Evaluating context-dependent efficiency of the marine transport industry

Before the outbreak of COVID-19, the maritime shipping industry was in a slump for almost ten years, because global trade volume did not catch up with the increase in vessels and capacities, leading to low ship prices and low freight rates (ISL, 2019). The industry found itself under the spotlight during COVID-19, as governments around the world imposed lockdown measures. The marine transport industry suffered an unprecedented hit when the global economy was disrupted. Previous literature has extensively studied COVID-19's impact on the marine transport industry, focusing on diverse topics such as stock price (Kamal et al., 2021), transport volume and capacity (Loske, 2020), and the health of seafarers (Doumbia-Henry, 2020). The COVID-19 pandemic nearly shut down the industry during the first half of 2020, but surprisingly ignited historically high demand for cargo during the second half of 2020. The freight rate soared in response to increasing demand, bringing many shipping companies out of the red and spurring great discussions. The efficiency of shipping companies has once again become a focus of researchers and people worldwide.

COVID-19 was not the only challenge the marine transport industry faced in recent years as the public blamed it for its negative environmental impact. The industry's carbon dioxide (CO₂) emission accounts for around 2.2% of global emissions (Smith et al., 2015). The negative effect impacting the environment and human health has raised public concern. In response, the International Maritime Organization (IMO), aiming to lower the negative impacts generated from shipping, has published the Initial Strategy, in which IMO set the goal of reducing industry emissions by at least 50% by 2050 (IMO, 2018). The European Union also enforced regulations to manage CO₂ emissions from maritime companies (Jessen, 2016). Because the concept of environmental, social and corporate governance (ESG) has become more and more popular among businesses, shipping companies must deploy strategies to lower their damage to the environment. Thus, finding the balance between the bottom line and the responsibility to protect the environment is now the most critical task for shipping companies. This trend also has shifted the focus of researchers from solely evaluating the bottom line or operating efficiency to taking the negative impact on the environment into account or providing solutions to reduce CO₂ emissions and other hazardous pollutants such as sulfur oxides and nitrogen oxides (Deng et al., 2021; Tran and Lam, 2022).

To reduce cost and improve competitive advantage, shipping companies frequently forge strategic alliances to partner with competitors (Lee and Song, 2010). In the 1990s, the first wave of strategic alliances was formed (Midoro and Pitto, 2000), and in the later years the alliance structure changed drastically. After several changes and reconfigurations, nearly all the largest shipping companies have since joined one of the largest three alliances (THE Alliance, 2M and Ocean Alliance), which dominate the marine transport industry and take up around 80% of the global container market share (Merk et al., 2018). For shipping companies, participating in strategic alliances can bring potential benefits (Lee, 2014; Lei et al., 2008; Mitsuhashi and Greve, 2009). In short, with the long history and the benefits brought to the companies within, strategic alliance is not only a critical feature of the marine transport industry but also a theme that cannot be ignored in marine transport industry studies.

Previous studies on the operational performance of the shipping industry have primarily focused on financial performance and overall company operations (Panayides et al., 2011; Venkadasalam et al., 2020; Ding et al., 2015; Yang and Yip, 2019), with limited consideration given to environmental impacts and related operational factors. As a result, environmental indicators have rarely been incorporated into efficiency analysis as inputs or outputs. International shipping accounted for approximately 3% of global carbon emissions in 2017, making it one of the industries with significant environmental impact. More than 50% of vessel operating costs are attributed to fuel consumption, and the widespread use of heavy fuel oil further exacerbates environmental damage (IRENA, 2019). With increasing environmental concerns and tightening regulations, shipping companies are now required to balance ecological sustainability with business growth, making environmental considerations an emerging challenge for operational efficiency.

In recent years, many studies have employed data envelopment analysis (DEA) to address undesirable outputs; however, most have focused on heavily polluting industries, with relatively few applications in the maritime sector. Traditional performance evaluations in shipping rarely account for negative by-products of production, such as pollution. Gong et al. (2019) indicate that environmental performance influences the preferences of customers and investors, and that environmentally oriented strategies are more attractive. Therefore, incorporating both traditional financial indicators (e.g. operating costs and revenue) and environmental indicators (e.g. CO₂ emission intensity) is essential for a more realistic assessment. This study addresses the long-standing gap in maritime research regarding the inclusion of undesirable outputs.

Conventional DEA applications in maritime and transportation research typically evaluate all decision-making units (DMUs) simultaneously, categorizing them into relatively efficient and inefficient groups (Ahn and Min, 2014; Bang et al., 2012; Gong and Kim, 2015; Mallikarjun, 2015; Venkadasalam et al., 2020). In such settings, inefficient DMUs do not influence the efficient frontier, and their efficiency scores are determined solely relative to efficient DMUs. Moreover, improvement paths are derived based on projections toward the efficient frontier under a radial framework. Consequently, inefficient DMUs play a limited role in the evaluation process, and the suggested adjustments may not always be feasible for certain companies, reducing the practical relevance of the results.

To address this limitation, this study adopts the context-dependent DEA model proposed by Seiford and Zhu (2003), which allows for the construction of multiple efficiency frontiers across different performance layers. This approach enables inefficient DMUs to play a more meaningful role in the evaluation process and allows companies to be classified into different efficiency tiers, thereby providing more realistic benchmarks and improvement targets. Unlike traditional DEA, which evaluates all DMUs under a single reference set, context-dependent DEA conducts efficiency assessments under multiple evaluation contexts. This layered structure helps reduce unrealistic improvement requirements and facilitates the development of more practical and stage-based policy implications.

The main objective of this research not only investigates the traditional aspect of operational efficiency pre- and post-COVID-19 outbreak but also considers the negative impact on the environment produced by shipping activities pre- and post-IMO 2020 regulation. It also seeks to test if joining a strategic alliance affects the overall efficiency, financial efficiency and the environmental efficiency of shipping companies.

The rest of this paper is organized as follows. Section 2 proposes the research hypotheses and reviews the previous literature on the marine transport industry, DEA and the application of DEA on efficiency evaluation of shipping companies, as well as examining the relationship between ESG strategies and maritime efficiency. Section 3 discusses the model settings as well as the variable definitions, sources and descriptive statistics. Section 4 reports the results of the empirical study of this research for global shipping companies. Section 5 concludes.

The DEA method is a linear technique for evaluating the productive efficiency of DMUs, constructed in the seminal work of Charnes et al. (1978) and extended from the one-input, one-output model (Farrell, 1957). DEA has since become a widely employed performance assessment tool. Researchers have conducted empirical studies to evaluate the productivity and efficiency of companies across different industries, including the marine transport industry. Hu et al. (2022) applied the input congestion DEA to compute efficiency scores and input congestions of 54 major marine transport companies worldwide during 2010–2019. Results of their paper showed that most companies face inefficiency and input congestion problems. Chao (2017) employed a multi-stage DEA model to assess the efficiency of liner shipping companies. Panayides et al. (2011) adopted the DEA method to test the efficiency of 26 leading maritime companies with results presenting that container carriers have higher efficiency than dry bulk and tanker carriers. Gutiérrez et al. (2014) computed the efficiency of 18 liner shipping companies in 2009 adopting the DEA approach. The model includes inputs of fleet capacity, labor and ships and outputs of container throughput handled and turnover.

Unlike the CCR (Charnes et al., 1978) and BCC (Banker et al., 1984) models, the slacks-based measure of efficiency in DEA (Tone, 2001) is a non-radical model. Later studies extended and adjusted the settings of this scheme to deal with different scenarios. For instance, Tone (2004) proposed a slacks-based model that allows undesirable outputs. Compared to those that analyzed the efficiency of shipping companies, more studies applied the SBM model to assess the efficiency of ports (Lee et al., 2014; Chang et al., 2018). Several other researchers also applied the SBM approach to measure port efficiency (Elsayed and Khalil, 2017; Wang et al., 2018). However, not many have evaluated shipping companies via the SBM approach, with the work of Gong et al. (2019) being one of the exceptions. Gong et al. (2019) tested different sets of inputs and outputs to analyze the ordinary efficiency, environmentally adjusted efficiency and environmental efficiency of leading shipping companies, suggesting that companies can improve their rankings by reducing their negative environmental impact. Wang et al. (2019) took undesirable outputs, including air emissions, solid wastes and water wastes, into account and adopted the super-SBM model and Malmquist productivity index to measure the efficiency of cruise shipping companies.

In addition, some studies have employed stochastic frontier analysis (SFA) to evaluate efficiency. Compared to the non-parametric approach of DEA, SFA specifies a production function and is capable of distinguishing random error from inefficiency, thereby providing an alternative framework for efficiency measurement. In the existing literature, Delfin-Ortega (2025) further integrates the metafrontier approach to examine technical efficiency and sustainability performance in the maritime industry under different technological conditions, thereby extending the application of SFA to environmental and sustainability issues. Gavalas (2016) applies SFA to assess the efficiency of shipping companies and finds that company size and resource utilization are key determinants of efficiency, with significant variation across companies. In the context of port efficiency, Cullinane and Song (2006) use SFA to evaluate the relative efficiency of European container ports, identifying economies of scale and operational models as critical influencing factors. More recent studies, such as Bui and Cho (2025), continue this line of research by analyzing the operational efficiency of Vietnamese container ports, revealing persistent efficiency disparities across ports and highlighting infrastructure and operational capabilities as important determinants. Overall, SFA-based studies primarily focus on measuring technical efficiency and identifying sources of efficiency variation, emphasizing the roles of scale, resource allocation and managerial practices.

Based on the above literature, although existing studies have extensively examined efficiency issues in the maritime and port sectors, several limitations remain. First, most studies focus on a single dimension of efficiency (e.g. operational or technical efficiency), with relatively limited consideration of environmental performance. Second, the application of SFA in handling multiple outputs is relatively complex. It often requires aggregating multiple outputs into a composite measure or specifying a multi-output functional form, which may introduce additional modeling challenges. Moreover, when a large number of input variables are included, multicollinearity among inputs may arise, potentially affecting the reliability of the estimation results (Coelli et al., 2005; Kumbhakar and Lovell, 2003). In addition, traditional DEA models may not fully capture the impact of undesirable outputs on efficiency evaluation. Therefore, to provide a more comprehensive assessment of shipping firms' performance, this study adopts the SBM-DEA model incorporating undesirable outputs. Without requiring the specification of a production function, the proposed approach simultaneously considers multiple inputs, desirable outputs and undesirable outputs and integrates both operational and environmental efficiency. This enhances the completeness, flexibility and practical interpretability of the efficiency evaluation.

With the growing importance of sustainable development, ESG (environmental, social and governance) has increasingly become a central element in the operational and strategic decision-making of the shipping industry. Andrikopoulos (2026) points out that existing studies generally recognize the environmental dimension – particularly carbon emissions and regulations issued by the International Maritime Organization (IMO) – as key factors influencing the performance of shipping companies. However, the study also highlights that most of the literature tends to focus on single dimensions, and the integration of environmental performance into efficiency evaluation remains limited. Regarding ESG disclosure, Tsatsaronis et al. (2024) find significant differences in the quality and transparency of ESG reporting across shipping companies, with larger and more international companies typically exhibiting higher disclosure standards. Among ESG components, environmental aspects (such as carbon emissions and energy usage) are the most comprehensively reported, indicating that the industry is gradually responding to decarbonization pressures and regulatory requirements.

Furthermore, the relationship between ESG and company performance is not strictly unidirectional. Le et al. (2026) argue that ESG investments may increase costs and exert pressure on operational efficiency in the short term; however, from a long-term perspective, they enhance competitiveness and risk management capabilities. In other words, a trade-off exists between ESG and efficiency, although companies may achieve both high efficiency and strong ESG performance through appropriate resource allocation. From an evaluation perspective, Wan et al. (2026) propose an integrated ESG assessment framework and identify environmental factors – particularly emissions and energy efficiency – as the core determinants of overall company performance, while governance mechanisms also play a crucial role in sustainable development. Similarly, Diamantara (2025) and Lee et al. (2023) emphasize that ESG has evolved from a disclosure-oriented concept into a key driver of corporate strategy and competitive advantage.

In addition, Lin et al. (2022) show that both financial and environmental factors are critical determinants of shipping company performance, and that the COVID-19 pandemic has altered the relative importance of these factors, highlighting the need for companies to balance efficiency and sustainability under uncertainty. Li et al. (2026) further summarize that the relationship between ESG and performance remains inconclusive, and that most studies lack quantitative approaches that integrate efficiency analysis with ESG considerations. In recent years, the shipping industry has shifted from a purely efficiency-oriented focus toward a broader emphasis on resilience and sustainability. Arslan and Akturan (2026) indicate that, under external shocks such as COVID-19, companies are transitioning from efficiency-driven strategies to resilience-oriented approaches, leveraging digitalization, ESG and human capital to enhance risk management and operational stability.

Overall, the existing literature on the shipping industry primarily examines ESG issues from the perspectives of disclosure, performance relationships and indicator development. Environmental factors – especially carbon emissions and energy usage – have become key determinants of company performance and competitiveness. However, the relationship between ESG and efficiency remains inconclusive, and most studies still focus on single dimensions. The integration of operational efficiency and environmental performance remains limited, indicating that further research in this area is needed.

Because of the unique and important role of the marine transport industry in the global economy, past studies have investigated different aspects of the industry. Jenssen and Randøy (2006) shed light on the causes of organizational innovation and how innovation influences the performances of Norwegian shipping companies. Lu et al. (2009) discovered that the financial performance of shipping companies in Taiwan is positively affected by corporate social responsibility performance. Because the marine transport industry causes substantial environmental impacts, many studies, based on results, have provided suggestions or strategies from different aspects seeking to help shipping companies reduce their damage to the environment (Wan et al., 2016, 2018).

Existing publications have approached the topic of strategic alliances within the marine transport industry with several focusing on the formulation of strategic alliance and discussing the mechanism of partner selections and the incentives to cooperate with other companies (Midoro and Pitto, 2000; Mitsuhashi and Greve, 2009). Slack et al. (2002) found that alliances have increased uniformity in the industry and have intensified companies' operations. Song and Panayides (2002) listed the potential benefits of alliances in five areas: financial, economic, strategic, marketing and operational goals. Strategic alliances also contribute to reducing the negative effect on the environment. Tan and Thai (2014) showed that regulations and environmental issues are often discussed topics within alliances. Qiu et al. (2018) suggested that an alliance of vessel sharing can dramatically reduce CO₂ emissions. Based on the research above, the following hypothesis is proposed.

The lockdown measures enforced by governments to fight the COVID-19 outbreak led to the side effect of a drop in global trade value. The performance of the marine transport industry deteriorated markedly as most industry activities fell or stopped entirely. Millefiori et al. (2021) pointed out that maritime mobility dropped as a result of the restrictions. The lockdown measures also changed consumer tendencies and demands as e-commerce surged in many countries (Kent and Haralambides, 2022). More and more consumer goods and packages were shipped to different places worldwide, leading to rising demand for container transport (Merk et al., 2022). The increasing demand then led to skyrocketing freight rates, extended waiting time and limited cargo space (UNCTAD, 2021). All these factors resulted in great financial performances for shipping companies in the second half of 2020. However, it is unclear whether the great performances in the second half can cover the losses from the first half. Based on the information, this research proposes the following hypothesis.

The marine transport industry is characterized by high capital intensity and economies of scale, requiring substantial investments in vessels and operational costs, while also facing risks associated with freight rate volatility and demand uncertainty. In this context, strategic alliances have become an important mechanism for shipping companies to enhance their competitiveness (Lee and Song, 2010). The three major global shipping alliances facilitate resource integration and operational coordination, enabling member companies to share vessel capacity, optimize route deployment and improve fleet utilization (Yang, 2020). Such cooperative arrangements help reduce operating costs, avoid redundant investments and enhance overall transport efficiency.

In addition, alliances enable companies to improve market responsiveness through information sharing and joint decision-making, thereby mitigating the negative impacts of operational uncertainty (Lee and Song, 2015; Loon et al., 2023). Prior studies suggest that strategic alliances can enhance company performance through economies of scale and resource-sharing mechanisms (Alhorr et al., 2012; Todeva and Knoke, 2005). Therefore, ceteris paribus, shipping companies participating in strategic alliances are expected to exhibit superior financial efficiency compared to non-alliance companies. Based on the research above, the following hypothesis is proposed.

The marine transport industry is highly energy-intensive, with its operations generating substantial CO₂ emissions. As a result, environmental efficiency has become an important dimension in evaluating company performance (Di et al., 2018; Wang et al., 2020). Strategic alliances facilitate resource sharing, information exchange and operational coordination among member companies (Parise and Henderson, 2001; Solesvik and Westhead, 2010). Through these mechanisms, alliance members are able to optimize route deployment, improve fuel efficiency and adopt cleaner technologies, thereby reducing CO₂ emissions. Existing studies further suggest that inter-company collaboration and strategic alliances contribute to reductions in CO₂ emissions and other environmental impacts (Qiu et al., 2018; Wong et al., 2018). Based on the above theoretical and empirical foundations, this study proposes the following hypothesis.

The first step of applying the DEA method is to set the scale of analysis by choosing the DMUs for this study. According to the Guiding Principles and Methodology for GICS (MSCP, 2020), the code categorizes companies into different sectors, industry groups, industries and sub-industries. This research picks companies in the Marine industry (Industry Code: 203030) since not all companies have disclosed financial or ESG reports, ending up with a sample of 38 companies during 2019–2020. Table 1 lists the companies included in this research and whether they were part of the three main alliances.

Table 1 also reveals substantial heterogeneity among the sample companies in terms of company size and carbon emissions. Company size is measured by total assets, ranging from approximately 400 million USD to over 55,000 million USD, indicating that the marine transport industry is highly capital-intensive. This also reflects notable differences in the operational scale of the sampled companies. Similarly, CO₂ emissions vary considerably, ranging from less than 0.3 million tons to over 45 million tons, highlighting differences in fleet size and operational models across companies.

From a geographical perspective, most strategic alliance members are concentrated in Asia and Europe and are typically large-scale shipping companies. In contrast, non-alliance companies are more likely to be small- to medium-sized or operate in niche markets. Moreover, larger companies tend to be associated with higher absolute carbon emissions, which is consistent with their larger operational scale. However, this also underscores a potential bias in performance evaluation if emissions are assessed solely based on absolute levels. Therefore, it is necessary to incorporate CO₂ emission intensity into environmental efficiency measures in order to more accurately evaluate companies' relative performance in carbon reduction.

This research reviews existing literature on the efficiency assessment of shipping companies and based on the purposes and data availability selects three inputs (including number of employees, number of vessels and operating cost), one desirable output (Revenue) and one undesirable output (CO₂ Emission Intensity). In the DEA model, the three inputs are separable inputs, while the desirable and undesirable outputs are non-separable. Table 2 summarizes the input and output variables and lists previous papers that used these variables.

Number of employees is a commonly used input indicator in related literature (Hu et al., 2022; Panayides et al., 2011). In this research, the number of employees includes employees of the whole company instead of only counting the ones for related businesses, because of data availability. The number of ships represents the operating capacity of the shipping company. Some studies used shipping capacity (Chao et al., 2018), while some used both the number of ships and shipping capacity simultaneously (Bang et al., 2012; Gong et al., 2019). This research selects number of vessels, because of data availability. Operating cost represents the cost of generating revenue for the entire operation. Companies seek to minimize operating cost to gain competitive advantages. Hsu et al. (2013) treated operating cost as an undesirable output. Revenue reflects the economic performance generated by resource input. While some studies used net sales (Hu et al., 2022) or profit (Lun and Marlow, 2011) to represent the economic performance of a company, revenue is a popular output indicator among studies on the efficiency of shipping companies.

To mitigate the influence of company size differences on the evaluation of true operational efficiency, operating costs are normalized by the number of vessels, while company revenues are normalized by the number of employees. In the marine transport industry, operating costs primarily arise from capital investment in vessels, as well as their management and maintenance, all of which are closely related to fleet size (Albertijn et al., 2011; Turan et al., 2009). Revenue per employee serves as a key indicator for capturing companies' underlying operational efficiency and productivity, highlighting the importance of human capital as a critical driver of value creation. The number of employees directly influences the overall profitability of shipping companies (Bryan, 2007; Tapaninen and Hilmola, 2025). Revenue is considered a desirable and non-separable output in this research.

CO₂ emissions are selected to represent the pollutant emitted by the shipping companies and are considered a non-separable undesirable output. Although CO₂ is not the only undesirable output generated by shipping companies, it is adopted as the sole undesirable output due to data availability constraints. However, relying solely on total emissions may introduce bias due to differences in company size, placing larger companies at a disadvantage in the evaluation. Therefore, to enhance comparability across companies of different sizes, this study further adopts CO₂ emission intensity as the undesirable output indicator, defined as CO₂ emissions per unit of revenue. This approach provides a more accurate representation of companies' environmental efficiency.

This research sets the three inputs as separable inputs since the shipping companies are able to change the input separately. Companies can modify the number of employees, the number of vessels or operating activities to improve their efficiency. These adjustments are not required to be done simultaneously, and thus this research sets these inputs as separable inputs. On the other hand, the two outputs of revenue and CO₂ emission intensity cannot be treated separately. When generating the desired output, the companies cannot avoid generating the negative output as well. Thus, this research set the two outputs as non-separable in the model.

The slacks-based measure in DEA (Tone, 2001) is a non-radial model. In response to increasing environmental conservation awareness at the time, Tone (2004) created a new scheme to deal with undesirable output based on the slacks-based measure. Considering the fact that producing desirable outputs can lead to simultaneously generating bad outputs that are not separable, Tone and Tsutsui (2006) defined separable inputs as $XS∈RM1×n$⁠. Non-separable desirable outputs and non-separable undesirable outputs are defined as $YNSG∈RS21×n$ and $YNSB∈RS22×n$⁠, respectively. All inputs and outputs have a value greater than 0. The new production possibility set P_NS is defined by:

The non-separable SBM model is defined as:

where m₁ is the number of separable inputs; n₂₁ and n₂₂ are number of non-separable good outputs and number of non-separable bad outputs, respectively; $α$ is the reduction proportion of inputs and outputs; $λ∈Rn$ is the intensity vector; and $sS−$⁠, $sNSG$⁠, and $sNSB$ are slacks of separable inputs, slacks of non-separable desirable outputs and slacks of non-separable undesirable outputs, respectively.

Based on Model (1), the SBM efficiency score is calculated and normalized between 0 and 1. When ρ* = 1, the DMU is efficient as it locates on the efficient frontier and all slacks are 0 (Morita et al., 2005). This research selects the method of SBM-DEA with undesirable outputs since it fits the research purpose of this research to consider negative outputs in evaluating efficiency. For this particular modeling situation, this research utilizes three separable inputs (number of employees, number of vessels and operating cost), one desirable output (Revenue) and one undesirable output (CO₂ Emission). The two outputs have a non-separable relationship and there are no separable outputs existing in the model.

The context-dependent DEA approach (Seiford and Zhu, 2003) follows the concept of dividing DMUs into separate efficient frontier levels. By removing the efficient DMUs, the remaining DMUs that were inefficient under the first frontier then form a second efficient frontier. The process is continued until there is no DMU left. The procedure of context-dependent DEA is presented as follows.

1. Execute the non-separable SBM model in Equation (1) by using all DMUs (all companies).

2. Remove the efficient companies (i.e. ρ* = 1) and execute the non-separable SBM model by using the inefficient companies to determine a new efficient frontier.

3. Repeat Step (2) to remove the efficient companies and form the next level of efficient frontier until there is no inefficient companies left.

Hu and Chang (2016) suggest that the efficiency of individual inputs and outputs can be calculated as the ratio of observed values to target values, thereby providing further insights into the efficiency performance of each component. In this study, disaggregated output efficiencies are employed to examine the improvement potential of both desirable outputs (revenue) and undesirable outputs (CO₂ emission intensity) in shipping companies, as well as their implications for financial and environmental efficiency. The formulation is given in Equation (2):

where o denotes the decision-making unit (DMU), and t represents year t. Since the target output is always greater than or equal to the observed output, the disaggregated output efficiencies score ranges between 0 and 1, which is consistent with the general definition of efficiency in the literature (Coelli, 1995).

The BCG matrix is a widely used tool for commercial analysis. The matrix categorizes companies into four quadrants based on market share and growth rate (BCG, 2022). Several studies have combined the DEA method and the concept of BCG matrix to map the efficiency positions of companies in the industry (Hu et al., 2022; Lin et al., 2005; Lu et al., 2021). This research applies the idea of the BCG matrix to check if companies in the higher tiers also perform better in CO₂ emission control. By discussing undesirable output efficiency separately, the companies that emit a large amount of CO₂ when generating revenue will have a lower BCG score, while the companies with higher undesirable output efficiency will be rewarded with a higher BCG score.

This research sets the efficiency tiers and the environmental efficiency as the x-axis and the y-axis, respectively. Environmental efficiency is defined as the ratio of the target emission amount to the actual emission amount. This research uses the median as the threshold. If a shipping company is in a tier equal or higher than the median and has higher environmental efficiency than the median in a year, then its annual BCG score is 4. If a shipping company is in a tier lower than the median, but has higher environmental efficiency than the median in a year, then its annual BCG score is 3. If a shipping company is in a tier equal to or higher than the median, but has lower environmental efficiency than the median in a year, then its annual BCG score is 2. If a shipping company is in a tier lower than the median and has lower environmental efficiency than the median in a year, then its annual BCG score is 1.

Mann and Whitney (1947) constructed the Mann–Whitney U test, which is a non-parametric test widely utilized to assess the differences in the distributions of two independent samples. Assume that two independent groups, X and Y, have $n1$ and $n2$ samples, respectively. The first step is to mix the samples of the two groups and then rank the samples by efficiency score, sorting from highest to lowest. The next step is to sum up the ranks of each group, getting the total ranks for the two groups, $R1$ and $R2$⁠, and then denote the higher of $R1$ and $R2$ as R. With $n1$⁠, $n2$⁠, and R, the value of U can be expressed as:

When $n1$ and $n2$ are larger than 10, the distribution of U approaches normal distribution. The Z-score can be expressed as:

If the absolute value of the z-score is above the critical value at a significant level, then the two groups have a significant difference in their total rankings.

The summary of statistics is presented in Table 3. The results show that there is no value equal to or less than zero for all input and output variables. Note that the two nominal variables, operating cost and revenue, have been deflated into real variables in the base year of 2019 by GDP deflators of the United States in order to perform a fair evaluation. To mitigate the influence of company size on the evaluation of operational performance and to more accurately reflect companies' true efficiency, this study adopts a scale normalization approach. Specifically, operating costs and revenues are normalized by the number of vessels and the number of employees, respectively, and are used as key indicators in the subsequent analysis. In addition, CO₂ emission intensity is employed as the measure of the undesirable output.

The correlation matrix of inputs and outputs is presented in Table 4. The outcomes of Table 4 indicate that almost correlations among the inputs and outputs are positive, and the p-values in the parentheses indicate that the results are statistically significant. It can be inferred from the correlation matrix that, on average, when increasing the input variables, the output variables also increase. A significant positive relationship is observed between the number of employees and the number of vessels, indicating that larger companies tend to operate larger fleets and employ more personnel, which is consistent with the scale characteristics of the shipping industry. Notably, carbon emission intensity is significantly negatively correlated with all input and output variables. This finding implies that companies with larger scale or more efficient resource utilization tend to generate lower emissions per unit of output, reflecting the presence of economies of scale and the environmental efficiency advantages of larger companies. Overall, no abnormal relationships are observed among the variables, suggesting good internal consistency of the data and supporting its suitability for subsequent analysis.

This research applies the context-dependent SBM-DEA to test the efficiency of sample shipping companies in 2019 and 2020. Table 5 shows the efficiency tiers and scores for shipping companies in 2019. After evaluating all the companies by Model (1), the efficiency scores ρ are reported in the column of Tier 1. Several companies (DMU #5, #17, #23, #28, and #35) have efficiency score of 1, indicating that they are efficient. This research then removes the efficient companies, reruns Model (1) to evaluate the remaining companies, reports the efficiency scores and repeats the process until there are no inefficient companies. In 2019, 5 companies were in Tier 1, 4 companies were in Tier 2, 6 companies were in Tier 3, 4 companies were in Tier 4, the 10 companies were in Tier 5, 8 companies were in Tier 6 and the 1 remaining company was in Tier 7. Table 6 reports the efficiency scores of companies in 2020 in the same format as Table 5. The result shows that 5 companies were in Tier 1, 4 companies were in Tier 2, 6 companies were in Tier 3, 8 companies were in Tier 4, 9 companies were in Tier 5, 5 companies were in Tier 6 and the 1 remaining company was in Tier 7. Table 7 summarizes the efficiency score for companies during 2019–2020.

A comparison of the results for 2019 and 2020 shows that the total number of efficiency tiers remained constant at seven, with no observable change. This indicates that the distribution structure of companies across different efficiency levels did not undergo significant shifts. The stability in the number of efficiency tiers also suggests that the efficiency gap among companies did not further widen. Although the COVID-19 pandemic may have affected individual companies to varying degrees, the overall efficiency structure of the shipping industry remained relatively stable, with no clear evidence of a systematic decline in efficiency. This finding implies that the shipping industry possesses a certain degree of resilience in the face of external shocks. Possible explanations include the continued support of global trade demand, as well as companies' ability to adjust operational strategies to cope with pandemic-related disruptions. In addition, rising freight rates and capacity adjustments during the pandemic may have helped some shipping companies maintain their operational efficiency, thereby mitigating fluctuations in the overall efficiency distribution.

A comparison between Tables 5 and 6 shows that member companies of THE Alliance, 2M and the Ocean Alliance (DMU #1, #3, #4, #10, #12, #13, #24, #26, and #37) exhibit a relatively concentrated distribution of efficiency tiers during 2019 and 2020, primarily falling between the second and fourth tiers, with limited variation overall. Further analysis indicates that, for most alliance members, changes in efficiency tiers are confined within one level, suggesting that these companies were able to maintain relatively stable operational efficiency during the pandemic period. This finding implies that, despite the impact of COVID-19 on the shipping industry, companies participating in strategic alliances demonstrated greater stability and resilience to external shocks. In contrast, non-alliance companies display a more dispersed distribution of efficiency levels. While their efficiency tiers were relatively concentrated in 2019, the distribution expanded in 2020 to include both higher and lower tiers, indicating a widening efficiency gap among companies. Overall, alliance companies are able to maintain more stable efficiency performance under external shocks, whereas non-alliance companies exhibit greater volatility. This difference may be attributed to the resource-sharing, information exchange and operational coordination mechanisms provided by strategic alliances, which enable member companies to adjust their operational strategies more effectively in response to environmental changes. These differences can be further observed through the response measures presented in Table 7.

The companies that achieved Tier 1 efficiency in both 2019 and 2020 (DMU #5, #17, #23, #28, and #35) are surprisingly all companies not in the main alliances. This finding suggests that, even in the absence of alliance-based resource support, certain companies are still able to maintain highly efficient operational performance. This study further investigates the characteristics and response strategies of these companies in 2020 to better understand how they sustained efficiency during the COVID-19 pandemic. A closer examination of their operational characteristics reveals that these companies are typically concentrated in specific market segments and exhibit a high degree of specialization and operational flexibility. For instance, several companies focus on the transportation of commodities such as oil, steel and coking coal (DMU #5, #17, #23, and #28). Compared to large shipping companies that rely heavily on global liner networks, these companies tend to face more stable demand conditions and are able to respond more rapidly to external changes through internal management adjustments, including workforce allocation, fleet utilization and transport planning.

In addition, some companies operate primarily within regional or niche markets, particularly in Asian trade routes (e.g. DMU #35), which makes them less vulnerable to international border restrictions and global supply chain disruptions during the pandemic. At the same time, the shipping industry is considered an essential sector, allowing these companies to maintain basic operations even under pandemic conditions, thereby supporting their efficiency performance. Overall, these findings suggest that, beyond the resource integration advantages provided by strategic alliances, company-specific characteristics – such as specialization, operational flexibility and market positioning – also play a crucial role in sustaining high efficiency under external shocks.

After generating the results using the non-separable SBM-DEA model with undesirable output and categorizing the companies into tiers using the context-dependent DEA method, this research practiced the one-way Mann–Whitney U test to examine the two hypotheses of this research. The four hypotheses to be examined include: Marine transport companies within the three main shipping strategic alliances exhibit higher overall efficiency than other companies; the overall efficiency of marine transport companies in 2019 is higher versus that of marine transport companies in 2020; marine transport companies within the three main shipping strategic alliances exhibit higher financial efficiency than other companies; and marine transport companies within the three main shipping strategic alliances exhibit higher environmental efficiency than other companies.

Table 8 reports the results of the one-way Mann–Whitney U test. The results for 2019 are statistically significant at the 10% level, while the results for the pooled sample (2019–2020) are significant at the 5% level. These findings indicate that, overall, alliance companies exhibit significantly higher efficiency than non-alliance companies. In contrast, the results for 2020 are not statistically significant, suggesting that the efficiency differences between alliance and non-alliance companies are not evident during the pandemic period.

Taken together, the empirical results provide partial support for Hypothesis 1. Prior to the COVID-19 pandemic (2019), alliance companies demonstrate a significant efficiency advantage, indicating that strategic alliances may enhance company performance through resource integration and operational coordination. However, during the pandemic (2020), the entire industry was affected by external shocks, leading to a convergence in efficiency performance and diminishing the statistical significance of differences between alliance and non-alliance companies.

Furthermore, the pooled results still show a significant difference, implying that alliance companies maintain an efficiency advantage on average over time. This finding suggests that while strategic alliances are effective in improving efficiency under normal operating conditions, their advantages may be partially offset during periods of large-scale external disruptions due to industry-wide fluctuations.

Table 9 presents the results of the one-way Mann–Whitney U test. The findings indicate that, for alliance companies, non-alliance companies and the overall industry, the efficiency differences between 2019 and 2020 are not statistically significant. This suggests that although the COVID-19 pandemic had a substantial impact on the shipping industry, it did not lead to a statistically significant decline in company efficiency. Therefore, Hypothesis 2 is not supported. Overall, the marine transport industry demonstrates a certain degree of resilience during the pandemic period, as company efficiency levels remain relatively stable. When these results are considered alongside those in Table 8, it can be observed that while no significant differences in overall efficiency exist across years, alliance companies still exhibit a significant efficiency advantage in specific periods (e.g. 2019). However, this advantage is not statistically significant during the pandemic period (2020), suggesting that external shocks may lead to a convergence in efficiency across different groups of companies.

This study conducts the Mann–Whitney U test based on financial efficiency derived from desirable outputs. The results indicate that, in both 2019 and 2020, alliance companies exhibit significantly higher financial efficiency than non-alliance companies. These empirical findings support Hypothesis 3. This result suggests that strategic alliances have a stable and significantly positive effect on financial performance, and that this advantage does not diminish during the pandemic. In other words, even under the market uncertainty and demand fluctuations caused by COVID-19, alliance companies are able to maintain superior financial efficiency. From a potential mechanism perspective, strategic alliances facilitate resource sharing and operational coordination, enabling member companies to improve vessel utilization, reduce empty capacity and distribute operating costs more effectively, thereby enhancing revenue efficiency. Furthermore, alliance members can jointly optimize route planning and capacity allocation, allowing them to respond more efficiently to changes in market demand and to mitigate the adverse effects of demand uncertainty on revenue during the pandemic period.

Table 11 presents the results of the Mann–Whitney U test based on environmental efficiency derived from undesirable outputs. The findings indicate that the difference in environmental efficiency between alliance and non-alliance companies is not statistically significant in 2019. However, during the pandemic year of 2020, alliance companies exhibit significantly higher environmental efficiency than non-alliance companies. In addition, the pooled results for the 2019–2020 period are statistically significant at the 10% level. Therefore, Hypothesis 4 is supported. These results suggest that, when undesirable outputs are taken into account, the advantages of strategic alliances become more pronounced during the pandemic. Compared with the findings from overall efficiency analysis, the environmental efficiency results further reveal that alliance companies not only maintain operational stability under external shocks but also demonstrate superior performance in reducing CO₂ emission intensity.

This phenomenon may be attributed to the information-sharing and coordinated decision-making mechanisms within strategic alliances, which enable member companies to optimize route planning and capacity allocation more effectively, for example, by reducing unnecessary sailings and improving vessel utilization, thereby lowering emissions per unit of output. Moreover, the implementation of the IMO sulfur cap regulation in 2020 may have further intensified the differences in environmental efficiency across companies. Under increasingly stringent environmental regulations, the resource integration and organizational flexibility provided by strategic alliances allow member companies to adjust their operational strategies more efficiently to comply with regulatory requirements. Overall, the results indicate that, when focusing on environmental efficiency based on undesirable outputs, strategic alliances have a positive impact on company performance, and this effect becomes more pronounced under the dual pressures of the COVID-19 pandemic and tightening environmental regulations.

During the COVID-19 outbreak, the alliances assisted shipping companies in implementing the Blanking Strategy, a strategy that cancels call of vessels to mitigate the impact of COVID-19 lockdowns. The alliance companies actively applied the blanking strategy, and with their operational coordination centers the companies can jointly adjust their transport capacity, seeking to minimize the impact of the decreasing demand caused by the pandemic. In the annual reports of alliance companies, the companies also credited the consistent performance to IT tools that improve the decision-making process and an internal risk management system that helps the companies react to any change and keep their operating activities smooth. However, a single company has limited ability to adjust according to the current situation. The alliances allow members to share slots and optimize resource deployment to quickly react to the fluctuated demand worldwide. Some research observed that the shipping alliances share common strategies on port calls, size of vessels and fleet deployments (Fedi et al., 2022; Notteboom et al., 2021), COVID-19 led to alliance companies making lower port calls and carrying higher cargo volume per call, which is a more economical way for shipping companies to operate under declining demand such as during the lockdown measures. Instead of sending ships to places under lockdown, the alliance companies were able to reduce unnecessary costs and deploy resources to other places to generate profit.

However, based on the empirical results of this study, although the resource integration and operational coordination mechanisms provided by strategic alliances enhance companies' adaptive capabilities, the efficiency differences between alliance and non-alliance companies are not statistically significant during the COVID-19 period. This finding suggests that, under large-scale external shocks, the entire industry faces common pressures, leading to a convergence in efficiency performance across companies. Such convergence may be attributed to heightened demand uncertainty and widespread operational constraints during the pandemic, which limit companies' ability to optimize resource allocation and operational adjustments, thereby weakening the relative advantages of strategic alliances. Overall, while strategic alliances may help mitigate the decline in efficiency, their advantages may not fully translate into statistically significant differences under extreme conditions.

When further examining disaggregated output efficiency from the perspective of financial performance, a different pattern emerges. As shown in Table 10, alliance companies exhibit significantly higher financial efficiency than non-alliance companies in both 2019 and 2020, with even stronger significance in the pooled sample. This result indicates that strategic alliances have a stable and persistent positive effect on financial performance, and that this advantage does not diminish during the pandemic.

Compared with the convergence observed in overall efficiency during the pandemic, the financial efficiency results reveal that alliance companies continue to maintain a clear advantage in economic output. From a mechanism perspective, strategic alliances facilitate resource sharing and operational coordination, enabling companies to improve vessel utilization, reduce empty capacity and lower operating costs. In addition, through joint route planning and capacity allocation, alliance members are able to respond more efficiently to demand fluctuations, thereby sustaining revenue performance and financial efficiency. Even under heightened uncertainty during the pandemic, alliance companies can mitigate operational risks through coordinated mechanisms, resulting in more stable financial performance compared to non-alliance companies.

Further analysis of environmental efficiency based on undesirable outputs reveals yet another distinct pattern. According to Table 11, alliance companies demonstrate significantly higher environmental efficiency than non-alliance companies in 2020, indicating that the advantages of strategic alliances become more pronounced under environmental and regulatory pressures. This phenomenon may be linked to the implementation of the IMO sulfur cap regulation in 2020. Under increasingly stringent environmental regulations, companies are required to adjust fuel usage and operational strategies to reduce emissions. Via information sharing and coordinated decision-making, alliance members can optimize route planning and capacity deployment more effectively, for example, by reducing unnecessary sailings and improving vessel utilization, thereby lowering emissions per unit of output.

The BCG score in this research considers the efficiency tier and the environmental efficiency. Table 12 presents the result of the Mann–Whitney U test comparing the BCG score. The result shows that the alliance companies have higher efficiency in 2020, while in 2019 there is no significant difference between the efficiencies of alliance companies and non-alliance companies. Compared with the results in Table 8, the findings in Table 12 suggest that companies place greater emphasis on environmental efficiency during the pandemic period. When environmental efficiency is taken into account, alliance companies outperform non-alliance companies. From another perspective, under increasingly stringent fuel regulations, companies participating in strategic alliances demonstrate superior efficiency performance compared to non-alliance companies. Under the dual pressures of the COVID-19 pandemic and tightening maritime fuel regulations, the importance of strategic alliances becomes more pronounced.

The reasons behind the alliance companies having higher BCG scores can be credited to the effect of strategic alliances. The knowledge-sharing activities and conferences that discuss environmental issues can help alliance companies implement technologies and strategies to monitor, control and reduce CO₂ emissions, and the resource-sharing agreements can allow companies to use optimized solutions under different scenarios to avoid unnecessary waste. This research's findings align with previous literature that suggested collaborations and strategic alliances can help reduce CO₂ emissions or other negative environmental effects (Qiu et al., 2018; Wong et al., 2018). Overall, the results of this study indicate that, in terms of overall efficiency, the advantage of alliance companies converges during the pandemic due to industry-wide shocks. However, when environmental efficiency and regulatory pressures are incorporated, the relative advantage of alliance companies re-emerges, as reflected in higher BCG scores.

This research seeks to understand how COVID-19, a shocking event that nearly stopped global trade but later provided opportunities to shipping companies, changed the efficiency of the marine transport industry, and how strategic alliances, as an approach often used by shipping companies to gain competitive advantages, affect these companies' efficiency. The empirical findings of this research show that the companies placed in Tier 1 in both 2020 and 2019 are either providing specialized services or focusing on the regional or niche markets. The result also means that most alliance companies were able to maintain their efficiency tier during the COVID-19 pandemic, while many non-alliance companies experienced greater fluctuations. These findings suggest that alliance companies demonstrate greater stability in efficiency under external shocks, while non-alliance companies exhibit higher performance volatility.

Further analysis reveals that while the resource integration and operational coordination mechanisms provided by strategic alliances enhance companies' adaptive capabilities – such as mitigating operational disruptions through blanking strategy, information system integration and capacity coordination – these advantages did not fully translate into statistically significant overall efficiency differences during the pandemic. However, when examining disaggregated output efficiency from the perspective of financial performance, a different pattern emerges. Strategic alliances demonstrate a stable and persistent advantage in financial efficiency, and this advantage does not diminish during the pandemic. This finding suggests that the economic benefits derived from resource sharing and operational coordination enable alliance companies to sustain profitability even in highly uncertain market environments.

When environmental efficiency, based on undesirable outputs, is further incorporated into the analysis, another distinct pattern is observed. The empirical results show that, during the pandemic year (2020), alliance companies exhibit significantly higher environmental efficiency than non-alliance companies, and similar findings are observed under the BCG composite indicator. This suggests that, when environmental performance is considered, the advantages of strategic alliances re-emerge during the pandemic.

This phenomenon may be associated with the implementation of the IMO sulfur cap regulation in 2020, as well as the growing emphasis on ESG principles. Under increasingly stringent environmental regulations, companies are required to adjust fuel usage and operational strategies to reduce emissions. In this context, strategic alliances facilitate information sharing and coordinated decision-making, enabling member companies to optimize route planning and capacity allocation more efficiently, for example, by reducing unnecessary sailings and improving vessel utilization, thereby lowering emission intensity.

Overall, the results of this study indicate that, in terms of overall efficiency, the advantages of alliance companies converge during the pandemic due to industry-wide shocks and are not fully realized. In contrast, in the dimension of financial efficiency, alliance companies maintain a stable and persistent advantage. Furthermore, when environmental efficiency and regulatory pressures are incorporated, the relative advantages of alliance companies re-emerge.

Based on these empirical findings, strategic alliances demonstrate significant advantages in both financial and environmental efficiency, suggesting that cooperative mechanisms can effectively enhance resource allocation efficiency and reduce emission intensity. Therefore, encouraging companies to participate in strategic alliances becomes an important issue for improving overall industry performance and promoting sustainable development.

From a policy perspective, several implications can be drawn. First, at the economic level, policymakers may provide tax incentives or subsidies to reduce the costs associated with alliance participation, thereby strengthening companies' incentives for resource sharing and operational coordination. Second, under increasingly stringent environmental regulations, strategic alliances may be incorporated into decarbonization policy tools, for example, through carbon trading mechanisms or emission reduction incentive schemes, to encourage companies to enhance environmental efficiency through collaboration. Finally, from an institutional perspective, regulatory authorities may establish cross-company information-sharing platforms and standardized operational frameworks to reduce institutional barriers to alliance participation and to facilitate more efficient route planning and capacity allocation. In summary, the integration of economic incentives, environmental policies and institutional support can effectively enhance companies' willingness to participate in strategic alliances, thereby promoting efficiency improvement and low-carbon transformation in the maritime transport industry.

This research serves as an analysis of the initial effects of COVID-19, as researchers a few years from now can provide a more comprehensive view of the effect it has brought to the world. The first limitation of this research includes the lack of annual and ESG reports of 2021 for most marine transport companies, and thus the sample may not cover the full effect of raising shipping demand post-COVID-19 outbreak. Second, not all marine transport companies disclose emission data or publish English versions of annual reports, leading to a restricted sample. Future studies can also apply the model settings of this research to study efficiency by considering other pollutants or side-effects such as NO_x, SO_x, sewage and solid waste, to name a few. The mechanism behind how strategic alliances help companies maintain efficiency rankings is also worth further investigation, underscoring that the mechanism may lead to more complete knowledge of the marine transport industry.