23 /274
23
臺大管理論叢
第
27
卷第
4
期
Incorporating the QFIIs’ trading strategies into regression (the IS_HH_HL model), the
coefficients of QFII_HH
t
*IS
t
*X
t3
and QFII_HL
t
*IS
t
*X
t3
are -0.055 (
t
= -1.47) and -0.075 (
t
=
-1.45), both negative and statistically insignificant. It is unlikely that the earnings
informativeness of income smoothing is affected by the QFIIs’ shareholding volatility.
However, it is worthy to mention that there were ceilings for each foreign investor’s
shareholding in individual firms as well as for total foreign shareholdings in individual firms
before the year 2000 in Taiwanese listed firms. Futhermore, these shareholding restrictions
have declined steadily over time and there was complete deregulation after 2003. Whether
these somewhat surprising results can be attributed to the restrictions of QFIIs’ ownership
will be discussed in the following section.
4.4 The Effect of QFIIs Regulations Examination
To examine the effect of regulations on QFIIs on the earnings informativeness for firms
with income smoothing, this study divides the entire observation period into two exclusive
subperiods, i.e., pre-deregulation (1997~2002) versus post-deregulation (2003~2007).
5
The
results are presented in Table 6.
6
From IS_QFII model in Table 6, the coefficients of
5 We note that using the two exclusive subsamples based on year 2003 is likely contaminated because of the IS
measure and increases the difficulty of interpretation about the association between income smoothing and
earnings informativeness after the QFIIs deregulation. Thus, this study uses both the variance of the residuals
from the regressions of the change in net income before extraordinary items and the change in operating cash
flow on variables, which are estimated by Lang, Raedy, and Wilson (2006) to measure the variability of
change in net income (ΔNI*) and the variability of change in operating cash flow (ΔCFO*) to understand
whether managerial accruals management has be changed after the QFIIs deregulation. We use two measures,
i.e., the variability of ΔNI* and the variability of ΔNI* over ΔCFO*, to examine whether the income
smoothing behaviors had been changed after the QFIIs deregulation. The untabulated results revealed that the
variability of ΔNI* is 0.058 and 0.104 in the pre-deregulation period versus post-deregulation period,
respectively. The mean ratio of the variability of ΔNI* to ΔCFO* measure is 0.898 and 1.195 in the pre-
deregulation period versus post-deregulation period, respectively. Both the variability measures are
statistically significant at the traditional level. We also turn back to look our initial discretionary accruals (DA)
estimation and find that average discretionary accruals is -0.0052 and 0.0018 in the pre- versus post-
deregulation period, respectively. The mean difference of discretionary accruals is statistically significant at
the 5% level (
t
= 2.28). These results to some extent revealed distinctive pattern of managerial income
smoothing behaviors after the QFIIs deregulation.
6 To control for the residuals may be correlated, we rerun Regression (1), (2), and (3) and compute standard
errors clustered by industry (Petersen, 2009). The results do not qualitatively change the primary findings.