Selection of effective risk mitigation strategies in container shipping operations

3.1 Identification of risk mitigation strategies

3.2 Measurement of the effectiveness of risk mitigation strategies

After identifying the strategies, we conducted a large scale questionnaire survey, namely, “mitigation-strategy survey”. This survey was conducted using a five-point Likert scale, where 1 meant “very inefficient” and 5 meant “very efficient”. The respondents were asked to select the level of effectiveness of the strategies based on their work experience.

The population was based on the list from the 2010 ROC National Association of Shipping Agencies in Taiwan, and all 116 container shipping companies in the list were included. Managers from three departments in each company were selected, including the information/documentation department, the physical/operations department and the financial/accounting department. This is because these three departments cover the main risk management issues that arise in container shipping operations. However, some companies did not have all three of these departments. After recalculating the population size, the final effective sample size was 342. After collecting the replies, the rank of these strategies could then be obtained.

3.3 Evaluation of the relative importance of risk mitigation strategies

There are a number of methods for multiple criteria decision making, yet there are some limitations for these methods such as some methods need a large scale of questionnaire replies, some of their purposes are not suitable for our research aim, and some of them need high computer language design skills and extensive quantitative data (Qu et al., 2017). As this study is relating to empirical research and has a relatively limited population to investigate, we decided using AHP as the method to evaluate the relative importance of risk mitigation strategies. AHP proposed by Saaty (1988) has been widely used to evaluate the relevant importance of decision criteria/alternatives in various industries including maritime and port (Ha et al., 2017). The basic concept of the AHP is to assist decision making through a hierarchical structure with different criteria and sub-criteria that are weighed through pairwise comparisons (Wang et al., 2015). Chang et al. (2014) proposed a structure with three risk categories (i.e. financial risk, reputation risk and safety and security risk), whereas the criteria in our study are adapted from this structure and amended as reducing financial loss, reducing reputation loss and reducing safety and security incident related loss. Four axioms of the AHP are assumed in its applications, including reciprocal comparison, homogeneity, dependence and expectations (Saaty, 1988). Reciprocal comparison means that when making paired comparisons, both members of the pair must be considered to judge the relative value. Homogeneity means that when the disparity is great, the elements are placed in separate clusters of comparable size giving rise to the idea of levels and their accommodation. Dependence means that the smaller elements depend on the outer parent elements to which they belong, in a large hierarchical cluster. Expectations are beliefs about the rank of alternatives derived from prior knowledge (Saaty and Kułakowski, 2016). The data were examined and matched to the four axioms.

While the AHP method can be used in many areas (Saaty, 1988; Ho, 2008), it has been criticised in a number of studies (Chang, 1996; Bana e Costa and Vansnick, 2008; Wang et al., 2015). Among the critics, the most common is uncertainty in terms of subjective perception (Chan and Kumar, 2007), which may result in an inaccurate measurement. Respondents may be confused, and hence, provide inconsistent answers when being asked to do pair comparisons or may also result in a lack of data when respondents fail to answer some questions (Wang et al., 2015). To overcome the weakness related to uncertainty, Zadeh (1965) proposed the fuzzy set theory, which fuzzifies the respondents’ perceived value by considering that human beings cannot always perceive exact values. Based on this, Laarhoven and Pedrycz (1983) proposed a fuzzy AHP method by adapting fuzzy numbers (e.g. triangular/trapezoidal) from the fuzzy set theory into the AHP method. With the advantage of fuzzy set theory, the fuzzy AHP, thus, overcomes the shortcomings of the AHP and has become a widely accepted method in multiple criterion decision making under uncertainty. In the maritime area, a number of studies have used the fuzzy AHP to carry out their investigations. For example, Ding (2010) addressed the critical factors, which affect customer value for shipping companies from a customer perspective. Ka (2011) used the fuzzy AHP to determine selection of locations for China’s dry ports. Yang and Chung (2013) applied the fuzzy AHP to find preferred ship flag registry locations among Taiwan, Hong Kong and China. This study, therefore, also uses the fuzzy AHP method to cope with the subjective perceptions of respondents. To keep the research task at a manageable scale, a set of the most important risk mitigation strategies are selected from the results of the “mitigation-strategy survey” to conduct a further survey evaluation, namely, the fuzzy AHP survey:

A fuzzy AHP analysis includes the following eight main phases (Buckley, 1985; Ding, 2010; Ding and Tseng, 2012):

1. Develop a hierarchical structure with three criteria and six alternatives:

In this study, the hierarchy structure included three major levels, namely, goal, criteria and alternatives. The goal refers to “mitigating risks in shipping operations”. Three criteria corresponding to three risk consequences were identified as “reducing financial losses”, “reducing reputation loss” and “reducing safety and security incident-related losses”. The alternatives included a set of the most important risk mitigation strategies selected from the results of the mitigation-strategy survey. The number of selected strategies is usually less than seven, as the brains of human beings cannot compare more than seven items at the same time:

• (2)Collect pairwise comparison matrix of decision elements:

Let xijh, h = 1,2,…, 12, be the relative importance given to reducing risk consequence i compared to reducing risk consequence j by expert h at the criteria level; let xsth, h = 1, 2,…, 12, denote the relative importance given to risk mitigation strategies s compared to risk mitigation strategies t by expert h at the alternative level:

• (3)Transform relative importance into a triangular fuzzy number (TFN):

TFN combines the minimum value (denoted by l), maximum value (denoted by u) and mean value (denoted by m) of the opinions of all experts. The meaning of TFN used in the fuzzy AHP is presented in Table II .

• (4)Build a fuzzy positive reciprocal matrix:

The TFN was used to build a fuzzy positive reciprocal matrix. To illustrate the TFN application, the results of one respondent are shown below. There are three criteria at the criteria level; thus, the fuzzy positive reciprocal matrix is a 3 × 3 matrix that can be generated by:

where B˜ijC represents the TFN of the relative importance of reducing risk consequence i over reducing risk consequence j, B˜ijC⊗B˜jiC=1,   ∀i, j=1, 2, 3,

where i = 1 or j = 1 means reducing financial loss;

i = 2 or j = 2 means reducing reputation loss; and

i = 3 or j = 3 means reducing safety and security-related incident loss:

• (5)Calculate the fuzzy weights of the fuzzy positive reciprocal matrices.

The method for calculating the fuzzy weights W̃ is separated into two steps: calculate the geometric mean Z̃_i and Z̃_s of the fuzzy comparison value of reducing risk consequence i and alternative s; and calculate the fuzzy weight W̃_i and W̃_s of the reducing risk consequence i and alternative s. At the criteria level, we use the criterion “reducing financial loss” as the example, where the geometric mean of TFN of the i-th criterion can be given by:

To simplify the notation, the fuzzy weight can be further denoted by:

• (6)Defuzzify the fuzzy weights to crisp weights:

After obtaining the fuzzy weights, we converted them into crisp ones using a centroid defuzzification method (Ali et al., 2012).

• (7)Standardise the crisp weights:

To facilitate the comparison of the relative importance between criteria, the obtained crisp weights (in Step 6) are standardised by:

SwiC=wiC∑i=1kwiC= for the criteria.

• (8)Calculate the integrated weight for each level:

After standardising the crisp weights, the integrated weight for each criterion is computed by taking into account the weight at the current level and its upper level. More specifically,

The integrated weights of each criterion at the criteria level are given by (note that the weight at its upper level is 1:

4.1 Respondents’ profile and validity and reliability test

According to Davis (2005), several common methods are used to enhance the level of validity, including careful identification of the measurement items from the literature and an expert interview to validate the identified items. The questions in the mitigation-strategy survey were designed based on the literature review and were validated through the seven face-to-face interviews to ensure a high level of validity.

After collecting the replies from the mitigation-strategy survey, we identified 62 (out of 88 replies) valid and 26 invalid feedbacks. The valid response rate was 18.13 per cent. The 62 respondents’ profile in the survey is presented in Table III . The results show that approximately 80 per cent of respondents has work experience more than 10 years. The respondents’ department include President/vice-president, information/document, financial/accounting and operation/shipping department. For the position, there is an approximate 80 per cent of respondents who are vice director or above. A reliability test was conducted for the questions on risk mitigation strategies, for which the Cronbach’s alpha was 0.872 (>0.8). It indicated that the designed questions on risk mitigation strategies were reliable.

4.2 Ranking of risk mitigation strategies

Table IV shows the results of the different risk mitigation strategies based on the data from the risk-mitigation survey. Based on the mean score, the top six strategies include “form alliances with other shipping companies” (mean score: 4.02); “use more advanced infrastructures and technologies (hardware and software)” (mean score: 3.92); “choose partners more carefully” (mean score: 3.87); “enter into long-term contracts with shippers” (mean score: 3.85); “collaborate with partners (e.g. port operators, inland transportation operators) through making joint long-term plans” (mean score: 3.85) and “flexible design of the timetable/schedule, e.g. include time buffers” (mean score: 3.81).

The strategy “acquire and merge with other shipping companies” (mean score: 2.95) had the lowest score among all mitigation strategies. As this strategy has a long-term, significant impact on shipping company operations, it often implies a high degree of uncertainty and may only be adopted in critical situations. On the other hand, it is interesting to observe that the recent popular practice of slow steaming had a relatively low score among these strategies. Slow steaming can reduce fuel consumption and absorb the idle capacity, which is an appropriate strategy for shipping lines when supply exceeds demand. However, it increases transit time and could cause extra inventory costs to shippers. The low score for slow steaming with the highest S.D. indicates that the respondents’ opinions were very different. Some of the respondents felt that slow steaming is a realistic strategy to reduce risks within container shipping operations, while the others hold an opposite opinion.

When considering the three different “cooperation” levels (Strategies 14, 15 and 16), the results show that collaboration level had the highest mean score with 3.85, following by coordination level (mean score: 3.77), and cooperation level has the lowest mean score with 3.65. This indicates that shipping companies will have better risk mitigation effects if the companies have a higher level of “cooperation” relationships with partners.

To evaluate the priority of these mitigation strategies under the three criteria (i.e. reducing financial loss, reducing reputation loss and reducing safety and security incident-related loss), the top six strategies were selected for a further step by conducting the fuzzy AHP. The reason for selecting the top six strategies is because the human being’s brain will be confused when comparing more than seven items (Saaty, 1977). In addition, several studies also suggested that to serve both consistency and redundancy to the AHP method, it is best to keep the number of criteria and alternatives at seven or less (Saaty and Ozdemir, 2003; Russo and Camanho, 2015). In this study, there are two strategies ranked at the seventh among the 18 ones, we thus selected six strategies. As shown in Table IV, the six selected strategies include: “form alliances with other shipping companies” (renamed as Strategy A for the fuzzy AHP analysis in Section 4.3); “use more advanced infrastructures (hardware and software)” (Strategy B); “choose partners very carefully” (Strategy C); “collaborate with partners (e.g. port operators, inland transportation operators) through making joint long-term plans (collaboration level)” (Strategy D);” “enter into long-term contracts with shippers” (Strategy E) and “flexible design of the timetable/schedule, e.g. include time buffers” (Strategy F).

4.3 Fuzzy AHP analysis

After building the hierarchy structure, paired comparisons were conducted for the fuzzy AHP survey. The purpose was to evaluate the relative importance of the different criteria and different alternatives in the fuzzy AHP model. The population size for the fuzzy AHP survey was still 342 in this study, (i.e. the same as that for the mitigation-strategy survey); however, the sampling process was different. To increase the return rate, all respondents to the first questionnaire survey were selected. In addition, some key managers from the non-responding list were also selected. Finally, a total of 114 questionnaires were sent out and 21 replies were received; including 12 valid ones and 9 invalid ones. The valid return rate was 10.53 per cent. Microsoft Office Excel software was then used to conduct the fuzzy AHP analysis. The results show the consistency ratio (CR) of each criterion to be 0.01, which is less than the standard acceptable value (0.1). Therefore, the data met the consistency requirement. After confirming the requirement.

The weights of the criteria and strategies were calculated by averaging the weight value of the 12 respondents’ perceived value. In addition, by combining the criterion priorities and the relevant alternative priorities, we were able to obtain an overall priority ranking of the decision alternatives shown in Table V .

Table V shows that the weights of the criterion “reducing financial loss” (0.424) and “reducing safety and security incident-related loss” (0.420) are much greater than “reducing reputation loss” (0.156). This indicates that the first two criteria are more important under the goal of mitigating risks in shipping operations. It is easy to understand that almost every company pays a lot of attention to reducing financial loss. However, “reducing safety and security incident-related loss” is also important in container shipping operations due to the dangerous work environment. Compared to retailer operations in which maintaining reputation and brand are of high priority (Dawar and Parker, 1994), container shipping operations tend to focus more on financial loss reduction and safety and security incident-related loss reduction.

Under the criterion “reducing financial loss”, Strategy E: “Enter into long-term contracts with shippers” and Strategy A: “Form alliances with other shipping companies” are the top two strategies for mitigating financial loss in container shipping operations. These two strategies can be used to tackle and reduce the risk caused by transportation demand uncertainty. Moreover, the global importance of Strategy E (0.108) was twice more than the one of Strategy F (0.048).

Under the criterion “reducing reputation loss”, Strategy A: “Form alliances with other shipping companies” was evaluated as the most important strategy. However, the variation of the weights of the six strategies under this criterion was insignificant, indicating their contributions to reducing reputation loss have no vast difference.

Under the criterion “reducing safety and security incident-related loss, Strategy A: “Form alliances with other shipping companies” was calculated as the most important risk mitigation strategy, and its global weight (0.089) doubled that of Strategy F (0.046).

To understand the importance of the mitigation strategies over all three criteria, we calculated the overall priority of each strategy, by calculating the sum of the global weights of each strategy under three criteria. The calculations of the overall priority of individual strategies are as follows:

In a similar way, the overall priority values of strategies B to F are obtained as 0.165, 0.162, 0.160, 0.197 and 0.112, respectively. Such a result reveals that from the overall perspective, the best risk mitigation strategies are strategies A and E, while the worst is strategy F. Strategies B to D of a priority value around 0.16 present a large distance to both best and worst strategies. It, therefore, can help ship lines to rationalise and justify their safety resources on different risk mitigation strategies with respect to the priority values.