This section is not mandatory, but can be added to the manuscript if the discussion is unusually long or complex.
5.1. Future Research Directions Derived from Individual Clusters
Cluster 1 (short ASCs) contains studies almost exclusively from the UK (e.g., Marsden et al., 2000; Ilbery and Maye, 2006) and EU countries (e.g., Renting et al., 2003; Blundel, 2002), while studies from other developed countries and developing countries are few. This predominance of studies from the UK and EU may be because short/alternative ASCs are not as developed in other areas than the UK and EU. However, this does not mean that there are no alternative ASCs in developing countries. With changes in consumer demand, different types of short ASCs are emerging in developing countries. Studies on alternative ASCs in developing countries may be a future research direction for this cluster.
Cluster 2 (sustainability in ASCs) focuses on the introduction of sustainability practices in large food companies (i.e., Nestlé) [109
] and on the sustainable performance/sustainable value chains of ASCs using the LCA method or scenario analysis [51
]. However, few of these studies have a theoretical basis. The one exception relies on agency theory [53
Therefore, the first future direction for sustainable ASCM researchers is to integrate theory, i.e., the resource-based view (RBV), resource dependence theory, dynamic capabilities, agency theory, social network theory, and organizational learning theory, into their work. For example, a recent study by Wilhelm et al. (2016) studied the double agency role of tier 1 suppliers (who serve as the agent for focal firms and the principle for sub-tier suppliers) of a focal company in a multi-tier SC using four cases in different food sectors, extending agency theory to multi-tier SSCM.
Second, although there are articles discussing certification and standards adoption in sustainable food SCM [111
], there is a lack of detail regarding how to implement sustainable initiatives in ASCs, i.e., detailed recommendations for how to incorporate sustainable criteria into supplier selection and development and discussion of how the adoption of certification is linked to firm performance. In the SSCM literature, two methods are being used to investigate certification adoption and firm performance. One is content analysis of sustainability reports [113
] and the other is secondary data analysis [114
]. Future research can adopt either but should focus on agricultural sectors by examining the effects of adoption of fair trade or Forest Stewardship Council certification on firm performance using secondary data analysis or the content analysis method.
For Cluster 3 (ASC modeling), we identified five future research directions based on the research gaps that we found. First, the uncertainties from new types of risks need to be taken into account when conducting research on ASC modeling for risk optimization. For example, macro-level risks, such as environmental risks and policy risks, which have a substantial influence on SC performance [115
], need to be addressed. Our review reveals that current studies focus more on modeling micro-level risks, such as demand management uncertainty [116
], supply management uncertainty [117
], production management uncertainty [118
], information management risk [119
], and food safety uncertainty (Wang et al., 2009), while macro-level risks have not been fully explored. Attempts to model weather risks have been made [72
], but attempts to model policy uncertainties have not. Therefore, a future direction for this cluster may be modeling farmers’ adaptations to climate change, such as farm-level decisions on production, selling, purchasing, and storage under different weather conditions, and modeling agricultural structural change under policy uncertainties, particularly in developing countries, where policies governing the agricultural sector (i.e., subsidy and environmental protection policies) change constantly.
Another risk that has been neglected is endogenous uncertainty related to collaborative activities [121
]. Specifically, uncertainties may arise from the opportunistic behaviors that occur during collaborations among supply chain stakeholders. For example, in the agri-food system, farmers in developing countries may be more likely to fail to fulfill their contracts due to the small-scale farming-dominated supply structure and fluctuating market price [87
], which makes the agricultural system more complex than the manufacturing sector. The only relevant paper we found was that by Burer, Jones and Lowe [72
], which examines contract practices between suppliers and retailers in the seed industry. Therefore, modeling small farmers’ behavior under different types of buyer-supply relationships (i.e., short-term contract, long-term contract, strategic alliance, and vertical integration) would be a useful research aim and would help cooperatives/firms/intermediaries to make decisions on the types of relationships they establish with small farmers in developing countries.
The second future research direction concerns the methodologies/models used in risk optimization. The existing research has adopted stochastic programming [122
] and robust programming [124
] to optimize agricultural decision-making under uncertainty. The models commonly applied include TSP, SP, and fuzzy-elements-added models; however, dynamic and stochastic processes are rarely addressed simultaneously [117
] using stochastic dynamic models (SDP) or multi-stage programming (MSP). Therefore, models such as SDP and MSP should be applied in the future to further relax the assumptions when designing models, assuming the uncertainty of both the stochastic and dynamic dimensions. For example, when modeling farm-level operating decisions, SDP or MSP can be adopted to optimize the decision by considering weather change/yield risk (stochastic dimension) and the price fluctuation/dynamic behavior of contracted farmers (dynamic dimension) simultaneously.
Third, it is found that the current literature proposes single-objective models for the related logistical problems [126
], while companies must actually balance multiple objectives in logistics management. Moreover, these objectives may be in conflict with each other, for example, profit vs. sustainability and quality vs. cost. Therefore, researchers need to develop multi-objective programming models to handle decision problems in logistics management.
Fourth, quantitative modeling methods for supply chain traceability [127
] and food waste in supply chain management [126
] are still lacking, despite the fact that there are many issues to be solved is these two areas that require modeling methods; examples include how agri-food firms determine the optimal level of investment of RFID/IT technology [66
] and the design and operation of pack houses to increase traceability [129
The last future research direction could be to extend the modeling from the supply chain perspective to the supply network perspective in the agricultural sector, and such research (supply network modeling) has been performed for manufacturing supply chains [130
] using social network analysis. The concept of a supply network has been noted in agri-food supply chain studies [133
]; however, few modeling studies adopting a supply network perspective have been performed. A recent paper [134
] develops a risk propagation model for agri-food supply chain risk management based on the susceptible-infected-remove (SIR) model, which models the evolution process of risk in a supply chain network. It is argued that an agri-food supply network has a large number of inter-connected nodes and is much more complex than a supply chain, and modeling of an agri-food network better captures reality than modeling of an agri-food supply chain. Therefore, more research could be conducted to model risks in the agri-food supply network.
It appears that Cluster 4 (Global ASC) is a well-developed research area in ASC studies, which means that a broad range of topics have been discussed using various methods. The effects of food safety certification have been extensively researched. However, it is not clear what the detailed mechanisms of global ASC management are. Future research can examine the role of agricultural cooperatives in managing the global ASC [135
]. Due to the difficulties of collecting data on social sustainability, the effects of the global ASC on women and minority groups remain under-researched [88
]. More research in this direction is warranted.
The themes of Cluster 5 (food safety/traceability) are of great importance in ASC studies; however, there are few articles in this cluster, and it is in a state of inertia. The mechanisms by which SC coordination affect ASC traceability are not well understood. Some authors (e.g., Lindgreen, 2003) began focusing on the role of trust (as a coordination mechanism) in shaping ASC traceability and transparency. Future research can explore the detailed relations among coordination mechanisms, SC traceability/traceability, and food safety.
Cluster 6 (ASC relationships) is a research area that continues to receive attention in ASC studies. However, we observe that resilience in ASC is under-explored. Within agricultural SCs, risks are inherent and varied due to a range of factors, including current climate sensitivity, the sensitive nature of biological processes, the complex structure of the industry, the pronounced seasonality of production and adverse changes in market prices, geographical separation between producers and end users, and the unique social and economic uncertainty of food and agriculture sectors, both domestic and international [138
]. Creating more resilient SCs may provide a better approach to managing and mitigating such risks and challenges facing businesses today and in the future. SC resilience encompasses the ability to prepare for unforeseen disruptions and to respond to, and recover from, them better than competitors do [139
]. Most of the studies on SC resilience have been conducted in a non-ASC context. In future, more research on ASC resilience should be conducted.
5.2. Future Research Directions Based on General Categories
In addition to the directions for future research derived from the six clusters, further directions are derived from general categories that cut across all the 188 articles reviewed and have implications for future research (Table 9
After reviewing the 188 articles, we found they tend to belong to the disciplines of operations management/supply chain management (OM/SCM), agricultural economics and food (AE), development studies, and operations research (OR). The studies also tended to use disciplinary methods (LCA, modeling, and case studies); the remaining disciplinary silos and the number of cross-disciplinary studies were small (only 9 articles in total). There is a need for a cross-disciplinary approach combining the strengths of individual disciplines. For example, OM and OR methods are complementary and can be used in the same ASC study. The case study methods of OM research and the econometric methods of AE research can be integrated to obtain more robust results and insights.
Regarding research methods, ASC studies are dominated by case studies and conceptual framework development research. Though such studies provide abundant case-based evidence of ASC practice, quantitative research methods such as surveys or secondary data analysis are also needed to provide statistical data to test hypotheses. Modeling is widely adopted by OR researchers in the ASC field. Longitudinal studies are rarely found in ASC research (only 9 articles in total used longitudinal data). Our literature review reveals that the dynamic evolution process of implementing ASC initiatives based on a multi-stakeholder approach offers additional insights into the adoption of ASC initiatives. Chris, et al. [141
] advised that longitudinal case study research can yield powerful in-depth insights (though limited in generalizability) and can address the problems caused by other methods, as the longer the period in which the phenomena were studied, the greater the opportunity there is to observe the sequential relationships of events, which is ideal for theorizing. Therefore, exploration of food SCM using longitudinal data will create new insights for ASC in the future.
Regarding the geographic areas of the studies, current ASC studies are concentrated in developed countries, including EU countries (45 articles), the UK (42), the USA (12), Australia (3), New Zealand (2), and multiple developed countries (9). The total number of articles that focused on developed countries was 129 (56.1%). The number of articles that focused on developing countries was 50 (21.7%, including both single country and multiple country studies). Fifty-three studies did not provide country information. Given the large number of developing countries, the percentage of studies devoted to developing countries and published in international journals was low. Developing countries that received the most attention in ASC research (only single-country studies were counted), included Kenya (5), South Africa (4), Indonesia (3), China (2), Turkey (2), Brazil (1), and India (1). It is clear that more ASC research needs to be conducted in developing countries in all six clusters.
Regarding the unit of analysis, ASC studies tend to be focused on small farmers/producers [1
], large companies [110
], consumers [43
], institutions, and government [50
]. There is a lack of research adopting the ASC as a whole as a unit of analysis. In general SCM research, multi-tier SCs are a new research topic [143
]; future ASC research should increasingly adopt whole chain or whole supply networks as a unit of analysis to obtain a comprehensive view of ASC topics.
ASC studies tend to be atheoretical (without theoretical basis) and are primarily descriptive; they provide empirical evidence for SCM studies using case studies/conceptual building/modeling (articles adopting case study methodology tend to be descriptive and are not theory-building-driven). Thus, the application of high level theories of the firm (e.g., RBV, Transaction Cost Economies (TCE), and agency theory) to develop middle-range theories in SCM [144
] remains rare. Future research should link the topic/phenomenon to grand theory and strive to develop and extend grand theories to an ASC context.
Finally, several emerging general SCM research topics are under-represented in ASC research but might be interesting to investigate in the future, e.g., ASC power dynamics and supply network analysis, the institutional environment of the global ASC, ASC risk management and resilience and the adoption of e-commerce in ASC. The incorporation of e-commerce into ASCs in China presents a good example of how the internet helps smallholder farmers obtain access to the global market [145
]. In the future, some of the general SCM research topics should be closely examined in an ASC context.