Identifying the characteristics of China’s maritime trading partners on the basis of bilateral shipping connectivity: a cluster analysis

Figures & data

Table 1. Descriptive statistics of the emergent 3 clusters- China.

Clusters	Variables	No. of countries	Minimum	Maximum	Mean	Std. Deviation
Cluster 1	Numtrans	97	0	1	0.73	0.445
	CommonConnect	97	1	35	11.45	7.950
	Position	97	9.2736	66.8730	36.9549	11.4618
	Shipsize	97	221	6750	2424.78	1539.298
	NmaxCarrier	97	1	30	6.03	5.112
Cluster 2	Numtrans	27	0	0	0.00	0.00
	CommonConnect	27	9	50	28.70	10.384
	Position	27	27.8209	70.6258	51.8639	10.3694
	Shipsize	27	8586	11,356	10,169.04	748.366
	NmaxCarrier	27	2	31	13.41	7.712
Cluster 3	Numtrans	31	0	0	0.00	0.00
	CommonConnect	31	21	77	51.55	15.573
	Position	31	49.94	95.9271	74.1333	12.7798
	Shipsize	31	13,092	19,224	16,716.52	2410.575
	NmaxCarrier	31	2	59	25.52	15.914

Notes: Variable definitions are as follows: ‘Numtrans’ is The minimum number of transhipments needed to ship a product between a country pair i and j; ‘CommonConnect’ is the total number of services with zero transshipment provided by third countries to both countries i and j (common direct connections); ‘Position’ is the geometric mean of the total number of direct services to both countries i and j; ‘Shipsize’ is the size of the largest ship (measured in TEUs) serving on the shipping route between the country pair and; ‘NmaxCarrier’ is the total number of shipping lines operating on the shipping routes between the two countries (this variable reflects the degree of competition on the route).

Table 2. Results of a one-way ANOVA and t-test of cluster analysis.

	China		Singapore		Hong Kong
	F-value	p-value	F-value	p-value	t-value	p-value
Numtrans	77.660	0.000	36.824	0.000	2.030	0.044
Commonconnect	183.716	0.000	118.092	0.000	−7.606	0.000
Position	124.434	0.000	80.744	0.000	−6.910	0.000
Shipsize	950.539	0.000	908.958	0.000	−9.085	0.000
NmaxCarrier	59.197	0.000	39.148	0.000	−6.999	0.000

Notes: See Table 1 for variable definitions.

Figure 1. Trading partners of China based on three clusters that reflect three levels of connectivity: low, medium, and high. (Source: Authors’ own compilation).

Table 3. Descriptive statistics of the emergent 3 clusters—Singapore.

Clusters	Variables	No. of countries	Minimum	Maximum	Mean	Std. Deviation
Cluster 1	Numtrans	99	0	1	0.73	0.448
	CommonConnect	99	1	35	10.68	8.209
	Position	99	8.8318	70.0999	35.8345	11.7057
	Shipsize	99	221	6750	2414.38	1554.690
	NmaxCarrier	99	1	36	6.39	5.532
Cluster 2	Numtrans	41	0	1	0.20	0.401
	CommonConnect	41	9	58	31.71	13.182
	Position	41	26.4953	91.3565	55.9587	13.6835
	Shipsize	41	8586	14,080	11,458.05	1903.902
	NmaxCarrier	41	2	39	15.07	9.650
Cluster 3	Numtrans	15	0	0	0.00	0.00
	CommonConnect	15	21	74	49.27	16.624
	Position	15	47.56	87.4300	73.2031	14.0595
	Shipsize	15	16,652	19,224	18,410.93	1145.326
	NmaxCarrier	15	2	60	24.93	17.417

Notes: See Table 1 for variable definitions.

Figure 2. Trading partners of Singapore based on three clusters that reflect three levels of connectivity: low, medium, and high. (Source: Authors’ own compilation).

Table 4. Descriptive statistics of the emergent 2 clusters—Hong Kong.

Clusters	Variables	No. of observation	Minimum	Maximum	Mean	Std. Deviation
Cluster 1	Numtrans	143	0.00	1.00	0.5524	0.4989
	CommonConnect	143	1.00	69.00	17.3497	13.7052
	Position	143	8.72	90.18	41.7254	15.6591
	Shipsize	143	221.00	14,080.00	5097.8182	4392.3737
	NMaxcarrier	143	1.00	45.00	8.7273	7.6987
Cluster 2	Numtrans	12	0.00	1.00	0.2500	0.4522
	CommonConnect	12	17.00	68.00	48.7500	14.1365
	Position	12	47.75	86.30	73.7239	11.6985
	Shipsize	12	16,652.00	16,652.00	16,652.0000	0.0000
	NMaxcarrier	12	12.00	12.00	26.2500	14.1493

Notes: See Table 1 for variable definitions.

Figure 3. Trading partners of Hong Kong based on two clusters that reflect two levels of connectivity: low, and high. (Source: Authors’ own compilation).

Table 5. Characterisation of clusters by the means of the LSBCI variables—China.

	Clusters
Variables	1	2	3
Numtrans	High	Low	Low
CommonConnect	Low	Medium	High
Position	Low	Medium	High
Shipsize	Low	Medium	High
NmaxCarrier	Low	Medium	High

Notes: See Table 1 for variable definitions.

Table 6. Characterisation of clusters by the means of the LSBCI variables—Singapore.

	Clusters
Variables	1	2	3
Numtrans	High	Medium	Low
CommonConnect	Low	Medium	High
Position	Low	Medium	High
Shipsize	Low	Medium	High
NmaxCarrier	Low	Medium	High

Notes: See Table 1 for variable definitions.

Table 7. Characterisation of clusters by the means of the LSBCI variables—Hong Kong.

	Clusters
Variables	1	2
Numtrans	High	Low
CommonConnect	Low	High
Position	Low	High
Shipsize	Low	High
NmaxCarrier	Low	High

Notes:See Table 1 for variable definitions.

Because only two clusters emerged in this case, the characterisation of the means as low and high, as opposed to medium, has been undertaken by comparison with the mean values for China and Singapore.

Table 8. Results of estimated ordered logistic regression models.

	Ordered logistic regression
Independent variables	Model 1 (China)	Model 2 (Singapore)	Model 3 (Hong Kong)
	No. of observations = 121 LR Chi2 (3) = 72.99 Prob > chi2 = 0.0000 Pseudo R-square = 0.3006 Log Likelihood = —84.9179	No. of observations = 119 LR Chi2 (3) = 63.78 Prob > chi2 = 0.000 Pseudo R-square = 0.2838 Log Likelihood = −80.4845	No. of observations = 119 LR Chi2 (2) = 25.91 Prob > chi2 = 0.000 Pseudo R-square = 0.3533 Log Likelihood = −23.7137
	Co-efficients	Co-efficients	Co-efficients
LPI	1.5213*** (0.5209)	1.5203*** (0.5476)	1.7152* (0.9831)
Log export	0.5994*** (0.2053)	0.0311 (01267)	−^5
Log import	0.0157 (0.1030)	0.2581** (0.1249)	0.4349** (0.1819)
Dependent variable = Clusters (1 = low connectivity; 2 = medium connectivity; 3 = high connectivity).			Dependent variable = Clusters (1 = low connectivity; 2 = high connectivity).

***p < 0.01, **p < 0.05, *p < 0.10.

Table

Dependent Variable	Clusters (i)	Clusters (j)	Mean Difference (i-j) (China)	Mean difference (i-j) (Singapore)
Numtrans	1	2	0.732^a (0.077)	0.532^a (0.077)
	1	3	0.732^a (0.073)	0.727^a (0.115)
	2	3	0.000 (0.093)	0.195 (0.125)
CommonConnect	1	2	−17.250^a (2.243)	−21.031^a (1.988)
	1	3	−40.095^a (2.127)	−38.590^a (2.966)
	2	3	−22.845^a (2.713)	−17.559^a (3.231)
Position	1	2	−14.908^a (2.515)	−20.124^a (2.318)
	1	3	−37.178^a (2.384)	−37.368^a (3.458)
	2	3	−22.269^a (3.042)	−17.244^a (3.766)
Shipsize	1	2	−7744.254^a (360.124)	−9045.665^a (301.361)
	1	3	−14,291.73^a (341.466)	−15,998.55^a (449.596)
	2	3	−6547.479^a (435.673)	−6952.885^a (489.654)
NmaxCarrier	1	2	−7.376^a (1.905)	−8.679^a (1.578)
	1	3	−19.485^a (1.807)	−18.539^a (2.354)
	2	3	−12.109^a (2.305)	−9.860^a (2.564)

^ap < 0.01. Notes: See Table 1 for variable definitions. The values in paranthesis are standard errors.