CHAPTER FOUR RESEARCH DESIGN
FIGURE 1 Event Periods Analysed
4.3 Measures and models for hypothesis testing
4.3.9 Hypothesis testing for H2c(i-iii) – substitution effect after the introduction of AASB
H2c(i-iii) predict that firms that report DO in the post-IFRS period are less likely to manage discretionary accruals and real activities to increase core earnings. Following the introduction of AASB 5 in 2005 that increased the scope of reporting items as DO, firms may engage in classification shifting using DO to increase core earnings as predicted in H1c. In that case, those firms are expected to engage in less accrual-based
earnings management and real earnings manipulation as these alternative methods are more costly than classification shifting. While H1c was tested by observing behaviour in post-IFRS period only, because there was an extremely low incidence of DO in the pre-IFRS period, H2c(i-iii) are tested using a pre-post sample, because there is scope to meaningfully measure differences in alternative earnings management behaviour after the reform. To test H2c(i-iii), I employ a difference-in-differences design similar to that for H2b(i-iii), within which measures of DO reporting are substituted for measures of AI reporting. I examine the differences in the accrual-based and real earnings management behaviour between firms that report DO in the post-IFRS period (treatment firms), and firms of similar characteristics that do not report DO post-IFRS (control firms). I therefore estimate the following model on a PSM sample of treatment and control firms (as per matching method for H1c) for the period 2002 to 2009:
UE_CEt= θ0+ θ1TREAT_DOt+ θ2DAt+ θ3SLSMANt+ θ4DISXMANt+ θ5IFRSt +
θ6TREAT_DOt*IFRSt+ θ7IFRSt*DAt+ θ8IFRSt*SLSMANt +
θ9IFRSt*DISXMANt+ θ10TREAT_DOt*DAt+ θ11TREAT_DOt*IFRSt*DAt +
θ12TREAT_DOt*SLSMANt+ θ13TREAT_DOt*IFRSt*SLSMANt +
θ14TREAT_DOt*DISXMANt+ θ15TREAT_DOt*IFRSt*DISXMANt +
Controls + εt (13)
Where the variables are as described below:
UE_CEt = Unexpected core earnings calculated as in Equation 3(a)
TREAT_DOt = An indicator variable = 1 if a firm operates in both the
pre-IFRS (30 June 2002-30 December 2005) and post- IFRS (31 December 2005-31 December 2009) periods, and reports DO in the post-IFRS period regardless of their pre-IFRS behaviour, and 0 if a firm reports no DO in post-IFRS irrespective of pre-IFRS behaviour.
DAt = Discretionary accruals in year t, calculated as residuals
from Equation (8).
SLSMANt = Sales manipulation (abnormal cash from operations),
calculated as the negative of residuals from Equation (9).
DISXMANi,t = Abnormal discretionary expenditure, calculated as the
negative of residuals from Equation (10).
IFRS1 = An indicator variable = 1 for observations in the period
starting 31 December 2005 to 2009, and 0 otherwise.
Control variables = Same variables as in previous test models.
Table 4.3 summarises the main effect for each of the alternative EM proxies and their respective interaction terms. The three-way interaction terms, TREAT_DOt*IFRSt*DAt,
TREAT_DOt*IFRSt*SLSMANt, and TREAT_DOt*IFRSt*DISXMANt are the key
variables that measure whether there is incremental difference in the manipulation of accruals and/or real transactions to inflate core earnings by firms that report DO in the post-IFRS period, relative to (i) their behaviour pre-IFRS, and (ii) the behaviour of control firms post-IFRS. If firms that report DO in the post-IFRS period use classification shifting to improve core earnings post-IFRS, then I expect them to engage in a lower degree of accrual-based and/or real earnings manipulation in that period. I therefore expect θ11,θ13 and θ15 to be negative.
TABLE 4.3
Interpretation of Coefficients for Main and Interaction Effects for Tests of H2c(i-iii)
TREAT_DOt Measures the unexpected core earnings of treatment
firms (i.e. firms that report discontinued operations in the post-IFRS period regardless of their pre-IFRS behaviour) relative to control firms (i.e. firms that do not report discontinued operations post-IFRS irrespective of pre-IFRS behaviour) pre-IFRS.
DAt Measures the association between unexpected core
earnings (UE_CEt) and income-increasing accruals
management for control firms in the pre-IFRS period.
SLSMANt Measures the association between UE_CEt and the
income-increasing management of sales activities for control firms in the pre-IFRS period.
TABLE 4.3 (continued)
DISXMANi,t Measures the association between UE_CEt and the
opportunistic reduction of discretionary expenditure for control firms in the pre-IFRS period.
IFRSt Measures the unexpected core earnings of control
firms in the post-IFRS period relative to pre-IFRS.
TREAT_DOt*IFRSt Measures the unexpected core earnings for treatment
firms in the post-IFRS period, relative to their behaviour pre-IFRS and the behaviour of control firms post-IFRS.
IFRSt*DAt Measures the incremental association between the
UE_CEt and discretionary accruals of control firms in
the post-IFRS period relative to pre-IFRS
IFRSt*SLSMANt Measures the incremental association between
UE_CEt and the income-increasing management of
sales activities for control firms in the post-IFRS period relative to pre-IFRS.
IFRSt*DISXMANt Measures the incremental association between
UE_CEt and abnormal discretionary expenditure of
control firms post-IFRS relative to pre-IFRS.
TREAT_DOt*DAt Measures the incremental association between
UE_CEt and accruals management for treatment firms
relative to control firms in the pre-IFRS period.
TREAT_DOt*IFRSt*DAt Measures the incremental association between
UE_CEt and accruals management for treatment firms
in the post-IFRS period, relative to their behaviour pre-IFRS and the behaviour of control firms post- IFRS.
TREAT_DOt*SLSMANt Measures the incremental association between
UE_CEt and abnormal operating cash for treatment firms relative to control firms in the pre-IFRS period.
TABLE 4.3 (continued)
TREAT_DOt*IFRSt*SLSMANt Measures the incremental association between
UE_CEt and sales transaction manipulation for
treatment firms in the post-IFRS period, relative to their behaviour pre-IFRS and the behaviour of control firms post-IFRS.
TREAT_DOt*DISXMANt Measures the incremental association between
UE_CEt and abnormal discretionary expenditure of
treatment firms relative to control firms pre-IFRS.
TREAT_DOt*IFRSt*DISXMANt Measures the incremental association between
UE_CEt and abnormal discretionary expenses for
treatment firms in the post-IFRS period, relative to their behaviour pre-IFRS and the behaviour of control firms post-IFRS.