Table 3

Addressing endogeneity using propensity score matching

Panel A: Pre-match propensity score regression and post-match diagnostic regression
Pre-matchPost-match
(1)(2)
Dependent variableINSDUM
LNS−0.689***0.034
(−61.248)(0.496)
K/S−0.680***0.042
(−12.673)(0.678)
Y/S−1.713***−0.620
(−5.794)(−0.888)
R&D/K−0.021***0.004
(−3.424)(0.549)
I/K−0.492***−0.028
(−8.062)(−0.594)
ROA0.050***0.009
(13.042)(0.167)
LEVERAGE0.677***0.179
(6.799)(1.481)
Age−0.710***−0.055
(−33.957)(−0.214)
Constant11.145***−0.234
(83.717)(−1.472)
Observation36,15717,471
Pseduo R20.2550.001
Panel B: Post-match difference in firm characteristics
VariableTreatedControlDifferencet-test
LNS11.32811.2630.060.43
K/S0.350.355−0.01−0.67
Y/S0.0850.0850.00−0.46
R&D/K0.1620.1460.020.3
I/K0.1520.1550.00−0.42
ROA5.14.9070.191.08
LEVERAGE0.1870.1840.001.06
Age3.3713.398−0.03−1.01
Panel C: Post-match regression analysis
Dependent variableDistance to defaultProbability of defaultCDS spread
(1)(2)(3)
DIR−0.012***  
(−3.758)  
CEOIR −0.010** 
 (−2.681) 
EMPIR  0.001
  (0.229)
LNS0.184***0.155***0.168***
(17.609)(13.608)(15.320)
K/S0.864***0.956***0.968***
(17.782)(17.578)(18.450)
Y/S7.046***6.798***6.941***
(25.612)(22.592)(23.726)
R&D/K0.027***0.029***0.031***
(5.437)(5.884)(6.152)
RDUM0.9000.7300.847
(1.509)(1.240)(1.437)
I/K0.0210.0400.047
(0.686)(1.273)(1.505)
ROA0.062***0.065***0.068***
(16.850)(16.342)(17.450)
LEVERAGE−6.022***−5.909***−5.877***
(−60.706)(−52.212)(−54.447)
Age−0.028−0.021−0.009
(−1.218)(−0.822)(−0.379)
Constant1.501**1.863***1.589***
(2.425)(3.040)(2.595)
Observations17,47113,39114,886
R-squared0.3400.3420.342
Fixed year effectsYesYesYes
Firm fixed effectsYesYesYes

Note(s): Table 3 reports the results of the propensity score matching procedure to investigate the effects of insider ownership on default risk. Panel A reports the parameter estimates from the logit model used to estimate propensity scores. The dependent variable INSDUM in column (1) and (2) of Panel A is an indicator variable set to one if the firm has insider ownership score more than median in a given year, zero otherwise. Panel A reports the pre-match propensity score regression and post-match diagnostic regression. Panel B reports the univariate comparisons of firm characteristics between treated and control firms the corresponding t statistics. Panel C reports multivariate results relating to default risk and insider ownership. The dependent variables are distance-to-default. t-statistics are reported in parentheses. Variables are defined in  Appendix and are winsorized at the 1–99% levels. ***, ** and * indicate statistical significance of coefficient estimates at the 1, 5 and 10% levels, respectively

Source(s): Table created by the authors

or Create an Account

Close Modal
Close Modal