Board gender diversity and corporate social responsibility

Chang, Yuan; Wu, Kun-Tsung; Lin, Shu-Hui; Lin, Chia-Jung

doi:10.1186/s40991-024-00095-x

International Journal of Corporate Social Responsibility

Table 10 Additional tests for the estimation of the relationship between board gender diversity and CSR performance

From: Board gender diversity and corporate social responsibility

	csrdummy	csrcumu	csrcont	csrovlp	scv	sroa	scvps
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
Panel A. Blau Index as Alternative Measure of Board Gender Diversity
fdblau	0.00794	0.0654	0.00698***	0.00108	0.561**	0.138	0.796***
fdblau	(1.11)	(1.44)	(2.78)	(0.35)	(2.25)	(0.58)	(4.88)
Panel B. Industry Adjustment of Female Director Ratio
fdrindjust	0.0000147	0.000272	0.0000852***	-0.0000131	0.00709**	-0.00556*	0.00422**
fdrindjust	(0.17)	(0.49)	(2.66)	(-0.45)	(2.14)	(-1.73)	(2.05)
Panel C. Two-stage Least Square Instrumental Variable Estimation (IV: Last-period Female Director Ratio)
fdr_HAT1	0.0000742	0.000349	0.0000740**	-0.0000196	0.00519	-0.00255	0.00748***
fdr_HAT1	(0.73)	(0.55)	(2.21)	(-0.57)	(1.37)	(-0.69)	(3.19)
Panel D. Two-stage Least Square Instrumental Variable Estimation (IV: Industry Average Female Director Ratio)
fdr _HAT2	0.00309***	0.0151***	-0.0000630	0.000212	0.00470	0.162***	0.132***
fdr _HAT2	(5.70)	(5.44)	(-1.02)	(1.07)	(0.25)	(6.29)	(7.20)
Panel E. Nonlinear (Quadratic) Effects of Board Gender Diversity on CSR Performance
fdr	0.0000877	0.00206	0.000202***	0.000174*	0.0121	0.0194***	0.0307***
fdr	(0.39)	(1.57)	(2.72)	(1.84)	(1.54)	(2.58)	(5.98)
fdrsq	0.000000619	-0.0000363	-0.00000332*	-0.00000494**	-0.000140	-0.000545***	-0.000617***
fdrsq	(0.11)	(-1.08)	(-1.67)	(-2.16)	(-0.72)	(-3.00)	(-5.05)
Panel F. Heckman Two-stage Estimation
fdd	0.251***	2.511***	0.0847***	0.0752***	-2.777**	5.138***	17.94***
fdd	(6.16)	(7.91)	(6.02)	(4.72)	(-2.48)	(4.49)	(9.64)

This table reports several additional tests of the effects of board gender diversity on firm’s CSR performance. Panel A shows the regression result of how the Blau Index (fdblau), recalculated for board gender diversity, affects CSR performance (only reporting the estimates of the main explanatory variable). Panel B reports the regression estimates of how the industry-adjusted female director ratio (fdrindjust) affects CSR performance (also reporting estimates of the main explanatory variable only). Panel C reports the result of two-stage least square instrumental variable estimation of whether board gender diversity influences CSR performance. In the first stage (not reported), the last-period female director ratio and control variables in regression equation (1) are used to predict the current female director ratio, resulting in the fitted value for the female director ratio (fdr_HAT1), which becomes the main explanatory variable for predicting CSR performance in the second stage. Control variables in the second stage estimation are the same as in regression equation (1) (estimates of control variables not reported). Panel D reports the results of two-stage instrumental variable estimation of whether board gender diversity affects CSR performance. In the first stage (not reported), the average female director ratio for the same industry and control variables in regression equation (1) are used to predict the current female director ratio, resulting in the fitted value for the female board member ratio (fdr_HAT2), which becomes the main explanatory variable for predicting CSR performance in the second stage. Control variables in the second stage estimation are the same as in regression equation (1) (estimates of control variables not reported). Panel E reports the results of a quadratic regression estimation of whether board gender diversity affects CSR performance, with the main explanatory variables including the female director ratio (fdr) and the square of the female director ratio (fdrsq). Panel F reports the results of a Heckman two-stage estimation of whether board gender diversity affects CSR performance. In the first stage, a probit model determines whether a firm has female board members, with explanatory variables including firm size, return on asset, independent director ratio, institutional investors’ shareholdings, and a dummy variable of whether a firm is family-controlled. After estimation, an bias-correction term (lambda) for self-selection bias is obtained and becomes an additional explanatory variable for the second stage. In the second stage estimation, explanatory variables include female director dummy variable (fdd), the additional selection bias correction term (lambda), and control variables that are the same as in regression equation (1). The Heckman two-stage estimation adopts maximum likelihood estimation (MLE)

Back to article page