)

(0.00)

(0.17)

(0.77)

LEV

0.01

-0.03

-0.02

-0.09

0.14

(0.33)

(0.00)

(0.07)

(0.00)

(0.00)

0.01

0.01

0.05

-0.01

-0.01

-0.00

(0.14)

(0.47)

(0.00)

(0.39)

(0.63)

(0.87)

0.02

-0.01

0.04

0.01

-0.01

-0.01

0.19

(0.06)

(0.24)

(0.00)

(0.25)

(0.57)

(0.21)

(0.00)

-0.01

0.04

-0.02

0.01

-0.01

-0.04

0.05

0.18

(0.52)

(0.00)

(0.03)

(0.23)

(0.33)

(0.00)

(0.00)

(0.00)

AREC

A REV

PPE

1

1

43

1

1

1

1

1

4.3 Empirical Results
Table 4.3 reports the results of the panel estimation of equation 2.1 using
pooled regression, fixed effects, and random effects models for the balanced panel
for the period 2000-2010. The estimates and fit for all three regression models are
good as evidenced by relatively high R2.
The empirical results show that earnings management and return on assets
are positively related as evident from the statistically significant positive
coefficient (hypothesis 1). Growth-based firms tend to be smaller in size and
because of their growth focus have an increased motivation to engage in earnings
management, which is validated by this empirical result. In all models the
coefficient for return on assets is positive and relatively high compared to other
independent variables, and is statistically significant at 1% level.
The empirical results also show a positive relationship between firm
leverage and earnings management (hypothesis 2). Firms that are highly
leveraged have the potential to suffer additional complications from having
quarterly earnings that are lower than expectations. An obvious case would be
that a significant decrease in their share price from a poor earnings report would
fundamentally alter their debt-to-equity ratio in a negative way. This could lead to
additional financing costs as the firm’s credit rating or ability to repay existing
debt could be jeopardized. The coefficient sign for leverage was expected across
all models, however it was not significant in the fixed effects model which is our
preferred model. Because o f the large sample size it is likely that heterogeneous

44

firm leverage has less of an effect on earnings management than was
hypothesized.
Among the control variables, total assets (TA) was hypothesized to have a
positive sign in hypothesis 3, with the rationale being that large firms have highly
compensated managers compared to smaller firms. Because of the large
compensation (especially performance-based compensation) that comes with the
management o f very large firms, it was hypothesized that as firm size increases,
discretionary accrual amounts would also increase as the motivation for managers
to engage in earnings management is greater. Our results do not validate this
hypothesis as we find a negative relationship between discretionary accruals and
total assets. As company size (as measured by total assets) increases, it appears
that earnings management in the form of discretionary accruals decreases
consistently among all three models. This could be explained by the fact that as
companies achieve a large size, they are audited more thoroughly, and also have
greater monitoring by individual and institutional shareholders and industry
analysts, removing the ‘flexibility’ that managers of smaller, less closely
monitored firms would have. It should be noted that this coefficient log of total
assets has a very low value, and only the fixed effects model yielded statistically
significant results.
The SOX variable in hypothesis 4 was posited to have a negative
relationship to discretionary accrual values. Introduction of regulations like the
Sarbanes-Oxley Act has been shown to decrease accrual-based earnings
management by a number of researchers (Graham et al. 2005, Cohen et al. 2007,
45

Cohen and Zarowin, 2010). Our results are at variance with these results. In all
three regression models, the SOX variable was positively related to discretionary
accruals and was statistically significant (at 1% level). A possible explanation for
the positive association could be that the Sarbanes-Oxley Act is more about
reforms at internal control of firms rather than earnings management.
As hypothesized in Chapter 2 (hypothesis 5), the business cycle dummy
has the predicted sign (positive) implying that earnings management varies with
phases of the business cycle; it is high during upward phases o f the business cycle
and lower during contractionary phases o f the business cycle. The variable is also
statistically significant. This result is similar to the results of Kang et al. (2010).
Firms indulge in higher earnings management (by maintaining large discretionary
accruals) to maintain earnings similar to their industry peers regardless o f their
actual performance. During recessionary periods, there is a contraction in the
economy as a whole, and this decrease is likely caused by macroeconomic events
outside of any specific industry. Since all firms experience a decrease in stock
value regardless o f their relative peer-related performance, managers will take this
opportunity to reverse the discretionary accruals they accumulated during the
previous period of economic expansion.

46

Table 4.3 Determinants of Discretionary Accruals in the United States —
2000 - 2010 .

The dependent variable is DA (discretionary accruals). SOX is a dummy variable
with a value o f 0 previous to the introduction o f the Sarbanes-Oxley Act and a
value o f 1 after. BUSjCYC is a dummy variable with a value o f 0 fo r any year
where there was a contractionary quarter during the year and a value o f 1during
years with four quarters o f economic expansion. ROA is earnings before interest
and taxes divided by total assets. Total Assets is the natural logarithm o f firm
total assets. LEV is the debt-to-equity ratio.

ROA
Log(Total
Assets)
LEV
Observations
(N)
R2
Adj-R2
Jarque-Bera
Normality
test of
residuals

0.045
0.045
96265602***

0.047
0.028
49217241***

*
*

BU SCYC

0.254
(0.084)***
0.072
(0.007)***
0.082
(0.005)***
0.098
(0.016)***
-0.025
(0.006)***
0.026
(0.019)
11964

Random
Effects
-0.089
(0.021)***
0.057
(0.006)***
0.089
(0.005)***
0.114
(0.014)***
-0.001
(0.002)
0.029
©

SOX

Fixed Effects

O

Constant

Expected Pooled
Sign
+/-0.090
(0.021)***
0.057
(-)
(0.006)***
+
0.089
(0.005)***
+
0.114
(0.013)***
+
-0.001
(0.001)
+
0.028
(0.011)*
11964

11964

0.045
0.045
96265602***

Hausman
35.04***
Test (x2
statistic)
Note:
Figures in brackets are standard errors.
***, **, * indicates statistical significance at 1%, 5%, and 10% respectively (twosided test).

47

We had presented empirical results of the models pooled, fixed effect, and
random effect. As discussed in Chapter 3 (Section 3.2), the pooled regression
model suffers from correlation of independent variables with the error term.
Moreover, the pooled regression model does not recognize or model
heterogeneity of firms in the panel which is a major limitation o f pooled
regression estimates. Fixed effects and random effects models recognize
heterogeneity among firms and are attractive from a statistical inference point o f
view. A comparison of the fixed effects and random effects models was done
using the Hausman test. The large and significant value of the Hausman statistic
shows that there is a significant difference between the coefficients of the two
models, and therefore the fixed effect model would be the more prudent choice.
4.4 Summary o f Results
The fixed effect model assumes that there are individual firm specific
effects correlated with the independent variables, and this makes intrinsic sense,
as yearly firm-specific discretionary accrual values are likely linked to the yearly
financial variables of the firm, in addition to the current macroeconomic
environment. The results show a strong relationship between the overall business
cycle and firm-specific discretionary accrual values. Specifically, discretionary
accruals increase during times of economic expansion, and decrease during
contractionary periods. The Sarbanes-Oxley Act appears to have had little effect
on discretionary accrual values, as these values have increased since the
introduction o f Sarbanes-Oxley. Additionally, as firm size increases (measured in
total assets) discretionary accrual values decrease. This is likely attributed to the
48

fact that large firms have higher quality boards, and audit committees meet with
greater frequency as well as have greater financial sophistication. This constrains
managerial propensity to engage in discretionary accruals-based earnings
management (Xie et al. 2003).

49

CHAPTER V
CONCLUSIONS

This study has been undertaken at a time of considerable global economic
uncertainty stemming from the 2008 recession centered in the United States and
Europe. Although this recession has many causes, one issue that has exacerbated
the crisis is a lack of transparency in the financial reporting o f firms. An
important issue in financial reporting is the extent to which managers manipulate
reported earnings. Following Healy (1985), accrual-based earnings management
measures continue to be a main focus of academic research. Compared to earlier
studies, our study focusses on the impact of business cycles on earnings
management in the United States over 11years using 1125 firms (2000-2010). The
results show that in economic expansionary times firms actively engage in
earnings management to maintain earnings levels comparable to their industry
peers. Similarly, during contractionary times when share prices fall across the all
industries (regardless of firm-specific performance), these accrual accounts will
be ‘washed’ clean. During recessions, shareholders expect poor earnings, and
having large negative discretionary accruals (to offset the positive ones created
during expansionary periods) has little effect on overall market sentiment, and
therefore individual firm share prices.
This study validates some o f the empirical results from prior earnings
management research but also gives some additional insights. As shown by
50

existing research, growth-oriented firms demonstrate higher discretionary accrual
amounts versus value-oriented firms on the aggregate. Additionally, larger firms
appear to have lower discretionary accrual amounts (as a percentage of total
assets). This result has also been found by many prior studies, namely that larger
firms are monitored more closely by auditors, shareholders, analysts, and
regulators. Additionally, large firms tend to have more sophisticated boards and
auditing committees, limiting opportunities by managers of large firms to engage
in earnings management.
Contrary to a number of prior studies, we found across all pooled
regression models that the introduction of the Sarbanes-Oxley Act (2002) has not
abated earnings management among US firms. Although the Sarbanes-Oxley Act
has likely created additional transparency in financial reporting, its limited effect
on discretionary accruals-based earnings management may simple be accredited
to the great difficulty in identifying discretionary accruals from financial
statements.
This study contributes to the existing literature on earnings management
by illustrating a highly significant positive connection between the level of
earnings management in the major US firms over a long period of time (20002010). An opportunity for future research would be a study of economic cycles
and discretionary accruals among firms using the ‘reversals’ development in
discretionary accruals by Baber et al., 2011 and Dechow et al., 2012.

51

BIBLIOGRAPHY
Aboody, D., Barth, M. E., & Kasznik, R. (1998) Revaluations o f fixed assets and
future firm performance: evidence from the UK, Journal o f Accounting
and Economics, Vol. 26, 149-178.
Baber, W. R., Kang, S. H., & Li, Y. (2011) Modeling discretionary accrual
reversal and the balance sheet as an earnings management consultant, The
Accounting Review, Vol. 86, 1189-1212.
Ball, R., & Shivakumar, L. (2008) How much new information is there in
earnings?, Journal of Accounting Research, Vol. 46,975-1016.
Bao, B. H., Chung, R., Niu, Y., & Wei, S. (2012) Real and accrual earnings
management around IPO’s: evidence from US companies, working paper.
Barber, B. M., & Lyon, J. D. (1996) Detecting abnormal operating performance:
the empirical power and specification o f test statistics, Journal o f Financial
Economics, Vol. 41, 359-399.
Barberis N., Shleifer A., & Vishny R. (1998) A model of investor sentiment,
Journal of Finance Economics, Vol. 49, 307-345.
Bartov, E., & Cohen, D. A. (2008) The ‘numbers game’ in the pre- and postsarbanes-oxley eras, Leonard N. Stem School of Business, New York
University, New York.
Bartov, E., Gul, F. A., & Tsui, J. S. L. (2001) Discretionary-accruals models and
audit qualifications, Journal of Accounting and Economics, Vol. 30, 421 452.
Beatty, A., Liao, W. S., & Weber, J. (2010) The effect of private information and
monitoring on the role of accounting quality in investment decisions,
Contemporary Accounting Research, Vol. 27, 17-47.
Bedard, J. (2006) Sarbanes oxley internal control requirements and earnings
quality, Universite Laval Ecole de Comptabilite, working paper.
Bens, D., Nagar, V., & Franco Wong, M.H., (2002) Real investment implications
of employee stock option exercises, Journal o f Accounting Research, Vol.
40, 359-393.
Bens, D., Nagar, V., Skinner, D.J., & Franco Wong, M.H., (2003) Employee
stock options, EPS dilution and stock repurchases, Journal of Accounting
and Economics, Vol. 36, 51-90.
Bhat, V. N. (1996) Banks and income smoothing: an empirical analysis, Applied
Financial Economics, Vol. 6, 505-510.
52

Biddle, G. C., & Hilary, G. (2006) Accounting quality and firm-level capital
investment, The Accounting Review, Vol. 81, 963-982.
Biddle, G. C., Hilary, G., & Verdi, R. S. (2009) How does financial reporting
quality relate to investment efficiency?, working paper.
Braun, M. & Larrain, B. (2005) Finance and the business cycle: international,
inter-industry evidence, The Journal o f Finance, Vol. 60, 1097-1127.
Burgstahler, D. & Dichev, I. (1997) Earnings management to avoid earnings
decreases and losses, Journal of Accounting and Economics, Vol. 24, 99126.
Chipalkatti, N. & Rishi, M. (2007) Do indian banks understate their bad loans?,
The Journal of Developing Areas, Vol. 40, 75-91.
Cohen, D. A., Dey, A., & Lys, T. Z. (2007) Real and accrual-based earnings
management in the pre- and post-sarbanes oxley periods, AAA 2006
Financial Accounting and Reporting Section (FARS) Meeting Paper.
Cohen, D. A., & Zarowin, P., (2010) Earnings management over the business
cycle, working paper, New York University.
Covas, F. & Den Haan, W. J. (2011) The cyclical behavior of debt and equity
finance, American Economic Review, Vol. 101, 877-899.
DeAngelo, L. (1986) Accounting numbers as market valuation substitutes: a study
of management buyouts of public stockholders, The Accounting Review,
Vol. 61,400-420.
Dechow, P. M., Hutton, A. P., Kim, J. H., & Sloan, R. G. (2012) Detecting
earnings management: a new approach, Journal of Accounting Research,
Vol. 50, 275-334.
Dechow, P.M., & Skinner, D., (2000) Detecting earnings management, The
Accounting Review, Vol. 70, 133-168.
Dechow, P. M., Sloan, R. G., & Sweeney, A. P. (1995) Detecting earnings
management, The Accounting Review, Vol. 70, 193-225.
DuCharme, L. L., Malatesta, P. H., & Sefcik, S. E. (2000) Earnings management:
IPO valuation and subsequent performance, University o f Washington
School of Business.
Edelstein, R., Liu, P., & Tsang D. (2009) Real earnings management and dividend
payout signals: a study for US real estate investment trusts, working paper.
Eil, D. & Rao, J. M. (2011) The good news-bad news effect: asymmetric
processing o f objective information about yourself, American Economic
Journal: Microeconomics, Vol. 3, 114-138.
53

Erickson, M., & Wang, S. W. (1999) Earnings management by acquiring firms in
stock for stock mergers, Journal of Accounting and Economics, Vol. 27,
149-176.
Fields, T.D., Lys, T. Z., & Vincent, L. (2001) Empirical research on accounting
choice, Journal of Accounting and Economics, Vol. 31, 255-307.
Graham, J. R., Harvey, C. R., & Rajagopal, S. (2005) The economic implications
of corporate financial reporting, Journal of Accounting and Economics,
Vol. 40, 3-73.
Greene, W. H., (2000) Econometric analysis, Upper Saddle River, NJ, Prentice
Hall.
Guay, W. R., Kothari, S. P., & Watts, R. L. (1996) A market-based evaluation of
discretionary accrual models, Journal of Accounting Research, Vol. 34,
83-105.
Guo, H., & Jiang, X. (2011) Accmals and the conditional equity premium, Journal
of Accounting Research, Vol. 49, 187-221.
Healy, P. M. (1985) The effect of bonus schemes on accounting decisions, Journal
o f Accounting and Economics, Vol. 7, 85-107.
Healy, P. M., & Wahlen, J. M. (1999) A review of the earnings management
literature and its implications for standard setting, American Accounting
Association, Vol. 13, 365-383.
Hirshleifer, D., Hou K., & Teoh, S. H. (2009) Accruals, cash flow, and aggregate
stock returns, Journal of Financial Economics, Vol. 91, 389-406.
Hong, X. (2001) The mispricing of abnormal accruals, The Accounting Review,
Vol. 76, 357-373.
Ikenberry, D., Lakonishok, J., & Vermaelen, T. (1995) Market Underreaction to
open market share repurchases, Journal of Financial Economics, Vol. 39,
181-208.
Jevons Lee, C. W., Yue Li, L., & Yue, H. (2006) Performance, growth and
earnings management, Review of Accounting Studies, Vol. 11, 305-334.
Jones, J. J. (1991) Earnings management during import relief investigations,
Journal o f Accounting Research, Vol. 29, 193-228.
Jones, K. L., Krishnan, G. V., & Melendrez, K. D. (2008) Do models of
discretionary accruals detect actual cases o f fraudulent and restated
earning? An empirical analysis, Contemporary Accounting Research, Vol.
25, 499-531.

54

Kang, Q., Liu, Q., & Qi, R. (2010) Predicting stock market returns with aggregate
discretionary accruals, Journal of Accounting Research, Vol. 48, 815-858.
Kang, S. H., & Sivaramakrishnan, K. (1995) Issues in testing earnings
management and an instrumental variable approach, Journal of
Accounting Research, Vol. 33, 353-367.
Kasznik, R. (1999) On the association between voluntary disclosure and earnings
management, Journal of Accounting Research, Vol. 37, 57-81.
Kieso, D. E., Kimmel, P. D., & Weygandt, J. J. (2005) Financial Accounting, 4th
Edition. John Wiley & Sons, Inc., New York, NY.
Kim, Y., Liu, Y., & Rhee, S. G. (2003) The effect of firm size on earnings
management, working paper, University of Hawaii, Honolulu.
Krishnan, G. V., (2002) Audit quality and the pricing of discretionary accruals,
City University o f Hong Kong, Hong Kong.
Kothari, S.P., Leone, A. J., & Wasley, C. E. (2005) Performance matched
discretionary accrual measures, Journal of Accounting and Economics,
Vol. 39,163-197.
Leuz, C., Nanda, D., & Wysocki, P. D. (2003) Earnings management and investor
protection: an international comparison, Journal of Financial Economics,
Vol. 69, 505-527.
Li, F. (2006) Annual report readability, current earnings, and earnings persistence,
Ross School of Business, University of Michigan, Ann Arbor, Michigan.
Li, S. F., McDowell, E., & Moore, E. A. (2009) Accrual based earnings
management, real transactions manipulation and expectiations
management: US and international evidence, working paper.
Liu, C. C. & Ryan, S. G. (2006) Income smoothing over the business cycle:
changes in banks’ doordinated management of provisions for loan losses
and loan charge-offs from the pre-1990 bust to the 1990s boom, The
Accounting Review, Vol. 81, 421-441.
Lo, K., (2007) Earnings management and earnings quality, Journal of Accounting
and Economics, Vol. 45. 350-357.
Martins, P. S., Solon G., & Thomas, J. P. (2012) Measuring what employers do
about entry wages over the business cycle: a new approach, American
Economic Journal: Macroeconomics, Vol. 4, 36-55.
McNichols, M. F., (2000) Research design issues in earnings management
studies, Journal of Accounting and Public Policy, Vol. 19, 313-345.

55

Naes, R., Skjeltorp, J. A., & Odegaard, B. A. (2011) Stock market liquidity and
the business cycle, The Journal of Finance, Vol. 66,139-175.
Rangan, S. (1998) Earnings management and the performance o f seasoned equity
offerings, Journal of Financial Economics, Vol. 50,101-122.
Ronen, J., & Sadan, S., (1981) Smoothing income numbers: objectives, means,
and implications, Addison-Wesley Publishing Company.
Ronen, J., & Yaari, V., (2008) Earnings management: emerging insight in theory,
practice, and research, 1st Edition, Springer Science and Business, New
York.
Roychowdhury, S. (2006) Earnings management through real activities
manipulation, Journal of Accounting and Economics, Vol. 42, 335-370.
Sabramanyam, K. R. (1996) The pricing of discretionary accruals, Journal of
Accounting and Economics, Vol. 22, 249-281.
Schipper, K., (1989) Commentary on earnings management, Accounting
Horizons, Vol. 3, 91-102.
Stock, J. H., & Watson, M. W., (2011) Introduction to Econometrics, Boston:
Addison-Wesley.
Teoh, S. H., Welch, I., & Wong T. J. (1998) Earnings management and the longrun market performance of initial public offerings, The Journal of Finance,
Vol. 53, 1935-1974.
Teoh, S. H., Wong, T. J., & Rao, G. R. (1998) Are accruals during initial public
offerings opportunistic?, Review of Accounting Studies, Vol. 3,175-208.
Teoh, S. H., & Wong, T. (2002) Why new issues and high accrual firms
underperform: the role of analysts’ credulity, Review of Financial Studies,
Vol. 15, 869-900.
Vander Bauwhede, H., Willekens, M., & Gaeremynck, A. (2000) Audit quality,
public ownership, and firms’ discretionary accruals, Catholic University o f
Leuven, Leuven, Belgium.
Veronesi, P. (1999) Stock market overreaction to bad news in good times: a
rational expectations equilibrium model, The Review of Financial Studies,
Vol. 12, 975-1007.
Walsh, P., Craig, R., & Clarke, F. (1991) ‘Big bath accounting’ using
extraordinary items adjustments: Australian empirical evidence, Journal of
Business Finance & Accounting, Vol. 18, p 173-189.

56

Wei, C. (2009) Does the stock market react to unexpected inflation differently
across the business cycle?, Applied Financial Economics, Vol. 19, 19471959.
Wilson, J. T. E. (2009) Financial reporting in the sarbanes-oxley era: have
reforms improved earnings quality?, Research in Business and Economics
Journal, Vol. l,p. 1-11.
Wooldridge, J. M. (2001) Applications o f generalized method of moments
estimation, The Journal of Economic Perspectives, Vol. 15, 87-100.
Wysocki, P. D., (2009) Assessing earning and accruals quality: US and
international evidence, MIT Sloan School o f Management, Cambridge,
Maine.
Xie, B., Davidson III, W. N., & DaDalt, P. J. (2003) ‘Earnings management and
corporate governance: the role of the board and the audit committee,
Journal of Corporate Finance, Vol. 9, p 295-316.

57