Factors of Digital Transformation in the Maritime Transport Sector

: This paper aims to present the model of factors which inﬂuence the digital transformation in maritime transport sector. The preliminary model is based on a literature review and interviews conducted to identify the relevant factors inﬂuencing the digital transformation of stakeholders operating in the maritime transport sector. In order to test the model, the survey was conducted on the sample of Croatian administrative (port authorities, ministry, harbormaster’s ofﬁces, etc.) and commercial stakeholders (freight forwarders, agents, terminal operators, etc.) operating in maritime transport sector. The collected data was analyzed using the partial least squares structural equation modeling (PLS-SEM) approach. The research has shown that organizational, technological, and environmental (TOE) factors affect the digitalization of the organizations in the maritime transport sector. As a result of digitalization, changes in business models are visible: organizations in maritime transport sector generate additional revenue from new sources, provide new services, and introduce new sales channels.


Introduction
An increasing number of practitioners and scholars are exploring the possibilities offered by digital technologies and digital transformation [1][2][3][4][5][6][7][8]. Digital transformation has increased consumer expectations and disrupted markets while at the same time putting pressure on traditional companies and traditional business models [9]. Digital transformation refers not only to the implementation of new technologies but also to shaping digital strategies and digital culture and creating a new business model [10][11][12][13].
The carriers, seaports, and shippers involved in maritime transport chains [14] have become increasingly dependent on information and communication technologies [15]. Digital transformation may positively affect the maritime transport chain in terms of optimized cargo handling, improved business processes, and minimized environmental impacts [16]. Furthermore, the digitalization of seaport business processes may enhance sealand supply chain performance [17,18]. Despite numerous benefits, digital transformation in the maritime transport sector lags other transport sectors [19].
The research problem can be observed as follows: A significant number of heterogeneous stakeholders operate in the maritime transport sector, often using incompatible information systems [20,21], the costs of establishing information interoperability are very high [3,22,23], there is a lack of awareness of the positive effects of digital technologies [24], laws and regulations often allow only paper data exchange [25], and cooperation among stakeholders operating in the maritime transport sector is at an insufficient level [23]. Despite these obstacles, there is an urgent need to move ahead and seize the opportunities of digital transformation in the maritime transport sector. Therefore, it is important to understand the current role and situation of digitalization, as well as factors that influence organizations in the maritime transport sector on their digital transformation journey.
In order to validate the preliminary research model as a next step, the authors interviewed six experts in the maritime transport sector. Regarding the interviews with experts, the authors considered different types of stakeholders (administrative and commercial) in order to make the results more relevant. To conduct the quantitative analysis, the authors defined the constructs. The constructs and respective measurement items (factors) are presented in Table 1. The abbreviation ST stands for success technological factors, SO for success organizational factors, and SE for success environmental factors. The abbreviation D stands for digitalization and BM for changes in a business model.

Constructs Items Sources
Digitalization D1: The organization cooperates with new partners with the aim of developing new digital solutions [55,101] D2: The organization has digitalized internal business processes [3,55] D3: The organization has digitalized external business processes [29,30,35,36,[110][111][112][113][114][115] Changes in business model BM1: The organization generates additional revenue from new sources as a result of the implementation of digital technologies [30,55,114,116] BM2: The organization has entered new markets as a result of digitalization and digital transformation [55,101] BM3: The organization provides new services as a result of digitalization and digital transformation [11,12,30,35,36,55,80,101,[110][111][112]114,115,[117][118][119][120] BM4: The organization has introduced new sales channels as a result of digitalization and digital transformation Interview BM5: The organization has introduced new ways of charging for services as a result of digitalization and digital transformation [30,55,114,116]

Methodology
The methodology combines qualitative and quantitative approaches [55,116,121]. Based on an extensive literature review [26], digital transformation influencing factors were identified and clustered using the TOE framework. These factors influence the level of digitalization in enterprises and are reflected in business models changes. To confirm the relevance of the influencing factors, interviews with six experts from different organizations were conducted, and a preliminary research model was designed. The authors interviewed the managers of the following organizations: the ministry of transport, shipping and logistic companies, a port authority and the enterprise that focuses on digitalization of stakeholders operating in maritime transport sector. After that, the authors designed the questionnaire and conducted a survey of 262 Croatian stakeholders to validate the research model. Figure 1 shows the research steps and outcomes.

Hypothesis 4 (H4). Digitalization has a positive impact on changes in a business model.
The preliminary research model of digital transformation in the maritime transport sector is shown in Figure 2. The authors designed a questionnaire and collected quantitative data on factors influencing the digital transformation of stakeholders operating in the maritime transport sector through an online survey. A five-point Likert type scale (1-totally disagree, 5totally agree) was used to measure the level of agreement with given statements on the questionnaire. For data analyses, the partial least squares structural equation modeling (PLS-SEM) method was used to test the model, using SmartPLS 3.3.9 (SmartPLS GmbH, Bönningstedt, Germany).
The authors contacted 262 Croatian stakeholders listed in relevant national databases, both commercial (freight forwarders, agents, terminal operators, etc.) and administrative (port authorities, harbour master offices, and the relevant ministries). Since seaports are an important link in the transport chain [122], and the connection of the seaport with the hinterland contributes to the competitiveness of the seaport and influences its development [123,124], the authors also considered rail and road carriers that are involved in the maritime transport chain. Experts who had leading positions within the organization and who had experience in the field of digital transformation responded to the surveys (one person from each organization). The complete set of data was collected in 2022 from 94 organizations in Croatia.

Descriptive Statistics
Out of the total of 262 invited enterprises, we received 122 responses. We took into account only fully completed surveys, 94 of them. Out of 94 respondents, 35.11% were administrative stakeholders, and 64.89% were commercial stakeholders. Table 2 shows the types and percentages of the different stakeholders. Among the commercial stakeholders, we received the largest number of responses from shipping companies, followed by freight forwarders and logistics operators. The category "other" includes: crewing (manning) agencies, the maritime training center for maritime education, and vessel management

Measurement Model Evaluation
In the first part of the PLS-SEM analysis, we tested the measurement model. We evaluated composite reliability and convergent validity and the reliability of measurement model indicators, and then we assessed discriminant validity and evaluated the composite model.

Evaluating Composite Reliability and Convergent Validity of Measurement Model
First, to measure the composite reliability of the measurement model, we calculated Cronbach's alpha, Dijkstra-Henseler rho_A, and composite reliability as shown in Table 3. All composite reliability values are measured in the interval from 0 to 1, where 1 means a complete reliability estimate. The measure of composite reliability (rho) is higher than Cronbach's alpha, which is recommended because rho estimates are usually closer to true reliability. Cronbach's alpha as rho has an acceptance limit of 0.7, and a match above 0.8 means good composite reliability. Table 3 shows that Cronbach's alphas for the latent variables are between 0.766 and 0.883, and rho is between 0.807 and 0.900. In recent years, Dijkstra-Henseler rho_A [121] has become increasingly popular. Dijkstra-Henseler rho_a above 0.707 is considered appropriate, meaning that the latent variable explains more than 50% of the variance in a construct [121]. Table 3 shows that all rho_A are above 0.707, indicating a reliable construct.
"Convergent validity is the extent to which the indicators belonging to one latent variable actually measure the same construct" and is estimated based on average variance extracted (AVE) [121]. As a composite reliability metric, AVE is between 0 and 1, where 1 represents a complete convergence estimate [125]. "An AVE larger than 0.5 has been suggested to provide empirical evidence for convergent validity, as the corresponding latent variable explains more than half of the variance in the belonging indicators" [121]. As shown in Table 3, the AVEs for all latent variables are between 0.646 and 0.809, which is appropriate and indicates convergent validity.

Evaluating the Reliability of Measurement Model Indicators
An assessment of the reliability of the indicator can be given based on factor loadings, where a factor loading above 0.707 is considered acceptable.
In the original measurement model, there were 38 indicators that we gradually eliminated. In each iteration, we eliminated one factor with the lowest factor loading. We continued this process until we came up with indicators with factor loadings above 0.707. It means that "more than 50% of the variance in a single indicator can be explained by the corresponding latent variable" [121].

Assessment of Discriminant Validity of a Measurement Model
"Discriminant validity entails that, two latent variables that are meant to represent two different theoretical concepts are statistically sufficiently different." To obtain empirical evidence for discriminant validity, the heterotrait-monotrait (HTMT) ratio should be considered [121]. The HTMT should be lower than 0.85 (stricter threshold) or 0.90 (more lenient threshold) or significantly smaller than 1". The HTMTs for the latent variables are shown in Table 4. No HTMT is higher than the strict criterion, 0.85, so HTMT indicates the discriminant validity of the measurement model. To provide an estimate of the composite model, we focus on estimating multicollinearity, weights, and composite loadings [121]. Multicollinearity occurs when two independent variables have a high correlation, which increases standard errors and test unreliability, and when there are difficulties in assessing the importance of variables depending on each other [125]. As a rule, problems with multicollinearity occur when the variance inflation factor (VIF) is above 4.0, or above 5.0 following the less stringent criterion [121,125].
Problems with multicollinearity can occur at both the measurement level and the structural model level [125], so SmartPLS separates VIF into inner VIFs for the measurement model and outer VIFs for the structural model. Table 5 shows the inner VIFs for the measurement model. All values are much lower than 4 (the maximum inner VIF is 1.482), which is why we do not expect problems related to multicollinearity. Table 6 shows the outer VIFs for the measurement model as well as the weights and loadings.
All VIFs are lower than 4 (the maximum outer VIF is 2.905), which is why we do not expect problems related to multicollinearity in the structural model. "While weights show the relative contribution of an indicator to its construct, composite loadings represent the correlation between the indicator and the corresponding emergent variable; a loading shows the absolute contribution of an indicator to its construct" [121]. If there is an indicator of a latent variable with a significantly smaller loading than other indicators, then it is necessary to assess whether it is appropriate to exclude that indicator. The validity of the indicator must also be taken into account. In the event of a change in the validity, we may choose to keep an indicator with a lower weight. As can be seen from Table 6, within the structural model, indicators of latent variables are similarly weighted/loaded.

Structural Model Evaluation
After completing evaluations of the measurement and composite models, according to which we consider the measured properties of the research model to be appropriate, we can proceed with the assessment of the structural model [121,125]. In the assessment of structural models, we focus on estimates of model fit as well as estimates of path coefficients, their importance, effect sizes (f 2 ), and coefficients of determination (R 2 ) [121].
First, it was necessary "to evaluate the overall fit of the estimated model through the bootstrap-based test of overall model fit and the Standardized Root Mean Square Residual (SRMR) as a measure of approximate fit to obtain empirical evidence for the proposed theory" [121]. For each iteration of the following steps, we used SRMR, the squared Euclidean distance (d ULS ), and geodesic distance (d G ) to verify that the model corresponds to a saturated structural model [121] (see Table 7). An SRMR below 0.08 (or in a more conservative version below 0.10) indicates acceptable model fit [121,126]. Based on the considered data, the SRMR coefficient is 0.08, which is still acceptable.
The values of d ULS and d G by themselves have no value for assessing the suitability of the model. The adjusted Bollen-Stine bootstrap should be implemented to estimate d ULS and d G , which in SmartPLS is marked as double, or perfect, bootstrapping. This procedure creates samples based on the distribution of confidence intervals for SRMR, d ULS, and d G . If d ULS and d G are within the 95% confidence interval, the model is considered appropriate [126]. The analysis of the research model showed that both values are within the 95% confidence interval.
The structural or internal model consists of latent variables and the relationships (arrows) between them. The weight written on the arrow that directly connects the two latent variables is the standardized regression coefficient. The statistical characteristics of individual paths are checked using a double, or complete, bootstrapping, and for a path to be statistically significant, it must have a p value lower than 0.05 [121]. The results of the bootstrapping analysis are shown in Table 8. All p values are lower than 0.05, which means that the relationships between the variables are statistically significant. Furthermore, as can be seen from Table 8, all path coefficients are positive. It can be concluded that the hypotheses have been confirmed. Table 9 shows the explained variance of a dependent construct (R 2 ) and the magnitude of an effect that is independent of sample size (f 2 ). When phenomena are already quite well understood, one expects a high R 2 . When the phenomena are not yet well understood, a lower R 2 is acceptable [121]. These results including the path coefficient and R 2 are presented in Figure 3. Given that, in our opinion, this is the first such study that evaluates the impacts of factors of digitalization on changes in business models, it is estimated that R 2 = 0.258 is an acceptable value. Table 9 shows f 2 , which indicates the practical relevance of an effect. The magnitude of the effect f 2 is independent of the magnitude of the sample value, and the effect size is considered weak for f 2 between 0.02 and 0.15, medium for f 2 between 0.15 and 0.35, and large for f 2 equal to or larger than 0.35 [121]. Table 9 shows that the structural model contains both medium and large effect sizes. Namely, in the structural model, the effect size of linking technological factors and digitalization is the weakest in the whole model (f 2 = 0.250). On the other hand, the effect size of linking organizational factors and technological factors is the strongest (f 2 = 0.675).

Discussion and Findings
Based on an assessment of the indicators, several organizational factors were assessed as the most reliable. One of the important organizational factors is the existence of awareness in the organization of how digital transformation can affect the business of the organization. If there is no awareness, the need for investment in employees, new digital technology, etc. will not be recognized, which will ultimately slow down or prevent digital transformation [24]. For this reason, organizational factors represent the base or first step towards digital transformation. Another factor is "the organization has enough human resources to introduce new digital technologies". In order to ensure a sufficient number of human resources, cooperation between the university and the private sector is needed (e.g., by investing in knowledge), which would facilitate the further development and implementation of digital technologies in maritime transport sector [26,127].
One of the factors is "The organization invests in employee knowledge in the context of digitalization and digital transformation". In this respect, changes in the structure as well as the culture of the organization lead employees to take on roles that have traditionally been outside their functions [26]. Therefore, employees in the maritime transport sector should be encouraged by managers to upgrade their knowledge through intern or extern workshops, seminars, etc., which consequently affects organizational agility [128]. However, it is equally important that employee education take place regularly. Therefore, one of the factors the importance of which was recognized by the respondents related to the continuity of training: "The organization conducts continuous training of employees in the field of digitalization and digital transformation (for example, the development of an internal academy with online training and training modules in individual departments)".
Regarding technological factors, three of them were assessed as the most reliable, two of which were "The organization regularly invests in modern technologies to develop its business and services" and "The existing technology in the organization allows the upgrade of modern digital technologies". In this respect, "necessary technical modifications depend on the state of existing technologies used in an organization and must be adapted according to the needs of the organization" [26]. The last technological factor is "The organization systematically manages the risks of the implementation of new digital technologies", which is related to risks regarding, e.g., the quality of project implementation by the contractor.
The external environment may influence the activities of the organization and its growth. In terms of environmental factors, three of them were assessed as the most reliable. One of them is "The business of the organization is tightly regulated or subject to special legal regulations", which is usually related to green transport technologies or technologies that are applied for efficient and safe operation [109]. Another environmental factor is "There is a compliance of the organization with standards (for example, ISO standards) and conventions". For example, "ISO/IEC 38500:2015 provides guiding principles for members of governing bodies of organizations on improved, and acceptable use of information technology within their organizations" [129]. The last environmental factor assessed as most reliable is "the organization conducts socially responsible business with the help of digitalization and digital transformation". For example, in maritime transport, the United Nations 2030 Agenda and Sustainable Development Goal 17 refer to significant regulatory development that triggered a diffusion of corporate social responsibility [130].
In addition to assessing the factors of digital transformation, the stakeholders who participated in our research were asked if their organizations had a formulated digital transformation strategy, and only 27% of organizations provided a positive response.
Furthermore, in 47% of organizations, general managers are responsible for leading the digital transformation. In other organizations, IT department managers, project managers, or digital transformation managers are responsible for digital transformation.
Stakeholders were also asked in which business area the digitalization has brought the most benefits, and the following areas were mentioned: sales, accounting, finance, cost management, procurement, reporting, the official procedures of arrivals and departures of ships, customs formalities, human resource management, and analytics. Furthermore, organizations use the following information technologies (in descending order, based on the number of responses): office programs, e.g., MS Office (93%); information systems for business support (69%); applications for communication with clients (48%); social networks, e.g., LinkedIn (48%); software solutions for business analytics (36%); online sales (14%); blockchain (1%); and other, e.g., geographic information systems (5%). The majority of organizations (60%) answered that they had increased productivity by introducing digitalization and digital transformation.
In addition, stakeholders pointed out the importance of digital transformation when it comes to sustainable business, especially the ecological aspect of sustainability. In this respect, some of them increasingly use green sources and have implemented various solutions in order to lower the harmful impacts of their business (such as fuel flow measuring systems to optimize fuel consumption). One of the stakeholders has also developed a research center that is focused exclusively on development of zero-carbon technologies and solutions.
Our research has shown that organizational, technological, and environmental factors affect the digitalization of organizations in the maritime transport sector. Digitalization includes cooperation between an organization and new partners to develop new digital solutions (including through participation in projects related to digitalization and/or digital transformation) and digitalized internal and external business processes. As a result of digitalization, organizations in the maritime transport sector generate additional revenue from new sources, provide new services, and have introduced new sales channels.
The contribution of this study is twofold. First, the results of the study enrich the body of knowledge in the field of digitalization and digital transformation in the maritime transport sector. The validated model of digital transformation offers other researchers an introduction to the investigated field and may provide a baseline towards future research designs. In this respect, this research offers a better understanding of the influencing factors (technological, organizational, and environmental) that affect the digitalization of organizations operating in the maritime transport sector and how these changes result in changes in business models (the way an organization operates and conducts business). Second, the model with identified influencing factors can help practitioners and decision makers in shaping their digital transformation and digitalization strategies.

Conclusions
In the maritime transport sector, stakeholders are at different stages regarding the digital transformation of their business. The motivation for this research stems from the lack of existing research focused on digital transformation in the maritime transport sector. The existing studies do not provide a comprehensive overview of digital transformation in the maritime transport or seaports. In this respect, this research presents a model of influencing factors on digital transformation in the maritime transport sector. For that purpose, the authors, as a first step, conducted a literature review and carried out interviews with six organizations. Based on that, the authors identified 11 technological factors, 13 organizational factors, and 6 environmental factors and defined digitalization through 3 items and changes in business models through 5 items. Furthermore, the authors collected quantitative data on factors influencing the digital transformation of stakeholders operating in the maritime transport sector through a survey methodology. In the first part of the PLS-SEM analysis, testing of the measurement model was performed. The authors evaluated composite reliability and convergent validity and the reliability of measurement model indicators and then assessed discriminant validity and evaluated the composite model. In the assessment of structural models, the authors focused on estimates of model fit, estimates of path coefficients, their importance, effect size (f 2 ), and coefficient of determination (R 2 ).
The research has several limitations, which may also serve as future research directions. First, the research findings were based only on a sample of 94 organizations in Croatia. The comparison of these findings with other countries (e.g., countries in which digital transformation leaders operate such as Holland at the Port of Rotterdam) could provide further insights regarding digital transformation in the maritime transport sector. Furthermore, the authors analyzed both commercial and administrative stakeholders. In this respect, further research could include only one group. In order to broaden the scope of the research, additional analysis of the impact of digitalization on the business model may be conducted, for example at the supply chain level. In this respect, virtualization of product supply may be included, which means selling items that are not even owned by the company through the digital integration of inventories (digital marketplaces). Furthermore, traceability and product safety can change the business model based on the securitization of food, etc. This study offers only partial insights into digital transformation, which is a complex and fast-evolving phenomenon.