Page 38 - 32.3

According to the theory proposed by Banker and Datar (1989), the weighting of one
performance measure in the compensation contract would decrease if the opacity of the
The Effect of Information Opacity on the Weighting of Performance Measures in the Compensation Contracts
information related to the performance measure is high. To verify the theory, we conduct
of CEOs: Evidence from U.S. Firms
an empirical study and predict that when the volume of stock-based information is low,
the weighting of the accounting-based performance measure increases, and the weighting
an empirical study and predict that when the volume of stock-based information is low, the
of the stock-based performance measure decreases simultaneously.
weighting of the accounting-based performance measure increases, and the weighting of
the stock-based performance measure decreases simultaneously.

2. Methodology and Research Design
2. Methodology and Research Design
To test our hypothesis, we collect the financial and stock data of U.S. listed firms
from 1993 to 2018 from Wharton Research Data Services (WRDS). Following the
To test our hypothesis, we collect the financial and stock data of U.S. listed firms
common practice of prior literature, we use return on assets (ROA) as the proxy for the
from 1993 to 2018 from Wharton Research Data Services (WRDS). Following the
accounting-based performance measure, and annual stock return (RET) as the proxy for
common practice of prior literature, we use return on assets (ROA) as the proxy for the
accounting-based performance measure, and annual stock return (RET) as the proxy for
the stock-based performance measure.
the stock-based performance measure.
To capture the information opacity of stock-related information, we use the bid-ask
To capture the information opacity of stock-related information, we use the bid-ask
spread (SPREAD) and number of analysts (ANALYSTS) as the proxies in our regression
spread (SPREAD) and number of analysts (ANALYSTS) as the proxies in our regression
analyses. The two regression models are specified as follows:
analyses. The two regression models are specified as follows:

ln(TP i ,t ) = α + α 1 SPREAD + α 2 ROA + α 3 RET + α 4 SPREAD × ROA ,t i
i
,t i
,t
,t
0
,t i
i
,t 
+ α 5 SPREAD × RET + α k Control V ariables + ε ,t i , (1)
,t
i
,t
i
i
ln(TP ,t i ) = β + β 1 ANALYSTS + β 2 ROA + β 3 RET + β 4 ANALYSTS × ROA ,t i (2)
,t i
,t i
0
,t i
,t i
,t i 
+ β 5 ANALYSTS × RET + β k Control V ariables + ε ,t i ,
,t i
,t i
A low volume of stock-related information suggests that the information opacity
A low volume of stock-related information suggests that the information opacity of
of stock information is high; and theoretically, SPREAD should be large. We expect the
stock information is high; and theoretically, SPREAD should be large. We expect the
weighting of the accounting-based performance measure to increase and the weighting of
weighting of the accounting-based performance measure to increase and the weighting of
the stock-based performance measure to decrease when SPREAD is large. Therefore, we
the stock-based performance measure to decrease when SPREAD is large. Therefore, we
predict the sign of α4 in equation (1) to be positive and the sign of α5 in equation (1) to be
predict the sign of α4 in equation (1) to be positive and the sign of α5 in equation (1) to be
negative. Conversely, a large number of stock analysts suggest that the information opacity
negative. Conversely, a large number of stock analysts suggest that the information
of stock related information is low. We expect the weighting of the accounting-based per-
opacity of stock related information is low. We expect the weighting of the accounting-
formance measure to decrease and the weighting of the stock-based performance measure
based performance measure to decrease and the weighting of the stock-based
to increase when ANALYSTS is large. Accordingly, we predict the sign of β4 in equation (2)
performance measure to increase when ANALYSTS is large. Accordingly, we predict the
sign of β4 in equation (2) to be negative and the sign of β5 in equation (2) to be positive.
to be negative and the sign of β5 in equation (2) to be positive.
Meanwhile, we also examine the information opacity of accounting information.
Although we propose that the volume of accounting information should be similar across
42

28