Do firms manipulate earnings when entering the bond market?
Pae, Sangshin "Sam" ; Quinn, Tina
INTRODUCTION
Substantial evidence indicates that managers engage in earnings
management. As evidenced by extensive corporate scandals, including
Enron, WorldCom and Xerox, it is common knowledge among investors,
analysts and regulators that earnings management exists. Previous
studies show that managers engage in earnings management to meet or beat
analyst forecasts, avoid losses and maintain earnings growth targets
(Burgstahler and Dichev, 1997; Degeorge et al., 1999). In addition, a
number of studies have documented that executives manipulate earnings
around firm-specific events such as initial public offerings (Teoh et
al., 1998a), seasoned equity offerings (Teoh et al., 1998b), violation
of debt covenants (DeFond and Jiambalvo, 1994; Dichev and Skinner, 2002)
and acquisition of other firms (Louis, 2004). However, there is little
evidence regarding whether and how firms manipulate earnings when
entering the bond market. The purpose of this paper is to investigate
whether firms manage earnings through either income increasing
discretionary (or abnormal) accruals or real operating decisions during
the period in which the debt is issued.
The incentives for earnings management arise because pricing and
non-pricing terms such as amount, maturity, collateral and covenants in
explicit contracts written between lenders and corporate borrowers are
affected by reported earnings. As Leftwich (1983) points out, the
conflict of interest between bondholders and shareholders actually is a
negative-sum game because it affects the firm's financing,
production and investment decisions. Therefore, managers of the firm,
who act in the best interest of shareholders, have incentives to
mitigate agency cost arising from debt contracting to maximize the firm
value.
Even though the literature has long recognized that managers can
take accounting actions or real economic actions to meet earnings
benchmarks or certain earnings threshold, real earnings management has
not received as much attention in the archival literature relative to
the attention given to accrual-based earnings management. Recent studies
(Graham et al., 2005; Gunny, 2005; Roychowdhury, 2006; Cohen et al.,
2008; Cohen and Zarowin, 2008; Chen et al., 2008; Kim and Sohn, 2009)
have documented that firms not only use accruals to manipulate earnings
but also conduct earnings management through real activities. Therefore,
it is important to examine whether management of the firms that issue
debt also engage in real economic actions to window-dress financial
reports when entering the bond market.
In this study, we use discretionary total accruals (DTACC) as a
proxy for accrual-based earnings management (Jones, 1991; Kothari et
al., 2004; Teoh et al., 1998a; Teoh et al., 1998b) and abnormal cash
flows from operations (CFO), abnormal discretionary expenses and
abnormal production costs as proxies for real earnings management
(Roychowdhury, 2006; Cohen et al., 2008). We perform cross-sectional
regression for every industry and year to estimate the discretionary
total accruals, abnormal CFO, abnormal discretionary expenses and
abnormal production costs. Then, we test whether these dependent
variables (discretionary total accruals, abnormal CFO, abnormal
discretionary expenses, abnormal production costs) of issuing firms are
higher or lower during the year of the issuance compared to performance
matched firms after controlling for other influencing factors.
Using data on a sample of public bond issuers from 1992 through
2002, we find evidence that discretionary total accruals increase prior
to the issuance and decline afterwards. In addition, we also find some
evidence that the sample firms are engaging in real earnings management
but not as strong as accruals. These results suggest that bond issuers
prefer to manipulate earnings via accruals compared to real activities.
This paper contributes to the literature on earnings management in
several ways. We extend ongoing research investigating the motivations,
characteristics and consequences of earnings management. Existing
earnings management research predominantly examines incentives of
managers related to stock market performance. For example, Teoh et al.
(1998) investigate managers' motivations to issue stocks at a
higher price to the market. In addition, Cheng and Warfield (2005) test
managers' incentives to manipulate earnings when their equity
incentives, such as stock-based compensation and stock ownership, are
relatively high. We contribute to the literature by documenting
incentives for managers related to the debt market. In addition, we use
both accrual-based and real activities to measure earnings management.
It is important to test both of these measures since recent evidence
suggests that managers use both accruals and real operating decisions to
manage earnings. Most of the prior studies on earnings management
investigate only discretionary accruals; however, as Graham et al.
(2005) pointed out, managers engage in real earnings management more
frequently than accrual-based manipulation.
The remainder of this paper is organized as follows. In section 2,
we develop hypotheses. Section 3 describes details of the sample
selection procedures, and section 4 presents the research design.
Empirical results are presented in section 5. Finally, we conclude in
section 6.
HYPOTHESES DEVELOPMENT
Prior empirical studies on earnings management show that managers
manipulate earnings prior to certain economic events such as initial
public offering (IPO), seasoned equity offering (SEO), violation of debt
covenants and acquisition of other firms. For example, Teoh et al.
(1998b) find that firms that conduct seasoned equity offerings manage
earnings through accounting accruals and that subsequent earnings and
stock return underperformance are correlated with the level of earnings
management during the equity issue period. Teoh et al. (1998a) also find
evidence showing earnings management before initial public offerings.
Louis (2004) finds strong evidence suggesting that acquiring firms
overstate their earnings in the quarter preceding a stock swap
announcement. In addition, Dichev and Skinner (2002) test a "debt
covenant" hypothesis--the idea that managers make accounting
choices to reduce the likelihood that their firms will violate
accounting-based debt covenants.
Leftwich (1983) pointed out that the conflict of interest between
debtholders and shareholders actually is a negative-sum game because it
affects the firm's financing, production and investment decisions,
and shareholders will support restrictions on such decisions when the
restrictions "lead to the highest firm value." Thus, since
managers are acting in the best interest of shareholders, managers have
incentive to lower the agency cost of debt as much as possible when
negotiating debt contract. In addition, if the firm performance is not
good prior to debt issuance, the firm might not be able to issue the
amount of funds it needs. Performance of the firm not only affects the
amount of funds it borrows but also affects various contract terms such
as maturity or collateral. Surprisingly, most earnings management
studies on debt focus on detecting the violation of debt covenants. To
our knowledge, there is no study that tests the association between
earnings management and the bond market.
Following previous earnings management studies (Teoh et al., 1998a,
1998b; Louis, 2004; Cheng and Warfield, 2005) and given
shareholders' incentive to mitigate agency cost of debt, we examine
whether the issuing firms exhibit unexpected high levels of
discretionary accruals compared to their performance matched non-issuing
firms. Thus, our first hypothesis related to accrual-based earnings
management is as follows (in the alternative form):
H1: Firms that issue bonds are likely to manipulate earnings
through income increasing accruals compared to non-issuing firms,
ceteris paribus, during the year of bond issuance.
Managers not only use accruals as an earnings management tool but
also engage in real operating decisions to manage earnings. Graham et
al. (2006) surveyed financial executives from a large number of public
U.S. firms and find that financial executives are willing to make small
or moderate economic sacrifices in representing the economic value of
the firm in order to obtain credibility in the market. They also find
that real earnings management is preferred to accrual-based earnings
management, which contradicts researchers' assumptions about the
higher likelihood of earnings management via accruals. Dechow and
Skinner (2000) posit that real earnings management methods used by
managers are (1) acceleration of sales, (2) alterations in shipment
schedules and (3) delay of research and development (R&D) and
maintenance expenditures. Other evidence also indicates that managers
engage in real transactions to manipulate earnings. Dechow and Sloan
(1991) examine whether CEOs in the final years of their tenure manage
discretionary investment to enhance short-term performance and find
evidence that the growth in R&D expenditures is reduced over this
horizon, but the reduction in R&D expenditures is mitigated through
CEO stock ownership. Roychowdhury (2006) provides evidence that firms
reporting small positive profits and small positive forecast errors
manage earnings through real activities. Therefore, merely testing
accrual-based earnings management is sufficient.
The advantage of using real earnings management instead of
accrual-based earnings management is that investors are able to
second-guess the firm's accounting policies; however, they cannot
readily challenge real economic actions that are taken in the ordinary
course of business. Thus, while it is more difficult to manage earning
via real actions rather than accruals, executives do use real earnings
management as documented in prior literature (Dechow and Skinner, 2000;
Graham et al., 2005). Therefore, to provide a more complete study of the
earnings management during the issuance of bonds, we also examine real
earnings management activities over the sample period.
Following Roychowdhury (2006), we use abnormal cash flow from
operations (CFO), abnormal discretionary expenses and abnormal
production costs as proxy measures for real earnings management (sales
manipulation, reduction of discretionary expenditure, and
overproduction, respectively). Thus, the next hypotheses regarding the
detection of real earnings management are as follows (in the alternative
form):
[H.sub.2]: After controlling for sales levels, firms that issue
bonds are likely to exhibit low abnormal CFO, ceteris paribus, during
the year of debt issue.
[H.sub.3]: After controlling for sales levels, firms that issue
bonds are likely to exhibit low abnormal discretionary expenses, ceteris
paribus, during the year of debt issue.
H4: After controlling for sales levels, firms that issue bonds are
likely to exhibit high abnormal production costs, ceteris paribus,
during the year of debt
issue.
SAMPLE SELECTION
Our initial sample consists of U.S. public companies that issued
bonds between January 1992 and December 2002. We use Securities Data
Company (SDC) New Issues database to obtain information on bond issuers.
For firms with multiple issuances in a given year, we only include the
largest offering amount to avoid overlapping data following Khurana and
Raman (2003). We exclude firms in financial industries because these
firms are closely regulated and have unique disclosure requirements
which make it difficult to manage earnings. In addition, issuing bonds
is more like a day-to-day operation rather than a financing activity for
financial firms.
We restrict the sample to all non-financial firms with available
data and require at least ten observations in each two-digit SIC
industry classification per year. For inclusion in the final sample, we
also require sufficient data to compute accrual-based measures (i.e.,
discretionary accruals) and real earnings management proxies (i.e.,
abnormal CFO, abnormal discretionary expenses, and abnormal production
costs). These requirements result in 5,696 bond issues for 420 firms
over the period between 1992 and 2002.
RESEARCH DESIGN
Accrual-Based Measure
Different measures have been used in prior studies to proxy for
earnings management. One of the most common metrics used to detect
earnings management is the magnitude of discretionary (or abnormal or
unexpected) accruals, which measures the discretion used by managers to
achieve their financial reporting goals. Following previous research
(Jones, 1991; Sloan, 1996; Teoh et al., 1998), we run the following
regression for a given year using non-issuers in the same two-digit SIC
code as the issuer in order to estimate normal accruals:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)
where [TAC.sub.i,t] is total accruals in year t for firm i,
[DELTA][REV.sub.i,t] is change in sales revenue from year t-1 to year t
for firm i, [DELTA][REC.sub.it] is change in accounts receivable from
year t-1 to year t for firm i, [PPE.sub.i,t] is property, plant, and
equipment in year t for firm i, [TA.sub.i,t-1] is total assets in year
t-1 for firm i, and [[epsilon].sub.i,t] is the error term in year to for
firm i. Specifically, total accruals is the change in noncash current
assets minus the change in operating current liabilities minus
depreciation, amortization, and depletion:
[TAC.sub.t] = [DELTA] [current assets (#4)--cash (#1)] - [DELTA]
[current liabilities (#5)--current maturity of long term debt (#44)] -
[DEP.sub.t] (#14) (2)
where numbers in parentheses are Compustat item numbers.
Discretionary total accruals (DTACC) are defined as the difference
between realized total accruals and normal accruals.
Real Activities Measure
We rely on prior studies to develop proxies for real earnings
management. Following Roychowdhury (2006) and Cohen et al. (2008), we
focus on three manipulation methods and their impact on the abnormal
levels of CFO, discretionary expenses and production costs. Sales
manipulation occurs when managers attempt to temporarily increase sales
through temporary price discounts or lenient credit terms. The
additional sales will boost current earnings but will result in lower
cash flows given sales level. Discretionary expense includes advertising
expense, selling, general and administrative (SG&A) expense, and
research and development (R&D) expense. Reducing such expenses will
immediately boost current earnings. Overproduction occurs when managers
produce more units so that fixed overhead costs could be spread over a
large number of units to lower fixed cost per unit.
Following Dechow et al. (1998), Roychowdhury (2006), and Cohen et
al. (2008), we estimate abnormal cash flows from operations, abnormal
discretionary expenses, and abnormal production costs by running the
following cross-sectional regression for every industry and year:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)
[DISEXP.sub.t]/[TA.sub.t-1] = [[beta].sub.0] + [[beta].sub.1]
(1/[TA.sub.t-1]) + [[beta].sub.2] ([REV.sub.t]/[TA.sub.t- 1]) +
[[beta].sub.3] [DELTA] [REV.sub.t-1]/[TA.SUB.T-1]) + [[epsilon].sub.t]
(4)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)
where [TA.sub.t-1] is total assets in year t-1, [REV.sub.t] is
sales revenue in year t, and [DELTA][REV.sub.t] is change in sales
revenue from year t-1 to year t. Specifically, CFO is obtained from
Compustat (#308), discretionary expense is computed as research and
development (R&D, #46) expense plus advertising (#45) expense plus
selling, general and administrative (SG&A, #189) expense. Production
cost is computed as cost of goods sold (COGS, #44) plus the change in
inventory (#3) level. For every firm year, abnormal CFO is the actual
CFO minus the normal CFO, abnormal discretionary expense is the actual
discretionary expense minus the normal discretionary expense and
abnormal production cost is the actual production cost minus the normal
production cost.
Cross-Sectional Regression Analysis
We use the multiple regression model to estimate the impact of bond
issues on abnormal (or discretionary) accruals. Specifically, we employ
the following regression to test the first hypothesis:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (6)
where [ISSUE.sub.i,t] is an indicator variable set equal to one if
the firm issued a bond in year t and zero otherwise, [SIZE.sub.it] is
natural logarithm of market value (in million dollars) at the end of the
fiscal year t, [MTB.sub.i,t] is market-to-book ratio at the end of the
fiscal year t, [ROA.sub.i,t] is proxy for firm performance at the end of
the fiscal year t and [LEV.sub.i,t] is leverage of the firm at the end
of fiscal year t. Specifically, market value is computed as price at the
end of fiscal year (#199) multiplied by common shares outstanding (#25),
market-to-book ratio is computed as market value of equity deflated by
book value of equity (#60), [ROA.sub.i,t] is computed as net income
before extraordinary items (#18) scaled by total assets (#6) and
[LEV.sub.i,t] is computed as sum of long-term debt (#9) and debt in
current liabilities (#34) divided by total assets (#6).
In testing all of our hypotheses, we use a matched-sample design
where each firm that issued bonds is matched to a control firm that did
not issue bonds. Following Kothari et al. (2004), we use two-digit SIC
codes and ROA in the same fiscal year to identify potential control
firms.
We use discretionary total accruals as a dependent variable rather
than using current abnormal accruals because all of the control
variables are related not only to the current portion but also to the
non-current portion. In addition, Richardson et al. (2005) find that
estimation error of accruals is significant for both current and
non-current assets and liabilities. Therefore, total accruals should
provide a more comprehensive measure of abnormal accruals.
The indicator variable, denoted as ISSUE, is set equal to one if a
firm issued bonds and zero for performance matched control samples. We
expect that this ISSUE variable, which is the main variable of interest,
will be significantly positive for bond issuers due to managers'
aggressive accounting manipulations prior to bond issuance. We use a
series of control variables based on the evidence in prior studies: firm
size, market-to-book ratio, firm performance and leverage. We use
natural log of market value denoted as SIZE to proxy for the size of the
firm. Positive accounting theory suggests that managers tend to manage
earnings to decrease political costs. Prior studies (Cheng and Warfield,
2005; Collins et al., 2007) use firm size to proxy political costs. In
addition, Kim et al. (2003) examine the relation of corporate earnings
management to firm size. They find that small-sized firms engage in more
earnings management to avoid reporting losses than do large-sized firms.
Warfield et al. (1995) indicate that riskier and high-growth firms have
more abnormal accruals. We use the ratio of the market value of common
equity to the book value of common equity, denoted as MTB, to proxy for
the growth potential. Return on assets (ROA) and leverage ratio (LEV)
are included to control for any potential impact of firm performance and
debt possession.
Next, we examine the relation between the real earnings management
and the issuance of bonds (second, third, and fourth hypotheses) by
estimating the following regression:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (7)
where [Y.sub.i,t] is either abnormal CFO, abnormal discretionary
expenses, or abnormal production costs. We expect the abnormal CFO to be
significantly negative, abnormal discretionary expenses to be
significantly negative, and abnormal production costs to be
significantly positive during the year of the bond issuance.
RESULTS
Descriptive Statistics
Table 1, Panel A shows the distribution of bond issues by year. The
table indicates that the frequency of bond issues tends to be stable
over time. Panel B of Table 1 reports the descriptive statistics for the
sample of 5,696 firm-year observations from 1992 to 2002. The average
sample firm has a market-to-book ratio (MTB; (Compustat #25 * Compustat
#199)/Compustat #60) of 2.800, return on asset (ROA; Compustat
#18/Compustat #6) of 0.0375 and leverage (LEV; (Compustat #9 + Compustat
#34)/ Compustat #6) of 0.343.
Panel C of Table 1 reports the Pearson correlation coefficients
among the variables. SIZE is significantly positively correlated with
MTB and ROA. This positive relationship indicates that (at least in our
sample) large firms have higher market-to-book ratio and are more
profitable. LEV is significantly negatively correlated with SIZE, MTB
and ROA. This negative relationship indicates that firms with high
leverage tend to be small, have a low market-to-book ratio and be less
profitable.
Multiple Regression Results
The results for testing accrual-based earnings management are
reported in Table 2. Table 2 presents the results from estimating
Equation (6) for a sample of bond issuers, where the sample is combined
with performance matched firms. For each sample and control firm, we
estimate cross-sectional regressions of discretionary total accruals
(DTACC) on ISSUE, the main variable of interest, and a series of control
variables based on the evidence in prior studies: firm size,
market-to-book ratio, return on assets and leverage ratio. As mentioned
earlier in the paper, the control firms are matched based on same
industry (two-digit SIC codes) and similar performance (ROA) following
Kothari et al. (2004). The results show that bond issuers have
significantly higher levels of discretionary total accruals compared to
non-issuers (controlled sample) during the year of issuance. Consistent
with the first hypothesis, the coefficient on bond issuers is positive
(0.014) and significant at the 5% level (t = 2.18). This result suggests
that bond issuers do engage in earnings management through income
increasing accruals compared to non-issuers.
The results for testing real earnings management are reported in
Table 3, which presents the results from estimating Equation (7). Column
1 of Table 3 provides evidence on Hypothesis 2. When the dependent
variable in regression (7) is abnormal CFO, the coefficient on ISSUE for
bond issuers is negative (-0.003) and significant at the 10% level (t =
-1.74). This result suggests that bond issuers exhibit lower levels of
abnormal CFO compared to performance matched firms which is consistent
with the hypothesis. This result also indicates that firms issuing bonds
manipulate earnings via real actions, such as using price discounts or
lenient credit terms to boost sales.
In the second column of Table 3, abnormal discretionary expense is
used as a dependent variable. However, we do not find any evidence that
the sample firms that issue bonds are reducing discretionary expense
during the issuance year. The coefficient on ISSUE for the sample is
positive (0.0327) but not statistically significant. This result
indicates that bond issuers are not using discretionary expense as an
earnings management tool to manipulate their earnings. We conjecture
that bond issuers are not using discretionary expense because it is
easily detected by creditors.
Column 3 of Table 3 provides evidence on Hypothesis 4. Bond issuers
exhibit high levels of abnormal production costs. The coefficient on
ISSUE is positive (0.021) and significant at the 5% level. This result
indicates that bond issuers engage in earnings management through
overproduction in order to report lower cost of goods sold (COGS).
Sensitivity Analysis
To further examine whether the results are indeed driven by income
increasing accruals and real actions, we conduct additional tests that
examine the pattern surrounding the event period. Specifically, we use
changes in dependent variables before and after the issuance by running
the following regressions:
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (8)
[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (9)
where [DELTA] [DTACC.sub.ix] is change in discretionary total
accruals, [DELTA][Y.sub.i,x] is either change in abnormal CFO, change in
abnormal discretionary expenses, or change in abnormal production costs,
and x indicates the change in time period of either (t-1 ~ t) or (t ~
t+1).
Table 4 presents the results for the changes in discretionary
accruals prior to and past bond issuance. The coefficient on ISSUE is
positive (0.013) and significant at the 1% level (t = 3.13) for the
first column, which indicates that discretionary accruals of bond
issuers increase by 0.013 on average from the year prior to issuance to
the year of issuance. The second column presents the changes in
discretionary total accruals from the year of issuance to the past year.
The coefficient on ISSUE is negative (-0.042) and significant at the 1%
level (t = -2.96). When combined with Table 2, these results show that
bond issuers manipulate earnings through income increasing accruals and
then accruals reverse after the issuance declining to the normal level.
This is consistent with our hypothesis that firms issuing bonds have
incentives to manipulate earnings via accruals similar to other firm
specific events such as IPO and SEO.
Panel A, B, and C of Table 5 present the results of changes in real
earnings management proxies surrounding the issue year. In Table 5 Panel
A, abnormal CFO of bond issuers shows a decreasing pattern prior to debt
issuance followed by an increase after the year of issuance. The
coefficient on ISSUE is negative (-0.004) for the period from t-1 to t
and significant at the 10% level. In addition, the coefficient on ISSUE
is positive (0.016) for the period from t to t+1 and significant at the
10% level.
Table 5 Panel B presents the results of changes in abnormal
discretionary expense. The results show that the level of abnormal
discretionary expense decreases by (0.029) on average and significant at
the 10% level before the bond issue.
Table 5 Panel C presents the results of changes in abnormal
production cost. The results show that abnormal production cost of bond
issuers increases prior to the issuance, which indicates that firms
increase their production level to report lower COGS.
Overall, the results suggest that firms that issue bonds use both
accrual-based and real activities to manipulate earnings. However, they
tend to manage earnings through income increasing accruals more heavily
than compared to taking real actions.
CONCLUSION
In this paper, we investigate whether firms that enter the bond
market manipulate earnings similar to firms entering the stock market
(Teoh et al., 1998a; Teoh et al., 1998b). In addition, we also examine
how these firms engage in earnings management (i.e., accrual-based
earnings management versus real earnings management). Based on all
firm-years with available data over the 1992-2002 period, we find
discretionary accruals of bond issuers are significantly higher than
non-issuers during the year of issuance. Further analyses show that bond
issuers increase their accruals prior to the issuance and then decrease
their accruals subsequent to the issue year. In addition, we find some
evidence that bond issuers engage in real earnings management. However,
the findings suggest that among three methods of real earnings
management, sales manipulation is much more prominent compared to other
real earnings manipulation methods. Overall, the results provide strong
evidence that bond issuers use both accrual-based and real actions to
manipulate earnings.
REFERENCES
Burgstahler, D., Dichev, I., 1997. Earnings management to avoid
earnings decreases and losses. Journal of Accounting and Economics 24,
99-126.
Chen, J., Rees, L., Sivaramakrishnan, S., 2008. On the use of
accounting vs. real earnings management to meet earnings expectations--A
market analysis. Working paper, SSRN.
Cheng, Q., Warfield, T., 2005. Equity incentives and earnings
management. The Accounting Review 80, 441-476.
Cohen, D., Dey, A., Lys, T., 2008. Real and accrual-based earnings
management in the pre- and post-Sarbanes-Oxley periods. The Accounting
Review 83, 757-787.
Cohen, D., Zarowin, P., 2008. Accrual-based and real earnings
management activities around seasoned equity offerings. Working paper,
SSRN.
Collins, D., Gong, G., Li, H., 2007. Corporate governance and
backdating of stock option exercise dates. Working paper, SSRN.
Dechow, P., Kothari, S., Watts, R., 1998. The relation between
earnings and cash flows. Journal of Accounting and Economics 25,
133-168.
Dechow, P., Skinner, D., 2000. Earnings management: reconciling the
views of accounting academics, practitioners and regulators, Accounting
Horizons 14, 235--250.
Dechow, P., Sloan, R., 1991. Executive incentives and the horizon
problem: An empirical investigation. Journal of Accounting and Economics
14, 51-89.
DeFond, M., Jiambalvo, J., 1994. Debt covenant violation and
manipulation of accruals. Journal of Accounting and Economics 17,
145-176.
Degeorge, F., Patel, J., Zeckhauser, R., 1999. Earnings management
to exceed thresholds. Journal of Business 72, 1-33.
Dichev, I., Skinner, D., 2002. Large-sample evidence on the debt
covenant hypothesis. Journal of Accounting Research 40, 1091-1123.
Graham, J., Harvey, C., Rajgopal, S., 2005. The economic
implications of corporate financial reporting. Journal of Accounting and
Economics 40, 3-73.
Gunny, K., 2006. What are the consequences of real earnings
management? Working paper, University of Colorado.
Jones, J., 1991. Earnings management during import relief
investigation. Journal of Accounting Research 29, 193-228.
Khurana, I., Raman, K., 2003. Are fundamentals priced in the bond
market? Contemporary Accounting Research 20, 465-494.
Kim, J., Chung, R., Firth, M., 2003. Auditor conservatism,
asymmetric monitoring and earnings management. Contemporary Accounting
Research 20, 29-48.
Kim, J., Sohn, B., 2009. Real versus accrual-based earnings
management and implied cost of equity capital. Working paper, SSRN.
Kothari, S., Leone, A., Wasley, C., 2004. Performance matched
discretionary accrual measures. Journal of Accounting and Economics 39,
163-197.
Leftwich, R., 1983. Accounting information in private markets:
Evidence from private lending agreements. The Accounting Review 58,
23-42.
Louis, H., 2004. Earnings management and the market performance of
acquiring firms. Journal of Financial Economics 74, 121-148.
Roychowdhury, S., 2006. Earnings management through real activities
manipulation. Journal of Accounting and Economics 42, 335-370.
Sloan, R., 1996. Do stock prices fully reflect information in
accruals and cash flows about future earnings? The Accounting Review 77,
289-315.
Teoh, S., Welch, I., Wong, T., 1998a. Earnings management and the
long-run underperformance of initial public offerings. Journal of
Finance 53, 1935-1974.
Teoh, S., Welch, I., Wong, T., 1998b. Earnings management and the
long-run underperformance of seasoned equity offerings. Journal of
Financial Economics 50, 63-100.
Warfield, T., Wild, J., Wild, K., 1995. Managerial ownership,
accounting choices, and informativeness of earnings. Journal of
Accounting and Economics 20, 61-91.
Sangshin (Sam) Pae, Arkansas State University
Tina Quinn, Arkansas State University
Table 1. Characteristics of Bond Issuance between 1992-2002
This table provides the characteristics of the sample. Panel A
presents event-year distribution of a sample of bond issues. Panel B
presents descriptive statistics. Panel C provides the value of
correlation between each of the variables used in subsequent tests.
To be included in this table, a firm-year observation must be
accompanied by sufficient data to compute the variables displayed
below. Therefore, the statistics for all variables are based on
5,696 firm-year observations. Firm-year observations are drawn from
the period between 1992 and 2002. ***, **, and * denote two-tailed
significance at the 0.01, 0.05, and 0.10 levels, respectively.
Variable definitions are as follows: SIZE=natural log of market
value at the end of the fiscal year. MTB=market-to-book ratio;
computed as market value of equity divided by total book value of
equity at the end of the fiscal year. ROA=return on assets; computed
as net income before extraordinary items scaled by total assets at
the end of the fiscal year. LEV=leverage ratio; computed as long-
term debt divided by total assets at the end of the fiscal year.
Panel A. Sample Distribution by Year
Bond Issues
Year Freq. % Cumul.
1992 269 4.72 269
1993 307 5.39 576
1994 400 7.02 976
1995 415 7.29 1,391
1996 497 8.73 1,888
1997 629 11.04 2,517
1998 600 10.53 3,117
1999 605 10.62 3,722
2000 623 10.94 4,345
2001 681 11.96 5,026
2002 670 11.76 5,696
Panel B. Descriptive Statistics
Bond Issuer
Variable Mean Median Std. Dev.
SIZE 7.8459 7.9119 1.7741
MTB 2.8006 2.0461 16.7396
ROA 0.0375 0.0392 0.0785
LEV 0.3428 0.3350 0.1574
Panel C. Pearson Correlations Matrix between Independent Variables
SIZE MTB ROA LEV
SIZE 1
MTB 0.1128 *** 1
ROA 0.1410 *** 0.0483 *** 1
LEV -0.0521 *** -0.0452 *** -0.4438 *** 1
Table 2. Levels of Discretionary Total Accruals
This table provides the results of multiple regression with the
dependent variable Discretionary Total Accruals (DTACC). To be
included in this table, a firm-year observation must be accompanied
by sufficient data to compute the variables displayed below.
Therefore, the statistics for all variables are based on 11,392
firm-year observations (bond issuer and performance-matched sample).
Firm-year observations are drawn from the period between 1992 and
2002. ***, **, and * denote two-tailed significance at the 0.01,
0.05, and 0.10 levels, respectively. Variable definitions are as
follows: SIZE=natural log of market value at the end of the fiscal
year. MTB=market-to-book ratio; computed as market value of equity
divided by total book value of equity at the end of the fiscal year.
ROA=return on assets; computed as net income before extraordinary
items scaled by total assets at the end of the fiscal year.
LEV=leverage ratio; computed as long-term debt divided by total
assets at the end of the fiscal year.
[DTACC.sub.i,t] = [[gamma].sub.0] + [[gamma].sub.1] [ISSUE.sub.i,t]
+ [[gamma].sub.2][SIZE.sub.i,t] + [[gamma].sub.3][MTB.sub.i,t] +
[[gamma].sub.4][ROA.sub.i,t] + [[gamma].sub.5][LEV.sub.i,t] +
[[xi].sub.i,t]
DTACC
Coefficient t-stat
Intercept -0.1049 -2.15 **
ISSUE 0.0135 2.18 **
SIZE -0.0006 -1.69 *
MTB 0.0012 2.26 **
ROA 0.0263 2.47 **
LEV 0.0687 2.12 **
Obs. 11,392
Adj. [R.sup.2] 0.03
Table 3. Levels of Real Earnings Management Proxies
This table provides the results of multiple regression with the
dependent variable abnormal CFO, abnormal discretionary expense, and
abnormal production cost. To be included in this table, a firm-year
observation must be accompanied by sufficient data to compute the
variables displayed below. Therefore, the statistics for all
variables are based on 11,392 firm-year observations (bond issuer
and performance-matched sample). Firm-year observations are drawn
from the period between 1992 and 2002. ***, **, and * denote two-
tailed significance at the 0.01, 0.05, and 0.10 levels,
respectively. Variable definitions are as follows: SIZE=natural log
of market value at the end of the fiscal year. MTB=market-to-book
ratio; computed as market value of equity divided by total book
value of equity at the end of the fiscal year. ROA=return on assets;
computed as net income before extraordinary items scaled by total
assets at the end of the fiscal year. LEV=leverage ratio; computed
as long-term debt divided by total assets at the end of the fiscal
year.
[Y.sub.i,t] = [[gamma].sub.0] + [[gamma].sub.1][ISSUE.sub.i,t] +
[[gamma].sub.2][SIZE.sub.i,t] + [[gamma].sub.3][MTB.sub.i,t] +
[[gamma].sub.4][ROA.sub.i,t] + [[gamma].sub.5][LEV.sub.i,t] +
[[xi].sub.i,t]
Abnormal CFO Abnormal Disc. Exp.
Coefficient t-stat Coefficient t-stat
Intercept 0.0277 1.16 -0.0232 -3.62 ***
ISSUE -0.0030 -1.74 * 0.0327 1.02
SIZE 0.0051 1.64 -0.0300 -1.76 *
MTB 0.0009 2.08 *** -0.0056 -1.79 *
ROA 0.0289 3.32 *** 0.0881 1.75 *
LEV -0.0701 -2.02 ** 0.0151 1.69 *
Obs. 11,392 11,392
Adj. [R.sup.2] 0.02 0.03
Abnormal Prod. Cost
Coefficient t-stat
Intercept -0.0118 -1.77 *
ISSUE 0.0208 2.19 **
SIZE 0.0040 2.08 **
MTB -0.0020 -2.47 ***
ROA -0.0617 -4.64 ***
LEV 0.0087 1.72 *
Obs. 11,392
Adj. [R.sup.2] 0.03
Table 4. Changes in Discretionary Total Accruals
This table provides the results of multiple regression with the
dependent variable Changes in Discretionary Total Accruals (?DTACC)
during pre-and post-issue. To be included in this table, a firm-
year observation must be accompanied by sufficient data to compute
the variables displayed below. Therefore, the statistics for all
variables are based on 11,392 firm-year observations (bond issuer
and performance-matched sample). Firm-year observations are drawn
from the period between 1992 and 2002. ***, **, and * denote two-
tailed significance at the 0.01, 0.05, and 0.10 levels,
respectively. Variable definitions are as follows: SIZE=natural log
of market value at the end of the fiscal year. MTB=market-to-book
ratio; computed as market value of equity divided by total book
value of equity at the end of the fiscal year. ROA=return on assets;
computed as net income before extraordinary items scaled by total
assets at the end of the fiscal year. LEV=leverage ratio; computed
as long-term debt divided by total assets at the end of the fiscal
year.
[[DELTA][DTACC.sub.i,t] = [[gamma].sub.0] +
[[gamma].sub.1][ISSUE.sub.i,t] + [[gamma].sub.2][SIZE.sub.i,t] +
[[gamma].sub.4][MTB.sub.i,t] + [[gamma].sub.4][ROA.sub.i,t] +
[[gamma].sub.5][LEV].sub.i,t] + [[xi].sub.i,t]
t-1 to t t to t+1
Coefficient t-stat Coefficient t-stat
Intercept 0.1054 2.09 ** 0.0252 6.62 ***
ISSUE 0.0126 3.13 *** -0.0417 -2.96 ***
SIZE -0.0015 -2.49 ** 0.0139 2.86 ***
MTB -0.0032 -3 70 *** 0.0008 2.26 **
ROA -0.0377 -2.09 ** -0.0388 -2 84 ***
LEV -0.1457 -2.03 ** 0.0021 1.66 *
Obs. 11,392 11,392
Adj. [R.sup.2] 0.13 0.10
Table 5. Changes in Real Earnings Management Proxies
This table provides the results of multiple regression with the
dependent variable Changes in abnormal CFO, abnormal discretionary
expense, and abnormal production cost during pre-and post-issue. To
be included in this table, a firm-year observation must be
accompanied by sufficient data to compute the variables displayed
below. Therefore, the statistics for all variables are based on
11,392 firm-year observations (bond issuer and performance-matched
sample). Firm-year observations are drawn from the period between
1992 and 2002. ***, **, and * denote two-tailed significance at the
0.01, 0.05, and 0.10 levels, respectively. Variable definitions are
as follows: SIZE=natural log of market value at the end of the
fiscal year. MTB=market-to-book ratio; computed as market value of
equity divided by total book value of equity at the end of the
fiscal year. ROA=return on assets; computed as net income before
extraordinary items scaled by total assets at the end of the fiscal
year. LEV=leverage ratio; computed as long-term debt divided by
total assets at the end of the fiscal year.
[DELTA][Y.sub.i,t] = [[gamma].sub.0] +
[[gamma].sub.1][ISSUE.sub.i,t] + [[gamma].sub.2][SIZE.sub.i,t] +
[[gamma].sub.3][MTB.sub.i,t] + [[gamma].sub.4][ROA.sub.i,t] +
[[gamma].sub.5][LEV.sub.i,t] + [[xi].sub.i,t]
Panel A. Abnormal CFO
t-1 to t t to t+1
Coefficient t-stat Coefficient t-stat
Intercept -0.0443 -2.23 ** 0.0185 3.39 ***
ISSUE -0.0042 -1.71 * 0.0162 1.91 *
SIZE 0.0090 2.50 ** 0.0017 1.76 *
MTB -0.0037 -4.35 *** -0.0015 -2.49 **
ROA 0.1219 1.99 ** 0.0519 1.78 *
LEV -0.0268 -1.82 * -0.0770 -2.19 **
Obs. 11,392 11,392
Adj. [R.sub.2] 0.02 0.03
Panel B. Abnormal Discretionary Expense
t-1 to t t to t+1
Coefficient t-stat Coefficient t-stat
Intercept 0.0516 2.41 ** -0.0535 -1.47
ISSUE -0.0285 -1.80 * -0.026 -1.09
SIZE 0.0310 1.72 * 0.1364 0.08 *
MTB 0.0066 1.76 * -0.0199 -1.76 *
ROA -0.0276 -1.65 * 0.0427 2.19 **
LEV -0.0110 -1.90 * 0.0977 2.49 **
Obs. 11,392 11,392
Adj. [R.sub.2] 0.04 0.04
Panel C. Abnormal Production Cost
t-1 to t t to t+1
Coefficient t-stat Coefficient t-stat
Intercept -0.0715 -1.31 -0.0822 -0.86
ISSUE 0.0106 1.79 * 0.0384 0.71
SIZE -0.0014 -1.73 * 0.0159 2.34 **
MTB 0.0005 1.87 * -0.0010 -1.91 *
ROA 0.0378 1.90 * -0.0374 -2.19 **
LEV 0.1565 2.79 *** -0.1239 -2.08 **
Obs. 11,392 11,392
Adj. [R.sub.2] 0.04 0.03
