27
more timely manner than ‘good news.’
32
The Basu (1997) model uses the following
equation:
EARN
it
= α
0
+ α
1
D
it
+ α
2
RET
it
+ α
3
D
it
*RET
it
+ u
it
(5)
The dependent
variable,
EARN
it
, is earnings per share before extraordinary items
scaled by stock price at the fiscal year-end of
t
-1 and
RET
it
is the 12-month compound
return ending three months after the fiscal year-end of
t
.
D
it
is an indicator variable
equaling one if
RET
it
is negative, and zero otherwise.
RET
it
is
used as a proxy for
economic gains (good news) when it is positive and for economic losses (bad news) when
it is negative.
α
2
captures the sensitivity of earnings to good news.
α
3
captures the
incremental sensitivity of earnings to bad news relative to good news (i.e., timely loss
recognition or, in particular, the asymmetric timeliness of loss recognition in earnings). I
am interested in examining the asymmetric timeliness of loss recognition in earnings
following
IFRS
adoption. Accordingly, I expand equation (5) as follows:
EARN
it
* = α
0
+ α
1
D
it
+ α
2
RET
it
+ α
3
D
it
*RET
it
+ α
4
IFRS
it
+ α
5
IFRS
it
*D
it
+ α
6
IFRS
it
*RET
it
+ α
7
IFRS
it
*D
it
*RET
it
+ u
it
(6)
Following Ahmed et al. (2010), I measure
EARN
it
*
, the dependent variable in
equation (6), as the residual from regressing
EARN
it
(defined above)
on industry and
country fixed effects.
33
In equation (6),
α
7
measures any change in the asymmetric
timeliness of loss recognition following
IFRS
adoption. I interpret a positive
α
7
coefficient as indicating an increase in asymmetric timeliness
of loss recognition
following
IFRS
adoption.
32
As indicated earlier, prior literature (e.g., Dietrich et al. 2007; Givoly et al. 2007) has shown that Basu
(1997) model suffers from several drawbacks. Even though, Basu (1997) model is considered the most
widely-used model to estimate the asymmetric timeliness of loss recognition.
33
This approach of measuring the dependent variable in equation (6) is also in line with Ball et al.’s (2011)
suggestion. In their recent paper, Ball et al. (2011) argue that the correlation between the ‘expected’
components of earnings and returns biases the estimate of the asymmetric earnings timeliness in Basu
(1997) model. To address this bias problem, they suggest fixed-effects regression as an example. Therefore,
by obtaining the residual from regressing earnings on country and industry fixed-effects, I control for the
‘expected’ earnings component across countries and industries.
28
Table C4 presents the results of a pooled regression based on equation (6). The
estimated
coefficient
α
7
is positive, although it is not significant. This finding suggests
that there is some modest evidence of an increase in the asymmetric timeliness of loss
recognition following
IFRS
adoption and contrasts with that of Ahmed at al. (2010) who
document a significant decrease in the asymmetric timeliness of loss recognition
following
IFRS
adoption.
Taken together, my preliminary results suggest that there is more timely loss
recognition for
PPE
following
IFRS
adoption. The frequency of impairment losses for
PPE
is significantly greater in the post-
IFRS
period
relative to the pre-
IFRS
period and
there is some modest evidence of an increase in the asymmetric timeliness of loss
recognition following
IFRS
adoption. Therefore, these results are consistent with my
prediction of having more timely loss recognition for
PPE
under
IFRS
strict impairment
rules.
Dostları ilə paylaş: