Table of Contents Table of Contents
Previous Page  28 /274 Next Page
Information
Show Menu
Previous Page 28 /274 Next Page
Page Background

公司盈餘平穩化行為與盈餘資訊性之關係-合格境外機構投資者角色之檢測

28

5. Robustness Test

5.1 Alternative Measure of Income Smoothing

The discretionary accrual estimation model might be misspecified (Kothari et al., 2005).

Thus, our income smoothing measure may suffer from potential measurement error

problems. To address this concern, we follow Chaney and Lewis (1998) and use the variance

ratio (VR) to measure income smoothing. The variance ratio is the ratio of the variance of

cash from operations to the variance of net income and is presented as follows:

VR

i

=

Var

(

CF

it

)

Var

(

NI

it

)

(4)

Var(CF

it

) and Var(NI

it

) are the sample variance of cash from operations and of net

income for firm

i

during the current and prior four sample years based on year

t

, respectively.

To control for industry and time effects, we also use a firm’s reversed fractional ranking of

variance ratio and refer to it as the VR variable. This study thus uses the VR variable to

replace the initial IS measure and reruns Equations (1), (2), and (3). The extracted results are

reported in Panel A of Table 7.

From Table 7, the coefficients of QFII

t

*VR

t

*X

t3

in the VR_QFII model are 0.005 (

t

=

0.58) and -0.005 (

t

= -1.82) in the pre-versus post-deregulation period, respectively. Only the

coefficient in the post-deregulation model is statistically significant at the 10% level. These

results again support the initial findings. It also reveals that the coefficients of QFII_

HH

t

*VR

t

*X

t3

and QFII_HL

t

*VR

t

*X

t3

are -0.038 (

t

= -0.83) and -0.092 (

t

= -2.09) in the post-

deregulation period, and that they are both negative and the latter coefficient is statistically

significant at the 5% level. In the pre-deregulation period, the coefficients of QFII_

HH

t

*VR

t

*X

t3

and QFII_HL

t

*VR

t

*X

t3

are -0.014 (

t

= -0.12) and -0.091 (

t

= -0.93), which are

both negative, yet statistically insignificant in the VR_HH_HL model. Except for the results

from VR_HH_HL model in the post-deregulation period, the findings again support that

deregulation accelerates the opportunistic role of QFIIs and reduces the informative

component of earnings for firms with income smoothing.