269 /342
臺大管理論叢
第
27
卷第
2S
期
269
and reduce risk. New CEOs result from non-routine turnovers have more concerns because
they do not have much track records with and trust from the board. The coefficient of routine
CEO turnover is not significantly related to change in reinsurance demand when compared to
insurers without CEO turnover in all Models. Because new routine CEOs are familiar with
the direction of the board, they tend to follow the original reinsurance decision as well. This
result is rejects Hypothesis 2. Other results of Table 4 show that corporate governance and
control variables which are statistically significant are similar to those in Table 3.
Table 3 Regressions of Reinsurance Demand on CEO Turnover
Dependent variables
∆
Reins ratio
∆
Reins_aff_ratio
∆
Reins_nonaff_ratio
Independent Variables
Estimate P value Estimate P value Estimate P value
Turnover (t-1)
0.043***
0.006
0.032**
0.041
0.011*
0.058
Mutual (t-1)
-0.007
0.915
0.013
0.851
-0.020
0.411
Duality (t-1)
-0.051***
0.007
-0.041**
0.031
-0.010
0.147
∆ Board size
0.001
0.883
0.000
0.972
0.000
0.758
∆ Independent director
0.030
0.474
-0.002
0.963
0.032**
0.035
Big 4 auditor (t-1)
0.007
0.731
-0.001
0.960
0.008
0.276
∆ Ln(na)
-0.279***
0.000
-0.290***
0.000
0.011
0.467
∆ Herfindahl
0.018
0.837
0.066
0.454
-0.048
0.125
∆ Geoherfindahl
-0.475***
0.000
-0.489***
0.000
0.014
0.702
∆ Leverage
0.034
0.769
-0.170
0.151
0.204***
0.000
∆ Underwriting risk
-0.298***
0.000
-0.215***
0.008
-0.083***
0.004
∆ 2yearLossDevelopment
0.091***
0.011
0.139***
0.000
-0.048***
0.000
∆ Coastal prem
-0.365**
0.014
-0.308**
0.040
-0.057
0.286
∆ Long-tail
0.249***
0.007
0.235**
0.012
0.014
0.666
∆ Tax_ex
-0.023
0.354
-0.026
0.315
0.002
0.797
∆ ROA
0.068
0.650
-0.022
0.883
0.090*
0.095
Group (t-1)
0.012
0.726
0.026
0.443
-0.014
0.235
Intercept
0.026
0.689
0.018
0.778
0.007
0.751
Hausman Test
14.29
13.08
28.84
R-Square
0.086
0.084
0.122
N
2,772
2,772
2,772
Note: The table shows the regression results of reinsurance demand on CEO turnover. “
∆
x
” means
change in
x
. Specifically it sugggets that
∆
x
i,t
means
x
i,t
minus
x
i,t
-1
. Please see definition of all
variables in Appendix A. ***significant at 1%, ** significant at 5%, * significant at 10%.