國
立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
28
C. Investors’ Investment Behaviors
We verify Hypothesis 1 by setting an OLS Linear Regression Model between corporate governance variables and investors’ shares holding after examining the relationships among variables, than we pay attention to whether there is a difference between 15 and 5 years’ data due to the probability of investors’ advanced realization in investing activity and implement of Information transparency and disclosure ranking.
To examine Hypothesis 1, we conduct model (1-1) to test the 15 years data in yearly frequency, model (1-1) is showed below:
(1-1)
Where Yi,t : The investors’ stockholdings ;
(1) INDIV: individuals Stockholdings,
(2) DOINDIV: domestic Individuals Stockholdings, (3) FOINDIV: foreign Individuals Stockholdings, (4) INSTI: institutions Stockholdings,
(5) FOINSTI: foreign Institutions Stockholdings
In model (1-1), we consider about there may exist non-linear relationship between investors’ stockholdings and management stockholdings (MGT) or board sizes
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
29
(BOARD), therefore, adding the square items of management stockholdings and board sizes to the model respectively. In fact, the class of corporate governance is depend on the cognition of management, so we think of the inside dominance of CEO (PARTCEO) may reduce the effect of management stockholdings (MGT) to our dependent variable, therefore adding the cross item of inside dominance of CEO and management stockholdings, and anticipating the coefficient is negative.
‧
INDIV DOINDIV FOINDIV INSTI FOINSTI
MGT 2.139*** 2.157*** -0.0178 -2.139*** 0.367*
t -statistics in parentheses * p<0.1, ** p<0.05, *** p<0.01
‧
variable to investors’ stockholdings first.The results of model (1-1) are showed in Table 8, whole individual investors’
stockholdings (INDIV) and domestic individual investors’ stockholdings (DOINDIV) are positive related to several governance variables we picked, such as management stockholdings, independent outside directors, and inside dominance of chairman or CEO, indicating that management stockholdings increasing, more outside supervisors, or part management of ultimate owners would raise individual investors’ faith in firms;
besides, leverage is positive related to domestic individual investors’ stockholdings too, but negative related to foreign individual investors’ stockholdings, representing that domestic investors prefer investing firms with higher leverage ratio, and foreign opposite. Domestic individual investors’ stockholdings is negative related to blockholders’ stockholdings, board size (but with non-linear relation), CEO duality, and firm size, indicating that individual investors would lose their interests in investment with blockholders’ stockholdings increasing, larger board sizes (but only a limit valid), CEO serves as the chairman, and larger firm sizes.
We find that despite for whole institutional investors’ stockholdings (INSTI) or foreign institutional investors’ stockholdings (FOINSTI), the results of governance variables are being opposite at all, illustrating that institutional investors prefer blockholders holding more shares, larger board sizes (but only a limit valid), and larger firm size, but not management holding more shares, more outside supervisors, part management of ultimate owners, or higher leverage ratio, nevertheless, management stockholdings and board sizes are in a non-linear relation with foreign
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
32
institutional investors’ holdings.
We conduct model (1-2) to test the 15 years data in monthly frequency to examine Hypothesis 1 again, model (1-2) is showed below:
(1-2) Where Yi,t : The investors’ stockholdings ;
(1) INDIV: individuals Stockholdings,
(2) DOINDIV: domestic Individuals Stockholdings, (3) FOINDIV: foreign Individuals Stockholdings, (4) INSTI: institutions Stockholdings,
(5) FOINSTI: foreign Institutions Stockholdings RTN (t-x)i,t : The Return of (t-x) period, where x=1, 2
As last model, we consider about there may exist non-linear relationship between investors’ stockholdings and management stockholdings or board sizes, so adding the square items of management stockholdings and board sizes to model (1-2) respectively; furthermore, thinking of the inside dominance of CEO may reduce the effect of management stockholdings to our dependent variable, therefore adding the cross item of inside dominance of CEO and management stockholdings, and anticipating the coefficient is negative. The returns of past periods are took into account, we adopt the returns of one and two months earlier than other variables respectively.
‧
INDIV DOINDIV FOINDIV INSTI FOINSTI
MGT 2.417*** 2.427*** -0.00971 -2.417*** 0.0383
t- statistics in parentheses * p<0.1, ** p<0.05, *** p<0.01
‧
INDIV DOINDIV FOINDIV INSTI FOINSTI
MGT 2.416** 2.426*** -0.00974 -2.416** 0.0397
t- statistics in parentheses * p<0.1, ** p<0.05, *** p<0.01
‧
The results of model (1-2) are showed in Table 9, Panel a is inputted the returns of past one period; Panel b is past two periods. The results show correspondences to several variables with yearly frequency data in Table 8, the domestic individual investors prefer firms with more outside independent directors, ultimate owners acting as management, blockholders holding a few shares, small board size, no CEO duality as well. However, there are some differences with yearly frequency data, management stockholdings is showed a non-linear relation with individual investors, but still positive related; investment in asset is negative related to domestic individual investors but positive related to foreign individuals and institutions, representing domestic individuals like to put their funds in firms with higher ratio of investment in asset, and foreign individuals and institutions opposite.
Outside independent directors is a variable have reverse effect on domestic and foreign institutional investors, it’s negative related to domestic institutional investors’
stockholdings but positive to foreign in either Panel a or Panel b, showing that foreign investors think highly of major outside independent directors. By observing the variables return (t-1) and return (t-2) in two panels, we find that only the return of past two periods is significant at 10% level on entirely individual and institutional investors market but with reverse directional impacts, individuals is negative related to the past returns, however, institutions is positive related to the past returns.
While examining whether the investors’ behaviors changed follow by the implement of governance mechanism, finally we use model (1-3) to test the 5 years data in monthly frequency to recheck our results before. In five years data, the variable of information transparency and disclosure ranking (ABC) is added into the model, because a series of corporate governance scandals happened after 2000, resulted in governance pay much attention to this area and proceeded establishing the
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
36
ranking. The model (1-3) is showed below:
(1-3)
Where Yi,t : The investors’ stockholdings ;
(1) INDIV: individuals Stockholdings,
(2) DOINDIV: domestic Individuals Stockholdings, (3) FOINDIV: foreign Individuals Stockholdings, (4) INSTI: institutions Stockholdings,
(5) FOINSTI: foreign Institutions Stockholdings RTN (t-x)i,t : The Return of (t-x) period, where x=1, 2
‧
INDIV DOINDIV FOINDIV INSTI FOINSTI
MGT 2.241*** 2.282*** -0.0404* -2.241*** 1.458***
t -statistics in parentheses * p<0.1, ** p<0.05, *** p<0.01
‧
Panel b. Return of Past Two Periods Variable
(1) (2) (3) (4) (5)
INDIV DOINDIV FOINDIV INSTI FOINSTI
MGT 2.241*** 2.282*** -0.0404* -2.241*** 1.458***
t -statistics in parentheses * p<0.1, ** p<0.05, *** p<0.01
‧
國立 政 治 大 學
‧
N a tio na
l C h engchi U ni ve rs it y
39
The results showed in Table 10 are similar with Table 9, several governance variables like management stockholdings, and blockholders’ stockholdings take opposite influence between individual and institutional investors. In addition, to judge whether the variable of information disclosure be took account of investment behaviors, we find that the grade of information disclosure only affect foreign institutional investors’ stockholdings, it shows a positive relation in Table 10.
The variable MGTPARTCEO is negative related to individual investors’ shares holding, illustrating that inside dominance of CEO may reduce the effect of management stockholdings to dependent variables, and vice versa. Finally, we discover individual investors always prefer to invest in small firms, and larger firms do institutional investors so far.