Port selection based on customer questionnaire: a case study of German port selection

Product-oriented thinking is very common in the industry; therefore, the existing literature on port selection mainly focuses on the position of container liners and seeks to optimize costs and other criteria such as port efficiency, location, service, and physical infrastructure. Taih-Cherng et al. (2003) evaluated the importance of various criteria in transshipment port selection from a container carrier’s perspective and revealed that port competitiveness mainly depends on carriers’ costs and efficiency in loading and discharging [1]. Guy and Urli (2006) examined how the combined importance of quality of infrastructure, cost, service, and geographical location affect shipping lines’ port choice [2]. Chang et al. (2008) examined the different factors affecting shipping lines’ port selections. They collected data through a survey of shipping companies and analyzed it from the varying perspectives of trunk liners and feeder service providers [3]. Nguyen (2011) constructed a model for port selection by minimizing total cost and two sensitivity analyses were used to evaluate different service patterns, and the efficiency of large vessels in the scope context of a logistics network [4]. Kadaifci et al. (2019) used a multi-criteria decision-making approach to examine the effects of transportation cost, geographical location, infrastructure, and technical conditions on port selection [5]. Nazemzadeh and Vanelslander (2015) analyzed port selection criteria through questionnaire surveys of three different industry groups [6]. It can be concluded that a common characteristic of these studies is the concern for cost saving. Decision-making was based on cost optimization and container carrier’s concerns regarding improving efficiency, service and geographical location convenience, and so on.

Other studies examine indicators such as hinterland connection, feeder connectivity, scale economies, cargo balance, and environmental issues. Bart et al. (2008) investigated how hinterland connections, port tariffs, feeder connectivity, and environmental issues affect deep-sea container operators’ port selections [7]. Tand et al. (2011) investigated the quality characteristics of port selection. Findings revealed that port efficiency and scale economies are the important dimensions influencing container carriers’ selection [8]. Based on a case study of the Spanish container trade, Veldman et al. (2013) introduced new variables such as coastline and inland transport cargo balance to expand upon earlier research [9]. Some studies established the criteria and sub-criteria system for container carrier’s port selection. Onut et al. (2011) examined the criteria and sub-criteria for port selection through a case study of port selection in the Marmara region of Turkey [10] .Chou (2010) examined the criteria and sub-criteria from shipping lines’ perspectives [11]. Button et al. (2015) investigated how incorporating subjective elements affects port selection [12]. In addition, a few papers have examined port selection from the viewpoint of container liners in specific environments. Yeo et al. (2014) analyzed port selection in uncertain environments [13]. Notteboom et al. (2017) investigated the complex link between port selection by shipping alliance members and changing organizational routines [14].

Based on the above mentioned literature review, it can be concluded that the related indicators were well reflected in port selection from the viewpoint of container liners. However, it remains a question as to whether customers match the container carriers’ choices and if their concerns are different. It is necessary to examine port selection from the viewpoint of customers especially under the current trend of container vessels’ upsizing.

There are very few papers that examine the effects of customers’ port selection choices. Tongzon (2009) examined the major aspects influencing port choice from the perspective of Southeast Asian freight forwarders [15]. Kashiha et al. (2016) investigated how geography and transportation costs influence shippers’ decisions regarding port selection [16]. Balci et al. (2018) studied terminal competition and selection criteria in the dry bulk market via a case study of Izmir, Turkey [17]. A common research approach for above mentioned literature was to rate the pre-selected indicators by sampling customers; indicator weight was calculated in a different manner. However, as all indicators were included in the pre-selection, the effects of subdividing indicators on customers’ port selections were not studied. Furthermore, the indicators’ weights (calculated by the mathematical model) were not verified, which is precisely what this paper intends to contribute to the literature.

A total of 315 questionnaires were issued, and 287 valid questionnaires were recovered. Among the 287 valid responses, 219 (76%) participants (shipping industry customers) had a global volume of less than 99,999 TEU/year; 32 (11%) had a global volume of 100,000 to 499,999 TEU/year; 16 (6%) had a global volume of 500,000 to 999,999 TEU/year; 11 (4%) had a global volume of 1,000,000 TEU/year, while volume size is not specified for 9 (3%) participants.

In terms of participants’ volume on the Asia–Europe route, 204 (71%) have a volume of 9999 TEU or less; 45 (16%) have a volume of 10,000 TEU to 49,999 TEU/year; 17 (6%) have a volume of 50,000 to 99,999 TEU/year; 13 (5%) have a volume of 100,000 TEU/year and above; the annual volume size was not specified by 8 (2%) participants.

In terms of the cargo volume distribution in Germany, there are 217 (76%) participants with German shipments of 4999 TEU/year or less; 24 (8%) with shipments between 5000 and 9999 TEU/year; 26 (9%) with shipments between 10,000 and 49,999 TEU/year; seven (2%) with shipments of 50,000 TEU/year and above; the volume of German shipments on the Asia–Europe route was not specified by 13 (5%) participants.

While the questionnaire covers customers of various sizes in the market, it mainly includes small- and medium-sized entities. This is consistent with the proportional distribution of various customers. Customers selected for the survey questionnaire have a wide range of representation in port selection (Fig. 2).

Traditionally, under the guidance of “product-oriented” thinking, container carriers believe that as long as the calling ports can contribute to reducing the network cost, improve delivery time, and maintain uniqueness, selection of such ports makes good business sense. However, through the questionnaire survey, a totally different concern was found regarding port selection. The level indicators proposed by participants in the questionnaire survey include freight competitiveness, service capability, cooperation and relationship, and “others” (which includes aspects such as container liners’ strategic compatibility, reputation, and financial situation). A number of subdivided indicators comprise the elements for evaluation of the level indicators. Correspondingly, these levels were also named as secondary indicators. Both types of indicators are key elements in customers’ port selection. Therefore, this is the evaluation indicator for assessing customer-oriented thinking. The composition and weights of the level indicators and secondary indicators are shown in Table 1.

In the questionnaire survey regarding port selection indicators, multiple choices were designed so as to examine customers’ concern on different indicators. Therefore, the sum of all options was used to calculate the weights of indicators. Among the participants, 259 (41.5%) selected “Freight Competitiveness” 223 (35.7%) selected “Service Capability” 122 (19.6%) selected “Cooperation and Relationship” while 20(3.2%) selected “Others” respectively.

In addition, 120 (42%) participants indicated that they would generally not change the core carrier; 75 would change their core carrier at any time, accounting for 26%; 52 would change their core carrier every year, accounting for 18%; 40 (14%) do not have fixed core carriers; The 120 participants who selected “basically unchanged” cover all global NVOCCs (non-vessel operating common carriers) with cargo volumes above 50,000 TEU/year in Germany, which indicates that global NVOCCs appear to be more stable than small- and medium-sized freight-forwarding companies.

The main ports available in Germany are the ports of Hamburg, Bremen, Bremerhaven, and Wilhelmshaven. According to the results of our survey, the port of Hamburg is still the preferred calling port for most customers. A total of 276 (79%) participants selected Hamburg; 37 (11%) selected Bremen; 30 (9%) selected Bremerhaven; only four (1%) selected Wilhelmshaven as their preferred calling port.

The reasons for selection were analyzed through the questionnaire survey. A total of 197 (38%) participants attributed the aforementioned choices to “Sailing Frequency,” while 123 (24%) participants attributed the reason to “Distribution Cost” Interviews revealed that many distribution centers are located near the port of Hamburg and its surrounding areas and switching to Wilhelmshaven and Bremerhaven will significantly increase the cost of logistics and distribution. A total of 115 participants (22%) attributed their selection to the “Collection and Distribution System”. (Fig. 3).

Further research on potential change of ports showed that 30 participants (11%) are willing to change to the port of Wilhelmshaven, but 98 (34%) participants clearly stated that they will not accept Wilhelmshaven; 159 (55%) participants indicated that a switch was “uncertain” or that it would happen “depending on the service”.

The survey revealed that when evaluating a potential change, customers pay the highest attention to “Better Freight Rate.” A total of 234 (43%) participants selected “Better Freight Rate,” 110 (20%) selected “Efficient Delivery and Transit,” 107 (20%) selected “Additional Detention and Demurrage,” 90 (16%) selected “Extra Free Storage Condition,” and seven (1%) selected “others”.

It seems that container carriers depend heavily on low-cost calling ports to achieve cost savings but have been forced to offer discounted freight rates and additional free detention, demurrage, and storage conditions, which does not create cost savings. Therefore, it is necessary to improve the traditional evaluation system based on the cost optimization of container liners and establish a new evaluation system of port selection based on customers.

As the raw data of the indicators were calculated using different dimensions and units, they need to be standardized for comparison. When the indicator is positive, its standardized formula is:

When the indicator is negative, its standardized formula is:

According to the above formulas, the processing of the raw data for Wilhelmshaven, Hamburg, Bremen, and Bremerhaven are presented in Table 3.

According to the questionnaire survey, the secondary indicators that constitute “Freight Competitiveness” are “Freight Rate” (40%), “Side Condition” (30%), “Volume Incentive” (20%), “Freight Stability” (5%), and “Quotation Efficiency” (5%), as set during the interviews and the fact that customers pay significant attention to the actual rate offer and related conditions, such as the credit terms, demurrage and detention terms, drop-off condition, and volume incentive when achieving the volume target of container carriers. Therefore, the weight of secondary indicators can be calculated as 16.6% for “Freight Rate,” 12.45% for “Side Condition,” 8.3% for “Volume Incentive,” and 2.075% for both “Freight Stability” and “Quotation Efficiency.”

Secondary indicators that constitute “Service Capability” are “Service Route Quality” (50%), “Cargo Tracing” (20%), “Transshipment Efficiency” (20%), and 5% for both “Emergency Response” and “IT System.” This is based on the interviews and the fact that customers choose different services provided by container liners mainly through the evaluation of transit time, frequency, on-time performance, and uniqueness of the service route. Cargo tracing, transshipment efficiency, emergency response in urgent cases, and the possibility and convenience of IT system connection and data exchange are also major considerations. Therefore, the weight of secondary indicators can be calculated as 17.85% for “Service Route Quality,” 7.14% for both “Cargo Tracing” and “Transshipment Efficiency,” and 1.785% for both “Emergency Response” and “IT System.”

Secondary indicators that constitute “Cooperation and Relationship” are “Global Cooperation” (40%), “Cooperative History” (40%), and “Business Claims” (20%). This is based on the interviews and the fact that customers pay almost equal attention to global cooperation rather than being restricted to specific routes and the assessment of historical cooperation and future expectations. The efficiency of business claims also has an impact on customer choice. Therefore, the weight of secondary indicators can be calculated as 7.87% for both “Global Cooperation” and” Cooperative History,” and 3.92% for “Business Claims.”

Other indicators include “Strategic Compatibility” (40%) and “Company Reputation” and “Financial Situation” (both 30%). This is also based on the interviews and the fact that customers pay much more attention to strategic consistency, value recognition, and the evaluation of the container liners’ reputation and financial situation. Therefore, the weight of secondary indicators can be calculated as 1.28% for “Strategic Compatibility” and 0.96% for both “Company Reputation” and “Financial Situation.”

According to the standardized data in Table 3 and the calculation formula

The final scores of the four alternative calling ports are presented in Table 4.

Based on the aforementioned evaluations, the order of preference for the German ports is as follows: Hamburg, Wilhelmshaven, Bremerhaven, and Bremen (Fig. 5, presents the order of preference).

The AHP is a decision analysis method proposed by American operations researcher T.L. Satty in the 1970s. It combines qualitative and quantitative analysis to solve complex multi-objective problems, and is widely applied in the scheme evaluation of port selection. Taih-Cherng et al. (2003) conducted a field survey of container carriers to analyze transshipment port selection from their perspective. Subsequently, Button et al. (2015) applied the AHP method to reveal the liners’ seaport choice assessments. The advantages of AHP lie in its capacity to deal with a wide range of qualitative and quantitative variables. Its disadvantages originate from untrustworthy responses, or divergent and contrary answers (Nazemzadeh and Vanelsander, 2015). There have been many other applications of AHP to transport problems, proving it to be a valid approach to the current analysis. Therefore, the AHP method was chosen to evaluate port selection criteria in this paper. The calculation steps are as follows:

Evaluation indicators number P, u = {u₁, u₂, ⋯⋯, u_p}.

Judging the value of the elements of the matrix reflects people’s understanding of the relative importance of each element, generally using a scale from 1 to 9 and its reciprocal method. However, when the importance of the mutual comparison factor can be explained through a meaningful ratio, the value of the corresponding element of the judgment matrix is taken as the ratio. That is, the judgment matrix is obtained by S = (u_ij)_p × p Table 5.

The maximum eigenvalue λ_max of the judgment matrix S and its corresponding eigenvector A are calculated using MATLAB software. This eigenvector is the order of importance of each evaluation factor, that is, the distribution of weight coefficients.

To perform the consistency test of the judgment matrix, it is necessary to calculate the consistency indicator.

Table 6 presents the results of consistency indicator RI_.

According to the indicator system, through the aforementioned scale method and the expert consultation method, eight experts in the field were selected and the degree of importance of the indicators was scored separately. The results were then further discussed and summarized internally, and the following 2 × 2 judgment matrix was obtained Table 7.

The maximum eigenvalue of the judgment matrix is calculated using MATLAB software. To check the consistency of the judgment matrix, the consistency indicator needs to be calculated:

The average random consistency indicator RI = 0.9. The random consistency ratio is

Therefore, the results of the hierarchical analysis ranking are considered to have satisfactory consistency, that is, the distribution of the weight coefficients is reasonable Table 8.

We use the analytic hierarchy method to find the index weight for each of the secondary indicators under the level indicators “Freight Competitiveness (B1),” “Service Capability (B2),” “Cooperation and Relationship (B3),” and “Other Indicators (B4).”

Under the level indicator “Freight Competitiveness (B1),” a judgment matrix S = (u_ij)_p × p is constructed for each sub-indicator. The results are shown in Table 9.

The maximum eigenvalue λ_max = 5.0133 of judgment matrix S is calculated using MATLAB software. To check the consistency of the judgment matrix, the consistency index needs to be calculated:

The average random consistency index is RI = 1.12. The random consistency ratio is

The results of the hierarchical analysis ranking are considered to have satisfactory consistency, that is, the distribution of the weight coefficients is very reasonable. The weights of the sub-indicators “Freight Rate (C1),” “Side Condition (C2),” “Volume Incentive (C3),” “Freight Stability (C4),” and “Quotation Efficiency (C5)” under the level indicator “Freight Competitiveness (B1)” are shown in Table 10.

Under the level indicator “Service Capability,” a judgment matrix S = (u_ij)_p × p is constructed for each sub-indicator, as shown in Table 11.

The maximum eigenvalue λ_max = 5.0022 of judgment matrix S is calculated using MATLAB software. To check the consistency of the judgment matrix, the consistency indicator needs to be calculated:

The average random consistency indicator is RI = 1.12. The random consistency ratio is \( CR=\frac{CI}{RI}=\frac{5.5506\mathrm{e}-04}{1.12}=4.9559\mathrm{e}-04<0.10 \).

Therefore, the results of the hierarchical analysis ranking are considered to have satisfactory consistency, that is, the distribution of the weight coefficients is very reasonable. The weights of the secondary indicators “Service Route Quality (C6),”“Cargo Tracing (C7),” “Transshipment Efficiency (C8),” “Emergency Response (C9),” and “IT System (C10)” under the first-level indicator “Service Capability (B2)” are shown in Table 12.

Under the level indicator “Cooperation and Relationship,” judgment matrix S = (u_ij)_p × p is constructed for each sub-indicator. The results are shown in Table 13.

The maximum eigenvalue λ_max = 3.0000 of the judgment matrix S is calculated using MATLAB software. To check the consistency of the judgment matrix, the consistency indicator needs to be calculated:

\( CI=\frac{\lambda_{\mathrm{max}}-n}{n-1}=\frac{3.0000-3}{3-1}=0 \)The average random consistency indicator is RI = 0.58. The random consistency ratio is

The results of the hierarchical analysis ranking are considered to have satisfactory consistency, that is, the distribution of the weight coefficients is very reasonable. The weights of the sub-indicators “Global Cooperation (C11),”“Cooperative History (C12),” and “Business Claims (C13)” under the first-level indicator “Cooperation and Relationship (B3)” are shown in Table 14.

Under the first-level indicator “Other Indicators(B4),” judgment matrix S = (u_ij)_p × p is constructed for each sub-indicator. The results are shown in Table 15.

The maximum eigenvalue λ_max = 5.0133 of judgment matrix S is calculated using MATLAB software. To check the consistency of the judgment matrix, the consistency indicator needs to be calculated:

The average random consistency indicator is RI = 1.12. The random consistency ratio is

Therefore, the results of the hierarchical analysis ranking are considered to have satisfactory consistency, that is, the distribution of the weight coefficients is very reasonable. The weights of the sub-indicators “Strategic Compatibility(C14)”,” Company Reputation(C15)”and “Financial Situation(C16)” under the first-level indicator “Others” are shown in Table 16.

Therefore, under the AHP, the weights and comprehensive weights of the level indicators and secondary indicators of port selection by customers are calculated and shown as Table 17.

The dimensions and units of each indicator are different and cannot be directly compared and calculated, so they need to be standardized before evaluation.

When the indicator is positive, its standardized formula is:

When the indicator is negative, its standardized formula is:

According to the above formula, the raw data of the port of Wilhelmshaven, Hamburg, Bremen, and Bremerhaven are processed, and the standardized data forms are shown in Table 18.

According to the calculation formula \( {Z}_i=\sum \limits_{j=1}^p{\omega}_j{x}_{ij}^{\hbox{'}} \), the scores of the ports of Wilhelmshaven, Hamburg, Bremen, and Bremerhaven are calculated and shown as Table 19.

The aforementioned AHP evaluation system reveals the priority of four alternative calling ports in Germany: Hamburg, Wilhelmshaven, Bremerhaven, and Bremen.

According to evaluation based on indicator weights (as set by questionnaire results), there was no difference in priority among the four alternative calling ports. The same results were found using the AHP evaluation system. However, under this same customer- oriented thinking, slightly different concerns are evident.

The indicator of “freight competitiveness” has a significantly higher weight than other indicators, as evidenced in the survey results. This reflects customers’ high attention of freight offer in the questionnaire survey. As dominant factors, freight rate, side condition, volume incentive, freight stability, and quotation efficiency attracted more attention in customers’ port selection. Weights of “Freight Competitiveness” and “Service Capability” are roughly equal under the AHP evaluation system, showing similar levels of concern for freight offer and customer service. Through further analysis of the secondary indicators, it was found that the main differences come from “Cargo Tracing” and “Transshipment Efficiency.” The implicit indicators affecting port selection may be overlooked by customers in the questionnaire survey, but were well considered under the AHP evaluation system. AHP results were both supplementary and confirmatory to the questionnaire conclusions.

Additionally, there is consistency in the main secondary indicators under both evaluation systems. Service route quality, freight rate, and side condition are the key subdividing indicators of customers’ port selection. These effects can also be verified in practice, as customers focus on transit time, frequency, on-time performance, and uniqueness in making port selections. Such choices were made while also considering the freight offer and its side conditions.