Table A1 Robustness checks for...

Table A1

Robustness checks for endogeneity issues with PLS-SEM and Gaussian copula method

	Model I			Model II			Model III			Model IV			Model V
Independent variables	β	L.C.I.	U.C.I.	β	L.C.I.	U.C.I.	β	L.C.I.	U.C.I.	β	L.C.I.	U.C.I.	β	L.C.I.	U.C.I.
Digital selling platforms	−0.004	−0.084	0.077	−0.098	−0.265	0.041	−0.003	−0.081	0.064	−0.101	−0.271	0.040	−0.091	−0.240	0.060
Public procurement	−0.046	−0.057	−0.007	−0.047	−0.058	−0.006	−0.035	−0.066	−0.013	−0.036	−0.069	−0.016	−0.067	−0.123	−0.051
Interaction term													0.032	0.002	0.114
Control variables
Revenue	0.393	0.341	0.438	0.393	0.342	0.442	0.393	0.347	0.445	0.394	0.343	0.443	0.393	0.339	0.441
Employees	−0.000	−0.046	0.046	0.000	−0.049	0.046	−0.002	−0.048	0.047	−0.001	−0.050	0.047	−0.000	−0.047	0.049
Employees trend	0.023	−0.010	0.046	0.023	−0.011	0.046	0.023	−0.010	0.044	0.024	−0.010	0.046	0.023	−0.009	0.046
Local units	−0.061	−0.080	−0.039	−0.061	−0.079	−0.040	−0.061	−0.082	−0.042	−0.061	−0.083	−0.042	−0.061	−0.081	−0.040
Innovation score	−0.112	−0.146	−0.081	−0.117	−0.150	−0.083	−0.112	−0.145	−0.082	−0.117	−0.148	−0.082	−0.117	−0.151	−0.084
Accredia certification	0.140	0.039	0.249	0.137	0.033	0.252	0.139	0.022	0.248	0.136	0.029	0.248	0.137	0.026	0.246
Gaussian copula for public procurement							−0.051	−0.266	0.080	−0.052	−0.263	0.085
Gaussian copula for digital selling platforms				0.046	−0.015	0.117				0.048	−0.015	0.115	0.043	−0.019	0.112
R squared	0.235				0.235		0.235			0.236			0.236

	Model I			Model II			Model III			Model IV			Model V
Independent variables	β	L.C.I.	U.C.I.	β	L.C.I.	U.C.I.	β	L.C.I.	U.C.I.	β	L.C.I.	U.C.I.	β	L.C.I.	U.C.I.
Digital selling platforms	−0.004	−0.084	0.077	−0.098	−0.265	0.041	−0.003	−0.081	0.064	−0.101	−0.271	0.040	−0.091	−0.240	0.060
Public procurement	−0.046	−0.057	−0.007	−0.047	−0.058	−0.006	−0.035	−0.066	−0.013	−0.036	−0.069	−0.016	−0.067	−0.123	−0.051
Interaction term													0.032	0.002	0.114
Control variables
Revenue	0.393	0.341	0.438	0.393	0.342	0.442	0.393	0.347	0.445	0.394	0.343	0.443	0.393	0.339	0.441
Employees	−0.000	−0.046	0.046	0.000	−0.049	0.046	−0.002	−0.048	0.047	−0.001	−0.050	0.047	−0.000	−0.047	0.049
Employees trend	0.023	−0.010	0.046	0.023	−0.011	0.046	0.023	−0.010	0.044	0.024	−0.010	0.046	0.023	−0.009	0.046
Local units	−0.061	−0.080	−0.039	−0.061	−0.079	−0.040	−0.061	−0.082	−0.042	−0.061	−0.083	−0.042	−0.061	−0.081	−0.040
Innovation score	−0.112	−0.146	−0.081	−0.117	−0.150	−0.083	−0.112	−0.145	−0.082	−0.117	−0.148	−0.082	−0.117	−0.151	−0.084
Accredia certification	0.140	0.039	0.249	0.137	0.033	0.252	0.139	0.022	0.248	0.136	0.029	0.248	0.137	0.026	0.246
Gaussian copula for public procurement							−0.051	−0.266	0.080	−0.052	−0.263	0.085
Gaussian copula for digital selling platforms				0.046	−0.015	0.117				0.048	−0.015	0.115	0.043	−0.019	0.112
R squared	0.235				0.235		0.235			0.236			0.236

Note(s): L.C.I. and U.C.I. stand for lower and upper confidence intervals

2,186 observations. Bootstrapping parameters include 2,000 samples, one-tailed test type, randomized seed, C.I. method Bias-corrected and accelerated bootstrap and C.I. at 90. The software used is SmartPLS 4. All the models have as D.V. "Export intensity" since the software does not allow calculating a single latent factor with both dummy and percentage items. Model I represents the PLS-SEM model with no interaction; Models II, III and IV are PLS-SEM models testing for endogeneity issues adopting the Gaussian copula (G.C.) method; all confirm no endogeneity issues. Model V tests with PLS-SEM the full model with the interaction term, keeping the G.C. just for digital selling platforms since the software in use does not allow measuring a G.C. in a relationship under moderation effect. The robustness checks confirm H2 and H3 but provide non-significant results for H1

Source(s): Authors’ own work

[ViewLarge]