48 / 330
銀行業資訊科技支出之價值攸關性
48
Table 5 The Value Relevance of IT Expenditure and Intensity
(1) Model 1.1
(2) Model 1.2
(3) Model 2.1
(4) Model 2.2
Dependent Var.
MV
it
MV
it
MV
it
MV
it
IT
it
2.034***
(0.001)
0.157
(0.661)
BV
it
0.906***
(0.000)
0.906***
(0.000)
0.905***
(0.000)
0.895***
(0.000)
NI
it
1.878***
(0.000)
1.870***
(0.000)
1.414***
(0.000)
BNI
it
1.877***
(0.000)
ITD
it
×NI
it
2.191***
(0.002)
ITD
it
0.002
(0.200)
-0.015**
(0.011)
Intercept
0.050***
(0.000)
0.050***
(0.001)
0.053***
(0.000)
0.056***
(0.000)
Adj-R
2
60.80%
60.80%
60.73%
62.67%
No. of obs.
2,952
2,952
2,952
2,952
a: All coefficients and p values are estimated based on double-clustered (year and firm) standard
errors. Refer to Tables 3 and 4 for variable definitions.
b: p values are presented in parentheses. *:
p
< 0.10; **:
p
< 0.05; ***:
p
< 0.01.
4.3 Robustness Checks
We apply another specification to evaluate the ability of IT intensity in improving
earnings’ persistence. In Table 6, one-year-ahead ROA (ROE, or Net income) is regressed on
the current IT intensity, current ROA (ROE, or Net income), and the related interaction
terms.
15
As shown, in Column (1), the coefficient of the interaction term between IT intensity
and current ROA (ITD
it
×ROA
it
) is positive and significant (coefficient = 0.006 and
p
=
0.000). Similarly, in Columns (2) and (3), both ITD
it
× ROE
it
and ITD
it
× NI
it
are significantly
positive (
p
< 0.001). Hence, this additional test further reveals that firms can improve the
persistence of their earnings by allocating more resources to IT infrastructures.
Some may argue that the effect of IT expenditures on performance measures should be
non-linear. Such conjecture is reasonable because the IT spending may be considerable in
certain cases. Therefore, following prior studies that focus on the non-linearity of a certain
variable (Belderbos, Carree, Diederen, Lokshin, and Veugelers, 2004; Davis, Soo, and
Trompeter, 2009), the squared terms of IT related variables are added to our original
regressions, and the results are reported in Table 7. In Column (1) where the one-year-ahead
15 The specification is adopted from Gu (2005).