|
Higher market power
|
Lower market power
|
---|
(1)
|
(2)
|
(3)
|
(4)
|
(5)
|
(6)
|
---|
\(BTD\)
|
\(SA\)
|
\(CASH\)
|
\(BTD\)
|
\(SA\)
|
\(CASH\)
|
---|
Panel A: Cross-sectional tests based on market power
|
Results based on 2SLS
|
\(TRC\)
|
0.089`
|
0.374
|
− 0.876
|
− 0.335**
|
− 0.801**
|
− 0.420**
|
|
(0.112)
|
(0.928)
|
(1.355)
|
(0.144)
|
(0.317)
|
(0.214)
|
Sample size
|
1224
|
1054
|
865
|
1240
|
1860
|
2564
|
Results based on Sharp RDD
|
\(SME\)
|
0.010
|
0.040
|
− 0.081
|
− 0.070***
|
− 0.180***
|
− 0.070**
|
|
(0.009)
|
(0.057)
|
(0.049)
|
(0.020) `
|
(0.045)
|
(0.030)
|
Sample size
|
1264
|
1249
|
776
|
1116
|
1331
|
2582
|
|
Higher ETCs
|
Lower ETCs
|
---|
(1)
|
(2)
|
(3)
|
(4)
|
(5)
|
(6)
|
---|
\(BTD\)
|
\(SA\)
|
\(CASH\)
|
\(BTD\)
|
\(SA\)
|
\(CASH\)
|
---|
Panel B: Cross-sectional tests based on ETCs
|
Results based on 2SLS
|
\(TRC\)
|
− 0.262*
|
− 0.668*
|
− 0.538**
|
− 0.165
|
− 0.331
|
− 0.433
|
|
(0.148)
|
(0.394)
|
(0.633)
|
(0.125)
|
(0.633)
|
(0.380)
|
Sample size
|
1586
|
1976
|
2460
|
625
|
828
|
593
|
Results based on Sharp RDD
| | | | | | |
\(SME\)
|
− 0.040***
|
− 0.113**
|
− 0.083***
|
− 0.027
|
− 0.046
|
− 0.070
|
|
(0.015)
|
(0.045)
|
(0.031)
|
(0.017)
|
(0.065)
|
(0.047)
|
Sample size
|
1712
|
1912
|
2328
|
571
|
819
|
668
|
- This table reports RDD results of cross-sectional tests for tax avoidance, financial constraints and cash dependence. Panel A presents cross-sectional tests based on market power. Panel B represents cross-sectional tests based on ETCs. The dependent variables are tax avoidance (\(BTD\)), financial constraints (\(SA\)) and cash dependence (\(CASH\)). Market power is measured as the gross profit margin. ETCs is calculated as entertainment and travel costs scaled by operation revenue. We divide the full sample into two groups respectively based on the medians of market power and ETCs to exercise the cross-sectional tests. \(TRC\) is an indicator variable that is equal to one if the firm's bank loan raise after TRRRCs. \(SME\) is an indicator variable that is equal to one if the firm's operation revenue was below the threshold of operation revenue. The discontinuity estimates are based on local linear regressions. Standard errors are in the parentheses. *, **, and *** represent statistical significance level at the 1%, 5% and 10% levels, respectively