Asset specificity, inter-firm ecosystem, firm adaptability and supply chain integration Open Access

Over the years, studies on supply chain integration (SCI) have been on the increase though less focus has been placed on developing economies (Khanuja and Jain, 2020; Ramanathan and Gunasekaran, 2014). These findings however contradict with those of Tiwari (2021) who revealed a decline in SCI studies between 2018 and 2020 which raises concerns. Globally, it could be due to global pressure to achieve sustainable development goals (Organization for Economic Cooperation and Development report, 2020; United Nations report, 2017) where firms that are not able to catch up are kicked out. The less focus on firms in developing countries could however be due to high levels of technological advancement which is key in SCI yet firms in these countries are not able to sustain it due to limited resources and uncertain benefits (Kanyoma et al., 2020; Nguyen et al., 2019).

In Uganda, integration and connectivity of operations are anchored on Uganda’s 2040 development agenda and the focus is to bring together internal and external supply chain operations. It is also anticipated that integrating supply chain operations will promote shared planning and joint goal-setting between the focal firms and their trading partners (Tukamuhabwa et al., 2021). These are possible where firms work together in their ecosystems.

However, creating an ecosystem environment where interdependencies exist in the food processing sector (FPS) in Uganda remains challenging because of the existence of free entry and exit with many of the firms not registered and unknown creating uncertain situations (Waiswa et al., 2021). The FPS is further characterized by changing market demands (Siagian et al., 2021). SCI is necessary in countries like Uganda that have limited sharing of information between partners regarding prices, production volumes and consumption patterns (Fowler and Rauschendorfer, 2019). This could be attributed to limited investment is specific assets. Current research shows low-capacity utilization among processors (Onward Resources International, 2016; Economic Policy Research Centre (EPRC), 2018; Munu, 2019). For example, Mukwano industries one of the leading producer of vegetable oil and fats in the country has capacity utilization ranging between 20 and 35% (Onward Resources International, 2016); the coffee industry has capacity utilization standing at 40%; fish processing is found to be less than 30%; dairy processors has an average capacity utilization of 66% and beef capacity utilization is less than 20%. At the industry level, capacity utilization in the dairy sector remains low between 40 and 60% (Makoni et al., 2014). Most of these firms located in rural areas far from their markets making coordination and transportation almost impossible especially in rain seasons (Fowler and Rauschendorfer, 2019). Farida et al. (2024) also found limited investment I human specificity to foster SCI and adaptability in uncertain situation. All these examples indicate the presence of poor or low levels of SCI in the FPS.

Because supply chains in developing countries are complex and uncertain the study utilizes the Transaction Cost Theory (TCT) and Complexity Adaptive System Theory (CAST) to understand the costs and complexities that emerge when firms interact with the market. What makes SCI complex is the multi-level interconnectedness and interrelated relationships that exist between the firm and its different partners (Jacobides et al., 2018; Christopher, 2011). These multi-level interactions create visibility and forecasting problems that come with costs (Jambulingam and Kathuria, 2020; Jose and Shanmugam, 2019). The TCT assumes that individuals are generally opportunistic and this is the main cause of costs in transactions which food processing firms (FPFs) need to understand. In managing these inter-firm relationships, complexities emerge due to differences in objectives, interests and values leading to further costs of monitoring and coordination (Williamson, 1985). To maintain these multi-level relationships, firms have to focus on the interdependencies (Helfat and Raubitschek, 2018). Thus, this study aimed at examining the influence of asset specificity, inter-firm ecosystem and firm adaptability on SCI. The rest of the paper is grouped as follows: Section 2 discussed the theoretical review and literature review; the methodology is discussed in Section 3. Findings are presented in Section 4 and Section 5 contains a discussion of findings and Section 6 looks at the recommendations and conclusion.

To understand the costs that come with integrating supply chain operations, scholars have used the TCET (Ketokivi and Mahoney, 2020; Turkulainen et al., 2017). As firms integrate with multiple suppliers in the market, they are faced with different kinds of costs like bargaining costs and switching costs (Huang and Huang, 2018; Williamson, 1971). This is because such firms are investing in specific assets like location specificity, dedicated specificity and human specificity to meet the different needs of the market (Williamson, 1985; Jambulingam and Kathuria, 2020). This also increases the cost of monitoring the behavior of partners, usage of assets for intended purpose and coordinating transactions in the different markets (Williamson, 1975, 1985; Yigitbasioglu, 2010). Christopher (2011) warns that external partners are connected to a web of many other firms and this may result into opportunistic tendencies, create visibility and forecasting challenges. This further makes supply chains complex and non-linear (Abbasi and Varga, 2021; Ciliberti et al., 2020). To address that, firms have to view SCI as a complexity concept by applying CAST with an objective of bringing order in non-linear situations (Urry, 2005).

The complexity adaptive system assumes that organizations operate in a changing environment which necessitates the need to learn and adapt to new ways of doing things (McMillan, 2008). Firms are at different levels of production and therefore for them to survive, collaborative relationships are created that allow them to share vital information. Each component and each intersection exhibit different skills, competencies and roles. To respond to such dynamic in the market, organizations co-exist as ecosystems interacting with various components like suppliers and customers in a web. An ecosystem relates to a collective arrangement of many different firms through which issues related to uncertainty in transactions are addressed (Ciliberti et al., 2020). Firms that join the ecosystem have different values, norms and cultures called attractors that they have to adjust to those of the collaboration (Goldstein et al., 2010).

Additionally, the CAST assumes that the success of any firm is highly dependent on the scope and quality of the network. This depends on how the firms align their attractors, depend on each other and also maintain the connectedness through feedback and learning. Goldstein et al. (2010) note that this is the heart of the complexity adaptive system theory. Complexity arises when two firms each with different thinking and information interact (Goldstein et al., 2010). Because of differences in objectives, opportunism in the ecosystems increases calling for adaptive strategies that focus on new ways of responding, copying and producing new behaviors previously unseen (McMillan, 2008).

Therefore, in this study, TCT is a theory of costs that are involved in SCI when firms invest in specific assets to foster market relationships while the CAST is to bring order and make firms adaptive to the dynamics that emerge in the market (Goldstein et al., 2010). Hence, we expect a direct link from asset specificity and SCI through the lens of TCT. The study also anticipates a direct impact of inter-firm ecosystem and firm adaptability on SCI through the lens of CAST. The interplay between asset specificity and SCI through firm adaptability is through the complementary role of CAST.

Asset specificity in this study has been operationalized focusing on dedicated specificity, human specificity and location specificity. Asset specificity drives integration in the supply chain through vertical integration (Ketokivi and Mahoney, 2020; Williamson, 1985). When firms invest in locations close to their trading partners, they achieve SCI (Bennett and Klug, 2012). When firms are in closer proximity, they frequently interact, share information (Balland et al., 2015) and quickly introduce products to the market (Huo et al., 2014). This reduces transaction costs and the reverse is true when they are located far from the market as they have to incur higher costs to monitor and coordinate the different markets (Ketokivi and Mahoney, 2020). Therefore transaction costs are high where firms are dealing with partners that are geographically dispersed necessitating firms to further invest in dedicated assets like delivery trucks, warehousing facilities and technology to track items (Jean et al., 2021).

The firms’ dedicated assets like technology drive integration especially where firms and their resources are geographically dispersed (Yu et al., 2020). This improves supply chain visibility and quickens decision making (Gu et al., 2020). The use of technological systems reduces mistakes in transactions, improves sharing of real time information and aligns the internal functions of a firm to the market (Oubrahim et al., 2023).

However, the level of SCI in developing economies has been limited due to limited skills and structures to implement advanced technology (Kalubanga and Namagembe, 2021). Also, Li et al. (2023) noted that increase in investment in specific assets can result into increased transactional costs with its lock-in effect. According to the TCT, costs are also a result of monitoring and enforcement (Williamson, 1985).

Human specificity results in task accomplishment (Song and Song, 2021). Accordingly, human specificity is the extent to which specialized human capital is attained through training. Zhao et al. (2023) revealed that training improves employee knowledge and skills promoting effective sharing within the supply chain. Relatedly, multi-skilled employees are in position to understand market demands and this has an influence on integration (Murfield et al., 2017). Strong firms should offer technical support to the weaker firms in the chain (Bhardwaj and Ketokivi, 2020). Such arguments have resulted to the formulation of the first hypothesis stated further in the text.

Developing an ecosystem is one way through which firms address issues related to uncertainty (Ciliberti et al., 2020). Though greatly explored in business (Rong et al., 2012); entrepreneurship (Sheriff and Muffatto, 2015) and innovation studies (Mikhailov et al., 2021), ecosystem studies in supply chain are few (Sheriff and Muffatto, 2015). Zhao et al. (2023) recommend that players should join associations to enable them communicate their interests. Inter-firm ecosystem will be studied focusing on: interdependence (Burford et al., 2021), stakeholder involvement and standard operating procedure (SOP) (Liu et al., 2019; Alinaghian et al., 2020).

Creating interdependence within a supply chain is critical to achieving shared outcomes (Kembro and Selviaridis, 2015). The drive behind interdependence is resource scarcity (van Dijk et al., 2021). Accordingly, actors create a common forum like WhatsApp groups for purposes of sharing common information. Ecosystems are characterized by opportunism but they diminish with interdependence (Kanyoma et al., 2018). However, interdependence thrives best where firms are of the same size (Selviaridis and Spring, 2018) with multi-level relationships (Adner, 2017; Kembro and Selviaridis, 2015).

In an ecosystem, partner involvement has also been identified as a key driver of SCI (Alam et al., 2014). Just like many scholars have recommended trust as critical issue in SCI, Shan et al. (2023) also posit that partner involvement will foster positive results in a supply chain when partners have trust in each other. It plays a critical role in understanding the firm’s operating supply chain environment (Magassouba et al., 2019). It may be through physical interaction or via social media platforms to allow information sharing and collaboration (Cheng and Krumwiede, 2018). Their involvement results in improved production schedules (Khanuja and Jain, 2020). Where stakeholders are not engaged, they adopt practices that are contrary to the firm’s objective (Mahapatra et al., 2019).

In addition, SOPs are crucial for firms that wish to integrate their operations (Rajaguru and Matanda, 2019). Ecosystems survive on established SOPs (Liu et al., 2019). These have to be rooted within a firm’s business processes (Bhardwaj and Ketokivi, 2020; Rajaguru and Matanda, 2019). They include the firms’ routines and rule of the thumb (Tukamuhabwa et al., 2021). These ensure compliance (Ciliberti et al., 2020). From the above discussion, the second hypothesis has been developed.

There is limited empirical research on firm adaptability. The focus has been on supply chain resilience using the triple-A (Sheel and Nath, 2019; Lee, 2004). Globalization has made supply chains complex and this has increased pressure on firms. This calls for adaptive strategies to coordinate processes (Jambulingam and Kathuria, 2020; Garrido-Vega et al., 2021). To study firm adaptability, learning and flexibility will be utilized.

Learning is one of the adaptive strategies devised to swiftly respond to changes in uncertain conditions (Lee et al., 2019). It varies across stages and so is integration across trading partners (Scholten et al., 2019). It is a critical driver for information sharing and as firms learn from each other, they quickly meet the market demand (Lisi et al., 2019). This prevents disruptions from occurring (Chen et al., 2021). However, external integration is only possible where firms learn, collaborate and share knowledge (Kumar et al., 2020).

Furthermore, uncertainties in today’s supply chains call for flexible strategies (Ivanov, 2021). This allows firms to easily respond to market pressures fostering collaboration and sharing of information (Despoudi et al., 2018). Mutebi et al. (2020) also found that participating firms find it easy to change and adjust to the market conditions. This discussion has given rise to the third hypothesis as reflected below.

In situations of uncertainty, firms embrace adaptive strategies to reduce transaction costs (Jambulingam and Kathuria, 2020). They invest in specific assets that enhance learning (Hernández-Espallardo et al., 2010). Through social interaction, assets improve learning resulting in change by all members in the network (Chen and Garg, 2017).

Human capital and learning act as a link to the information in the market (Liu, 2017). Where individuals have the required skills and expertise, they identify problems, errors and suggest solutions (Palacios et al., 2014). Such expertise must be trained and encouraged to socialize with other external actors to enhance learning (Liu, 2017).

Individuals respond to changes in the internal and external environments of business operations by investing in technology to enhance production and quantity mix (Shukor et al., 2021). During the COVID-19 pandemic, firms with stronger digital infrastructure fared better than those without (Gu et al., 2020).

The location of a firm and its trading actors also has an impact on the level of learning (Chen et al., 2021). Accordingly, when firms are located near each other, they learn from the network. The hypothesis below has been derived from the above discussion.

Supply chains with more trading partners have been found to be adaptive because of the existence of many players with varying strategies and ability to learn from one another (Garrido-Vega et al., 2021). A highly dense network is a sign that the ecosystem firms are adaptative and flexible to each other (Statsenko et al., 2018; Adner, 2017). The existence of the firm’s routine policies and processes allows individuals to learn new ways of doing business (Silvestre et al., 2020).

Stakeholder involvement is a prerequisite for learning to take place. Stakeholders learn through interaction, sharing experiences and collaborating creating a sense of certainty (Scholten et al., 2019). It has also been found to be a source of flexibility given its ability to enhance coordination and improve access to market information (Khanuja and Jain, 2020). During such interface engagements, new ideas are shared (Ramirez-Portilla et al., 2017). Xiao et al. (2019) and Khanuja and Jain (2020) posit that synchronization of data and access to customers’ real sales data are possible with partner involvement. On the contrary, Singh and Rao (2016) advance that learning is an outcome of interaction between firms.

Mandal et al. (2017) encourages firms to develop flexible actions and policies to allow synchronization and exchange of information related to product delivery, forecast and inventory status. This influences compliance with acceptable measures and reduces opportunism (Yunus and Tadisina, 2016). This discussion has led to the development of the hypothesis below.

Firms have invested in rare assets and this has put them in an adaptive position by analyzing the supply chain opportunities (Aslam et al., 2018). Investing in specific assets as well as SCI have been found to be strategic decisions in the firm’s operation (Kumar et al., 2020).

Technologies needed for integration are always evolving and the ease of integrating the different supply chain partners depends on how easily and fast the players manage this transition (Wang and Wei, 2007). Investment makes firms adaptive by improving on communication and also attaining product modification (Aslam et al., 2018; Sheel and Nath, 2019).

Employees with the right skills and competences are vital in enabling technological usage to make adjustments. This allows real-time sharing of information-across the entire supply chain (Turkulainen et al., 2017; Gawankar et al., 2019). It further improves learning and later influences integration (Sheel and Nath, 2019).

Furthermore, when trading partners are located very far from each other, the pressure to adapt and cope up with the changing market conditions is high (Jean et al., 2021). Jean et al. (2021) further reveal that firms can employ IT systems to make firms more adaptive to the changes in the market. The above discussion has given rise to the hypothesis below.

Inter-firm ecosystem has been found to be successful where members are willing to align their visions and develop a high level of commitment to the norms of the ecosystem (Letaifa, 2014). The success of aligning the firm’s policies and procedures with those of its trading partners is embedded in members’ flexibility to adjust and accommodate changes (Mutebi et al., 2020).

Developing and adopting shared cultures and norms across firms is critical in making firms adaptive (Shahzad et al., 2017). Through firm adaptability, individuals learn the different policies of the different firms in the ecosystem making them able to align interests and be committed to their success (Selviaridis and Spring, 2018).

It is further believed that involving partners into the firm’s operations results in a better view of the overall operations of the firm through shared information (Xiao et al., 2019). In addition, firms tap into the resources and capabilities of the external partners especially where players are flexible and willing to learn and adjust from the routine. Khanuja and Jain (2020) reveal that stakeholder involvements allow firms to be flexible by sharing real-time information on volume, mix and velocity. The hypothesis discussed further in the text has been developed from the above discussion.

The above hypotheses are illustrated in Figure 1.

This section looks at the research design, study population, sample size, instrument development and data collection and data analysis.

The study used a quantitative research methodology as used by previous scholars (Ganbold et al., 2020; Shukor et al., 2021). This research adopted a quantitative methodology using hypothesis testing. A cross-sectional research design was adopted. Given the nature of the FPS, firms were stratified depending on the categories of products they produce.

Using data from Uganda National Bureau of Statistics (UBOS), a population of 345 food processing firms found in Kampala District was taken. Using Krejcie and Morgan (1970) table of sample size determination, a statistical sample of 186 firms was arrived at. However, following recommendation Ganbold et al. (2020) and Goodhue et al. (2012), a sample of 103 firms was sufficient and fair for analysis. The unit of analysis was a firm. This study is anchored on the recommendations by Barratt and Barratt (2011). The researchers recommended that a study on several organizations at different levels of development and economies of scale be conducted. Data was therefore collected from various FPFs with two respondents selected per firm giving a total of 206 responses. A disproportionate stratified simple random sampling technique was used to choose the firms for inclusion in the study. This is because the FPS are in different categories dealing in different activities like Fish and Meat processing (28), manufacturers of bakery products (104), vegetable and animal oil processing (5), manufacture of animal feeds (30), processing and preserving of fruit and vegetables (12), dairy processing (14), grain milling (128) and coffee and tea processing (24) based on the UBOS statistics (2011). Different strata were developed and later applied simple random sampling in selecting the desired sample from each stratum.

The target respondents were managers given their strategic role in the firm’s supply chain operations. These are believed to have the required knowledge on both the firms external and internal operations (Cresswell and Clark, 2011; Palinkas et al., 2015).

The instrument had items that covered variables like asset specificity, inter-firm ecosystem, firm adaptability and SCI. These were measured on a 6-point Likert scale with verbal anchors ranging from very untrue (1)–extremely true (6). To achieve content validity, items from previous scholars were adopted with modification. To ensure anonymity and confidentiality of respondents, the questionnaire was designed in a way that there was no space for filling in respondent’s details like name.

SCI was measured using internal integration and external integration (supplier integration and customer integration) despite controversies in literature on the measures (Khanuja and Jain, 2020). Internal integration was measured using six items: three items were from the works of Wong et al. (2011), two items from Delic et al. (2019) and one item was from Barratt and Barratt (2011). External integration was studied using ten items: three items from the works of Wong et al. (2011), three items from Ataseven et al. (2020) while four items were adopted from the works of Kanyoma et al. (2018).

Asset specificity was measured by human specificity, location specificity, dedicated specificity (Wang et al., 2019; Anderson and Weitz, 1992).

Firm adaptability was measured by learning and flexibility. Learning was studied using twelve items: two were from the works of Prieto-Pastor et al. (2018), five were from Guo et al. (2020), four were from Kumar et al. (2020) and one from Bouncken and Fredrich (2015). Flexibility was measured by six items: three were from Ding et al. (2012) while the other three were from Kumar et al. (2020).

Inter-firm ecosystem was measured by interdependence, stakeholder involvement and SOPs. Interdependence was measured by 12 items: four items were from Shahzad et al. (2017) three were from the works of Bongomin et al. (2018) while five were developed from the works of Kembro and Selviaridis (2015). Stakeholder involvement was measured by nine items: four were from the works of Cheng et al. (2018) with the rest being obtained from literature review (Cozzolino et al., 2017). SOP was measured by thirteen items. Out of the thirteen items, six were from Prieto-Pastor et al. (2018) while seven were from Mutebi et al. (2020).

Hard copies of structured questionnaires were physically delivered to every identified firm. After a week, respondents were followed up through telephone calls to inquire if they had any challenge in answering the questions or had completed the questionnaire. Completed questionnaires were then picked for further management. Respondent and firm anonymity were maintained to reduce the social desirability bias and also non-response to certain questions.

Despite the existence of other versions of SmartPLS-SEM, the study used SmartPLS-SEM version 3.0 for data modeling because it was accessible to the researchers and could be used to obtain the required results. SPSS version 21 was used to test for firm and respondent characteristics, normality, skewness, kurtosis and reliability. Smart PLS was used to test for Variance Inflation Factor (VIF), Average Variance Extracted (AVE), Confirmatory Factor Analysis (CFA) and structural model results. CFA revealed that all measurement items were significantly associated with the variables they measured (Table 1).

Based on our scale, all respondents agreed that the items were a true representation of what was happening in their organizations given our mean values (see Table 1). A formative-reflective model was used to measure the constructs and their indicative items.

On SCI, respondents were meant to reveal whether their organizations were coordinating and sharing information about their supply chain processes both internally and externally with suppliers and key customers. Out of the nine items that measured external integration, eight were retained with loadings between 0.478 and 0.631 (β = 0.609; p = 0.000) while all the six items that measured internal integration were retained with 0.487–0.778 and β = 0.528; p = 0.000. Following recommendations by Hair et al. (1995), indicators of 0.4 and above should only be deleted if it will improve composite reliability or AVE. This implies that all constructs were significant to SCI. However, the success of SCI depends more on external integration capabilities as revealed by earlier scholars like Kanyoma et al. (2020) and Ataseven et al. (2020).

Asset specificity with its measures aimed at examining the level of investment in customized facilities by the firms to foster adaptability in uncertain situations and to also promote SCI. The measures of asset specificity were dedicated specificity, human specificity and location specificity. Dedicated specificity had eight items all loading between 0.506 and 0.816 with β = 0.611; p = 0.000. Human specificity had six items and all were retained loading between 0.476 and 0.751 with β = 0.204; p = 0.000. Lastly, location specificity measured how far or near the firms located their facilities in the market. Five items were used to study location specificity. All items were retained with loadings between 0.656 and 0.808 and β = 0.391; p = 0.000. These findings indicate that FPFs are more focused at investing in dedicated facilities and locating them near the market (suppliers and customers) to achieve frequent sharing of information and quick response to market pressures than investing in improving the skills and competencies of their human resource.

Inter-firm ecosystem measures were interdependence, standard operating procedures and stakeholder involvement. Interdependence had twelve items. Eight items were retained with loadings between 0.512 and 0.765 while four items were dropped because they loaded below 0.4. According to Hair et al. (2017), items with outer loadings below 0.4 should be removed due to their effect on content validity. Standard operating procedure had nine items out of which, two were dropped because they loaded below 0.4 while seven were retained with loadings between 0.629 and 0.809. Lastly, stakeholder involvement was measured using eight items out of which six were retained with loadings between 0.409 and 0.741. In addition, of all the inter-firm ecosystem measures, respondents contended that firms embraced SOP (mean = 4.8) while dealing with their suppliers and customers more than the rest of the constructs (Table 1). Though all measures were significant, still SOP was found to be a better measure of inter-firm ecosystem than the other two factors. Therefore, in uncertain situations FPFs embrace their routines, they stick to the rules of the game, goals and objectives.

Firm adaptability aimed at investigating whether in uncertain situations FPFs adopted learning and flexibility strategies. It was found that firms adopted learning as an adaptive strategy more than flexibility (mean = 4.629). Learning items were ten but only six were retained with loadings between 0.483 and 0.633 while the six items that measured flexibility were retained with loadings between 0.609 and 0.740. Therefore, flexibility was found to be a better measure of firm adaptability (β = 0.569; p = 0.000).

Preliminary data was analyzed using SPSS version 21. As reflected on Table 2, the bakery sub-sector accounted for the highest composition of the sector (32.7%), followed by grain milling (31.2%), with the least falling under processing and preserving of fruit and vegetables (2.3%). This means that this study is informed by data from the bakery sub-sector and the grain milling sub-sector. This is consistent with NDPII report (2020) which found that grain milling is one of the largest sub-sector of the food processing industry in Uganda.

The number of employees was used to determine the size of the firm. Medium-sized firms employing between 51 and 100 people accounted for 160 (60.8%), followed by small-sized firms that employ between 5 and 50 (33.5%). This implies that small and medium firms are the biggest employers. Most firms were found to have been in business between 6 and 10 years (63.1%) followed by those between 1 and 5 years (25.7%). The analysis of respondent profiles revealed that most respondents were within the age bracket of 33–48 years (47.5%) with the least being those of 68 years and above (1.5%). This implies that the sector requires more energetic people and given the nature of operations, most people leave the sector as they advance in age. Most respondents had also worked with the firm for a period between 2 and 3 years (44.7%) which further indicates that as people join the sector, they leave to join other sub-sectors.

To determine the reliability, convergent and discriminant validity of the measurement items and the study variables was conducted. Convergent reliability was tested using Cronbach alpha and composite reliability (see Table 3). For both tests, the values were above 0.7 indicating that the data was reliable as advanced by Alam et al. (2014). AVE values were above 0.5 confirming the presence of convergent validity (Rajaguru and Matanda, 2019).

Discriminant validity (see Table 4) was measured using the heterotrait–monotrait method. All the values were below the set threshold of 0.9 (Namagembe, 2020). Data was also tested for normality and the first test was skewness and Kurtosis (see Table 3). Our data was in line with Namagembe (2020) who revealed that the skewness values of less than 2 and kurtosis values of less than 7 shows that the data is normal hence normality of data was established. Further, there was no evidence of multicollinearity between the variables because the Variance Inflation Factor (VIF) values were less than 5 (Table 3).

Our structural model results are provided in Table 5. The model was designed to test the relationship between the latent variables of the study using PLS-SEM following model criteria by Hair et al. (2017). From the literature search, the study anticipated the independent variables to have a positive effect on the dependent variable. From analysis (Figure 2), it was found that asset specificity and inter-firm ecosystem had a positive but insignificant effect on SCI (β = 0.016; p < 0.804; β = 0.094; p < 0.201) respectively which implies that H1 and H2 was not supported. Firm adaptability had a positive significant effect on SCI (β = 0.537; p < 0.000) giving support to H3. Also, all independent variables asset specificity and inter-firm ecosystem are associated positively with the mediator variable giving support to H6 and H7. Following Barron and Kenny’s four-stage procedure, a mediation test was carried out (Baron and Kenny, 1986). Accordingly, the following were done: 1. The independent variables were regressed on the mediator variable, step 2 involved regressing the mediator variable (firm adaptability) on the dependent variable. Thirdly, independent variables were then regressed on the dependent variable. Lastly, mediation analysis was conducted with all the variables involved in the analysis. From that analysis, all paths were found significant implying that firm adaptability mediated the relationships as per the model. The results from the mediation analysis were as follow; a partial mediation was established between asset specificity and SCI since the influence of asset specificity on SCI was found significant (β = 0.341; p-value = 0.004) while a full mediation effect was found in inter-firm ecosystem and SCI since its influence ceased with the introduction of a mediator (β = 0.114; p-value = 0.237).

The section provides the findings on the effect of asset specificity and inter-firm ecosystem on SCI mediated by firm adaptability.

Hypothesis 1 provides an exciting theoretical opportunity for the application of TCT in studying SCI contrary to Farida et al. (2024). The current study found that asset specificity had a positive effect on SCI though not significant (β = 0.016; p-value = 0.804) which contradicts with Farida et al. (2024) who found positive significant results. The current findings also vary with those of Farida et al. (2024) in the level of influence (β = 0.365; p-value = 0.003). This could be attributed to the introduction of the inter-firm ecosystem variable in the model. Comparing the two studies, we note that in the current study, when inter-firm ecosystem factors are increased, firms would reduce investment in specific assets causing its effect on SCI to further go down. This is in line with Bygballe et al. (2023) who found that dependence reduces investment in specific assets since it promotes sharing of scarce resources. Also, Tseng et al. (2023) found that it was not the location of the firm that influenced their investment in specific assets but rather the level of partner involvement.

Our findings, however, imply that though FPFs have invested in specific assets like locating firms near the market to achieve coordination and quick response to the market, the objective has partially been achieved. From TCT, this indicates that firms are still incurring high costs to monitor and coordinate supply chain operations with trading partners (Williamson, 1985). This is contrary to Akidi et al. (2022) who found that partners conveniently integrate their operations with firms that are located within their vicinity to reduce transactional costs. They also achieve sharing of market information through face-to-face meetings (Balland et al., 2015). The high costs could also be due to high switching costs resulting from having multiple partners. Kataike (2019) in the study that focused on the dairy sector in Uganda using the TCT revealed that farmers maintained multi-level relationships with many dairy firms which increased the cost of monitoring and coordination.

In line with the findings, Garces Rivera and Pfeiffer (2018) found that partners were still incurring extra costs transporting coffee beans to the market. In addition, the pig sub-sector is suffering from poor warehousing facilities, especially in the upstream supply chain and the sector has been sidelined for a long time (Ouma et al., 2016). Abdulsamad and Gereffi (2016) also found little investment in cold chains as some processors continue to deliver pasteurized milk in non-refrigerated trucks.

Some firms have made investments in standardized warehouses, technology and transport facilities to coordinate inventory levels. However, these efforts have not resulted in sharing of information. This implies that much as investment in specific resources has been made, they are either still low or not appropriate to foster coordination and sharing of information. Therefore, players need to rethink their investment decisions. This is consistent with Kataike (2019) who revealed that much as dairy firms have made investments in milk cooling tanks, milk cans and road tankers, they are not enough.

Previous scholars revealed that investment in human resources results in better working relationships (Huo et al., 2014). Coffee farmers have also received training on coffee stem quality (Garces Rivera and Pfeiffer, 2018). However, there are still contradictions across industries. Lakuma et al. (2021) in the ERPC report revealed that there were limited personnel in the fish processing sector to manage the value chain and called upon Uganda National Bureau of Standards to train persons to cover the gaps. Low levels of education among Community Animal Health Workers (CAHW) in the piggery sector (meat processing) limit sharing of information related to drug usage and type (Ouma et al., 2016). These findings are in line with β = 0.204; p-value = 0.000 for human specificity. This implies that much as firms have made investment is specific assets, the influence of human specificity still remains low.

In line with the TCT, this study assumes that investment in specific assets results in integration. The TCT assumes that vertical integration is possible and successful where firms invest in specific assets that act as hostages to reduce the cost of transactions. However, the existence of opportunism among the players limits investment (Williamson, 1985). Therefore, much as under this hypothesis the objective to invest in specific assets has not been successfully achieved, in uncertain situations, firms are in a position to survive and also share information as reflected in H4 and H6. This explains critical role of the mediation variable.

Hypothesis 2 contradicts the CAST assumption that ecosystems survive by sharing resources through symbiotic interactions (Goldstein et al., 2010). It also contradicts Statsenko et al. (2018) who advanced that in an ecosystem where partners are connected, it fosters sharing of information. Our results reveal that inter-firm ecosystem and SCI were insignificantly related (β = 0.094; p-value = 0.201) implying that players in the FPS are independent of each other in accomplishing tasks, they don’t fully involve trading partners and have also not fully embraced the use of standard operating procedures. It could therefore imply that these firms are only embracing ecosystem factors under uncertain situations as explained in H5 and H7. This hinders sharing of operational and strategic information and joint inventory planning. This explains why firms have resorted to informal sources of input supply (World Bank, 2018).

The dairy sector is fragmented with independent transporters and local traders who collect milk from farmers creating insufficient milk and under-capacity utilization in the formal sector. It’s believed that suppliers/farmers prefer dealing with the informal players because they offer better prices compared to processors limiting coordination (Abdulsamad and Gereffi, 2016).

The reason for the low levels of interdependence can be explained by Selviaridis and Spring (2018). Accordingly, interdependence and sharing information may not be possible where firms are at different levels of production in terms of size. The existence of opportunism noted in our results also compromises interdependence limiting sharing of information (Kanyoma et al., 2018).

In line with Obwona et al. (2014), our findings revealed limited use of websites, emails and social media platforms like WhatsApp and Facebook to communicate and share information with trading partners. The results also revealed the absence of common goals, vision and objectives. This is contrary to previous studies that advanced that having a shared vision determines the survival and the nature of relationships in an ecosystem (van Dijk et al., 2021). Absence of a shared vision results in compromising the industry standards. In the piggery sector, there is limited adherence to standards and this limits integration (Ouma et al., 2016).

There is limited adherence to the available standards in the fish sector and this affects alignment of processes from the suppliers end to customers end (Lakuma et al., 2020). Waiswa et al. (2021) recommends that dairy farmers need to revise existing policies. Where Memorandum of Understanding (SOPs) have been adopted, there have been complaints about the services provided (Joughin, 2019). Consistent with our finding, most firms in our study did not belong to an ecosystem (association) (Makoni et al., 2014; Waiswa et al., 2021). This limits supply chain visibility and results in forecast errors.

We also note and confirm Rajaguru and Matanda’s (2019) finding that failure to align interests, vision and goals in an ecosystem though there exist rules and policies may not result in integration with partners.

On partner involvement, the results found that firms had limited involvement of their trading partners in supply chain masters. These were never consulted whenever adjustments in prices and orders were made. This is consistent with Tumushabe (2020). Accordingly, dairy farmers have no say on the price of milk. Of all the processors in the dairy sector, it is only JESA that offers fixed prices to its suppliers for a period of one year (Toye, 2012). Mahapatra et al. (2019) noted that when partners are not involved in ecosystem activities, they adopt practices that are contrary to the firm’s objective.

Controversies in our results could be due to role ambiguity and the many players that are involved in the ecosystem limiting information sharing (Waiswa et al., 2021; Garces Rivera and Pfeiffer, 2018). Garces Rivera and Pfeiffer (2018) contend that role clarity is the cornerstone for better relationships.

Hypothesis 3 results unveiled a significant relationship between firm adaptability and SCI (β = 0.537; p-value = 0.000). This is in line with the findings by Farida et al. (2024) with (β = 0.321; p-value = 0.004) under the same context. However, in the current study, the adaptability levels of firms are improving and this could be attributed to the existence of inter-firm ecosystem factors. The study finds that in uncertain situations, supply chain partners change their attitude towards market situations by changing inventory management practices based on the knowledge acquired through interactions. They connect processes, share inventory status information to allow quick response to market needs.

Our study finds support and contradictions in existing literature. Lisi et al. (2019) posit that as firms interact in uncertain situations, they learn how to respond to the market dynamics. However, some scholars posit that integration influences learning (Silvestre et al., 2020).

Furthermore, firms make adjustments in production schedules to accommodate changes from both suppliers and customers in uncertain markets. This enhances the coordination of inventory schedules throughout the entire supply chain. Mutebi et al. (2020) contend that the ability to adjust to the market pressure allows firms to coordinate and share market information. The findings further reveal that in uncertain situations, supply chain actors learn from their mistakes and adopt new skills, and knowledge on how to manage supply chain operations like order processing and deliveries to customers. This helps firms to prevent disruptions in the supply chain (Chen et al., 2021).

However, H3 contradicts with Garrido-Vega et al. (2021) who found that in a complex and uncertain environment, firms find it difficult to adapt due to information asymmetry.

Hypothesis 4’s findings reveal a positive association between asset specificity and firm adaptability (β = 0.300; p-value = 0.000) which is in line with Farida et al. (2024). However, in their study the influence of asset specificity on firm adaptability was (β = 0.416; p-value = 0.000) a bit higher than in the current study. A lower influence in the current study could be attributed to the introduction of inter-firm ecosystem factors more so standard operating procedures (β = 0.431; p-value = 0.000). The more firms engage in meetings as a routine, the more they don’t invest in specific assets because they understand the challenges involved and therefore reduce investment (AlHares et al., 2018).

The current findings imply that in uncertain situations, firms envisage an increase in the cost of trading partners and invest in adaptive assets. However firms are investing more in human specificity (mean = 4.714) and dedicated specificity (mean = 4.714) to be adaptive than in location specificity (mean = 4.396). Investment in dedicated assets enables firms to stock enough inventories, and quickly make adjustments to respond to market demand. Our findings are consistent Waiswa et al. (2021) who recommended establishment of rural milk collection and cooling infrastructure to increase proper storage of milk. Our findings however contradict with Obwona et al. (2014) who found limited investment in technology limiting adaptability levels.

Investments have been in human skills in the coffee industry making farmers adaptive (Garces Rivera and Pfeiffer, 2018). JESA has also been noted to train its dairy suppliers (Toye, 2012). Training imparts skills that allow partners to change their strategies based on the market conditions (Liu, 2017). Contrary to our mean = 4.714 for human specificity, some scholars have noted inadequate skills in the manufacturing sector (Uganda Bureau of Statistics, 2011/2012; Obwona et al., 2014).

This study further finds that some players are located close to the market and are therefore spending minimal costs and time transporting raw materials and finished stock which is consistent with Farida et al. (2024). In Uganda, some processors in the dairy sector have invested in milk collection centres acting as coordination centres (Abdulsamad and Gereffi, 2016). This enhances the ability to adjust production and delivery schedules to customers.

On the contrary, most processors in Uganda are located far from their suppliers and this increases transaction costs (Garces Rivera and Pfeiffer, 2018). Despite such inconsistences, our findings are backed up by some scholars. Chen et al. (2021) posit that the location of a firm is vital in the learning process. Based on our study, this is due to continuous interactions and knowledge exchanges that take place.

Hypothesis 5 findings show a significant relationship between inter-firm ecosystem and firm adaptability (β = 0.278; p-value = 0.000). This is consistent with our theoretical foundation that ecosystems are complex webs with many actors who depend on each other in uncertain situations to evolve (Goldstein et al., 2010). They adopt the same working norms and communicate with partners via social network platforms as well as holding physical meetings to agree on price changes (mean = 4.803). The use of both media platforms and meetings was supported by Barratt and Barratt (2011) and Ganbold et al. (2020). The findings reveal that there is interdependence among firms (mean = 4.511). This is in line with Garces Rivera and Pfeiffer (2018) and Goldstein et al. (2010) who found that interdependence is the source of survival in an ecosystem. This reduces opportunism as firms embrace shared objectives (Kanyoma et al., 2018; Goldstein et al., 2010).

Furthermore, ecosystem partners learn from experiences shared during meetings and via social media platforms enhancing their ability to adjust their attitude towards supply chain operations. It is believed that involving external players in market analysis puts firms in a better position to quickly adjust their schedules in response to market needs. These findings resonate with Scholten et al. (2019) who posited that as firms interact, they learn from experiences and create a sense of certainty in their operations. Also AlHares et al. (2018) posit that better decisions are made after meetings.

In addition, FPFs have to understand that though suppliers are not “the spider in the web” connecting all players in the supply chain, they determine the success of the ecosystem (Garces Rivera and Pfeiffer, 2018). Despite that, coffee farmers have been sidelined and their involvement has been limited to the early stages of coffee farming. This hypothesis could be as a result of the many players in the sector (Waiswa et al., 2021; Goldstein et al., 2010). The more the trading partners, the more adaptive the players as they share strategies (Garrido-Vega et al., 2021; Goldstein et al., 2010).

Hypothesis 6 shows a significant and positive influence (β = 0.186; p-value = 0.000). Accordingly, though asset specificity has a positive influence on SCI (β = 0.016; p-value = 0.804), its impact is made significant through firm adaptability which implies that a partial mediation was established. The findings reveal that though asset specificity increases transaction costs and limits sharing of information and other resources, in uncertain situations, firms have no option but to invest in specific assets and work with their external partners. They also take advantage of their already existing resources to remain adaptive to the market and get the necessary market information. According to Bhardwaj and Ketokivi (2020) and Palacios et al. (2014) such adaptive skills and competencies put a firm in a better position to solve the day-to-day market challenges. Akidi et al. (2022) posit that training reduces partner bias allowing sharing of information.

In addition, the results reveal that asset specificity impacts SCI by improving coordination, monitoring of inventory through their trained staffs and being able to swiftly adjust processes to meet the market needs. These assets put firms in an adaptive position as they critically analyze the supply chain opportunities (Aslam et al., 2018). Furthermore, they make supply chains flexible (Wang and Wei, 2007). Being flexible, allows sharing of information (Despoudi et al., 2018). During training between JESA and its suppliers, there is a lot of interaction that takes place (Toye, 2012).

Hypothesis 7 findings revealed the existence of a full mediation of firm adaptability on the effect of inter-firm ecosystem on SCI (β = 0.184; p-value = 0.000). This implies that for inter-firm ecosystem to have an influence on supply chain integration, it has to be through firm adaptability especially in uncertain situations. These results reveal that firms are making efforts to involve external partners in most of the supply chain activities, encouraging them to belong to associations, putting in place policies and standards and creating an environment where support is given to them in case of any challenge to improve interactions and sharing information especially in uncertain situations. Waiswa et al. (2021) believes that policies aimed at reducing the numbers of players in the sector need to be adopted in hard economic times.

The outcomes of this research provide critical insights into the present literature on asset specificity, inter-firm ecosystem and firm adaptability on SCI.

The study contributes to previous scholars of SCI by using the TCT and CAST because few studies have taken SCI to be a complex concept as well as a TCT concept. The study confirms recommendations by Song and Song (2021) that indeed SCI can be looked at as a TCT concept. Integrating the firms’ operations comes with costs and the complexity created by investing in specific assets and bring in new firms can be managed when firms are adaptive. This study is the first to design a model that combines the three independent variables to study supply chain integration.

From our findings, we believe managers need to focus on adaptive strategies given the uncertain nature of their operating environment. There is a need for joint collaboration between the players in the sector to increase investment in specific assets and improve capacity utilization. This will reduce supply chain costs and also achieve better forecasts. Managers are warned that developing a strong ecosystem culture where rules, norms and standards exist does not foster SCI probably because it breeds grounds for opportunistic tendencies. Therefore there is need to create an environment where opportunism is minimized and this could be through trust.

To policy, more government efforts are needed to achieve shared outcomes. The FPS contributes to the economy by feeding over 34.9 million to 40.3 million Ugandans. Government needs to come up with policies to ensure that all players in the sector are registered in their ecosystems to ease coordination and extend support to them. This will allow them to share information and achieve production at full capacity to feed the growing population. Government should also come up with fair policies that boast the small and medium firms in the sector because these have been identified to dominate the sector and they employ more people than the large firms. With this strategy, government will be able to address the rampant unemployment in the country.

From our results, all direct relationships were positive except the relationship between the inter-firm ecosystem and SCI. This research concludes that there could be other factors that could strengthen our results like creating an environment where trust thrives and reducing opportunism as these build confidence to invest in specific assets. Williamson (2010) found that investing in specific assets requires that firms develop relationships based on trust.

Our study comes shortly after the COVID-19 pandemic that may have affected investment in human specificity and dedicated specificity. This requires investigation after economies normalize. During the pandemic, investing in human skills to run the installed systems was necessary. The study found that inter-firm ecosystems could have an influence on investment in the firm’s specificity assets. Therefore other scholars could try and investigate this in a similar context. Lastly, this study took a quantitative approach that focused on perceptions, future studies can take a qualitative approach to get deeper insights into the factors that impede ecosystems from thriving in the FPS. There is a need to get a deeper understanding of why the level of investment in specific assets remains low in the FPS. SCI is identified as a behavioral variable that needs more time to be studied meaning that a longitudinal study must have been the most appropriate. Thus, a longitudinal survey may be used to examine the impact of the independent variables on SCI.