42 / 330
銀行業資訊科技支出之價值攸關性
42
NI
it
= Net income of bank
i
for the year
t
(Compustat Annual data
NI
)
IT
it
= Information-technology related costs of bank
i
for the year
t
BNI
it
= Net income plus information-technology related costs of bank
i
for the year
t
,
i.e., NI
it
plus IT
it
All the above variables are deflated by lagged total assets. Model 1.1 is constructed
based on the typical Ohlson model, in which the market value is explained by the book value
and net income. If IT conveys relevant information about banks’ values that have not been
reflected in book values and net incomes, b
3
is expected to be significant and positive. In
addition, as a supplementary test to examine whether IT expenditures are considered merely
an “expense” or not
8
, in Model 1.2, the net income is further separated into net income
before IT expenditures (BNI) and IT expenditures (IT). If b
3
in this Model is above -1 or not
significant, it suggests that IT expenditures may contribute to long-term economic benefits
that exist longer than one year.
9
Hence, they are not considered to be merely an expense item.
8 In contrast to Model 1.1, Model 1.2 more directly tests the coefficient of the IT expenditure shown in the
income statement. In other words, the expected positive coefficient of IT expenditure in Model 1.1 triggers
a researcher’s conjecture that the accounting treatment for IT expenditure is not appropriate; otherwise, it
should not be significant. The conjecture is further supported by the expected insignificant coefficient of IT
expenditure in Model 1.2. The necessity of clearly ascertaining whether the market views the IT outlay as a
merely expense lies in that (1) if the management of banks does not know that the market “correctly”
considers only part of the IT outlay as the current expense, he (she) may make inappropriate decisions on
IT investment for fear of impairing the market value, and that (2) the accounting standard-setters usually
consider an accounting treatment that corresponds with the viewpoints of capital market participants to be
appropriate, and hence the accounting literature traditionally pays attention to such issue (e.g., Aboody and
Lev, 1998; Aboody, Barth, and Kasznik, 2004). We acknowledge that the result and inference from Model
1.2 is partially implied by that from Model 1.1, but we wish to make more clear implications to accounting
standards by doing so.
9 When analyzing the possible direction of the coefficient on a type of outlay that may give rise to long-run
benefits, consider a simplified example that is based on the purchases of a pencil (costing $100) and a car
(costing $100) in the present accounting period (say, this year). As in most cases, assume the pencil
generates $120 cash inflows during this year, while the car generates $40 per year for the subsequent 3
years. For simplicity, ignore discounting and the market value equals the anticipated benefits less the
incurred outlays. So, at the end of year 1, the market value for the pencil-purchase is 120 - 100 = $20,
while that for the car-purchase is 40 × 3 - 100 = $20. Then, in the regression, “Market value = Cash inflows
(some form of assets) + outlays,” it can be written as “$20 = b
1
× (benefit from the pencil outlay, $120) + b
2
× (the pencil outlay, $100), and “$20 = b
3
× (benefit from the car expenditure, $40) + b
4
× (the car
expenditure, $100). If the benefit is defined as some form of cash assets, it is more likely that b
1
= b
3
= 1.
So, b
2
= -1, and thus b
4
= -0.2. The fundamental reason for the less negative coefficient of car expenditure
is that the benefits generated by the car have not been fully recognized in the cash assets during the current
period. Due to the possibility that the benefits generated by IT expenditure may not be realized until the
subsequent periods, it is likely that no accounting earnings are recognized. But, because the market has
anticipated (at least partially) such benefits, in the regression “Market value = Cash inflows (some form of
assets) + IT outlays,” the coefficient of IT might be positive due to the related cash inflows possibly being
zero. In combination, the coefficient of IT expenditure may be positive or less negative. Which possibility
dominates is not clear, and the conflicting possibilities may also drive the coefficient to be insignificant.