Table 4 Robustness tests for addressing...

Table 4

Robustness tests for addressing endogeneity

Panel A: 2SLS regression
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Variables	1^st stage	1^st stage	1^st stage	2^nd stage	2^nd stage	2^nd stage	2^nd stage	2^nd stage	2^nd stage
Variables	CSR	ENV	SOC	ROA	ROA	ROA	ROE	ROE	ROE
Ind_CSR	0.009***
Ind_CSR	(7.348)
Ind_ENV		0.558***
Ind_ENV		(16.114)
Ind_SOC			0.464***
Ind_SOC			(11.082)
CSR_P				0.008			−0.137
CSR_P				(0.099)			(−0.639)
CSR_P×CSR_COM				0.085***			0.108**
CSR_P×CSR_COM				(3.537)			(2.159)
ENV_P					−0.058			−1.002**
ENV_P					(−0.453)			(−2.190)
ENV_P×CSR_COM					0.094***			0.109**
ENV_P×CSR_COM					(3.745)			(2.147)
SOC_P						−0.054*			−0.060
SOC_P						(−1.716)			(−0.619)
SOC_P×CSR_COM						0.052***			0.056**
SOC_P×CSR_COM						(3.845)			(2.062)
CSR_COM				−0.044***	−0.047***	−0.028***	−0.053**	−0.053**	−0.028*
CSR_COM				(−3.625)	(−3.779)	(−3.943)	(−2.133)	(−2.121)	(−1.954)
Constant	−0.411***	−0.077	−0.108	−0.442***	−0.455***	−0.470***	−0.807***	−0.976***	−0.812***
Constant	(−7.803)	(−0.808)	(−1.128)	(−4.110)	(−4.245)	(−4.322)	(−4.804)	(−5.233)	(−4.727)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Wu–Hausman test				0.002	0.001	0.001	0.004	0.002	0.002
Weak instrument test				0.000	0.000	0.000	0.001	0.000	0.000
Observations	12,285	12,278	12,278	12,278	12,278	12,278	12,278	12,278	12,278
Adjusted-R²	0.290	0.844	0.834	0.562	0.562	0.562	0.414	0.414	0.414

Panel A: 2SLS regression
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Variables	1^st stage	1^st stage	1^st stage	2^nd stage	2^nd stage	2^nd stage	2^nd stage	2^nd stage	2^nd stage
Variables	CSR	ENV	SOC	ROA	ROA	ROA	ROE	ROE	ROE
Ind_CSR	0.009***
Ind_CSR	(7.348)
Ind_ENV		0.558***
Ind_ENV		(16.114)
Ind_SOC			0.464***
Ind_SOC			(11.082)
CSR_P				0.008			−0.137
CSR_P				(0.099)			(−0.639)
CSR_P×CSR_COM				0.085***			0.108**
CSR_P×CSR_COM				(3.537)			(2.159)
ENV_P					−0.058			−1.002**
ENV_P					(−0.453)			(−2.190)
ENV_P×CSR_COM					0.094***			0.109**
ENV_P×CSR_COM					(3.745)			(2.147)
SOC_P						−0.054*			−0.060
SOC_P						(−1.716)			(−0.619)
SOC_P×CSR_COM						0.052***			0.056**
SOC_P×CSR_COM						(3.845)			(2.062)
CSR_COM				−0.044***	−0.047***	−0.028***	−0.053**	−0.053**	−0.028*
CSR_COM				(−3.625)	(−3.779)	(−3.943)	(−2.133)	(−2.121)	(−1.954)
Constant	−0.411***	−0.077	−0.108	−0.442***	−0.455***	−0.470***	−0.807***	−0.976***	−0.812***
Constant	(−7.803)	(−0.808)	(−1.128)	(−4.110)	(−4.245)	(−4.322)	(−4.804)	(−5.233)	(−4.727)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Wu–Hausman test				0.002	0.001	0.001	0.004	0.002	0.002
Weak instrument test				0.000	0.000	0.000	0.001	0.000	0.000
Observations	12,285	12,278	12,278	12,278	12,278	12,278	12,278	12,278	12,278
Adjusted-R²	0.290	0.844	0.834	0.562	0.562	0.562	0.414	0.414	0.414

Panel B: GMM regressions
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	ROA	ROA	ROA	ROE	ROE	ROE
L.ROA/l.ROE	0.040	0.020	0.132	−0.019	−0.125	0.707
L.ROA/l.ROE	(0.55)	(0.28	(0.92)	(−0.08)	(−0.51)	(0.31)
CSR	−0.648***			−0.0162**
CSR	(−2.71)			(−2.36)
CSR×CSR_COM	0.508***			1.731***
CSR×CSR_COM	(2.66)			(2.76)
ENV		−0.220***			−0.730***
ENV		(−3.07)			(−2.90)
ENV×CSR_COM		0.172***			0.701***
ENV×CSR_COM		(2.61)			(2.90)
SOC			−0.241**			−0.775**
SOC			(−2.37)			(−2.45)
SOC×CSR_COM			0.204**			0.657***
SOC×CSR_COM			(2.28)			(2.69)
CSR_COM	−0.199*	−0.064	−0.062*	−0.753***	−0.303**	−0.226*
CSR_COM	(−2.45)	(−1.62)	(−1.70)	(−2.62)	(−1.99)	(−1.94)
Constant	0.128	−0.134	−0.177	1.572	0.296	0.296
Constant	(−0.21)	(−0.47)	(−0.65)	(1.52)	(0.29)	(0.29)
Baseline controls	YES	YES	YES	YES	YES	YES
Observations	10,915	10,915	10,915	10,915	10,915	10,915
Instruments	17	17	17	17	17	17
AR1	0.000***	0.004***	000***	0.057**	0.189	0.189
AR2	0.401	0.371	0.673	0.440	0.245	0.245
Hansen J.	0.736	0.511	0.385	0.170	0.399	0.399

Panel B: GMM regressions
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	ROA	ROA	ROA	ROE	ROE	ROE
L.ROA/l.ROE	0.040	0.020	0.132	−0.019	−0.125	0.707
L.ROA/l.ROE	(0.55)	(0.28	(0.92)	(−0.08)	(−0.51)	(0.31)
CSR	−0.648***			−0.0162**
CSR	(−2.71)			(−2.36)
CSR×CSR_COM	0.508***			1.731***
CSR×CSR_COM	(2.66)			(2.76)
ENV		−0.220***			−0.730***
ENV		(−3.07)			(−2.90)
ENV×CSR_COM		0.172***			0.701***
ENV×CSR_COM		(2.61)			(2.90)
SOC			−0.241**			−0.775**
SOC			(−2.37)			(−2.45)
SOC×CSR_COM			0.204**			0.657***
SOC×CSR_COM			(2.28)			(2.69)
CSR_COM	−0.199*	−0.064	−0.062*	−0.753***	−0.303**	−0.226*
CSR_COM	(−2.45)	(−1.62)	(−1.70)	(−2.62)	(−1.99)	(−1.94)
Constant	0.128	−0.134	−0.177	1.572	0.296	0.296
Constant	(−0.21)	(−0.47)	(−0.65)	(1.52)	(0.29)	(0.29)
Baseline controls	YES	YES	YES	YES	YES	YES
Observations	10,915	10,915	10,915	10,915	10,915	10,915
Instruments	17	17	17	17	17	17
AR1	0.000***	0.004***	000***	0.057**	0.189	0.189
AR2	0.401	0.371	0.673	0.440	0.245	0.245
Hansen J.	0.736	0.511	0.385	0.170	0.399	0.399

Panel C: Quality of matching
	Firms with CSR committee (treated)	Firms without CSR committee (control)	Treat - control	t-stats
Pscore	0.4107522	0.4107508	1.35e−06	0.0002
ROA	0.0344014	0.0419765	0.0075751	2.0459
Matching variables
LEVERAGE	0.2174666	0.2235537	−0.0060871	−0.9089
SIZE	15.03716	15.01636	0.0207981	0.3506
GROWTH	0.0933761	0.0907836	0.0025926	0.2153
B_SIZE	2.193252	2.16697	0.0262819	1.7275*
B_IND	0.3489718	0.3498637	−0.0008918	−0.0903
IQ	77.12744	78.24271	−1.115266	−3.5371***
GDP	2.955151	2.935896	0.0192549	0.2032

Panel C: Quality of matching
	Firms with CSR committee (treated)	Firms without CSR committee (control)	Treat - control	t-stats
Pscore	0.4107522	0.4107508	1.35e−06	0.0002
ROA	0.0344014	0.0419765	0.0075751	2.0459
Matching variables
LEVERAGE	0.2174666	0.2235537	−0.0060871	−0.9089
SIZE	15.03716	15.01636	0.0207981	0.3506
GROWTH	0.0933761	0.0907836	0.0025926	0.2153
B_SIZE	2.193252	2.16697	0.0262819	1.7275*
B_IND	0.3489718	0.3498637	−0.0008918	−0.0903
IQ	77.12744	78.24271	−1.115266	−3.5371***
GDP	2.955151	2.935896	0.0192549	0.2032

Panel D: PSM regression
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	ROA	ROA	ROA	ROE	ROE	ROE
CSR	0.015			0.057
CSR	(0.973)			(1.613)
CSR×CSR_COM	0.072***			0.114**
CSR×CSR_COM	(3.017)			(2.117)
ENV		−0.005			0.005
ENV		(−0.697)			(0.253)
ENV×CSR_COM		0.034***			0.050*
ENV×CSR_COM		(2.635)			(1.829)
SOC			−0.004			0.018
SOC			(−0.496)			(0.865)
SOC×CSR_COM			0.051***			0.081***
SOC×CSR_COM			(3.817)			(2.779)
CSR_COM	−0.051***	−0.029***	−0.038***	−0.080***	−0.042**	−0.062***
CSR_COM	(−3.966)	(−3.140)	(−4.577)	(−2.922)	(−2.191)	(−3.730)
Constant	−0.010	−0.011	−0.007	0.119	0.122	0.134
Constant	(−0.216)	(−0.233)	(−0.154)	(1.164)	(1.195)	(1.310)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes	Yes
Baseline controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	2,870	2,870	2,870	2,870	2,870	2,870
Adjusted-R²	0.151	0.146	0.151	0.0947	0.0904	0.0951

Panel D: PSM regression
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	ROA	ROA	ROA	ROE	ROE	ROE
CSR	0.015			0.057
CSR	(0.973)			(1.613)
CSR×CSR_COM	0.072***			0.114**
CSR×CSR_COM	(3.017)			(2.117)
ENV		−0.005			0.005
ENV		(−0.697)			(0.253)
ENV×CSR_COM		0.034***			0.050*
ENV×CSR_COM		(2.635)			(1.829)
SOC			−0.004			0.018
SOC			(−0.496)			(0.865)
SOC×CSR_COM			0.051***			0.081***
SOC×CSR_COM			(3.817)			(2.779)
CSR_COM	−0.051***	−0.029***	−0.038***	−0.080***	−0.042**	−0.062***
CSR_COM	(−3.966)	(−3.140)	(−4.577)	(−2.922)	(−2.191)	(−3.730)
Constant	−0.010	−0.011	−0.007	0.119	0.122	0.134
Constant	(−0.216)	(−0.233)	(−0.154)	(1.164)	(1.195)	(1.310)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes
Firm FE	Yes	Yes	Yes	Yes	Yes	Yes
Baseline controls	Yes	Yes	Yes	Yes	Yes	Yes
Observations	2,870	2,870	2,870	2,870	2,870	2,870
Adjusted-R²	0.151	0.146	0.151	0.0947	0.0904	0.0951

Note(s): This table presents results from various estimators addressing potential endogeneity concerns associated with CSR and FP. Panel A of the table reports 2SLS regression estimates. The average industry CSR, including its components, is used as the instrument for the firm’s CSR score. The predicted values of CSR, ENV and SOC (e.g. CSR_P, ENV_P, and SOC_P) estimated in the first-stage regression are used as independent variables in the second stage of the 2SLS (Columns 4–9). Panel B reports the estimates from the GMM model, an alternative approach for controlling endogeneity. Variable L. ROA/L, ROE indicates the lagged ROA and lagged ROE, respectively, with the baseline controls included as instruments. Panel C reports the quality of matching between the control and treated groups to confirm that the two samples' variables are not statistically different. Panel D reports estimates from propensity score matching (PSM) using the control and treatment groups to address concerns about sample selection bias. The regression models are estimated using year and firm fixed effects. Moreover, standard errors are adjusted for heteroscedasticity and serial correlation using firm-level clustering. t-statistics are reported in the parenthesis while the *, ** and *** denote statistical significance at the 10%, 5% and 1% levels, respectively

[ViewLarge]

Sharing Unavailable