40 / 330
銀行業資訊科技支出之價值攸關性
40
contended by Bancorp Inc. in its 2008 annual report). Though, few prior studies have
explored this conjecture. Anderson, Banker, and Ravindran (2001) focus on the disclosure
regarding the IT expenditure made in preparation for Y2K during 1998 and 1999, and posit
that the firm value is affected more by the industry level of Y2K spending. They document a
positive relation between firm value and the amount by which a firm’s Y2K spending
exceeded the industry median level. Additionally, using a balanced scorecard (BSC)
framework, Kim and Davidson (2004) put forward the assumption that, depending upon the
level of IT, the relationship between IT expenditures and banks’ performance is significantly
different. For banks that maintain a high IT level, IT expenditures appear to have increased
market share and improved revenue and profits, while little of such an effect is observed for
those with lower IT levels. Accordingly, we posit that IT intensity enhances the banks’ ability
to keep up with customers’ needs in a timely manner, and hence retain a longer-duration
customer relationship that results in a more persistent pattern of earning. Because prior
studies indicate that a more persistent earning is usually reflected in a greater current
earnings’ multiplier in the determination of stock price or market value (Kormendi and Lipe,
1987; Easton and Zmijewski, 1989; Collins and Kothari, 1989), we put forward the
following argument.
H2: IT-intensive banks are related to higher earnings multipliers.
3. Research Design
3.1 Data and Sample Description
The sample includes all publicly-listed commercial banks for which annual data were
available for some or all years during the period of 2001–2010. Financial data are obtained
from COMPUSTAT, whereas IT expenditures are manually collected from Form 10-K
reports, as they are not readily available in any database. Table 1 displays a typical disclosure
pertaining to IT expenditures provided in banks’ Form 10-K reports. After deleting
observations with missing values and outliers that are above or below 0.5% of the original
sample, the final sample includes 2,952 observations.