Advertisement

Relationship Between Shipping Accessibility and Maritime Transport Demand: the Case of Mainland China

  • Liquan Guo
  • Zhongzhen Yang
Article

Abstract

The relationship between shipping accessibility and maritime transport demand is studied based on the relationship between production and consumption and stochastic frontier analysis. First, the consumption market of the products produced in a region is classified into the local market, non-local domestic market and foreign market. Then, to obtain the potential export volumes, the production function, extended linear expenditure system model, and multi-regional input-output model are built to calculate the regional production, the consumption in local market, and the consumption in non-local market, respectively. Second, the impact of potential export volumes and shipping accessibility on maritime transport demand is studied based on the stochastic frontier gravity model. Finally, the data from 2006 to 2012 for eight domestic regions and eight overseas regions were used to conduct an empirical study. The findings indicate that the maritime transport demand is positively correlated with shipping accessibility and that the region with higher accessibility has the greater demand. Furthermore, adding shipping lines in Quanzhou Port had the largest positive effect on Chinese exports.

Keywords

Production function Consuming market Shipping accessibility Maritime transport demand Stochastic frontier analysis 

1 Introduction

As the most important freight transportation mode among continents and non-bordering countries, the ocean shipping capacity of a country directly affects its trade competitiveness (Yang et al. 2014a). Since the “Reform and Opening” policy was enacted in the 1980s, China’s seaborne trade shipments have increased from 44 million tons in 1980 to 2.48 billion tons in 2013, and its share of the world’s seaborne trade shipments has risen from 1.2% to 26% (WTO 2014). During the same period, China’s total imports and exports have expanded from 40 billion US dollars to 4.16 trillion dollars, and China’s ratio of the world’s total imports and exports has grown from 0.9% to 11% (China Statistical Year Book 2014). It can be seen that China’s shipping industry and foreign trade encourage each other, and both of them have grown considerably. According to the interactive relationship, the maritime transport demand is a key issue in determining the maritime transport supply. Based on the maritime transport demand, the decision makers can determine the scale of the port groups, optimize the port distribution and shipping line network, and reasonably construct and operate the ports and their hinterland access system. The maritime transport demand concerning foreign trade arises mainly from the potential imports and exports, and the potential exports of a region depend mainly on its production capacity because the products can only be sold in local markets, non-local domestic markets and foreign markets.

The foreign trade shipping accessibility is another key factor affecting the maritime transport demand between a region and its main trading partners (Bensassi et al. 2015). The foreign trade shipping accessibility illustrates the ease with which the export goods can be transported from production sites to consuming market sites in overseas regions. Higher foreign trade shipping accessibility indicates that it is more convenient for export goods to be transported to the foreign markets, and thus, the actual maritime transport demand is higher.

In the above context, the main focuses of this paper are to study the methodology of determining the potential maritime transport demand, to analyze the relationship between shipping accessibility and maritime transport demand, and to assess whether the shipping accessibility fits the maritime transport demand in China. For this purpose, we first classify the consumption market of the products produced in a region and then use the production function, consumption function, and input-output model to calculate the potential export demand with the disaggregate manufacturing industry data. Then, the stochastic frontier gravity model is proposed to measure how much the potential export demand can become the actual export volumes under the influence of shipping accessibility for each sector.

The remainder of the paper is organized as follows. Section 2 presents a literature review related to the relationship between transportation and trade. Section 3 describes the method for determining the potential export demand and the structure of the stochastic frontier gravity model. Section 4 introduces the definition and measure of shipping accessibility. Section 5 presents an empirical study and an analysis of the results. Section 6 concludes with the key findings and limitations of this study.

2 Literature Review

Numerous studies have investigated the relationship between transportation and trade, among which a number of papers have analyzed the relationship between connectivity or accessibility and international trade (According to Calatayud et al. (2016), connectivity and accessibility can be used interchangeably or as synonyms). For example, Thill and Lim (2010) evaluated the performance of an intermodal network by an accessibility indicator, which can be defined as the attractiveness of the place in question, considering the trade and interaction opportunities offered by other locations and the impedances to reaching these locations on a transportation network. Alamá-Sabater et al. (2013) used a spatial autoregressive model to analyze the role of transport connectivity in interregional trade flows in Spain, and the results indicated that interregional trade among Spanish regions benefits from the transport networks of its neighbors. Ramli and Ismail (2014) analyzed the impact of infrastructure provisions on trade costs and access to markets and concluded that improvements to basic infrastructure increases the accessibility of goods from producers to consumers.

Meanwhile, several studies have focused on the impact of transport cost on bilateral trade, finding that the transport cost is an important determinant between a country and its main trading partners. For instance, Baier and Bergstrand (2001) studied the relative effects of transport cost reduction, tariff liberalization, and income growth on the growth of world trade among several OECD countries between the late 1950s and late 1980s. The empirical results suggested that income growth, tariff-rate reductions, and transport cost reductions explain approximately 67%, 25%, and 8% of the average world trade growth of their post-World War II sample, respectively. Martínez-Zarzoso et al. (2003) analyzed the impact of transport costs on Spanish ceramic exports and found that higher transport costs significantly deter trade and that distance is not a good proxy for transport costs in the ceramics sector. Hummels (2007) argued that the rapidly declining cost of transportation has been a critical input into a second era of economic globalization. Carrasco et al. (2009) studied the impact of transport cost reductions on the export behavior of firms, accounting for the role of entry costs and other firm characteristics. Márquez-Ramos et al. (2011) studied the impact of maritime networks and the structure of their services on maritime transport cost and analyzed the relationship between maritime transport cost and maritime trade. Chi (2016) explored the long-run impacts of gross domestic product, exchange rate, and transport costs on bilateral air and ocean freight flows between the US and China using export and import data over the period of 2003: Q1 to 2014: Q2.

Other papers have demonstrated the importance of infrastructure and its efficiency on trade facilitation. For example, Bougheas et al. (1999) utilized an augmented gravity model and data from European countries to provide empirical evidence that there is a positive relationship between the level of infrastructure and volume of trade. Limao and Venables (2001) investigated the dependence of transport costs on geography and transport infrastructure and showed that transport infrastructure is a key determinant of transport costs, particularly for landlocked countries. Sánchez et al. (2003) and Clark et al. (2004) analyzed the relationship between port efficiency and international trade and found that an improvement in port efficiency from 25% to 75% reduced maritime transport costs by more than 12% and would result in a 25% increase in bilateral trade. Wilmsmeier and Hoffmann (2008) analyzed the impacts of port infrastructure and liner shipping connectivity on intra-Caribbean freight rates and the relationship among liner shipping services, port infrastructure endowment and liner shipping freight rates. Portugal-Perez and Wilson (2012) derived four new indicators for more than 100 countries over the 2004–07 period to estimate the impact of aggregate indicators of “soft” and “hard” infrastructure on the export performance of developing countries. The estimates indicated that trade facilitation reforms improve the export performance of developing countries. Xu et al. (2012) performed an empirical study to measure the effect of transport costs and port efficiency on China’s exports. Fabling et al. (2013) investigated the impact of port infrastructure on exporter behavior and found that firms use the new infrastructure in conjunction with the existing port to mitigate capacity constraints and/or access a greater range of transport options. Bensassi et al. (2015) used an augmented gravity model to analyze the relationship between logistics infrastructure and Spanish regional exports and noted that the number, size and quality of logistical facilities positively influence export flows. Coşar and Demir (2016) studied how internal transportation infrastructure affects Turkish access to international markets.

Although previous studies have investigated the relationship between transportation and trade and have found that reductions in transport cost and improvements to transport services have a considerable influence on trade facilitation, certain problems remain unresolved. The majority of the studies used an aggregate model to investigate the relationship between GDP or total exports and transport cost, and none have focused on the sensitivity of the export demand of different products to transport cost. To the best of our knowledge, only two papers have used a disaggregate model to analyze the elasticity of the export demands of different sectors with regard to transport costs. Martínez-Zarzoso and Suárez-Burguet (2005) used a gravity model for sectoral imports for five Latin American countries from the European Union and found that a large distance and poor importer’s infrastructure increased transport costs considerably, whereas a higher volume of trade had the opposite effect. Martinez-Zarzoso et al. (2008) analyzed the determinants of transport costs and investigated their influence on international trade with a sample of disaggregate trade data. The results showed that trade flows are negatively correlated with transport costs, and high value-added industries (e.g., household appliances and vehicle parts) are more sensitive to changes in transport cost than low value-added industries (e.g., agro-industry and ceramics). However, these two papers did not adequately consider the impact of potential export demand on actual export volumes. The potential export demand is the upper limit of the actual export volumes, and it will affect the actual export volumes together with transport cost. Thus, this paper contributes to the literature by providing a method for determining the potential export demand and by investigating the impact of potential export demand and shipping accessibility on the actual maritime transport demand for each sector. Our findings could be used to help policymakers, port agencies, and exporters to design effective strategies for improving the development of China’s foreign trade and shipping industry.

3 Sector Outputs and Consumption Volumes

3.1 Sector Outputs

The Cobb-Douglas production function can be used to calculate the output of each sector (Cook and Zhu 2014), and the logarithm of the function can be expressed as
$$ \ln {Y}_{ri t}=\ln {V}_{ri}+{\alpha}_{ri}\ln {K}_{ri t}+{\beta}_{ri}\ln {L}_{ri t} $$
(1)
where r, i and t represent domestic region, sector and the year, respectively; Y denotes the output of the sector; V denotes the total factor productivity; K denotes the capital input; L denotes the labor input; and α and β represent the output elasticity of capital input and labor input, respectively.

3.2 Consumption Volumes

As shown in Fig. 1, the market for the products produced at a site may contain a local market, non-local domestic market and foreign market (Huang 2008). Moreover, Eq. (2) can be used to measure the relationship between the sector outputs and the volumes consumed in the markets:
$$ {Y}_{rit}={T}_{1 rit}+{T}_{2 rit}+{T}_{3 rit} $$
(2)
where T 1rit denotes the volumes consumed in the local market, T 2rit denotes the volumes consumed in the non-local domestic market, and T 3rit denotes the potential export volumes. The method to calculate T 1rit and T 2rit and the actual export volumes based on potential export volumes is provided below.
Fig. 1

Relationship between production and consumption

3.2.1 Volumes Consumed in the Local Market

The consumption function is a suitable method for studying the household consuming expenditure (Wang and Liu 2015). Therefore, we use the consumption function, which is based on the extended linear expenditure system (ELES) model, to calculate the total volumes consumed in the local market. The ELES model analyzes the impact of household disposable income on consumption and takes the household consuming expenditures as the interactive behavior (Wang et al. 2014). The ELES model is as follows:
$$ {O}_{rg t}={b}_{rg t}+{U}_{rg}\left[{I}_{rt}-\sum \limits_l{b}_{rlt}\right],\kern0.5em g,l=1,2,3.\dots, n,g\ne l $$
(3)
where g and l represent expenditure categories, O denotes household consuming expenditure, b denotes basic demand, U denotes the marginal propensity to consume, and I denotes disposable income. Eq. (3) can be rewritten as
$$ {O}_{rg t}=\left[{b}_{rg t}-{U}_{rg}\sum \limits_l{b}_{rlt}\right]+{U}_{rg}{I}_{rt}+{\varepsilon}_{rg t}={d}_{rg}+{U}_{rg}{I}_{rt}+{\varepsilon}_{rg t} $$
(4)

Here, d rg and U rg can be estimated with the least squares estimation. Then, the total volumes of each sector consumed in local market could be determined based on the corresponding relationship between the products of each sector and the household expenditure category.

3.2.2 Volumes Consumed in the Non-Local Domestic Market

The multi-regional input-output model can be used to calculate the total volumes consumed in non-local domestic regions (Zhao and Kockelman 2004), and the commodity flows between domestic regions in the model can be expressed as
$$ {T}_{2 rit}=\sum \limits_q{x}_{rqit}\kern0.5em r,q\in \left\{1,2,\dots, R\right\} $$
(5)
where x rqit denotes the product flow of sector i from production site r to domestic site q in year t. Moreover, based on the row model of the multi-regional input-output model, the method to calculate x rqit is as follows:
$$ {x}_{rqit}=\sum \limits_{j=1}^n{a}_{qijt}\sum \limits_{m=1}^R{x}_{qmjt}+{W}_{qit}\kern0.5em \forall i $$
(6)
where a qijt denotes the Leontief technological coefficient at domestic site q in year t and W qit denotes the final consumption of sector i at domestic site q in year t. The output of a sector consumed in the non-local domestic market (T 2rit ) can be calculated using Eq. (6) and the Leontief inverse matrix obtained from multi-regional input-output tables.

3.2.3 Actual Export Volumes

Many factors may affect the actual export volumes of each sector (e.g., the potential export volumes, import demand of overseas regions, trade barrier, and commodity price). The gravity model and stochastic frontier gravity model are the two popular models used to calculate the bilateral trade volumes (Lu and Zhao 2010). The gravity model was first introduced to the field of international trade by Tinbergen (1962), and the gravity model supposed that the bilateral trade volumes are proportional to the GDP and inversely proportional to transport impedance. The gravity model cannot indicate the subjective trade resistance and cannot determine the impact of policy shifts on bilateral trade volumes. To overcome the inherent defects of the gravity model, Aigner et al. (1977) proposed the stochastic frontier gravity model based on the trade efficiency, trade potential and trade performance. The trade efficiency, which is similar to the popular concept of technical efficiency in production economics (Drysdale et al. 2000), can be used to measure the policy and the other factors that cannot be easily observed or quantified. The trade potential is the maximum trade volume between two countries. The trade performance is the actual trade volume, which is determined by the trade efficiency and trade potential. The stochastic frontier gravity model is as follows:
$$ {T}_{rsit}=f\left(T,\beta \right)\cdot \exp \left({v}_{rsit}-{u}_{rsit}\right) $$
(7)
$$ {TE}_{rsit}=\exp \left[{X}_{rsit}\right]/\exp \left[f\left(T,\beta \right)+{v}_{rsit}\right]=\exp \left(-{u}_{rsit}\right) $$
(8)
where T rsit is the trade volume of sector i from domestic site r to overseas site s in year t, f(T, β) + v rsit denotes the trade potential and T is a vector of explanatory variables that represent key factors influencing actual trade volumes (mainly containing potential trade volumes in a domestic region, import volumes in overseas regions, population, and shipping accessibility); β is a vector parameter to be estimated; v rsit is a random error, which is distributed as a normal random variable with mean zero and variance \( {\sigma}_v^2 \); TE rsit denotes the trade efficiency; and u rsit denotes the trade inefficiency variable, which is assumed to follow a half-normal distribution or exponential distribution. The logarithm of Eq. (7) is
$$ \ln {T}_{rs it}={\beta}_{0i}+{\beta}_{1i}\ln {T}_{3 rit}+{\beta}_{2i}\ln {Pop}_{rt}+{\beta}_{3i}\ln {T}_{sit}+{\beta}_{4i}\ln {Pop}_{st}+{\beta}_{5i}\ln {A}_{rs}+{v}_{rs it}-{u}_{rs it} $$
(9)
where β0iβ5i are the parameters to be estimated; Pop denotes the population; T sit denotes the total import volumes of sector i in overseas site s in year t; and A rs denotes the shipping accessibility from the domestic site r to overseas site s.
A number of factors may affect the trade efficiency (e.g., linguistic differences, degree of democratization, customs duties, commodity price, and quality of transport infrastructure). Because the linguistic differences and degree of democratization cannot be easily quantified, we use the customs duties, commodity price and quality of transport infrastructure to measure the trade efficiency (TE rsit ):
$$ {TE}_{rsit}={\lambda}_{0i}+{\lambda}_{1i}{QPI}_{rt}+{\lambda}_{2i}{QPI}_{st}+{\lambda}_{3i}{PRATIO}_{rsit} $$
(10)
where λ0iλ3i are the estimated parameters; QPI rt and QPI st denote the quality of the port infrastructure at domestic site r and overseas site s, respectively; and PRATIO rsit denotes the ratio of the commodity price in overseas site to the C.I.F price of China’s export goods, which is expressed as
$$ {PRATIO}_{rs it}=\frac{p_{sit}}{p_{rit}+{C}_{rs}+{tariff}_{rs it}} $$
(11)
where p sit denotes the commodity price of sector i in the overseas site s in year t; p rit denotes the local commodity price of sector i in domestic site r in year t; C rs denotes the generalized transport cost from domestic site r to overseas site s; and tariff rsit denotes the custom duties of sector i from domestic site r to overseas site s in year t. The local commodity price, generalized transport cost and tariff constitute the C.I.F price of China’s export goods.

4 Shipping Accessibility

Figure 2 illustrates the entire trip of export goods being transported to overseas regions from a domestic production site, which is comprised of three parts: a highway/railway trip to the gateway ports, in-port operation and an ocean shipping trip to the gateway ports in the overseas regions. The ocean shipping trip is the dominant trip (Tsiotas and Polyzos 2015), and its quality of services will greatly influence the convenience of the entire trip. Here, we define the convenience of transporting export goods from the domestic production site to the overseas site as the shipping accessibility. The measure of the shipping accessibility based on the entire trip is provided below.
Fig. 2

Example of available paths for export goods to overseas sites

4.1 Discrete Path Choice

The export goods can be transported to overseas regions by multiple paths. The shipper will choose the alternative paths based on their utilities (a type of discrete path choice behavior). An alternative path is more likely to be chosen if it has a higher utility. If city A in Fig. 2 is chosen as the spatial unit, there will be three alternative paths available from city A to overseas site s, namely, R1: city A → port C → port F; R2: city A → port D → port F; and R3: city A → port E → port E. If the utilities of the three paths are ordered as R1 > R2 > R3, then the choice probabilities are R1 > R2 > R3. Because each alternative path is independent, the multinomial logit model can be used to calculate the choice probabilities as follows (Tang et al. 2011; Palacio et al. 2016):
$$ {P}_{ms}^k=\exp \left(\theta {V}_{ms}^k\right)/\left({\sum}_{k\in {E}_{ms}}\exp \left(\theta {V}_{ms}^k\right)\right) $$
(12)
where \( {P}_{ms}^k \) is the choice probability of path k from city m to overseas site s and \( {V}_{ms}^k \) is the direct utility of path k, which is determined by the generalized transport cost. If the direct utility of the path with the highest generalized transportation cost is assumed to be zero, then \( {V}_{ms}^k={Q}_{ms}-{C}_{ms}^k \), where \( {C}_{ms}^k \) is the generalized transportation cost of path k from city m to overseas site s and Q ms is the largest value of the generalized cost among all paths. E ms is the set of available paths from city m to overseas site s, and θ is an estimated parameter, which is often set to be 1 in empirical studies.

With the probabilities of each path choice known, the weighted average generalized transportation cost of all paths from city m to overseas site s can be calculated as \( {C}_{ms}={\sum}_{k\in {E}_{ms}}{C}_{ms}^k\cdot {P}_{ms}^k \).

If domestic region r contains several cities, the generalized transport cost from region r to overseas site s (C rs ) can be regarded as the weighted average generalized for each city in region r to overseas site s. Because a higher generalized transport cost results in lower shipping accessibility, the shipping accessibility from domestic region r to overseas site s can be calculated as A rs  = 1/C rs . C rs is expressed as follows:
$$ {C}_{rs}={\sum}_{m\in {B}_r}{C}_{ms}\cdot \frac{Pop_m}{\sum_{m\in {B}_r}{Pop}_m} $$
(13)
where Pop m is the population of city m and B r is the city set in domestic region r.

4.2 Generalized Transport Cost of a Path

The generalized transport cost of path k from a city to overseas sites is as follows:
$$ {C}_{ms}^k={C}_{ms}^{k, land}+{C}_{ms}^{k, port}+{C}_{ms}^{k, sea} $$
(14)
where \( {C}_{ms}^{k, land} \) is the generalized cost of land transportation, \( {C}_{ms}^{k, port} \) is the generalized cost of port operation, and \( {C}_{ms}^{k, sea} \) is the generalized cost of ocean shipping.

4.2.1 Generalized Cost of Land Transportation

According to Zhang and Huang (2009), the land transport modes from hinterland to gateway ports can be classified into four type, namely, highway, railway, inland waterway, and domestic coastal shipping line. For the whole country of China, the modal splits of inland waterway and domestic coastal shipping line are relatively low (approximately 1%) based on China Ports Year Book, thus it is rational to suppose that the export goods from domestic sites are transported to gateway ports by highway and railway only. If port l is on path k, then
$$ {C}_{ms}^{k, land}=\eta \cdot \left(\phi \frac{D_{ml}^{road}}{{\overline{V}}^{road}}+{\omega}_1{D}_{ml}^{road}\right)+\left(1-\eta \right)\cdot \left(\phi \frac{D_{ml}^{rail}}{{\overline{V}}^{rail}}+{\omega}_2{D}_{ml}^{rail}\right) $$
(15)
where η is the highway modal split, ϕ is the value of time, \( {D}_{ml}^{road} \) is the shortest path distance on highway network from city m to port l, \( {\overline{V}}^{road} \) is the average highway running speed, ω1 is the unit transport cost of one TEU by highway, \( {D}_{ml}^{rail} \) is the shortest path distance on the railway network from city m to port l, \( {\overline{V}}^{rail} \) is the average railway running speed, and ω2 is the unit transport cost of one TEU by railway.

4.2.2 Generalized Cost of Port Operation

The generalized cost of port operation is determined by the port charges, frequency of ship calls, and scale of the shipping line network for the port. If port l is on path k from city m to overseas site s, then \( {C}_{ms}^{k, port} \) can be calculated as follows:
$$ {C}_{ms}^{k, port}=\phi \left(1/{\sum}_{h\in {H}_l^s}{f}_{ls}^h\right)+{F}_l $$
(16)
where \( {f}_{ls}^h \) is the frequency of shipping line h from port l to overseas site s, \( {H}_l^s \) is the set of available shipping lines from port l to overseas site s, and F l is the charges at port l.

4.2.3 Generalized Cost of Ocean Shipping

The shipping cost should be determined by the price of the shipping lines. Here, we use the average price and average running time of all shipping lines as the shipping cost and shipping time from gateway port to overseas regions, respectively (Wang and Cullinane 2008; Caschili et al. 2014), which can be calculated as
$$ {C}_{ms}^{k, sea}=\phi \left({\sum}_{h\in {H}_l^s}{T}_{ls}^h/{n}_l^s\right)+{\sum}_{h\in {H}_l^s}{F}_{ls}^h/{n}_l^s $$
(17)
where \( {T}_{ls}^h \) is the vessel running time of shipping line h from port l to overseas site s, \( {F}_{ls}^h \) is the price of the corresponding shipping line, and \( {n}_l^s \) is the number of shipping lines.

5 Empirical Study

In this section, we first divide the study area and aggregate the products of each sector. Then, we collect the data concerning the macro economy and sectors (which will affect the sector outputs and the total volumes consumed in the three markets), and the data for the land-sea intermodal network. Finally, we analyze the results and present some policy implications.

5.1 Study Area Division and Product Aggregation

The China Multi-Regional Input-Output Tables (2007 edition) provide the commodity flows among domestic regions. These tables can be used to divide mainland China into eight main regions (see Table 1) and determine the total volumes of products consumed in the non-local domestic market. We classify the countries that have traded with China into eight overseas regions (i.e., Japan/S.Korea, Southeast Asia, Middle East, North America, Latin America, Oceania, Europe, and Africa), which are the destinations for Chinese export goods. Moreover, we aggregate the 97 types of export goods in the China Customs Statistics Yearbook into the products of 17 main sectors based on the product aggregation method of Lee et al. (2011) and the sectoral aggregation in the Multi-Regional Input-Output tables (Table 6).
Table 1

Eight main regions in China

Region

Contained provinces

Northeast

Heilongjiang, Jilin, Liaoning

Beijing-Tianjin

Beijing, Tianjin

Northern coast

Hebei, Shandong

Eastern coast

Jiangsu, Shanghai, Zhejiang

Southern coast

Fujian, Guangdong, Hainan

Central

Shanxi, Henan, Anhui, Hunan, Jiangxi, Hubei

Northwest

Inner Mongolia, Qinghai, Sinkiang, Gansu, Shanxi, Ningxia

Southwest

Sichuan, Guangxi, Chongqing, Tibet, Yunnan, Guizhou

5.2 Data Preparation

5.2.1 Data Concerning Macro Economy and Sectors

Table 2 shows the statistics and sources of the data that will affect the sector outputs and total volumes consumed in the three markets over the 2006–2012 period.
Table 2

Summary statistics and sources of the variables

Variables

Mean

Std. Dev.

Min

Max

Obs

Sources

Log of sector output (lnY rit )

7.59

1.23

3.11

10.09

952

China Industrial Economic Statistical Yearbook

Log of capital input (lnK rit )

14.30

1.42

8.93

20.15

952

China Regional Economic Statistical Yearbook

Log of labor input (lnL rit )

3.29

1.08

0.35

5.88

952

Disposable income (I rt )

82,081

29,915

30,666

148,431

56

China Statistical Yearbook

Household consumption expenditure (O rgt )

7330

6857

647

37,952

448

Log of domestic region exports (lnT3rit)

21.71

1.81

16.51

25.95

952

UN Comtrade Database

Log of overseas region imports (lnT sit )

23.92

1.74

17.75

27.61

952

Log of bilateral exports (lnT rsit )

18.89

2.20

10.65

24.82

7616

Log of the population of domestic regions (lnPop rt )

9.53

0.68

7.89

10.49

56

China Statistical Yearbook

Log of the population of overseas regions (lnPop st )

12.61

0.99

10.44

13.89

56

World bank

Quality of port infrastructure in domestic regions (QPI rt )

4.2

0.29

3.63

4.47

56

The global competitiveness report (World Economic Forum)

Quality of port infrastructure in overseas regions (QPI st )

4.37

1.19

2.41

5.98

56

The commodities’ export price of each sector in domestic regions (p rit )

18,226

44,299

400

256,559

952

UN FAO Database and Industrial Development Organization

The commodities’ local price of each sector in overseas regions (p sit )

24,243

66,849

262

615,820

952

Tariff (tariff rsit )

7.04

5.58

0

25.53

7616

WTO International Trade Statistics

5.2.2 Land-Sea Intermodal Network

The top ten Chinese seaports in terms of foreign trade container throughput in 2012 (i.e., Shanghai, Shenzhen, Ningbo, Qingdao, Xiamen, Tianjin, Dalian, Guangzhou, Lianyungang, and Quanzhou Ports) are chosen as the gateway ports. The foreign trade container throughput these ports accounts for nearly 94% of the national total, and the ten gateway ports belong to five major port cluster in China. With the slowing down of both the economy and trade growth over recent years, the ports have been facing the excessive intra-cluster and inter-cluster competition (Yang et al. 2014b). For example, Shenzhen port and Guangzhou port are facing considerable challenges of neighboring Hong Kong free port in South China although they grew much faster than Hong Kong port over the past decades (Song 2002; Wang et al. 2012; Homsombat et al. 2016). Shanghai port is now suffering from the threat of Ningbo-Zhoushan port after the two ports were integrated in 2015 (Notteboom and Yang 2017).

Now we will build the land-sea intermodal network covering the domestic production sites, the rival gateway ports and overseas site, in which the inland transport network from the origin to the gateway ports is comprised of China’s main highway and railway lines (see Figs. 3 and 4), and the ocean shipping network from the gateway ports to overseas sites consists of the main container shipping lines of the world’s top five liner companies (i.e., MAERSK LINE, CMA CGM, MSC, EVERGREEN, and COSCO). Regarding the inland transport network of ports’ hinterland access, the total lengths of Chinese main highway and railway lines have reached 42.3 million km and 97.6 thousand km in 2012, respectively. Based on the Chinese Port Year Book (2012) and the website of China-Highway (www.china-highway.com), the parameters η, \( {\overline{V}}^{road} \) and ω1 are set as 84%, 80 km/h and 5 RMB/km, respectively. Based on the website of China-Railway (www.12306.cn), the parameters \( {\overline{V}}^{rail} \) and ω2 are set as 60 km/h and 1.98 RMB/km, respectively. Moreover, since the total import or export volumes of the 17 main sectors have been converted into the corresponding import or export volumes of the containerized cargos with the containerization ratio in Table 6, here we set the time value the products of the 17 main sectors to be a uniform number (i.e. 40 RMB/h) based on previous study (Shi et al. 2014). In the ocean shipping network, the number, average price and average shipping time of the main lines are sourced from the official website of the world’s top five liner companies (Table 7), and the port charges of the gateway ports are sourced from JCTRANS.
Fig. 3

Chinese main highway network in 2012

Fig. 4

Chinese main railway network in 2012

5.3 Empirical Results

5.3.1 Potential Export Volumes in the Eight Domestic Regions

First, the parameters in Eq. (1) are estimated with the sector data in Table 2 (see Table 8). The estimated results indicate that the goodness of fit of the production function of all sectors in the eight domestic regions is greater than 0.8, which indicates that the capital input and labor input can significantly explain the output for all sectors. Second, E 1 -E 8 are used to represent the eight types of household consumption, namely, food, clothing, residence, household facilities and services, transport and communications, education and cultural recreation, health care and medical services, and miscellaneous goods and services (the corresponding sectors of the household consumptions are shown in Table 6). Then, the ELES models are estimated with the least squares method (the results are shown in Table 9). The T U values of the ELES models of E 1 -E 8 in the eight domestic regions are greater than the critical value, which means that all ELES models pass the statistical test. Moreover, nearly all of the goodness-of-fit values of the ELES models in the eight domestic regions are greater than 0.8 (except E 1 and E 3 in the northwest, E 3 and E 4 in the northeast, and E 4 in Beijing-Tianjin), which means that disposable income can significantly explain household consumption. Finally, the China Multi-Regional Input-Output Tables (2007 edition) are used to calculate the Leontief Inverse Matrix between the 17 sectors in the eight domestic regions (the results are not listed due to space limitations).

The total output, total volumes consumed in the local market and non-local domestic market, and potential export volumes in the eight domestic regions in 2013 are predicted with the estimated production functions, ELES models and the Leontief Inverse Matrix (Fig. 5). The potential export volumes on the eastern coast and southern coast are higher than the potential export volumes in the six other regions; the northwest region has the lowest potential export volumes, accounting for only 1/20 of the highest volumes in the eastern coast region. Furthermore, for the ratios of volumes consumed in the local market, volumes consumed in the non-local domestic market and potential export volumes to the total output, the results indicate that the ratios of volumes consumed in the local market in the northwest region, volumes consumed in the non-local domestic market in the northern coast region, and potential export volumes in the southern coast region are the highest (accounting for 44%, 66%, and 40% of the corresponding total outputs, respectively), whereas the ratios of volumes consumed in the local market in the eastern coast region, volumes consumed in the non-local domestic market in the southern coast region, and potential export volumes in the central region are the lowest (accounting for 24%, 26%, and 6% of the corresponding total outputs, respectively).
Fig. 5

Outputs, volumes consumed in domestic market and potential export volumes in 2013

5.3.2 Relationship between Shipping Accessibility and Maritime Transport Demand

Impact of Shipping Accessibility on Maritime Transport Demand

The stochastic frontier gravity models are estimated with the maximum likelihood estimation algorithm based on the data in Table 2 (the results are shown in Table 3). The parameters of the stochastic frontier gravity models of the 17 sectors and their γ values (which represent the ratio of the error caused by trade efficiency to total error) pass the significance test, which illustrates that the stochastic frontier gravity model can be used to study the impact of shipping accessibility on maritime transport demand.
Table 3

Estimation results of the stochastic frontier gravity models

Sector

β 0i

β 1i

β 2i

β 3i

β 4i

β 5i

λ 0i

λ 1i

λ 2i

λ 3i

γ

1

0.35*

0.91***

−0.03*

0.03*

0.8***

1.13***

1.96***

0.1**

0.28***

−0.52***

0.52**

(1.66)

(77.3)

(−1.7)

(1.77)

(31.7)

(30.1)

(6.42)

(2.29)

(4.18)

(−21.1)

(2.36)

2

−10.8***

0.97***

−0.02*

0.31***

0.33***

0.18*

2.91*

2.83***

0.49*

−0.92***

0.67***

(−10.7)

(24.8)

(−1.73)

(5.7)

(4.24)

(1.68)

(1.68)

(2.92)

(1.65)

(−3.81)

(9.4)

3

−16.3***

1.09***

0.01*

0.22***

0.39***

0.58***

1.91***

1.45***

0.28**

−0.37***

0.81***

(−16.3)

(50.3)

(1.68)

(6.17)

(3.4)

(7.48)

(3.67)

(3.73)

(2.01)

(−2.9)

(5.07)

4

−13.9***

0.91***

−0.02*

0.58***

0.53***

0.87***

−0.11*

0.7***

0.46*

−0.2***

0.99***

(−20.5)

(65.6)

(−1.65)

(41.3)

(15.8)

(20.8)

(−1.89)

(3.51)

(1.65)

(−2.63)

(90.5)

5

−14.7***

0.86***

−0.05*

0.79***

0.44***

1.21***

−0.09*

0.57***

0.54***

−0.3***

0.09***

(−15)

(27.5)

(−1.95)

(30.9)

(11.5)

(51.6)

(−1.86)

(7.57)

(3.08)

(−4.25)

(4.1)

6

−4.64***

1***

0.03*

0.15***

0.62***

0.1***

13.15***

1.51***

1.28***

−1.11***

1***

(−118)

(76.5)

(1.80)

(47.2)

(142)

(120.1)

(16.8)

(11.48)

(7.18)

(−19.5)

(7.9)

7

−3.83**

0.86***

−0.06*

0.53***

0.62***

2.08***

−1.1*

0.9***

1.22***

−0.72***

0.95***

(−2.4)

(34.9)

(−1.76)

(9.01)

(9)

(17.3)

(−1.69)

(7.2)

(7.03)

(−8.4)

(42.3)

8

−2.91***

0.88***

−0.03*

0.11***

0.7***

1.08***

−2.31*

0.98**

2.13**

−2.59***

0.92***

(−2.9)

(33.5)

(−1.69)

(4.85)

(14.9)

(13.4)

(−1.78)

(2.23)

(2.27)

(−2.92)

(27.5)

9

−9.27***

0.95***

−0.02*

0.29***

0.51***

0.6***

1.84***

0.8**

0.22***

−0.36***

1***

(−9.09)

(36.8)

(−1.77)

(6.97)

(22.4)

(14.1)

(7.58)

(2.03)

(4.3)

(−23.4)

(47.2)

10

−11.1***

0.98***

0.12*

0.53***

−0.14***

0.22***

−0.08*

0.15**

0.1*

−0.05**

0.87***

(−11.1)

(25.3)

(1.85)

(13.9)

(−3.28)

(3.03)

(−1.76)

(1.97)

(1.68)

(−2)

(5.39)

11

−14.2***

0.94***

−0.06*

0.69***

0.35***

0.98***

0.97*

0.57***

0.88***

−0.51***

1***

(−18)

(50.6)

(−1.89)

(12.4)

(5.34)

(20.3)

(1.9)

(4.87)

(3.87)

(−6.67)

(89)

12

−17.9***

1.04***

0.02*

0.35***

0.01*

1.19***

2.7*

1.12***

0.24*

−0.17*

0.97***

(−20.6)

(48.9)

(1.75)

(16.7)

(1.93)

(21.3)

(1.83)

(3.04)

(1.65)

(−1.73)

(103.4)

13

−21.6***

0.93***

−0.02*

0.89***

0.17***

0.55***

30.45*

0.18**

3.1*

−2.39*

1***

(−27.7)

(53.1)

(−1.72)

(38.2)

(5.58)

(13.3)

(1.93)

(2.18)

(1.85)

(−1.93)

(173.5)

14

2.69**

0.78***

−0.06*

0.19***

1.09***

2.29***

−0.46*

0.52***

1.1***

−0.58***

0.11**

(2.21)

(33.3)

(−1.82)

(4.12)

(27.6)

(38.2)

(−1.84)

(11)

(10.6)

(−13.2)

(2.05)

15

−33.3***

0.89***

−0.01*

1.35***

0.41***

0.86***

21***

5.42***

0.21**

−0.07**

1***

(−30)

(47.1)

(−1.71)

(41)

(10.1)

(17.1)

(4.29)

(4.56)

(3.47)

(−2.41)

(65.3)

16

−16.3***

1.01***

0.06*

0.33***

0.46***

0.37***

5.98***

1.33***

3.98***

−0.95***

0.94***

(−36.3)

(75.7)

(1.85)

(16.6)

(16.1)

(10.8)

(7.25)

(14.7)

(2.05)

(−13.9)

(51.6)

17

−14.1***

0.93***

−0.02*

0.59***

0.55***

0.93***

−2.58*

1.51***

1.75***

−1.65***

0.91***

(−18)

(58.8)

(−1.72)

(31.5)

(21.7)

(9.9)

(−1.88)

(6.1)

(4.49)

(−7.16)

(50.8)

\( \gamma ={\sigma}_u^2/\left({\sigma}_v^2+{\sigma}_u^2\right) \); the data in the brackets is t-ratios; *denotes the significant at 10% level; ** denotes the significant at 5% level; *** denotes the significant at 1% level

The parameter calibration indicates that the maritime transport demand (the proxy for actual export volumes) is positively correlated with the shipping accessibility for all 17 sectors (the parameters in column 7 in Table 3 are all greater than 0), whereas the effects of the impacts are different. Six sectors have coefficients greater than 1, with sector 14 (manufacture of transport equipment) being the most affected by shipping accessibility (the coefficient is 2.29). The coefficients of eleven sectors are smaller than 1, with sector 6 (processing of timber) being the least affected (the coefficient is only 0.1). Moreover, concerning the other four control variables in Eq. (9), the potential export volumes (T 3rit ) have the greatest impact on maritime transport demand (the parameters in column 3 in Table 3 are all close to 1), followed by the population in the overseas regions (Pop st ) and the import demand of the overseas regions (T sit ) and then the impact of the population in the domestic regions (Pop rt ), for which the parameters in column 4 in Table 3 are all close to 0.

Moreover, Table 3 also presents the parameters that affect the trade efficiency. The quality of the port infrastructure (QPI rt and QPI st ) positively affects the trade efficiency, whereas the ratio of commodity price in the overseas region to the C.I.F price of China’s export goods is negatively correlated with trade efficiency. These results suggest that raising the export price for overseas regions will decrease the trade efficiency from China to its trading partners, whereas decreasing the price of China’s export goods and custom duties, improving the quality of the port infrastructure, and increasing the shipping accessibility (i.e., reducing the transport cost) will promote the trade efficiency from China to overseas regions and further expand the maritime transport demand.

Relationship between Shipping Accessibility and Maritime Transport Demand

We calculated the shipping accessibility from China to overseas regions. Then, the stochastic frontier gravity model is used to calculate the maritime transport demand from the domestic regions to the overseas regions in 2013 based on the potential export volumes (the results are shown in Fig. 6 and Table 4). Figure 6 presents the shipping accessibility, potential and actual maritime transport demand, which illustrates that the potential and actual maritime transport demand are in line with the decreasing trend of the accessibilities for the eight domestic regions (i.e., in the region with high maritime transport demand, the shipping accessibility is also high). The reason for this trend is that regions with high maritime transport demand (e.g., eastern coast) have a well-developed hinterland access system, and the gateway ports in these regions have a larger capacity and more frequent vessel calls from main liner shipping lines (e.g., the highway network density in the Yangtze River Delta is four times the national highway network density, and the Shanghai Port and Ningbo-Zhoushan Port are the top 2 ports in the world). All of these factors will directly expand the shipping accessibility from these domestic regions to overseas regions.
Fig. 6

Shipping accessibility and maritime transport demand in 8 domestic regions in 2013

Table 4

Actual maritime transport demand and shipping accessibility

From\To

Japan/S.Korea

Southeast Asia

Oceania

Middle East

North America

Europe

Latin America

Africa

China

4.9/478

3.51/589

2.26/163

1.93/284

1.69/867

1.47/1108

1.46/232

0.73/184

Northeast

5.07/19

2.79/20

1.86/6

1.73/11

1.58/34

1.36/42

1.28/9

0.65/7

Beijing-Tianjin

5.79/29

3.41/31

2.17/9

1.93/15

1.74/50

1.47/62

1.42/12

0.73/10

Northern coast

5.87/41

3.61/45

2.26/13

1.98/22

1.76/70

1.48/87

1.46/18

0.75/14

Eastern coast

6.57/191

4.14/220

2.55/60

2.12/106

1.81/317

1.56/406

1.56/86

0.79/69

Southern coast

5.13/141

4.17/197

2.57/54

2.08/94

1.77/280

1.57/363

1.6/76

0.78/61

Central

5.26/28

3.83/35

2.39/10

2.02/16

1.74/53

1.51/67

1.51/14

0.76/11

Northwest

3.7/8

2.77/10

1.92/3

1.69/5

1.51/17

1.32/21

1.3/4

0.66/3

Southwest

3.97/22

3.29/30

2.18/9

1.85/15

1.61/47

1.43/60

1.43/13

0.71/10

Data are arranged as shipping accessibility/maritime transport demand (Ten thousand TEU)

Table 4 shows the actual maritime transport demand and shipping accessibility from eight domestic regions to the overseas regions. In general, the descending order of the shipping accessibility from China to the eight overseas regions is: Japan/S.Korea, Southeast Asia, Oceania, Middle East, North America, Europe, Latin America, and Africa; the shipping accessibility to Japan/ S.Korea is the highest (4.9), whereas the shipping accessibility to Africa is the lowest (0.73). However, the descending order of the maritime transport demand is different from the descending order of shipping accessibility. The maritime transport demand from China to Europe is the highest (11.08 million TEU), followed by the maritime transport demands from China to North America, Southeast Asia, and Japan/ S.Korea; the maritime transport demand to Oceania is the lowest (0.96 million TEU).

The regional maritime transport demand and shipping accessibility exhibit nearly identical trends as the national maritime transport demand and shipping accessibility. The maritime transport demand from the eight domestic regions to an overseas region also exhibit the same trend, namely, the maritime transport demands in the eastern coast and southern coast are the highest, followed by those in the northern coast, central, Beijing-Tianjin, and southwest, whereas the demands in the northeast and northwest are the lowest. However, the shipping accessibility to Japan/ S.Korea from the northeast are considerably higher than those from the southwest and northwest.

5.3.3 Sensitivity Analysis - IMPACTS of Shipping Lines on Maritime Transport Demand

To analyze the impact of the number of shipping lines on maritime transport demand, we design five scenarios in which the selected gateway ports open 10 more shipping lines to every overseas region (i.e., a total of 80 shipping lines will be newly opened). These shipping lines are as follows: 1) S1: adding new lines at Dalian Port (northeast); 2) S2: adding new lines at Tianjin Port (Beijing-Tianjin); 3) S3: adding new lines at Qingdao Port (northern coast); 3) S4: adding new lines in Lianyungang Port (eastern coast); and 5) S5: adding new lines in Quanzhou Port (southern coast). The growth rates of the regional maritime transport demand under different scenarios are shown in Table 5.
Table 5

The growth rates of regional maritime transport demand under different scenarios

Scenario

China

northeast

Beijing-Tianjin

Northern coast

Eastern coast

Southern coast

Central

Northwest

Southwest

S1

0.58%

3.94%

1.61%

2.28%

0.13%

0.35%

2.86%

3.62%

2.69%

S2

1.07%

1.71%

3.27%

2.77%

0.37%

0.82%

3.37%

4.15%

3.19%

S3

2.62%

3.07%

0.30%

1.71%

1.96%

2.35%

5.01%

5.50%

4.60%

S4

2.09%

2.64%

0.18%

3.80%

1.39%

1.87%

4.43%

5.11%

4.20%

S5

3.25%

3.46%

0.62%

4.62%

2.51%

1.61%

5.53%

5.85%

5.37%

It can be seen that China’s maritime transport demands differ considerably in the different scenarios. Adding new shipping lines at Quanzhou Port (S5) is anticipated to have the largest encouraging effect, increasing the Chinese maritime transport demand by 3.25%, followed by S3, S4, S2 and S1. The reason for the differences in the effects of each scenario is that the land transport costs to the various ports (i.e., that to Quanzhou Port is the lowest, and that to Dalian Port is the highest). Therefore, when Quanzhou Port opens new shipping lines, the shipping accessibility from China to overseas regions may change by the largest amount, thus encouraging Chinese maritime transport demand mostly.

Moreover, the maritime transport demand in the northeast is expected to increase by 3.94% in S1 (Dalian Port), that in Beijing-Tianjin may increase 3.27% in S2 (Tianjin Port), that in northern coast may increase 1.71% in S3 (Qingdao Port), that in eastern coast may increase 1.39% in S4 (Lianyungang Port), that in southern coast may increase 1.61 in S5 (Quanzhou Port). In terms of regional development, adding new shipping lines at Dalian Port is expected to have the largest effect, whereas opening new lines in Lianyungang Port is expected to have the smallest effect; this trend occurs because the shipping lines in eastern coast region are dense (and thus, the marginal effect is small), whereas the shipping lines in the northeast region are less (and thus, the marginal effect is large).

Therefore, an increase in the number of shipping lines at Quanzhou Port would be the best choice for the development of the entire country. However, increasing the number of shipping lines at Dalian Port would be the optimal scheme for regional balanced development (e.g., the Northeast Area Revitalization Plan).

6 Conclusions

Based on the relationship between production and consumption, this paper proposes a method to determine the potential export volumes and analyzes the relationship between shipping accessibility and maritime transport demand with the stochastic frontier gravity model. A better understanding of the relationship may be helpful to transport policy making with the aim of encouraging China’s shipping industry and foreign trade. With this purpose in mind, we estimate the stochastic frontier gravity model of 17 sectors using trade and transport data from 2006 to 2012. The results show that the maritime transport demand is positively correlated with shipping accessibility for all 17 sectors. The relationship between the shipping accessibility and maritime transport demand from eight domestic regions to the overseas regions indicated that the potential and actual maritime transport demand align with the decreasing trend of the accessibilities, namely, the eastern coast region has the highest shipping accessibility and the highest potential and actual maritime transport demand. Moreover, according to the analyses of the impacts of the shipping lines in gateway ports on the maritime transport demand, increasing the shipping lines at Quanzhou Port will result in the greatest benefit for China’s maritime transport demand, and this conclusion verifies that Quanzhou Port plays an important role in the twenty-first century maritime silk road.

The methodology and main results of our paper may help decision makers understand the differences in shipping accessibility and maritime transport demand from eight domestic regions to the overseas regions and provide a theoretical basis for them to develop more useful strategies to promote the development of the Chinese shipping industry and foreign trade. A limitation of this paper is that we have not calibrated the parameter θ, which can reflect the importance of generalized transport cost to shipping accessibility. Future work will focus on developing a calibration method for this parameter, and then further improve our study. Moreover, since the shipping prices change dramatically and have high volatility, and they are often affected by the situation of shipping market and the price strategy of the international shipping companies (Zeng and Qu 2014), then the impacts of shipping price volatility on shipping accessibility and maritime transport demand should be considered. We will deal with this problem in our future study. Furthermore, here only the gateway container ports are investigated for simplicity, whereas in practice certain export goods need to be transported to overseas sites through specialized terminals or ports (Zhuang et al. 2014), then the port specialization could be one of the main issues for future analysis.

Notes

Acknowledgements

This research is supported by the key project of National Natural Science Foundation of China (Grant No. 71431001) and the Fundamental Research Funds for Central Universities of China (Grant No. 3132016303).

References

  1. Alamá-Sabater L, Márquez-Ramos L, Suárez-Burguet C (2013) Trade and transport connectivity: a spatial approach. Appl Econ 45(18):2563–2566CrossRefGoogle Scholar
  2. Aigner D, Lovell CK, Schmidt P (1977) Formulation and estimation of stochastic frontier production function models. J Econ 6(1):21–37CrossRefGoogle Scholar
  3. Baier SL, Bergstrand JH (2001) The growth of world trade: tariffs, transport costs, and income similarity. J Int Econ 53(1):1–27CrossRefGoogle Scholar
  4. Bensassi S, Márquez-Ramos L, Martínez-Zarzoso I, Suárez-Burguet C (2015) Relationship between logistics infrastructure and trade: Evidence from Spanish regional exports. Transp Res A Policy Pract 72:47–61CrossRefGoogle Scholar
  5. Bougheas S, Demetriades PO, Morgenroth EL (1999) Infrastructure, transport costs and trade. J Int Econ 47(1):169–189CrossRefGoogle Scholar
  6. Calatayud A, Palacin R, Mangan J, Jackson E, Ruiz-Rua A (2016) Understanding connectivity to international markets: a systematic review. Transp Rev 1–24.  https://doi.org/10.1080/01441647.2016.1157836
  7. Carrasco R, Albarran P, Holl A (2009) Transport infraestructure, sunk costs and firms' export behaviour. Universidad Carlos III de Madrid. Departamento de EconomíaGoogle Scholar
  8. Caschili S, Medda F, Parola F, Ferrari C (2014) An analysis of shipping agreements: the cooperative container network. Netw Spat Econ 14(3–4):357–377CrossRefGoogle Scholar
  9. Chi J (2016) Exchange rate and transport cost sensitivities of bilateral freight flows between the US and China. Transp Res A Policy Pract 89:1–13CrossRefGoogle Scholar
  10. China Statistical Year Book (2014) The total imports and exports of China (in Chinese). http://navi.cnki.net/KNavi/YearbookDetail?pcode=CYFD&pykm=YINFN&bh. Accessed 12 May 2015
  11. Clark X, Dollar D, Micco A (2004) Port efficiency, maritime transport costs, and bilateral trade. J Dev Econ 75(2):417–450CrossRefGoogle Scholar
  12. Cook WD, Zhu J (2014) DEA Cobb–Douglas frontier and cross-efficiency. J Oper Res Soc 65(2):265–268CrossRefGoogle Scholar
  13. Coşar AK, Demir B (2016) Domestic road infrastructure and international trade: Evidence from Turkey. J Dev Econ 118:232–244CrossRefGoogle Scholar
  14. Drysdale P, Huang Y, Kalirajan KP (2000) China’s trade efficiency: Measurement and determinants. In: Drysdale P, Zhang Y, Song L (eds) APEC and liberalization of the Chinese economy. Asia Pacific Press, Canberra, pp 259–271Google Scholar
  15. Fabling R, Grimes A, Sanderson L (2013) Any port in a storm: impacts of new port infrastructure on exporter behaviour. Transport Res E Log 49(1):33–47CrossRefGoogle Scholar
  16. Homsombat W, Ng A, Fu X (2016) Regional transformation and port cluster competition: the case of the pearl river delta in South China. Growth Chang 47(3):349–362CrossRefGoogle Scholar
  17. Huang JL (2008) Market size and industrial growth in Chinese provinces. China Econ Q 7(4):1317–1334 (in Chinese)Google Scholar
  18. Hummels D (2007) Transportation costs and international trade in the second era of globalization. J Econ Perspect 21(3):131–154CrossRefGoogle Scholar
  19. Lee TC, Wu CH, Lee PTW (2011) Impacts of the ECFA on seaborne trade volume and policy development for shipping and port industry in Taiwan. Marit Policy Manag 38(2):169–189CrossRefGoogle Scholar
  20. Limao N, Venables AJ (2001) Infrastructure, geographical disadvantage, transport costs, and trade. World Bank Econ Rev 15(3):451–479CrossRefGoogle Scholar
  21. Lu XD, Zhao QW (2010) China’s export potential and determinants. J Quant Tech Econ 10:21–35 (in Chinese)Google Scholar
  22. Márquez-Ramos L, Martínez-Zarzoso I, Pérez-García E, Wilmsmeier G (2011) “Special issue on Latin-American research” maritime networks, services structure and maritime trade. Netw Spat Econ 11(3):555–576CrossRefGoogle Scholar
  23. Martínez-Zarzoso I, García-Menéndez L, Suárez-Burguet C (2003) Impact of transport costs on international trade: the case of Spanish ceramic exports. Marit Econ Logist 5(2):179–198CrossRefGoogle Scholar
  24. Martínez-Zarzoso I, Suárez-Burguet C (2005) Transport costs and trade: empirical evidence for Latin American imports from the European Union. J Int Trade Econ Dev 14(3):353–371CrossRefGoogle Scholar
  25. Martinez-Zarzoso I, Pérez-García EM, Suárez-Burguet C (2008) Do transport costs have a differential effect on trade at the sectoral level? Appl Econ 40(24):3145–3157CrossRefGoogle Scholar
  26. Notteboom T, Yang Z (2017) Port governance in China since 2004: Institutional layering and the growing impact of broader policies. Res Transp Bus Manag 22:184–200CrossRefGoogle Scholar
  27. Palacio A, Adenso-Díaz B, Lozano S, Furió S (2016) Bicriteria optimization model for locating maritime container depots: application to the port of Valencia. Netw Spat Econ 16(1):331–348CrossRefGoogle Scholar
  28. Portugal-Perez A, Wilson JS (2012) Export performance and trade facilitation reform: hard and soft infrastructure. World Dev 40(7):1295–1307CrossRefGoogle Scholar
  29. Ramli MKR, Ismail NW (2014) Role of infrastructures in explaining trade costs in ASEAN-5. Eng Econ 25(2):119–129CrossRefGoogle Scholar
  30. Sánchez RJ, Hoffmann J, Micco A, Pizzolitto GV, Sgut M, Wilmsmeier G (2003) Port efficiency and international trade: port efficiency as a determinant of maritime transport costs. Marit Econ Logist 5(2):199–218CrossRefGoogle Scholar
  31. Shi R, Hu Z, Zhou Y, Liu P (2014) Research on railway freight market share based on the whole process of transport. Procedia Soc Behav Sci 138:298–304CrossRefGoogle Scholar
  32. Song DW (2002) Regional container port competition and co-operation: the case of Hong Kong and South China. J Transp Geogr 10(2):99–110CrossRefGoogle Scholar
  33. Tang LC, Low JM, Lam SW (2011) Understanding port choice behavior—a network perspective. Netw Spat Econ 11(1):65–82CrossRefGoogle Scholar
  34. Thill JC, Lim H (2010) Intermodal containerized shipping in foreign trade and regional accessibility advantages. J Transp Geogr 18(4):530–547CrossRefGoogle Scholar
  35. Tinbergen J (1962) Shaping the world economy: suggestions for an international economic policy. The Twentieth Century Fund, New YorkGoogle Scholar
  36. Tsiotas D, Polyzos S (2015) Analyzing the maritime transportation system in Greece: a complex network approach. Netw Spat Econ 15(4):981–1010CrossRefGoogle Scholar
  37. Wang K, Ng A, Lam J, Fu X (2012) Cooperation or Competition? Factors and Conditions Affecting Regional Port Governance in South China. Marit Econ Logist 14(3):386–408CrossRefGoogle Scholar
  38. Wang L, Béland D, Zhang S (2014) Pension fairness in China. China Econ Rev 28:25–36CrossRefGoogle Scholar
  39. Wang Y, Cullinane K (2008) Measuring container port accessibility: An application of the Principal Eigenvector Method (PEM). Marit Econ Logist 10(1):75–89CrossRefGoogle Scholar
  40. Wang YF, Liu SL (2015) Dynamic effect of government spending structure on household consumption——empirical analysis based on DSGE model. Syst Eng Theory Pract 35(2):300–307 (in Chinese)Google Scholar
  41. Wilmsmeier G, Hoffmann J (2008) Liner shipping connectivity and port infrastructure as determinants of freight rates in the Caribbean. Marit Econ Logist 10(1):130–151CrossRefGoogle Scholar
  42. WTO (2014) The seaborne trade shipments of each country. http://stat.wto.org/Home/WSDBHome.aspx?Language. Accessed 25 July 2015
  43. Xu LZ, Fang SC, Wang SY (2012) Analysis on transport costs and China's exports. Syst Eng Theory Pract 32(5):1057–1067 (in Chinese)Google Scholar
  44. Yang D, Ong GP, Chin ATH (2014a) An exploratory study on the effect of trade data aggregation on international freight mode choice. Marit Policy Manag 41(3):212–223CrossRefGoogle Scholar
  45. Yang J, Wu X, Luo M (2014b) Regional economy, transportation infrastructure and container port system evolution process: a comparative study of China and the U.S. in 1979-2010. Econ Geogr 34(2):80–85Google Scholar
  46. Zeng Q, Qu C (2014) An approach for baltic dry index analysis based on empirical mode decomposition. Marit Policy Manag 41(3):224–240CrossRefGoogle Scholar
  47. Zhang R, Huang K (2009) Path analysis of influential elements of collecting and dispatching system for container port. J Tongji Univ Nat Sci 37(1):57–62 (in Chinese)Google Scholar
  48. Zhao Y, Kockelman KM (2004) The random-utility-based multiregional input–output model: solution existence and uniqueness. Transp Res B Methodol 38(9):789–807CrossRefGoogle Scholar
  49. Zhuang W, Luo M, Fu X (2014) A game theory analysis of port specialization—implications to the Chinese port industry. Marit Policy Manag 41(3):268–287CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Transportation Engineering CollegeDalian Maritime UniversityDalianPeople’s Republic of China
  2. 2.Faculty of Maritime and TransportationNingbo UniversityNingboPeople’s Republic of China

Personalised recommendations