DEPARTMENT OF TECHNOLOGY MANAGEMENT AND ECONOMICS DIVISION OF SUPPLY AND OPERATIONS MANAGEMENT CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2022 www.chalmers.se Report No. E2022:050 Horizontal Collaboration of Outbound Transport Between Retailers A Multiple Case Study Master’s thesis in Supply Chain Management Fredrik Andersson Hampus Nilsson REPORT NO. E 2022:050 Horizontal Collaboration of Outbound Transport Between Retailers A Multiple Case Study FREDRIK ANDERSSON HAMPUS NILSSON Department of Technology Management and Economics Division of Supply and Operations Management CHALMERS UNIVERSITY OF TECHNOLOGY Gothenburg, Sweden 2022 Horizontal Collaboration of Outbound Transport Between Retailers A Multiple Case Study FREDRIK ANDERSSON HAMPUS NILSSON © FREDRIK ANDERSSON, 2022. © HAMPUS NILSSON, 2022. Report no. E2022:050 Department of Technology Management and Economics Chalmers University of Technology SE-412 96 Göteborg Sweden Telephone + 46 (0)31-772 1000 Gothenburg, Sweden 2022 Horizontal Collaboration of Outbound Transport Between Retailers A Multiple Case Study FREDRIK ANDERSSON HAMPUS NILSSON Department of Technology Management and Economics Chalmers University of Technology SUMMARY Horizontal collaboration is defined as collaboration between actors on the same level of different supply chains. Literature regarding horizontal collaboration involving an independent third party is indecisive and therefore, needs further investigation. Furthermore, this thesis adds to the existing number of reasons why retailers want to collaborate horizontally and which obstacles must be considered. The purpose is therefore, to investigate how horizontal collaboration could be implemented for retailers. This is fulfilled through finding enablers and obstacles to horizontal collaboration between retailers and evaluating three different collaborative structures. An independent third party is further considered in these structures and lastly, the impact on sustainability is discussed. To fulfill the purpose a multiple case study has been performed. Seven retailers in different industries have been interviewed and each retailer was considered a separate case which are compared with each other. Furthermore, six experts and three representatives from an IT- service provider in the field of logistics have been interviewed to provide an external perspective on collaboration between retailers. The findings show enablers and obstacles within data sharing, transport requirements, service improvements, coalition formation, motives for collaboration, legal and information sharing technology. Three structures of horizontal collaboration between retailers were both found in literature and suggested by retailers. These are collaboration through a marketplace, collaboration in clusters or a partnership. The role of an independent third party should be to facilitate a platform in a marketplace, coordinate and lead the collaboration in clusters or initiate partnerships and have a neutral role. Horizontal collaborations will bring economies of scale which reduces the number of trucks on the roads. This will lead to cheaper transport solutions and a reduction of pollution, noise and congestion caused by transport vehicles. This thesis was limited to only outbound transportation and did not consider mathematical solutions to route planning of outbound transports. The thesis was also limited to European retailers. Furthermore, neither inbound transports nor return flows were considered. . Keywords: Horizontal Collaboration, Retailers, Independent Third Party, Outbound Transport. ACKNOWLEDGEMENT This Master Thesis was conducted during the spring of 2022 at the division of Supply and Operations Management. The thesis was written as the final examination to receive the master's degree in Supply Chain Management at Chalmers University of Technology. We would like to show our appreciation to our supervisors at the collaborating company for helping us conducting this thesis and providing us with valuable resources. We also want to thank our supervisor from Chalmers, Ala Arvidsson, for her support during the project. She has provided us with continuous feedback and valuable discussions regarding the topic of horizontal collaboration. Lastly, we would like to thank the interviewees participating in the master thesis and providing valuable information to the thesis. Gothenburg, May 2022 Fredrik Andersson Hampus Nilsson Contents 1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Purpose and Research Questions . . . . . . . . . . . . . . . . . . . . . . . . 3 1.3 Delimitations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.4 Thesis Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 2 Literature Review 7 2.1 Sustainable Supply Chain Management . . . . . . . . . . . . . . . . . . . . 7 2.2 Horizontal Collaboration in Transport . . . . . . . . . . . . . . . . . . . . 8 2.2.1 Intensity of Collaboration . . . . . . . . . . . . . . . . . . . . . . . 9 2.2.2 Consolidation of Goods . . . . . . . . . . . . . . . . . . . . . . . . . 10 2.2.3 Leadership . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.4 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 2.2.5 Actors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 2.3 Enablers for Horizontal Collaboration . . . . . . . . . . . . . . . . . . . . . 13 2.3.1 Operational . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 2.3.2 Strategical . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.3.3 Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.4 Obstacles for Horizontal Collaboration . . . . . . . . . . . . . . . . . . . . 15 2.4.1 Operational . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.4.2 Strategical . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.4.3 Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.5 Third Party Involvement in Horizontal Collaboration . . . . . . . . . . . . 19 2.6 Overview of Collaboration Between Retailers . . . . . . . . . . . . . . . . . 22 3 Method 24 3.1 Research Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2 Data Collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3.3 Data Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.4 Ethical Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.5 Research Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.5.1 Validity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.5.2 Reliability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 4 Empirical Findings 29 4.1 Retailer Perspective on Horizontal Collaboration . . . . . . . . . . . . . . . 29 4.1.1 Retailer A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 4.1.2 Retailer B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4.1.3 Retailer C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.1.4 Retailer D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 4.1.5 Retailer E . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.1.6 Retailer F . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 4.1.7 Retailer G . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 4.2 Comparison of Retailers . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.2.1 Operational . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 4.2.2 Strategical . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 4.2.3 Market . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 4.3 Experts from the Logistics Field . . . . . . . . . . . . . . . . . . . . . . . . 44 4.3.1 Data Sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.3.2 Transport Requirements . . . . . . . . . . . . . . . . . . . . . . . . 44 4.3.3 Coalition Formation . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.3.4 Motives for Collaboration . . . . . . . . . . . . . . . . . . . . . . . 45 4.4 Representatives from the IT-Service Provider . . . . . . . . . . . . . . . . . 45 4.4.1 Data Sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 4.4.2 Transport Requirements . . . . . . . . . . . . . . . . . . . . . . . . 46 4.4.3 Coalition Formation . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.4.4 Motives for Collaboration . . . . . . . . . . . . . . . . . . . . . . . 47 4.4.5 Legal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.4.6 Information Sharing Technology . . . . . . . . . . . . . . . . . . . . 48 5 Discussion 49 5.1 Enablers and Obstacles from Multiple Perspectives . . . . . . . . . . . . . 49 5.1.1 Data Sharing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 5.1.2 Transport Requirements . . . . . . . . . . . . . . . . . . . . . . . . 50 5.1.3 Service Improvements . . . . . . . . . . . . . . . . . . . . . . . . . 51 5.1.4 Coalition Formation . . . . . . . . . . . . . . . . . . . . . . . . . . 52 5.1.5 Motives for Collaboration . . . . . . . . . . . . . . . . . . . . . . . 53 5.1.6 Legal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.1.7 Information Sharing Technology . . . . . . . . . . . . . . . . . . . . 54 5.2 Dimensions for Horizontal Collaboration . . . . . . . . . . . . . . . . . . . 55 5.2.1 Intensity of Collaboration . . . . . . . . . . . . . . . . . . . . . . . 56 5.2.2 Consolidation of Goods . . . . . . . . . . . . . . . . . . . . . . . . . 57 5.2.3 Leadership . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 5.2.4 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 5.2.5 Actors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 5.3 Sustainability of Horizontal Collaboration . . . . . . . . . . . . . . . . . . 59 5.3.1 Economical . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 5.3.2 Environmental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 5.3.3 Social . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 6 Conclusions 63 6.1 Key Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 6.2 Limitations and Future Research . . . . . . . . . . . . . . . . . . . . . . . 65 References 67 A Appendix - Expert Interview Questions i B Appendix - Case Interview Questions ii C Appendix - IT-Service Provider Interview Questions iii 1 Introduction This thesis has been written in collaboration with an IT-service provider within logistics. The following chapter provides a background, purpose and research questions, delimitations and the structure of the thesis. Initially, the background elaborates upon the potential of horizontal collaboration and why it needs further investigation. The purpose and research questions describe what is expected to be accomplished with the thesis and why this needs to be investigated. Delimitations describe which areas have been excluded from the project. Lastly, the structure for the remainder of the thesis is presented. 1.1 Background Road transportation of goods within the European Union has increased with 6.9% dur- ing the period 2015-2019 in terms of tonne-kilometres (Eurostat, 2021). In 2020 the total number of transported goods decreased by 3.9% (ibid.). However, this decrease was mainly during the second quarter which was when Covid-19 struck and therefore the increasing trend is assumed to continue. The transportation industry is characterised by many small actors and a consequence of this is low vehicle utilisation and ineffective route planning (Karam et al., 2021). Combining these two trends results in an unnecessarily high num- ber of road transport vehicles and an increase in transportation costs. Furthermore, the increase of vehicles in road transportation cause congestion and an increase in carbon emis- sions. During the COP26 meeting in Glasgow, the United Nations declared that they will keep their goal from the Paris Agreement to not let the global temperature increase above 1,5 °C (United Nations, 2021). The emissions of road transportation represent about 21% of the total emissions in EU (European Commission, n.d.-b). Therefore, reducing road transportation emissions could be a contributing factor to achieving the environmental goal. One approach to reduce the number of underutilised vehicles is horizontal collaboration. The concept means that actors on the same level of different supply chains co-operates (Basso et al., 2019). Horizontal collaboration does not include cooperation upstream or downstream in the supply chain, but between actors with the same role in different supply chains. One large EU-project on this topic was Collaboration Concepts for Co-modality (Cruijssen, 2012). A common typology for horizontal collaboration was developed through a set of seven dimensions: intensity of collaboration, direction of consolidation, leadership, scope and intensity, scope, competition, assets and objectives, carrier and/or shipper and number of partners (ibid.). However, some of the dimensions have almost the same prop- erties and are therefore merged. The dimensions which will be used in this thesis are: intensity of collaboration, consolidation of goods, leadership, scope and actors. These will be further explained in section 2.2. Cruijssen et al. (2007c) states that horizontal collaboration in transport enables economies of scale, reduces congestion and emissions. One type of horizontal collaboration is joint route planning which entail that organisations share resources when transporting goods. 1 This has been found to reduce costs by up to 30% (Cruijssen et al., 2007a). Economies of scale lead to more goods being transported in each vehicle and thus price per unit is decreased. Due to higher vehicle utilisation, fewer vehicles will be needed. This will lead to less congestion and less emissions. Horizontal collaboration enhances a company’s pos- sibility to focus on core activities and also give logistic service providers the possibility to broaden their services (Cruijssen et al., 2007b). Bull Sletholt et al. (2020) describe that the Swedish government has given Trafikverket the assignment to investigate horizontal collaboration during an eleven-year period. The purpose of this investigation is to find if and how horizontal collaboration can contribute to achieving the environmental goals. As of today, it has been found that there are issues regarding laws not being compatible with horizontal collaboration (Bull Sletholt et al., 2020), this is further described in section 2.4.3. It has also been found that there is a lack of facilitators and that trust between organisations is important (ibid.). Trafikverket has conducted a pilot project investigating a possible collaboration between ICA and SSAB. ICA is the largest Swedish food retailer (DLF et al., 2021) and SSAB is a Swedish steel manufacturing firm. Arvidsson (2017) describes that ICA and SSAB planned to share train wagons for transportation. It allowed them to increase fill-rate and ICA was able to change transportation mode from road to rail which was beneficial for the environment. These environmental benefits were the main result of the collaboration since savings only were marginal (ibid.). However, it was described that the collaboration faced difficulties. ICA had requirements on lead times which affected the transportation schedule for both actors. The cooperation between ICA and SSAB also demanded a new technical solution regarding how to load the goods. Several solutions were tested but all of them had drawbacks. An example of a solution was to stack containers from ICA on top of the goods from SSAB. However, these containers needed to be specially designed due to loading height and therefore deviate from the ISO-standard (ibid.). Arvidsson (2017) further mentions one example where horizontal collaboration was imple- mented successfully. The collaboration involved Scania, LKAB and SSAB. Scanias goods had temperature requirements which were not needed for SSABs and LKABs goods. The temperature problems were solved through investments in new containers. SSAB and LKAB had to adapt their schedule to fit Scanias since they were the organisation with the highest requirements on lead times. The collaboration lead to increased control of the inbound flow and higher frequency of deliveries. It was further concluded by Arvidsson (2017) that actors need to share sensitive information for the collaboration to be successful and therefore, trust and transparency are necessary. It has been shown that actors worry about how the risks and gains are allocated in horizon- tal collaboration (Cruijssen et al., 2007b). For a collaboration to be viable, the organisa- tions must believe that these allocations are made fairly. Most frequently either an ad hoc solution or the Shapely value method is used for fair allocation (Guajardo & Rönnqvist, 2016). Ad hoc solutions are usually developed to satisfy a specific issue and therefore tend to perform well while measured towards this issue. Shapely value is a method that aims 2 to compensate for increased efforts. However, it does not always perform well in terms of fairness (ibid.). Furthermore, many allocation methods are not developed to solve practi- cal problems. Instead they are developed based on existing theoretical frameworks (ibid.). Hence some of the frameworks may only work in theory but not in practice. This is a complex problem and one of the barriers to participating in a horizontal collaboration. Another potential barrier to horizontal collaboration is competition laws. Organisations in competition must act independently and consequently be affected by competitive pressure from other actors (European Commission, n.d.-a). Collaboration between two direct com- petitors could violate this law. Furthermore, the fear of information disclosure has been found to be the biggest impediment to horizontal collaboration (Cruijssen, 2012). Even though trust and transparency have been proven necessary, it is also an obstacle. Basso et al. (2019) mention other barriers related to the design of the relationship, operations and planning, behaviour and the market. Arvidsson (2017) concluded that organisations often use other alternatives to reduce costs in their logistics system than being a part of a horizontal collaboration. Even though there are benefits to horizontal collaboration, ob- stacles still exist and need to be overcome. It was concluded by Karam et al. (2021) that an independent third party can be used to mitigate obstacles. An independent third party could manage sensitive information and allocate the benefits and risks from the collaboration fairly (ibid.). The importance of an independent third party that manage sensitive information has also been emphasised by Cruijssen (2012). The management of sensitive information from an independent third party allow organisations to be more transparent without commercial risks associated with information sharing (ibid.). However, an independent third party acting as a coordinator of the collaboration has been found both beneficial and unnecessary. Arvidsson (2017) concluded that mainly individuals within the organisations managed and pushed for the collaborations to be maintained. Therefore in this context, there was no need for a third party. Cruijssen et al. (2007b) mention a survey where it was found that an independent third party was not necessary for the coordination of horizontal collaborations. On the other hand, Leitner et al. (2011) suggest that in most cases horizontal collaboration needs an independent third part. Because of the inconsistency regarding third party involvement, further investigation is needed. 1.2 Purpose and Research Questions The purpose of this thesis is to investigate how horizontal collaboration could be imple- mented for retailers. To fulfill this purpose, the main enablers and obstacles must be found. Thereafter, a structure must be found to understand how horizontal collaboration should be implemented. The involvement of an independent third party is a topic without a clear answer and therefore, the role of a third party will be investigated. Lastly, the impact on sustainability of horizontal collaboration between retailers needs to be investigated to ensure long term usability. 3 The common definition of a retailer is a firm that sells goods to end consumers or the public. According to Ramanathan et al. (2014) retailers are pushing for improving the environmental impact of their supply chains, developing sustainability strategies and dis- cussing information sharing between organisations. Recently, retailers have begun priori- tising warehousing costs and more frequent deliveries (Ballot & Fontane, 2010). Hence, retailers are facing challenges that could be solved with horizontal collaboration and will therefore be investigated. Since retailers are selling goods to end consumers without pro- ducing their own goods, the outbound quantity is smaller or equal to the inbound quantity. Therefore, the outbound flow of goods should be more dispersed than the inbound flow. Empty running trucks have been found to increase further downstream in a supply chain, with the lowest fill-rate between distribution centres and physical stores (ibid.). The po- tential of collaboration between supply chains at this level should be higher than at other levels of the supply chain. A simplified version of both the current information and goods flow in supply chains is depicted in Figure 1. Figure 1: Information and physical flow in current supply chains for retailers. Solid line represent physical flow and dotted line represent information flow. To implement horizontal collaboration between retailers, it is important to understand the enablers and obstacles. Therefore, the first research question will be: RQ 1: Which are the main enablers and obstacles for horizontal collaboration between retailers? Thereafter, once it is possible to understand the enablers and obstacles, one must un- derstand how it could be implemented. There are multiple suggestions in the literature regarding how this could be performed. Karam et al. (2021) suggest collaboration through an electronic market platform, through an urban consolidation centre or entering a part- nership with another actor. Basso et al. (2019) on the other hand suggest that horizontal collaboration between more than two or three partners is hard to implement. To under- stand how retailers would prefer horizontal collaboration, the second research question will thus be: RQ 2: How could horizontal collaboration be structured for retailers? 4 Through the introduction of an independent third party, a reduction of carriers can be possible due to the consolidation of the physical flows. The potential flow of information and goods between retailers, customers and a third party in a collaboration can be seen in Figure 2. Figure 2: Information and physical flow in supply chains for retailers involving a third party. Solid line represent physical flow and dotted line represent information flow. Because of the inconsistencies in the literature regarding third party involvement, clari- fication is needed. Verstrepen et al. (2009) conclude that the motives for horizontal col- laborations often are case-specific which creates a heterogeneous range of collaboration initiatives. Karam et al. (2021) mention a pilot study where it was found that 70% of transport service providers were more interested to join a collaboration network if it was managed by an independent third party. Therefore, this could be a more consistent way to develop and manage collaborations. The definition of a third party in this thesis will be a neutral coordinator, which according to Karam et al. (2021) could be for example a research institution or IT-provider. The main reasons to involve a third party were to manage confidential information and for fair treatment (ibid.). Therefore, a third party seems to bring benefits to the collaboration but it is important to understand what its role should be. Hence, the third research question will be: RQ 3: How could a third party be positioned in horizontal collaboration between retailers? According to Bull Sletholt et al. (2020), improving transportation efficiency through hor- izontal collaboration could reduce both energy consumption and the cost of transporting goods. For the solution to be fully sustainable the framework of sustainable supply chain management should be considered (Carter & Rogers, 2008). Kale et al. (2007) explain how previous attempts to develop a market for collaboration were mainly marketed toward environmental sustainability and this was not successful. Furthermore, research regarding 5 horizontal collaboration and sustainability have historically had either an economical or environmental focus (Chen et al., 2017). Therefore, further consideration regarding the sustainability impact of horizontal collaboration between retailers is needed. Hence, the fourth research question will be: RQ 4: How will sustainability be impacted by horizontal collaboration between retailers? 1.3 Delimitations The thesis will not provide a general framework for implementation. The thesis will not investigate any mathematical solutions to route planning problems. Neither inbound logis- tics nor back-hauling will be investigated. Only the geographical location of Europe will be considered since the market can have different characteristics elsewhere. 1.4 Thesis Structure The remainder of this thesis will be structured into five further sections. Next, Section 2 will cover the literature which has created an understanding of horizontal collaboration. In Section 3, the method for this thesis will be presented. Thereafter, the data from interviews with experts, the IT-service provider and the retailers will be presented and compared in Section 4. All these will be further discussed in Section 5 and here the research questions will be answered. Lastly, conclusions will be drawn in Section 6 together with suggestions for future research. 1 Introduction 2 Literature review 3 Method 4 Empirical Findings 5 Discussion 6 Conclusions 6 2 Literature Review In this chapter, an overview of existing literature on the subject horizontal collaboration is presented. The chapter includes sustainable supply chain management, horizontal col- laboration in transport, enablers and obstacles for horizontal collaboration, third party involvement in horizontal collaboration and an overview of collaboration between retailers. 2.1 Sustainable Supply Chain Management Carter and Rogers (2008) describe that the triple bottom line consist of three dimensions: economic, environmental and social performance. For a supply chain to be considered sustainable, all three dimensions must be considered (Stefaniec et al., 2020). Economical performance of a supply chain refers to the efficiency in movement of people and goods in a system and how well it contributes to societal economic development. Social performance refers to how well a system reaches both individuals and societal needs in terms of health, safety and equality. Environmental performance of a supply chains refer to its ability to manage goods while minimising resource consumption and maximising land-use. Further- more, it should neither affect the ecological systems nor pollute the environment. Carter and Rogers (2008) developed a framework for sustainable supply chain management based on the triple bottom line, which can be seen in Figure 3. The aim is to increase the sustainability performance of an organisations and its supply chain. The framework includes the triple bottom line dimensions and supporting facets. Risk management, trans- parency, strategy and culture are the supporting facets in sustainable supply chain man- agement. Risk management consider risks in terms of likelihood and outcome based on the triple bottom line (ibid.). Carter and Rogers (2008) describe the importance of trans- parency and that it could be improved both vertically and horizontally in supply chains. By increasing transparency, stakeholder feedback could be used to improve operations. Furthermore, audits could be developed to increase transparency of a supplier and reduce transaction costs for many buyers to the supplier. Another supporting facet is related to strategy and culture where the triple bottom line should be considered when these are developed (ibid.). Combining these strategies is not always easy since it will impact the culture of an organisation. 7 Figure 3: Framework including dimensions and supporting facets of sustainable supply chain management developed by Carter and Rogers (2008). Historically, most of the research regarding horizontal collaboration and sustainability have been focusing on the economical or environmental effects (Chen et al., 2017). The social aspects of horizontal collaboration such as personal development and child labour have not been addressed to the same extent (ibid.). This is also mentioned by Sudusinghe and Seuring (2022), who further describe that stability and performance is the most discussed subject in relation to economical performance. Training and education are mostly discussed in literature for social performance and waste management and emission in relation to environmental performance (ibid.). However, Ahi and Searcy (2013) describe that different terms of sustainability in supply chains cover different parts of the triple bottom line. Green supply chain management is mainly focusing on environmental performance, and in some cases also economical performance (ibid.). Sustainable supply chain management on the other hand, includes all three dimensions of the triple bottom line. 2.2 Horizontal Collaboration in Transport Horizontal collaboration is defined as a collaboration between actors at the same level of different supply chains (Basso et al., 2019). An example of horizontal collaboration is when retailers collaborate in their distribution of goods. There are many different aspects to consider when discussing horizontal collaboration. Cruijssen (2012) have created a typology for horizontal collaboration consisting of seven dimensions. However, in this thesis only five dimensions will be considered and those are: intensity of the collaboration, consolidation of goods, leadership, scope and actors. 8 2.2.1 Intensity of Collaboration The first dimension of horizontal collaboration is the intensity. Lambert et al. (1999) have developed a framework for different types of supply chain relationships and their intensities. This framework is developed for vertical collaboration. However, it is argued by Cruijssen (2012) that it is directly applicable to horizontal collaboration. There are five types of intensity in the framework and the outskirts are arm’s-length relationship or joint venture (Lambert et al., 1999). The five different types of relationships can be found in Figure 4. Figure 4: Five types of relationships inspired by Lambert et al. (1999). An arm’s length relationship tends to occur when there is a rapid increase in demand on the market. An example of an arm’s length relationship is horizontal subcontracting, i.e., an actor provides extra capacity for another when there is a temporarily increase in demand (Spiegel, 1993). However, they could still be in competition with each other. The communication between the actors is spontaneous and there is commonly a lack of joint commitment (Cruijssen, 2012). In the type I relationship actors consider themselves as partners. A partnership is usually defined as a relationship between two actors where the result is of higher value than what would have been achieved individually (Lambert et al., 1999). Usually, a type I relationship has a short horizon and occurs between individual departments or functions in different organisations (ibid.). Type I relationships also include coordination of operational activities but to a limited extent. Examples of coordinated activities can be back-hauling or joint distribution (Cruijssen, 2012). In type II relationships, the extent of the collaboration increases. These relationships have a longer horizon and several departments in both organisations are involved (Lambert et al., 1999). Cruijssen (2012) mentions collaboration in multi-modal transport and cross-docking warehouses as examples of commitments in type II relationships. Type III relationships are closer to a joint venture and do not have an end date. In type III relationships organisations view each other as extensions to their own organisation (Lambert et al., 1999). Actions occurring in these relationships are joint investments and network integration (Cruijssen, 2012). A joint venture goes beyond a relationship between organisations and agreements of shared risks and benefits are common. The main difference between a type III relationship and a joint venture is the sharing of profits. This might not be fully developed in a type III relationship but is necessary to be considered in a joint venture. 9 2.2.2 Consolidation of Goods The second dimension of horizontal collaboration is consolidation of goods (Cruijssen, 2012). Through horizontal collaboration, orders from different actors can be consolidated into bigger shipments (Cruijssen et al., 2007a). According to Cruijssen (2012) three differ- ent variations of consolidation is possible. In the first approach, two full truckloads travel in opposite directions. It is possible to use the same vehicle to collect the second load after the first one has been dropped off and it reduces empty re-positioning. In Figure 5 the first type of consolidated shipments is depicted. Figure 5: Shared back-haul described in Cruijssen (2012). Another type of consolidation is a milk-run. This is used when less than truckload ship- ments are sent back and forth. Additional actors could be added to the network for re-routing of the shipments. This can reduce the travel distance with partially utilised vehicles and thus improve transport efficiency. A milk-run is depicted in Figure 6. Figure 6: Milk-run shipping described in Cruijssen (2012). The last approach to consolidate goods is through multi-modal shipping. When many small shipments have similar destinations they can be consolidated into larger shipments. Potentially, the larger shipment will be large enough to utilise another, more efficient mode of transportation, i.e. rail or sea. A multi-modal solution where retailers consolidate all their shipments at a warehouse is shown in Figure 7. 10 Figure 7: Multi-modal shipping described in Cruijssen (2012). The squares represent actors and the triangles represent consolidation centres. 2.2.3 Leadership The third dimension is leadership. According to Cruijssen (2012), three types of leaderships have been identified: convened collaboration, primus inter pares collaboration and inter pares collaboration. Convened collaboration has a neutral third party that facilitates the collaboration. The third party should make it easier for actors to collaborate. The author also mention that with this leadership it is hard for the involved actors to control the collaboration. In a primus inter pares collaboration, an actor has been elected to manage the collaboration. The author mentions that this type of leadership makes it hard for other actors to influence it. However, the collaboration is dependent on a successful leadership. The third leadership, inter pares collaboration, can be seen as a neutral and open leadership. All actors have the same mandate in the collaboration and can therefore affect the outcome. However, this leadership demands everyone to share information. It is also resource demanding to make the collaboration successful. 2.2.4 Scope The fourth dimension of horizontal collaboration is the scope of the relationship. Zinn and Parasuraman (1997) have developed a typology explaining the relationship in a strategic alliance. It is based on the two factors scope and intensity. Scope is measured as the range of services and the intensity is measured by the level of integration. Through the combination of these two factors, four different types of alliances are formed: integrated, focused, extensive and limited. These can be seen in Figure 8. 11 Figure 8: Strategic alliances described in Zinn and Parasuraman (1997). There are a couple of things to consider when deciding on how to design the scope and intensity. An integrated cooperation can reap the largest cost savings in the overall system. However, in these cases it can be hard to leave the cooperation since the organisations commonly are dependent on each other. Limited cooperation is on the other hand easier to implement but do not bring as large cost savings. It is hence important to know the desired outcome before entering the cooperation (Zinn & Parasuraman, 1997). 2.2.5 Actors Cruijssen (2012) describes three different types of set ups for actors in horizontal collabora- tion. In the first one only shippers collaborate. In the second approach, transport providers are collaborating to coordinate a better flow of goods. Finally, in the third approach both shippers and transport providers collaborates in a network. However, Cruijssen (2012) mentions that the third approach requires a lot of effort to be feasible, but has largest potential to improve efficiency. Furthermore, in this approach it could be required to have an independent third party involved to coordinate. Another dimension that needs to be considered in these collaborations are the numbers of actors involved. Cruijssen (2012) mentions that if the number of involved actors increase, the coordination cost also increases. The author further mentions that no optimal number of actors have been found, but an increase of actors will lead to more opportunities for collaboration. However, too many could also become an obstacle. 12 2.3 Enablers for Horizontal Collaboration There are several drivers for horizontal collaboration and these will be declared in this chapter. Operational, strategical and market drivers will be covered. Operational drivers are results of activities in an organisation. Strategical drivers derive from strategical de- cisions within an organisation and market drivers come from outside the organisations control. 2.3.1 Operational A summary of the operational enablers to horizontal collaboration and which actors that are affected by these can be found in Table 1. Table 1: Operational enablers to horizontal collaboration and the actors effected by them. Enabler Affected Actor Lower Transport cost Carriers, Shippers Lower Carbon Emissions Carriers, Shippers Improved Distribution Efficiency Carriers, Shippers A frequently mentioned driver for horizontal collaboration is lower transportation cost. Indirect costs connected to distribution can get reduced through consolidation of goods. Cruijssen et al. (2007c) mentions handling costs and purchasing costs as two examples of these indirect costs which are affected. In an early article on the subject, it was found that the potential savings from joint route planning could be approximately 30% (Cruijssen et al., 2007a). Frisk et al. (2010) found savings of 14.2% and Arvidsson (2017) found savings of 1.3%. However, these cases had different circumstances. Cruijssen et al. (2007a) con- clude that the largest savings could be found when many shippers with similar businesses collaborate. Further synergistic effects can be found when order sizes are small, customers dispersed and delivery windows narrow. In the case described by Frisk et al. (2010) eight different shippers collaborated to find the savings. However, it was stated that it is unusual for this many shippers to collaborate in transportation (ibid.). Another driver for horizontal collaboration is the reduction of carbon emissions. Cruijssen (2012) describe that more efficient modes, such as rail or sea transportation, often need a larger scale of operations. Therefore, horizontal collaboration creates the possibility for organisations to collectively create large enough shipments for a more efficient transport mode. In the pilot study between ICA and SSAB (Arvidsson, 2017), 860 000 kg carbon dioxide would be saved per year if the solution would rely solely on rail transport. The author describes that this corresponds to 590 people flying from Stockholm to Bangkok. In cases where a change of transport mode is not possible, horizontal collaboration can still reduce emissions. Frisk et al. (2010) found a 20% reduction of emissions through con- solidation of goods in trucks without changing the mode of transportation. 13 A problem in transport highlighted by Karam et al. (2021) is re-positioning of empty con- tainers. The flow of goods between two locations is often imbalanced. Therefore, utilising horizontal collaboration to create milk-runs or fill the same truck in both directions can improve the transport efficiency (Cruijssen, 2012). Consolidation of goods can also lead to shorter lead times for the customers (Karam et al., 2021). 2.3.2 Strategical A summary of the strategical enablers to horizontal collaboration and which actors that are affected by these can be found in Table 2. Table 2: Strategical enablers to horizontal collaboration and the actors which are effected by the enablers. Enabler Affected Actor Service Improvements Carriers, Shippers Focus on Core Activities Shippers A driver for actors involved in horizontal collaboration is the possibility to widen range of services. Karam et al. (2021) describe horizontal collaboration in a network structure. When a network structure is used, the range of services which can be offered to the cus- tomers increases. This was depicted in a survey by Lin et al. (2007) where it was found that shippers could consider to purchase services they need from an online market. This online market could be set up in a network structure as described by Karam et al. (2021). The two major services considered to be bought on the online market was online tracking of shipments and requests for quotes (Lin et al., 2007). Cruijssen et al. (2007b) and Cruijssen et al. (2007c) describe that entering an alliance with another party has the potential to increase the quality of services due to close collabora- tion. Cruijssen et al. (2007c) further mention that the shipper can specialise themselves to satisfy the specific customer needs. The collaboration between two carriers can further improve the transportation efficiencies since they can make use of each others strengths. When these actors collaborate, an overall faster and better transport service could be de- livered to the shippers (Cruijssen, 2012). Horizontal collaboration allows organisations to reduce costs in non-core activities and fo- cus more on their core activities (Cruijssen, 2012). According to Cruijssen et al. (2007c) it will allow organisations to reduce their purchasing costs of supporting items. Karam et al. (2021) further argues that organisations most likely will have an easier time collaborating with non-core activities since these are not what differentiate the businesses. 14 2.3.3 Market A summary of the market enablers to horizontal collaboration and which actors that are affected by these can be found in Table 3. Table 3: Market enablers to horizontal collaboration and the actors effected by them. Enabler Affected Actor Joint Development Projects Shippers Sharing of Logistical Knowledge Carriers, Shippers, Third Party Other identified drivers for horizontal collaboration are related to the market. Karam et al. (2021) and Cruijssen et al. (2007c) mention that horizontal collaboration has potential to improve R&D. Karam et al. (2021) describe that it enables access to new knowledge. This is achieved through connecting with other actors and learning from each other. Moreover, horizontal collaboration has potential to improve R&D since several actors can financially contribute and share risks. This sharing of risk can allow for organisations to make larger investments in projects they would otherwise not have access to. Collaboration could also result in the possibility for organisations to learn best practices (Cruijssen et al., 2007c). Both Cruijssen et al. (2007c) and Karam et al. (2021) describe that horizontal collaboration opens up possibilities to enter new markets. The collaboration could enable new insights for those involved in the collaboration which will open new possibilities. Furthermore, strategic alliances between organisations can give access to new geographical markets (Gulati et al., 1998). 2.4 Obstacles for Horizontal Collaboration Even though horizontal collaboration shows a lot of promise, there are a few examples of successful implementation (Arvidsson, 2017). Identified barriers for horizontal collab- oration in transport will be presented in this section, following the same structure and definitions as Section 2.3. 2.4.1 Operational A summary of the operational obstacles for horizontal collaboration and which actors that are affected by these can be found in Table 4. Table 4: Operational obstacles to horizontal collaboration and the actors effected by them. Obstacles Affected Actor Flow of Goods Carriers, Shippers Data Sharing Shippers, Third Party 15 Some obstacles found in literature are related to goods and data sharing. Karam et al. (2021) mention imbalanced flow of goods on a national level as an obstacle for horizontal collaboration. Since the volume of goods often is heavier in one direction, it could be hard to find a partner that can fit the collaboration and thus balance the flow (ibid.). Bull Sletholt et al. (2020) has also risen fill-rate as an obstacle for horizontal collaboration. The problem is to define fill-rate because of its multiple parameters. The last impediment found regarding to the flow of goods is how to optimise the system (Pan et al., 2019). It is hard to find an optimal solution for whole system and therefore establishing the best transportation routes can be problematic (ibid.). There are also obstacles related to the flow of information. When sharing data, it must be secured and trustful (Basso et al., 2019). Karam et al. (2021) describe that inaccurate information can result in decisions made on the wrong data. The consequence of this is inaccurate estimations of the outcomes and business decisions. It is further mentioned that a lack of information and communication technology can result in a poor flow of information. Another question related to the flow of information and the interest in data sharing is who owns the data (Bull Sletholt et al., 2020). The owners of transport related data are carriers and shippers and these need to be convinced that their data can provide value at a system level. Furthermore, the data must not be traceable to the actor who shares the data. 2.4.2 Strategical A summary of the strategical obstacles for horizontal collaboration and which actors that are affected by these can be found in Table 5. Table 5: Strategical obstacles to horizontal collaboration and the actors effected by them. Obstacles Affected Actor Coalition Formation Shippers, Third Party Cost Allocation Shipper, Third Party Practitioner Knowledge Third Party Negotiation Position Shippers, Third Party Trust Shipper, Third Party A barrier for horizontal collaboration is how the collaboration should be structured. The most common motives for a shipper to horizontally collaborate is cost reduction, service enhancement or better market positioning (Cruijssen et al., 2007b). Recently, resilience of the supply chain and environmental improvements have also become more common motives (Pan et al., 2019). According to Cruijssen et al. (2007b), finding a reliable coordinator who is trusted by all parties is a barrier to overcome. Further, it will be important to consider how many parties should be included in the collaboration. It was found by Frisk et al. (2010) that eight companies could collaborate horizontally. However, Basso et al. (2019) argues that it is rare for a collaboration to include more than two or three actors. It is also 16 important to acknowledge a leader for the collaboration before it is initiated (Basso et al., 2019). Closely related to the coalition formation barrier is cost allocation which can be solved through cooperative game theory (Guajardo & Rönnqvist, 2016). However, these methods can become fairly complex and therefore be hard to understand. This was found by Frisk et al. (2010) where the equal profit method seems to perform better than others due to its simplicity. Geographical location has proven to be an important factor for cost savings (ibid.). Therefore, when parties understand the importance of this factor, more advanced methods of allocation can become favoured (ibid.). Bull Sletholt et al. (2020) conclude that the most important thing is that the cost allocation problem gets managed but not how it is managed. This is further validated by Karam et al. (2021) who believe that it is more important that the cost allocation method is transparent and straightforward rather than sophisticated and theoretical. Furthermore, a set of methods should be applied instead of a single method since this will allow the participants to negotiate and decide which method would fit their collaboration the most (ibid.). The set-up costs for collaboration are perceived to remove large parts of the savings (Cruijssen, 2020). However, changes in the environmental regulations by the EU or development of a collaboration model by an independent third party could reduce the perceived set up costs for collaborations (ibid.). Horizontal collaboration could thus become more lucrative for participants. Furthermore, practitioner knowledge is an obstacle described by Basso et al. (2019). It is important to have people with knowledge of how to design the collaboration (ibid.). The author recommend to outsource the designing of the collaboration to an expert with experience within the area, i.e. an academic expert. Another option can be to acquire the knowledge from a fourth party logistics provider, i.e. a management consultant. These can be hired to design and run an advanced supply chain solution (Cruijssen, 2012). However, the fourth party arrangement does not suit all collaborations and should be kept at an arm’s length relationship. Otherwise a collaboration expert who is independent from the parties would be more suitable (ibid.). According to Basso et al. (2019) it is important to consider the collaborating partners negotiation power before the collaboration and dynamical over time. The reason is that the party with better negotiation position will most likely benefit more from the collaboration than the party with a worse position for negotiation (ibid.). In the case by Frisk et al. (2010), rules for the collaboration were negotiated before the collaboration was initiated. It was concluded that this was one of the reasons for the successful collaboration. Karam et al. (2021) suggest a similar solution where the benefit sharing mechanism should be negotiated before the collaboration is initiated. However, opposed to Frisk et al. (2010), a set of mechanisms should be negotiated instead of just one. With this method smaller actors can easier use the method which is the most fair for everyone. Cruijssen (2012) emphasises that it is important that the negotiation end in a win-win situation. Otherwise, the relationship will be unsustainable in the long term. 17 Trust between actors is also an obstacle for horizontal collaboration. When there is distrust in a horizontal collaboration, transaction costs and other costs related to managing the collaboration will increase significantly (Ik-Whan & Taewon, 2006). In the worst scenarios it can result in an opportunistic behaviour (Cruijssen et al., 2007b). According to Basso et al. (2019) trust needs to exist toward the system, partners and data. If the involved actors believe that the relationship is beneficial, there will be trust of the system. It is mentioned by Karam et al. (2021) that there can be distrust towards the coordinator or other actors in a network. This type of distrust can make partners hesitant to share information with the coordinator (ibid.). Trust between partners refer to the belief that other parties will act for collective benefits. Market position, objectives, structure and similarities of flows all influence the trust between partners (Pan et al., 2019). Distrust in this regard has been found the toughest to overcome since it often led to issues with information sharing (Karam et al., 2021). Lastly, trust towards data refer to the belief that shared data by other parties are true to the real value (Basso et al., 2019). It has been found that organisations with a low level of digitisation often have more inaccurate data than organisations with a higher level of digitisation (Karam et al., 2021). Furthermore, data can be measured differently in different organisations (Bull Sletholt et al., 2020). Therefore, if two organisations want to collaborate but have different definitions on for example fill-rate, they can begin to distrust the data since it is collected differently. 2.4.3 Market A summary of the market obstacles for horizontal collaboration and which actors that are affected by these can be found in Table 6. Table 6: Market obstacles to horizontal collaboration and the actors effected by them.. Obstacles Affected Actor Legal Barriers Shippers Information Sharing Technology Carriers, Shippers, Third Party Legal barriers are another obstacle for horizontal collaboration. Basso et al. (2019) de- scribes collaboration rules and how they vary between countries. For example, competi- tion laws regulate the collaboration between competitors to avoid negative effects for con- sumers. GDPR is another legal barrier regulating how data can be collected and shared (Bull Sletholt et al., 2020). Collusion is a further risk in collaborating horizontally (Basso et al., 2019). It is important that the collaboration does not create an unfair advantage for any parties. Laws regulating competition mainly affect bigger companies and does not influence smaller companies to the same extent (Karam et al., 2021). Small and medium sized companies do not affect the possibility to compete to the same extent and is there- fore often not affected and restricted by the laws as bigger companies. However, it would still be illegal for them to coordinate price or capacity (ibid.). Furthermore, horizontal collaboration is also affected by local laws. Two examples in Sweden are the Public Access 18 to Information Act and Security Act (Bull Sletholt et al., 2020). Obstacles related to information sharing technology are also identified. Basso et al. (2019) describe that sensitive information need to be shared for a collaboration to be feasible. However, companies are not always open to share sensitive information. It was found by Cruijssen (2012) that the biggest impediment to horizontal collaboration is the fear of sensitive information being leaked. This hesitancy towards information sharing creates a problem since sharing information is important to create an efficient transport system (Pan et al., 2019). Lastly, Karam et al. (2021) raise IT-systems for information sharing as an obstacle for horizontal collaboration. Furthermore, insufficient IT-systems can result in incomplete or inaccurate data which create inefficiencies in the systems. 2.5 Third Party Involvement in Horizontal Collaboration Cruijssen (2012) refers an independent third party as being neutral, transparent and trusted. The author concludes that an independent third party must be present for the collaboration to be successful. A further benefit of an independent third party is the avoid- ance of competition laws. Otherwise, it could make the collaboration not work as efficient as possible and lead to unfair work distribution in the long run. Cruijssen (2012) mentions that an independent third party should act as a safeguard of confidential data. It should also be responsible of legal foundations and handle contracts. Furthermore, it could be beneficial to involve an independent third party in the start up of a collaboration since this could ease the process of initiation of collaboration (ibid.). Rossi (2012) mentions uneven power positions as a reason for why collaborations fall apart. The author also states that the involvement of an independent third party could mitigate this since it avoids any party getting an unfair advantage. However, Cruijssen (2012) mentions a wide uncertainty in literature regarding the involvement of an independent third party in horizontal collabo- rations. Three different types of horizontal collaborations are presented by Karam et al. (2021). The first type is an electronic marketplace platform where companies can build short term collaborations for transportation. These collaboration are often without any formal doc- umentation and connect shippers, carriers and logistic service providers. The third party facilitates the platform in this case. The information flow and physical flow in a market- place structure can be seen in Figure 9. 19 Figure 9: Dotted line represents the information flow and the solid line represents the physical flow in a marketplace. The second type is close collaboration between few actors. These are characterised with long strategical alliances where the third party might not even exist. Collaborating part- ners have formal contracts and agreements to ensure that they are on the same page. The information and physical flow in such a structure can be found in Figure 2. Lastly, urban consolidation centres are mentioned as a type of horizontal collaboration. Organisations ship their goods to a consolidation centre outside of a city which they share with multi- ple other organisations. The goods are then reloaded to smaller last-mile vehicles. The structure can be seen in Figure 10. Figure 10: Representing the physical flow in a cluster. 20 A Swedish example of a successful horizontal collaboration is Skaraborg Eco shuttle (Arvids- son, 2017). It is a collaboration between a Swedish retailer and a carrier. The retailer has contracted the carrier to operate a train for incoming goods from a harbour to a dry-port. Other shippers with an interest in transporting goods along this distance can join the col- laboration and consolidate their shipments to improve fill-rate. Introducing a new partner to the collaboration is challenging and takes about 2-3 years until they are fully operating. The main obstacles are related to trust since the new partners need to be fully trusted. The result for organisations who join is reductions of up to 10% in costs and 80% of pollution. Vanovermeire and Sörensen (2014) mention several examples of third party involvement. An example is between Kimberly-Clark and Unilever-HPC in the Netherlands. Both par- ties are consumer goods producers with shipments mostly to the same customers. The aim of the collaboration was to increase the delivery frequency and thus being able to send less goods separately, but still having a high fill-rate in the truck. They collaborated through sharing a warehouse managed by an independent third party logistics provider. A report was published regarding cost allocation and how the Shapely value method was used in the Unilever-HPC case (Vanovermeire et al., 2012). Through creating an incentive for flexibility with Shapely value, a better result was found in the collaboration since more opportunities of consolidation arose. Another example described by Vanovermeire and Sörensen (2014) is between the plastic manufacturer JSP and the metal forger Hamerwerk who both sent trucks from Germany to Czech Republic. In this case, they involved a third party logistics provider for goods transportation and an independent third party to guide the collaboration. The collaboration led to a higher delivery frequency, environmental ben- efits, and reduction of inventory-in-transit costs. Vanovermeire and Sörensen (2014) describe how the University of Antwerp has created an organisation called TriVizor with the main goal of being an independent third party in horizontal collaborations. TriVizor is tasked with coordinating and contracting shippers and transport providers when collaboration is demanded. However, it is mentioned that the process is time consuming and inconvenient. It was stated that they in the future want to develop a cost allocation method which encourage behaviour that improves the collaboration. Another example of a third party is Seven Senders. Their goal is to connect retailers with parcel carriers and thus create a network where retailers can send shipments with a low cost and short delivery times. They act as a control tower for parcel delivery in Europe where they can create a more optimised network for shipments than what would otherwise be possible (Wurst & Graf, 2021). Furthermore, the European Commission have provided projects for horizontal collaboration (Cruijssen, 2020). One of these projects was named NEXTRUST with the goal to develop a logistical network for both horizontal and vertical collaboration. The project aimed to integrate shippers, logistic service providers and intermodal operators as equal parts in the network to both build higher volumes in each shipment but also increase the quantity shipped by rail or sea. 21 2.6 Overview of Collaboration Between Retailers Ramanathan et al. (2014) have researched the possibilities for collaborations among supply chain partners with the goal of achieving environmental sustainability and improve busi- ness performance. Two approaches were found for reduction of negative environmental impact. A company can either force their suppliers to reduce emissions or try to reduce its own emissions through improved logistics, production or raw material. A retailer can change their order sizes to achieve an overall improved environmental effect. Increasing the batch size will reduce the energy consumed in transport but will increase inventory carrying costs (ibid.). Furthermore, the main reason for the increasing demand of green supply chains have come from pressure by different stakeholders (ibid.). It was concluded by Ramanathan et al. (2014) that both intra- and inter-organisational collaboration is nec- essary to achieve green supply chain management. In an interview with a large food retailer in the UK it was found that environmental issues are driving consumer choices (Hingley et al., 2011). Therefore, solving these issues could be used for marketing purposes (ibid.). Furthermore, Frankel et al. (2002) mention that another enabler for entering a collaboration for retailers in the grocery industry is to drive innovation. Innovation was interesting since the collaboration could lead to better results than working separately. However, the same retailer which was mentioned in Hingley et al. (2011) was hesitant to the suggestion that collaborative approaches, for example vertical and horizontal collaboration, could reduce the cost pressures in the current economic envi- ronment. Furthermore, the retailer was confident that their management of suppliers was superior to their competitors. If this is the general case for retailers, it will be hard to collaborate since no one would want to compromise nor have someone else manage their supply chain. Lastly, smaller actors with less power in their respective supply chain might be more willing to collaborate (Hingley et al., 2011). In some cases it can be beneficial for retailers to collaborate in their physical distribution (Stephens & Wright, 2002). The main reason is that retailers usually do not consider physical distribution as their core activity. Thus it can be more efficient from a system point of view to outsource the logistical function (ibid.). An interesting aspect brought up by Stephens and Wright (2002) is the diminishing returns from supply chain innovation caused by the rapid spread of best practises within the area. Problems with the relation between retailers and suppliers including trust and power have been documented in vertical collaboration (Aastrup et al., 2007). Business relationships of this type can bring competitive advantages but require a positive notion of trade-off between benefits and sacrifices. The operational benefits in terms of know-how and time to market are the most important. The sacrifices to consider is giving up full control and negotiation power (ibid.). Furthermore, retailers seem to collaborate better with suppliers who are only working with them and no other retailers (Hingley et al., 2006). Improving trust, collaboration and co-operations have been found to improve the relationship between retailers and suppliers (ibid.). 22 Ballot and Fontane (2010) conclude that pooling of supply networks for retailers lead to different benefits depending on the size of the retailers. Larger retailers usually have higher fill-rate of their trucks than smaller retailers. Therefore, larger retailers do not find as large benefits from supply network pooling as smaller retailers. Further, it was found that economical benefits and ecological benefits can be acquired at the same time. In order to increase the possibility for horizontal collaboration without the involvement of a third party, logistical clusters can be utilised (Sheffi et al., 2019). A logistical cluster is defined as a geographical location where both shippers and carriers are located (ibid.). This set up eases the process of finding possible partners for horizontal collaboration due to a physical proximity, which enables consolidation without a long detour from the planned route. However, the physical proximity is not enough for the shippers to collaborate, a joint value proposition also needs to be developed. Trust and social interactions in these scenarios were also found to be a important for successful collaboration (ibid.). More formal interactions and contracts should not be absent since these are important to secure fair resource sharing and avoid opportunistic behaviour. 23 3 Method The following chapter describes the method that was used through the thesis. It contains the research approach, data collection, data analysis, ethical considerations and research quality. 3.1 Research Approach A multiple case study was conducted in order to find answers to the research questions. Yin (2009) describes that a case study is most advantageous when the research questions con- tain “how”. The author also mention that the researchers should not influence the events and have a real-life context that could bring value to the thesis. All these requirements were fulfilled and therefore a case study approach was suitable. Since the research questions aimed towards inter-organisational horizontal collaboration, the study needed to involve more than one retailer to provide a reliable answer. Therefore, a multiple case study was more appropriate to use than a single case study. Bryman and Bell (2011) mention that a multiple case study can be used when the researchers aim to find similarities or differences between cases. Yin (2009) recommends that a multiple case study should be designed to include six to ten cases. Two to three of them should find literal replications while the others should identify theoretical replications. However, due to limitations in both time and resources, only two literal replications were considered. These were fashion retailers with a global customer base and who both managed an e-commerce solution. Further, six different industries were examined to find theoretical replications. Every case involves a retailer with business towards end consumers. In Figure 11 the used method can be seen. Before the purpose and research questions were formulated, the topic of horizontal collaboration was explored through a literature review and interviews with experts. When the purpose and research questions had been formu- lated, the data collection phase began. A qualitative approach to collect data was used and hence, answers during interviews, context and opinions were preferred over statistics (Bryman & Bell, 2011). Empirical data was coded and compared in a cross-case anal- ysis. In parallel to the cross-case analysis, interviews with an IT-service provider were performed. The findings from the cross-case analysis were compared with the data from the expert interviews, the literature review and the IT-service provider interviews. Finally, conclusions were drawn based on the comparison between the data collected. 24 Figure 11: Qualitative research approach. 3.2 Data Collection A qualitative research approach is appropriate to use when researchers are interested in the interviewees point of view (Bryman & Bell, 2011). A semi-structured approach provides structure to the empirical data which will help the researchers perform a multiple case study. Structured data will be easier to compare than unstructured data. Therefore, interviews in this thesis have been performed with a semi-structured approach. Interviews were held with experts from the field to develop a thorough understanding of what has been done previously within the area of horizontal collaboration in Sweden. These interviews were not recorded but notes were taken. Furthermore, these interviews were semi-structured but did not follow the same structure as the case interviews. The expert interviews followed the structure which is provided in Appendix A and these questions were sent to the experts before the interview. An overview of the experts who were interviewed can be seen in Table 7. All interviews were performed online and in Swedish. Table 7: Overview of interviews made with experts from the academia. Expert Organisation Role Time I University Docent 45 min II & III Non-profitable Research Institute I Head of investigation 60 min IV Non-profitable Research Institute II Senior researcher 30 min V & VI Authority Programme director & Project manager 60 min Convenience sampling was used to find participating retailers since the opportunity to speak with many retailers in different industries presented itself. The interviews were semi-structured and the questions can be seen in Appendix B. Each interviewee was asked for permission to be recorded. If an interview was not recorded, it was either due to a lack of consent or technical difficulties. The interview was chosen to be recorded to 25 capture both the answers but also how they were framed (Bryman & Bell, 2011). Fur- thermore, recording an interview allow the interviewers to concentrate on the interview instead of taking notes. However, Bryman and Bell (2011) mention that interviewees often become more cautious of their wording and Jacobsson and Åhlström (2007) describe how answers can become more politically correct. Although these are valid counterarguments, the benefit of recording outweighs the drawbacks. An overview of the interviews can be found in Table 8. Individuals with key positions within logistics such as head of logistics or distribution manager at the participating organisations have been interviewed. A risk while interviewing retailers is a lack of knowledge regarding horizontal collaboration. The interviews therefore started with a quick overview of the subject to make sure that the interviewees had a sufficient understanding. Table 8: Overview of interviews from representatives of retailers. Code Company Time Language Online or Offline Recorded Representative I A 30 min Swedish Online No Representative II A 30 min Swedish Online Yes Representative III B 30 min Swedish Offline Yes Representative IV C 40 min English Online No Representative V D 30 min Swedish Offline Yes Representative VI E 40 min Swedish Online Yes Representative VII F 30 min Swedish Online Yes Representative VIII G 30 min Swedish Online Yes The interviews at the IT-service provider were sampled through convenience sampling and the interviews were semi-structured. All three representatives at the IT-service provider were field experts in the logistics industry. The questions were not sent to the interviewee before the interview to get more spontaneous answers. The questions which were asked can be found in Appendix C. Furthermore, the interviews were recorded if permission was given by the interviewee. An overview of the interviews can be found in Table 9. Table 9: Overview of interviews from representatives of IT-service provider. Code Time Language Online or Offline Recorded Representative IX 50 min Swedish Offline Yes Representative X 50 min Swedish Offline Yes Representative XI 40 min Swedish Offline Yes 26 3.3 Data Analysis The process to analyse the collected data was inspired by the steps of qualitative analysis provided by Williamson (2002). The pursued steps were: 1. Transcript interviews 2. Categorise data 3. Brainstorm ideas 4. Organise ideas 5. Validate theories with literature The data collected from the interviews was transcribed with the help of the transcript-tool in Word Online. This approach was chosen since there is a high risk of human errors while transcribing an interview (Bryman & Bell, 2011). These errors are mainly connected to the ability of concentration while transcribing a whole interview. It was mentioned by Bryman and Bell (2011) that a one hour interview could result in five to six hours of transcription. However, a software tool is not always fully correct. Therefore the transcribed file was checked by the researchers to ensure data quality. When the data transcription was completed, the data was coded and categorised. The information was sorted for patterns to be found in the data. While brainstorming, the categories of information were viewed from different perspectives. The ideas which sprung from the brainstorming were organised following the structure of: operational, strategical and market. The ideas and theories were compared with the literature review to validate their feasibility. The empirical data from both the experts and representatives from the IT-service provider was used to get an external perspective to horizontal collaboration between retailers. An important aspect to consider while conducting the research is the level of analysis (Bryman & Bell, 2011). Bryman and Bell (2011) mention that the level of analysis could be conducted on an individual, group, organisational or societal level. This is further emphasised by Yin (2009) who describes that a key reason to define the level of analysis is to know where the boundaries should be drawn. Through this thesis, the level of analysis was organisational. To find differences between industries and a broader perspective, key individuals from companies in different industries has been chosen. 3.4 Ethical Considerations Bryman and Bell (2011) mention harm to participants, a lack of informed consent, invasion of privacy and deception as the most important principles to consider when conducting business research. During the data collection, all these were considered. Questions regard- ing sensitive topics could cause stress to the interviewee. To avoid this potential harm, the 27 interviewee was informed that they could choose to not answer questions if it made them uncomfortable. The interviews were not recorded if the interviewee did not consent. Fur- thermore, for the commercial safety of the interviewees and participating organisations, they were kept anonymous through the thesis. To avoid any invasion of privacy, only questions regarding the concept of horizontal collaboration and the organisations opinion towards it was asked. Lastly, to avoid any deception, the interviewee was given a clear description of the concept of horizontal collaboration and how their participation would contribute to the research. 3.5 Research Quality The validity and reliability of the collected data will be discussed in this section. 3.5.1 Validity Validity can be addressed from three angles: construct, internal and external validity (Yin, 2009). Construct validity addresses the validity of the method used to collect the data. It is important that the studied concepts are clearly defined and that operational measures are connected to these concepts. The concept of horizontal collaboration has been defined and discussed between multiple stakeholders to the thesis. Furthermore, the data was collected from multiple cases which further enhance the construct validity. Internal validity address how the data was analysed. However, Yin (2009) mention that internal validity is often a concern when conducting an explanatory study. Since this is more of an exploratory study with the aim of creating a foundation for future horizontal collaboration, internal validity was not considered. External validity address whether the findings of the study can be generalised or not. To find interviewees, convenience sampling was used which reduces external validity of the results since it is hard to know if the sample represent the full population. However, to mitigate this uncertainty multiple cases and sources of data was used. 3.5.2 Reliability Reliability is discussed to minimise potential errors or biases throughout the study. For the data collection to be standardised, a set of questions were developed before the interviews and these were asked during all of them. Furthermore, the interviews were recorded when permitted and transcribed afterwards. Content created during the thesis was stored for the possibility to go back and revisit any information previously gathered. A common critique toward case studies is that the social setting changes over time and they can therefore be hard to replicate (Bryman & Bell, 2011). The interviews were performed during a short time frame and therefore randomness related to the time frame could be captured. 28 4 Empirical Findings In this section empirical data from the interviews with the retailers will be presented. Thereafter, answers from the retailer interviews are categorised and compared and lastly, the experts and the IT-service provider’s point of view on the categories will be presented. 4.1 Retailer Perspective on Horizontal Collaboration In this section, the retailers perspective on horizontal collaboration will be presented. Table 10 lists the retailers with their respective industry, whether they are local or global and if they operate through brick and mortar or e-commerce. A local retailer operates only in Sweden while a global retailer operates in multiple countries. If many organisations are part of a group who share income statement, then they are referred to as subsidiary. If the organisations within the group do not share income statement, they are referred to as affiliated. Table 10: An overview of organisations that are interviewed. Retailer Industry Local or Global E-commerce or Brick and Mor- tar A Pharmaceutical Local Both B Fashion Global E-commerce C Fashion Global Both D Grocery Local Both E Sports Global Both F Cosmetics Global Both G Convenience Global Brick and Mortar 4.1.1 Retailer A Retailer A have been involved in a pilot project regarding horizontal collaboration. It was a collaboration between retailer A and another subsidiary of the group and it was only performed in one city. The goal was to deliver last-mile goods simultaneously to customers of both companies. However, the customers had to make two separate purchases and the goods were consolidated in the transport phase. It was active during six months but re- cently put on hold due to updates in the communication system. Overall, the customers appreciated the offer since they got shipments from two actors in the same delivery. An ob- stacle during the project was the lack of a standard system for data sharing. Furthermore, they are planning to re-initiate the project when the updates have been implemented. Further, two other possibilities of collaboration had been examined by retailer A. The first involved a sports equipment retailer with a warehouse nearby retailer A’s warehouse and they bought transport services from the same carrier. Due to problems with timing 29 of pick-ups, uncertainty regarding the outcome and a negative attitude from the carrier, the project was never launched. Another examined collaboration was together with two manufacturers of medicine. It was found that all three actors delivered daily to the same dispersed stores. Therefore, they wanted to merge these shipments to improve transport efficiency. An obstacle was resistance from the carriers which hindered the collaboration to launch. Carriers do not open up so much for collaboration. ... They would rather sell to two parties and have double purchases. (Representative II, personal communication, translated from Swedish) Retailer A has different transport solutions for delivery to the stores and last-mile de- livery. Stores receive deliveries five times a week from their central warehouse and the transportation is outsourced to a third party logistics provider. It is negotiated together with other subsidiaries to increase the volumes. Except for these negotiations, there are no further collaboration between these subsidiaries since they have different systems for tracking, tracing and back-hauling. Another reason is due to temperature requirements for distribution of pharmaceuticals. Their e-commerce solution on the other hand is taken care of by other smaller transport providers specialised in last-mile transportation. The retailer usually purchase their last-mile transport per parcel instead of a fully outsourced solution. To optimise these parts of the distribution network, carriers utilise milk-runs. The e-commerce warehouse is located with proximity to other customers to the carrier and they can therefore optimise their fill-rate in this area. To share the volume and forecast data with an independent third party was not considered a problem, since transportation of goods is not part of their core business. However, there needs to be an incentive for them to share the data. I can not see any issues with it, it is not our business to transport goods from A to B. Our business is in our stores, websites and to meet the customer. ... However, there must still be an aim with it, and it must be a benefit that comes with it. (Representative II, personal communication, translated from Swedish) An enabler for retailer A to participate in horizontal collaboration is to further develop their customer offerings. This was a benefit they found in the pilot project and they were able to reach new customers with no previous relation to retailer A. Through increased economies of scale, more frequent pick-ups could be possible and thus, lead times to cus- tomers could be reduced. However, there were some scepticism towards the involvement of a third party from retailer A. It was described that a third party logistics provider al- ready accumulate volumes and optimise a network and thus, an independent actor doing a similar activity would be unnecessary. Another possible enabler for retailer A to horizon- tally collaborate would be to reduce transportation costs. The recent price increase in fuel costs was mentioned as one reason for them to further investigate cooperative solutions in transportation. 30 It would of course be cost reductions, I would say. It is, as you know, also depending on how it looks on the world market with fuel prices and so on. (Representative II, personal communication, translated from Swedish) Lastly, the environmental factors were mentioned as an enabler to further collaborate. However, there must not be any reduction of transport quality or delivery precision com- pared with the current solution. In the pharmaceutical industry, the Patient Safety Act regulates what type of data can be shared. It should, for instance, not be possible to connect the content of a package with its wrapping. During the pilot with the other subsidiary, laws regarding which data they could share became troublesome. Today, retailer A shares data for both volume and forecasts with their contracted carrier. It was further highlighted that the competition law could be an obstacle when retailers collaborate. 4.1.2 Retailer B Retailer B has historically been a subsidiary in a company group of fashion retailers. In the group, they negotiated the shipping contracts together with other subsidiaries in order to increase volume. It was stated that the collaboration worked well since they were part of the same group and thus had the same structures and no allocation problems. The carriers did not have any problems with this structure of collaboration. However, the group have since then broken apart and they are therefore not collaborating anymore. Retailer B is sending goods form their central warehouse to consumers in multiple coun- tries. Transportation from the central warehouse to the local warehouse is performed with one carrier and the last-mile transportation is managed by a local carrier with knowl- edge in the individual market. This is usually negotiated as a full solution provided by carriers. In some countries the transport solution requires specific adjustment to fit the environment. However, they deviate from buying transport per parcel when large enough volumes to the destination are accumulated. Frequently, volume become a deciding factor since larger volumes create a more advantageous negotiation position and better solution can be bought. Back-hauling is structured around different warehouses in Europe. The back-hauling solution is fully outsourced and cover everything from check and control to payback. Thereafter, it is sent back to the central warehouse and it is thus ready to be re-sold. The retailer was open to share data regarding shipments. If it is business critical data, I believe, it will be hard to share it. If it is data that can be anonymised in order to create a volume, then it should surely be easier to do it somehow. (Representative III, personal communication, translated from Swedish) 31 However, it was emphasised that the content must be anonymous and not contain business critical information. Information that can be used to increase volume was not regarded business critical and is shared with carriers today. A concern regarding the willingness of carriers to participate in horizontal collaboration was mentioned. Selling transportation is their core business and if their customers begin to collaborate, they would be affected and sell less transports. Furthermore, the decision of who to collaborate with, what information that should be shared and why should come from higher positions in the organisation. Retailer B is open for a future horizontal collaboration but consider it important to know their partners and their motives. The retailer had a positive attitude towards third party involvement acting as a coordinator of the collaboration. However, retailer B was sceptical towards practical implementation of a third party. The businesses must have an aligned view of the collaboration to avoid different opportunistic behaviours. Trust is an impor- tant facilitator for them to be comfortable in a collaboration. It was also important for the retailer that a plan for how to end the collaboration was developed before it was ini- tiated. The potential for collaboration was believed to be largest when transporting from a warehouse to the market. However, a concern regarding the possibility to find partners with transport needs between the same nodes at the same time was mentioned. An enabler for the retailer to participate in horizontal collaboration is higher frequencies of deliveries. I think a driver could be to get a higher frequency and faster deliveries to the market. Yes, and then it would obviously lead to lower costs. (Representative III, personal communication, translated from Swedish) Through horizontal collaboration, a full truck load can be accumulated faster and thus enable more frequent shipments and shorter lead times to customers. Shipping larger volumes lead to economies of scale which reduce the shipping costs per package. It need to be an easy way to understand if there is a demand from other actors to send goods to similar locations. However, how the goods should be shipped also need to be considered. It is not in the shippers core business to transport the goods and therefore, carriers need to be open to collaboration. Scepticism regarding how horizontal collaboration should work in a real context exists however. Coordination, information and volume gathering is what many carriers have as their core business today. 4.1.3 Retailer C Retailer C has examined a potential collaboration project together with an e-commerce company and a coordinator of horizontal collaborations. Due to operational problems re- garding which carrier and trucks were used and when the shipments would be collected, retailer C withdrew from the collaboration. During the project with the e-commerce retailer and coordinator, issues regarding cost allocation and coordination of the collaboration was encountered. It was important for retailer C that the quality of transportation was not below their standards and therefore, they wanted to keep the transport provider already in 32 use. When the coordinator wanted to use a cheaper solution with poorer quality, retailer C lost interest in collaborating. However, retailer C is willing to collaborate in the future. To ensure the quality of their transports, they want to have direct contact with the freight forwarder which was not possible if they wanted to collaborate with the coordinator and e-commerce company. Furthermore, contractual and insurance issues was reported as ob- stacles to the collaboration. The retailer had a central warehouse for flat packages and another warehouse for hanging packages due to different requirements. The hanging packages needed to be transported with a hanger and thus required special transport. The flat goods could be transported as general goods otherwise. Furthermore, these two types of packages could be combined while transported, flat packages at the bottom and hanging at the top. They also have regional distribution warehouses in countries with high sales, managed by third party logis- tic providers. The outsourced solution includes warehousing and last-mile transportation. Items which were not sold in stores was sent and managed at specific warehouses in Europe. The items are checked, refreshed if needed and sent to outlet stores. The retailer was restrictive to sharing data since they have a high security level. They could agree to share encrypted data with a third party but not with a competitor. Fur- thermore, data regarding specific customer addresses was fully restricted. Data that could be encrypted and shared is transport information and package specifications, i.e., volume and weight. The high security standard was also found internally in the organisations and a majority of employees had limited access to information. Clustering of parties was suggested as a possible solution of how to coordinate a horizontal collaboration. It would enable retailers with similar requirements on transportation to col- laborate. An independent third party was suggested as a coordinator who could manage multiple clusters and divide the gains fairly. The lack of a good information sharing system was a barrier for horizontal collaboration according to retailer C. A good system should have an easy on-boarding, the possibility to collaborate in clusters and good visualisation of results. Furthermore, last minute changes were described as a potential problem for the system, since these are common but change the prerequisites for the transportation. Lastly, it would be beneficial to have the opportunity to prioritise which carrier to use and manage other requirements. A global trend which can increase the need for collaboration is increasing container prices. Suppliers are often entitled to transport goods to their local ports. However, after the goods arrive at the port, they are managed by personnel at the port which is financed by retailer C. However, there are a lack of coordination of goods at the port and consolidation of goods from different customers rarely happens. Lastly, inbound flows were mentioned to be more immature than outbound flows and could therefore have higher potential for collaboration. Economies of scale was the main enabler for retailer C to horizontally collaborate. This would lead to cheaper transportation, improved range of services and a reduction in lead time to customers. Faster deliveries are sometimes needed and this forces the package to 33 be transported in an inefficient way. Thus, it raises the prices and have an environmental impact which could be reduced through collaboration. 4.1.4 Retailer D The retailer has been involved in a potential project with other grocery retailers. Several actors withdrew from the project while retailer D still wanted to proceed with the project. The driving part in this project was a carrier who wanted to consolidate the goods. In the grocery industry, goods have temperature and lead time requirements which does not apply to other industries. Transport requirements is therefore an obstacle when grocery retailers collaborate with other industries. Thus, retailer D only considered competitors in the same industry as possible actors to collaborate with. Half of the shipments are last-mile deliveries and this is provided at the largest geographical areas by retailer D. The other half are click and collect deliveries, i.e. ordered online and collected at a store. The first one has the largest potential for horizontal collaboration. The whole distribution chain is expensive and it was estimated that it stands for a third of the marginal loss. Even if they bill the customer for part of the transport, they do not pay for the whole transport since it would become too expensive for the customer. It is common in the grocery industry that companies have their own vehicle fleet for transportation. This is unique in the industry while other industries often share carriers. Retailer D emphasises that collaboration in the grocery industry should be possible, but the benefits of better price, more frequent deliveries and an increased service level must be secured. Horizontal collaboration do require investments in software, route planning systems and functionality, which is an obstacle for the initiation of a collaboration. Retailer D was very restrictive about sharing their data. It was important for them to have trust in the collaborating partner and even more so for the coordinator of the collaboration. It is very sensitive so a collaboration model like this, we need to have trust for the partners who participates and who creates the prerequisites. If it now is as we say today a driving part is the carrier, then we can very much share according to the right contractual agreements and with the right trust. (Representative V, personal communication, translated from Swedish) Sharing data to a third party logistics provider was not seen as a problem if they had trust in the partner and contracts that regulates the collaboration. However, sharing data directly to a competitor will never be considered, and thus there is need for a neutral third party. No, not a chance. ... Who would own the data in this case? It would be a very though situation where a neutral part is necessary, I would believe. (Represen- tative V, personal communication, translated from Swedish) 34 Retailer D saw the potential to reach customers in new geographical areas with their offerings through collaboration. Collaboration between retailers with customers in the same area would allow them to send more frequent deliveries with higher vehicle utilisation. Hence, it enables customers in dispersed areas of Sweden to use their last-mile service. Developing this transport solution with a competitor and a neutral third party would not be a problem since the retailer do not consider transportation as part of their core business. The neutral third party should be the leader of the collaboration while the retailers are flexible enough for the collaboration to be efficient. The main enablers for retailer D to participate in horizontal collaboration was cost savings and improved distribution efficiency. We need profitability on the channel and then I must say that the effective- ness and cost savings is the main reason for us. (Representative V, personal communication, translated from Swedish) Environmental benefits were acknowledged and considered a side-benefit but would not be sufficient for retailer D. Furthermore, it was emphasised that if the retailer did not get the results they expected from a collaboration, they would not hesitate on withdrawing and continue on their own. 4.1.5 Retailer E Retailer E is part of a company group with affiliate stores in several countries. A collab- oration between the affiliates in Europe was considered in a project but due to problems regarding the allocation of costs, it never launched. Cost allocation also became a prob- lem in collaboration between different national affiliates since everyone wanted the biggest advantage. There were no obvious reason why international collaborations are harder to implement than national collaborations, but politics and individuals involved in the collaborations were possible explanations. Since the collaboration was examined, a high turnover of individuals had happened. Today, the contact between the international affili- ates is worse. Their supply chain can be divided in two flows. One flow is from suppliers in Asia, through their central warehouse in Europe, to warehouses in each country. The transportation from Asia to Europe is consolidated to increase volumes and reduce price per unit. Furthermore, the second flow is nationally transported goods to affiliated stores. At this level, there are collaboration between affiliated stores to reach some economies of scale. Retailer E trans- ports a majority of its goods through their national warehouse. This is their preferred solution since it gives them control of the transports. In some cases, their suppliers have a more efficient distribution network than retailer E and therefore, they could be allowed to transport their goods directly