Ports and Shipping

Empirical analysis and findings

Structural equation model

As the measurement model and reliability tests have confirmed validity and reliability, the structural equation model is proceeded. The parameter estimations, model fit indices, and the results of hypotheses proposed in Section 2.2 are presented and discussed.

The SEM, including estimated standardised factor loadings and regression coefficients, is presented Fig. 2, along with their respective paths. All 10 factors loadings are above the recommended level of 0.70 and are statistically significant. The structural model has a good fit with chi-square (χ²) of 73.78, and the ratio of χ² and degrees of freedom (that is, 73.78/30 = 2.46 < 3) is within the required level recommended by Bollen and Long (1993). The Adjusted Goodness of Fit Index (AGFI) is 0.99, meaning the estimated model predicted 99% of the variances and co-variances in the observed data. Moreover, other fit index measures, such as CFI (0.98) and TLI (0.97), are well above the minimum requirements. Finally, the root mean-square error of approximation (RMSEA = 0.08), as well as the standardised root mean square residual (SRMR = 0.02), confirmed a good fit of the SEM.

Structural Equation Model.

*p < 0.05, **p < 0.01, ***p < 0.001;

Model –fit: χ² (30) = 73.78, AGFI = 0.99, CFI = 0.98, TLI = 0.97, RMSEA = 0.08, SRMR = 0.02;

Note that the correlation curve between LPIEC and LPIQT represents correlation among their error terms.

After confirming the fitness of the proposed SEM, we estimated the hypothesised relationships between the latent constructs. The estimates of the hypothesised relationships and their significance are presented in Table 4. Before establishing the mediation relationship among the variables (that is in H1d, H1e, H1f and H2c), the direct association between independent and dependent variable as well as its association with the mediation variable was confirmed.

Table 4 Results of Structural Equation Modelling

Hypotheses	Regression paths	Std. Estimates	S.E.	C.R.	Remarks
H1 (a)	QPI → LP	0.66***	0.04	12.93	Supported
H1 (b)	QPI → ST	−0.10	0.40	−1.60	Not supported
H1 (c)	QPI → NE	0.17**	0.25	2.80	Supported
H1 (d)	QPI → LP → NE	0.45***	0.20	9.31	Supported
H1 (e)	QPI → ST → NE	0.01	0.04	1.06	Not supported
H1 (f)	QPI → LP → ST → NE	−0.04	0.10	−1.83	Not supported
H2 (a)	LP → ST	0.66***	0.65	8.99	Supported
H2 (b)	LP → NE	0.67***	0.38	10.53	Supported
H2 (c)	LP → ST → NE	−0.06	0.21	−1.81	Not supported
H3	ST → NE	− 0.09	0.03	− 1.95	Not supported

Std. Estimate standardized estimates, S.E. standard error, C.R. critical ratio
*p < 0.05, **p < 0.01, ***p < 0.001

Based on the statistical significance of the regression coefficients depicted in Table 4, this study finds support for H1 (a), H1 (c), H1 (d), H2 (a) and H2 (b). The other hypotheses are not supported. Thus, quality of port infrastructure has a positive effect on logistics performance and the national economy. Logistics performance has a positive effect on seaborne trade and the national economy. The effects of quality of port infrastructure and logistics performance on the national economy were found to be significant. While the mediation effect of port infrastructure quality on the national economy via logistics performance is found significant, the mediation effect via seaborne trade was found to be insignificant. Also, the mediation effect of logistics performance on the national economy through seaborne trade was found to be insignificant. To further investigate whether the findings are similar for both developed and developing economies, a multi-group SEM is formed and estimated in the next section.