378 / 414
審計委員會權益薪酬之決定因素
378
Therefore, a larger
CEO ownership dummy
means better corporate governance.
We then add these two measures into one variable so that we can obtain a score that
represents a firm’s internal governance.
5
Following He and Tian (2013), we use
Coverage
and
InsOwn
to proxy for external
governance, explained below:
Coverage
: the natural logarithm of the number of analysts following for the fiscal year.
InsOwn
: Institutional holding percentage for the firm, the arithmetic mean of the four
quarterly institutional ownerships reported through 13F.
We divide the observations of these two variables into ten groups based on the
magnitudes and score them from 1 to 10, where 1 is the score for the smallest group and 10
for the largest. Then we add the scores in order to obtain a score showing the quality of the
firm’s external governance.
The study includes other control variables that may affect a firm’s equity-based
compensation. Following Fich and Shivdasani (2005), we include the market-adjusted,
12-month stock return (
STOCK_RETURN
) to control for the preference for stocks or options
arising from the stock price performance of the company, annual R&D expense over sales
(
R&D
) to control for a company’s growth opportunities, percentage of annual cash retainer
to total compensation of the entire audit committee (
CASHCOMP
) to control for the impact
of other compensation components, and the natural logarithm of the firm’s sales (
LNSALES
)
to control for the firm’s size. In addition, we use the percentage of interest expense to sales to
proxy for the liquidity constraints (
LIQ_CONSTRAINTS
) of a firm based on Core and Guay
(2001) study. Finally, we control for board size (
BOARDSIZE
) and the 2-digit SIC industry
dummy in the model.
4.2.5 Empirical Model
To sum up, we use the binary logistic regression model below to test the association
between the occurrence of equity-based compensation for audit committee members and the
firm’s agency conflict, level of overlapping audit committee and compensation committee
members, and proportion of audit committee members who are top executives receiving
compensation from other companies. The dependent variable (
EQUCOMP
) takes the value
of 1 if the company gives equity-based compensation to at least one of its audit committee
5 Albuquerque (2009) also includes the
number
of
meetings
and
interlock
to construct the corporate
governance variables. We do not include these two variables because including them causes a large drop
in our sample.