The role of public policies in the digitalisation of the agri-food sector. A systematic review

ABSTRACT

This paper examines the roles that public policies should adopt in the process of digitalisation of the agri-food sector according to existing knowledge. For this purpose, the authors examined the rationale of the public policies’ involvement from three perspectives: the justifications for policy interventions, what should be accomplished with the policies and how the policy interventions should be conducted. A systematic review of the literature enabled the identification of the two main interrelated roles of public policies. First, public policies have an important role in ensuring equitable digitalisation of the agri-food sector. In this role, they are expected to protect the public interests, the interests of farmers and other supply chain actors that have weak bargaining power or are less organised. The other main role of public policies is to be receptive to the opportunities presented by the development and implementation of digitalisation and to enable necessary innovation. The interrelation of the two roles underlines the importance of encouraging innovation to help the agri-food sector in meeting major challenges while avoiding the unintended consequences of the technological development for weaker parties. The findings of this paper serve as a basis for further research on the specific characteristics that governments and policymakers should possess to operationalise the expected policy roles.

KEYWORD:

• smart farming
• digital agriculture
• public sector
• food systems
• digital transition

1. Introduction

In the succession of changes to the ways in which our food is produced, we have entered a phase dedicated to the digitalisation of the agri-food sector. Digitalisation of the economic sector means integrating digital technologies into its processes to make lasting changes to the implemented practices (Leviäkangas, 2016). In the context of the agri-food sector, digitalisation refers to the trend of using digital technologies and data created on-farm (e.g. by machines and sensors about the status of soil, crops, animals and work processes) and off-farm (e.g. economic information, transactions, weather data and satellite data) in the decision-making process of the stakeholders such as farmers, input and output suppliers and policymakers (Klerkx et al., 2019). Therefore, digitalisation of the agri-food sector should not be considered as a process that involves only individual production units, but rather spans over multiple segments of the value chain and ultimately creates new links between the actors of food systems (Shepherd et al., 2020).

Digitalisation of the agri-food sector is generally viewed as a positive means to strive towards sustainable development goals. The technological development is expected to increase the efficiency of food production at the firm and value chain levels (Klerkx et al., 2019) and provide social and environmental benefits in food systems (Rose & Chilvers, 2018). These expectations are also vividly reflected in the policy documents envisioning potential futures of agriculture, food production and food systems (Barrett & Rose, 2020; Lajoie-O’Malley et al., 2020). For example, the use of emerging technologies could help to achieve stronger links between farmers and the end-consumers (Elijah et al., 2018; Shepherd et al., 2020) and in general improve the transparency of the food value chain (Fielke et al., 2020; Lesser, 2014). Similarly, farmers’ adoption of data-based technologies has the potential of improving their market power in relation to the input suppliers of the value chain (Pham & Stack, 2018).

Several possible problems and risk areas have been highlighted alongside the potentially positive effects of digitalisation. According to the largest umbrella organisation of farmers in the EU, increased data exchange will raise several challenges in the domains of “privacy, data protection, intellectual property, data ownership, relationships of trust/power, storage, conservation, public data, and usability” (Copa-Cogeca, 2016). Achieving seamless and effective interoperability between technologies of different manufacturers, guaranteeing data security and thereby increasing trust remain a challenge (Busse et al., 2014; Tzounis et al., 2017; Wolfert et al., 2017). The lack of interoperability between different machines, equipment and software places farmers at the risk of being locked into a relationship with one specific hardware or software developer without the real possibility to change to another system or aggregate data from different systems (Rotz et al., 2019). In addition, data infrastructure and connectivity in farms need to be further developed and geared towards sharing larger amounts of data (Kritikos, 2017). Upgrading farmers’ skills to become “informed data consumers as well as co-creators and curators of data” to utilise the potential of data and information is crucial for better decision-making (Jakku et al., 2019).

In its efforts to ensure the agri-food sector’s compliance with societal demands (reduced environmental pressure, improved product and process traceability, etc.), public policies are expected to encourage technological development (Busse et al., 2014; Reichardt & Jürgens, 2009; Sørensen et al., 2010). The policy documents of the Food and Agriculture Organisation of the United Nations (FAO), the Organisation for Economic Co-operation and Development (OECD) and the World Bank view the adoption of new technologies in agriculture as a way to increase the productivity of the agri-food sector while also mitigating its environmental impact (Lajoie-O’Malley et al., 2020). In aiming to reduce bureaucracy (Lawson et al., 2011) and overcome market failures (Eastwood et al., 2017), the public policies could promote the availability of new technologies and mitigate excessive market concentration of technology providers (Birner et al., 2021). Public policies are also major contributors to well-functioning innovation systems by assuring quality regulations and institutions concerning data ownership, privacy and liability (Busse et al., 2014; Eastwood et al., 2017; Jakku et al., 2019; Regan, 2019; Wiseman et al., 2019); they support delivering upgraded research and extension services (Eastwood et al., 2017). New technologies have significant potential in terms of measuring and comparing the results of different agricultural and environmental practices and policies in the agri-food sector and assessing the progress of moving towards set goals (M.-H. Ehlers et al., 2021), a notion that is also reflected in several strategic initiatives of the EU, for example, Farm to Fork Strategy (European Commission, 2020a).

As previously suggested, a significant body of knowledge on the relationship between public policies and digitalisation of the agri-food sector has been created in recent years. However, several uncertainties remain. More knowledge is needed about the ways public policies can and should protect the public interests in the process of digitalisation (Fielke et al., 2019; Klerkx et al., 2019; Regan, 2019). There is a need for more knowledge about the ways agricultural policy may affect technological adoption (Groher et al., 2020) and how to sustainably balance the interests of the stakeholders with different characteristics (e.g. small and large farms, numerous farms and concentrated processing industry; Klerkx et al., 2019). Despite the significant potential digital technologies have in the policy process, several missed opportunities remain (M.-H. Ehlers et al., 2021). The use of digital technologies is commonly decided on an ad hoc basis and lacks systematic evaluation of the new opportunities on offer (OECD, 2019). In line with this, studies have revealed that the ways digital technologies impact the policy process (e.g. in monitoring and surveillance) need further investigation (Fielke et al., 2019; Jakku et al., 2019).

The common thread in the aforementioned knowledge gaps can be defined as a lack of knowledge about the roles that public policies should play in the process of digitalising the agri-food sector.¹ This is assuming that a “role” is understood as something that can be studied from three perspectives: justifications for, aims of, and ways of carrying out policy interventions. Reducing the lack of information about the role of public policies and studying it from three distinguishable perspectives offers an opportunity to adopt a structured approach to analyse the expectations regarding the relationship between public policies and digitalisation of the agri-food sector in the existing literature. Reviewing, summarising and synthesising this existing knowledge using the systematic review methodology further adds to the integrity of the research by helping avoid the exclusion of any relevant information. This kind of comprehensive and structured approach has been missing from the thematic literature so far. Therefore, this study seeks to help resolve some knowledge gaps and add a new angle to the conversation. This study examines the following research questions (RQs) about the expected role of public policies as reported in existing academic literature: 1) What are the justifications for policy interventions? 2) What should be done via policies? 3) What should be the ways for carrying out the policies? Furthermore, the results of the three RQs were used to identify further knowledge gaps regarding the role of public policies in the digitalisation of the agri-food sector.

2. Materials and methods

According to Munn et al. (2018), a systematic review is a “robust, reproducible, structured critical synthesis of existing research”. This definition is complemented by Higgins and Green (2008), who argue that the aim of a systematic review is to “seek to collate all evidence that fits pre-specified eligibility criteria in order to address a specific research question”. According to Boaz et al. (2002), the main difference between a systematic review and common reviews is that the former must be “carried out to agreed standards”, which means that a review protocol must be compiled in the beginning and followed throughout the research. According to Davies and Crombie (1998), as referenced by Tranfield et al. (2003), a review protocol is a “plan that helps to protect objectivity by providing explicit description of the steps [taken during the research]” and the protocol is composed of RQs, sample descriptions, search strategy of sources and criteria for including and excluding sources.

Following Tranfield et al. (2003), a preliminary scoping study was conducted to decide on the criteria that should be used to determine the sample of the reviewed papers and keywords to be used. The study included a general search and screening of the academic literature on socio-economical aspects that are involved in the digitalisation of the agri-food sector involving studies from the fields of agriculture, sociology, economics, innovation, ethics and public policies.

Based on the scoping study, we decided to use three criteria to determine the sample of the reviewed papers. Only peer-reviewed papers and publications in English were included. Furthermore, the scoping study highlighted the need to establish a criterion that would reflect the societal context considered in the papers included in the sample. The preliminary study indicated that when the involvement of public policies was scrutinised in the context of non-democratic countries, the expectations for the state involvement in the market relations and process of development of regulations and policies are significantly different than in the context of democratic countries. The need for a criterion recognising societal context was further enforced by the extensive nature of the three perspectives that have been used herein to investigate the rationale of the public policies’ involvement (justifications for involvement, aims of the policies and ways for carrying out the actions); each of the perspectives is very much dependent on societal agreements about the divisions of the tasks between the public and private sectors, scope of policymakers’ responsibility and so on. For these reasons, only articles analysing the situation in the “electoral democracies” as assessed by Freedom House were included in the sample (Freedom House, 2020).

The preliminary scoping study revealed that the keywords of the papers that study or touch upon the relationship between public policies and digitalisation of the agri-food sector can be divided into three groups related to innovation, technology and the agri-food sector. Furthermore, the extensive character of the RQs also dictates the need to include a wide array of keywords in the search strings. Accordingly, three sub-types of keywords were used in the search string: 1) Innovation, 2) Technology and 3) Agri-food sub-types (see Table 1). The composition of keywords under each of the sub-types was determined based on common terms used in the papers that were screened in the preliminary study. Altogether, the wide range of the search string (3 sub-types and 14 keywords) helped to ensure that any relevant papers were not missing from the search results. The RQs implicitly dictate the need to also identify a sub-type that would relate to the public sector, policy and governance. During the preliminary study, it was discovered that the bulk of papers identifying public policies as relevant elements in the process of digitalisation did not include public-policy-related concepts in the title, abstract or keywords. Therefore, findings related to public policies may often have a secondary nature from the perspective of the sources included in the sample, which did not decrease the value of the findings from the perspective of the current study.

Table 1. Keywords based on the sub-types

Table

Agri-food sub-type	Innovation sub-type	Technology sub-type
Agri-food Agriculture E-agriculture Farming Food processing Food production	Digital innovation Digitalisation Innovation Innovation system Open innovation	Digital ICT Information and communications technology

Scopus and Web of Science databases were used to identify relevant sources. According to the key word sub-types, the following search string was used in Scopus and its analogue in a different format in Web of Science to identify relevant papers:

(TITLE-ABS-KEY (innovation* OR “Innovation system” * OR “Digital Innovation” * OR “Open innovation” * OR digitalisation*) AND TITLE-ABS-KEY (digital* OR “Information and communications technology” * OR ict*) AND TITLE-ABS-KEY (agri-food* OR agriculture* OR farming* OR e-agriculture* OR “Food Production” * OR “Food processing” *)) AND DOCTYPE (ar OR re) AND PUBYEAR > 1950 AND (LIMIT-TO (LANGUAGE, “English”)).

The search was conducted on 10 November 2020 and, after the removal of duplicates between databases, produced 316 findings. The subsequent process towards the final selection of papers is summarised in Figure 1.

Figure 1. Process of compiling the final selection of papers.

First, titles and abstracts were screened to eliminate papers that clearly had no relation to the analysis of the role of public policies and that dealt with non-electoral democracies.

After this, 106 papers remained, and the researchers read their full texts. Papers were included that address the relationship between public policy and the digitalisation of the agri-food sector by referring to the tasks of public policies (employing terms like “policy interventions”, “funding schemes” and “regulations”) or by referring to the tasks of the entities responsible for developing and implementing public policies (employing terms like “governments”, “policymakers” and “regulators”) or by highlighting ways that entities are subject to public policies (e.g. public extension suppliers). The second type of papers that was included addressed specific shortcomings regarding digitalisation that according to common knowledge should be addressed by public policies (e.g. “need for trusted information and advice networks”, “require strong, science-based evidence” and “surveillance”). In addition, the process of eliminating papers dealing with non-electoral democracies continued. At least two researchers read each of the papers and in case there were disagreements on whether to include a specific paper, at least three researchers were involved in the discussion and the final decision. Finally, the initial papers were also screened for relevant references, which were then added. A total of 47 papers remained, which were then analysed to answer the RQs.

The yearly distribution of the papers in the initial search and final selection is heavily skewed towards years 2017–2020, which reflects the increase in relevance of the topic of digitalisation of the agri-food sector in academic literature (Figure 2). Furthermore, the fact that the proportion of the articles of the last few years is comparable in the initial search results (rather generic search terms) and final selection (articles specifically related to public sector role) is an indication that the relevance of the topic of public policies in digitalisation has been increasing at the same pace as that of the topic of digitalisation in general.

Figure 2. Yearly distribution of papers in the initial search and final sample.

In parallel to identifying the relevant papers, relevant data were identified and cleaned. The first task was to identify the paragraphs of relevance in each of the papers included in the final selection. At this stage, the approach was to include as much data as possible. Extracts that would directly or indirectly refer to public policy and its relationship with the process of digitalisation of the agri-food sector were produced from every paragraph. Second, based on the respective RQs the specific extract was addressing, the list of extracts was divided between the RQs 1–3. Some initial extracts were applied to multiple RQs simultaneously. In this case, the extracts were further split so that specific sentences would be best matched with the respective RQs. At the end of this stage, the researchers had 69 extracts, 23 of which were assigned to RQ1, 18 to RQ2 and 28 to RQ3. Third, to facilitate the search for common denominators between the extracts, the core idea of each of the extracts from the perspective of the previously assigned RQ was summarised in 1–3 sentences (see, Tables 1, 2 and 3 in the Annexe for the detailed overview). Finally, these condensed extracts were aggregated according to a common theme they address (thematic coding). In this process, the emphasis was on creating aggregations that would address the specific RQ as concretely as possible. The aim was to strike a balance between generalisation and specificity that would enable the researchers to review the findings while also maintaining relevance. This process produced 14 common themes (five for RQ1, six for RQ2 and three for RQ3), which are introduced and discussed in the following sections.

3. Results

This study scrutinises the roles that, according to existing knowledge, public policies should adopt in the process of digitalisation of the agri-food sector. For this purpose, the authors examined the rationale of the public policies’ involvement from three perspectives: the justifications for policy interventions, what should be done via policies (what policy interventions are needed) and how should policy interventions be carried out. A systematic review of the literature allowed the authors to identify common threads from the three perspectives, which enabled them to generalise the expectations placed on the public policies. This search for common denominators resulted in the identification of two main interrelated roles of the public policies in the digitalisation of the agri-food sector: the role of protector of weaker parties and that of enabler of innovation to develop and deploy new technologies (Figure 3). The nature of both roles is introduced in the following sections. While the two roles have been presented separately here, they are significantly interrelated. This means that the individual justifications, aims and ways for carrying out the policies can support the fulfilment of both roles in some cases, which is consistent with previous research on the similar topic (Borrás & Edler, 2020). The decisions to include the perspectives of policies under one role or another were made based on the focus of the paper from which the specific perspective was found, which does not necessarily exclude the impact on the other role. For example, while the need to facilitate dialogue between the stakeholders is presented as part of the effort to enable innovation, this surely also contributes to the role of protecting the interests of weaker parties.

Figure 3. Roles that public policies should fulfil in the digitalisation of the agri-food sector.

3.1. Protector of weaker parties

One of the main recurring sentiments of the reviewed papers is that public policies have an important role of ensuring equitable digitalisation of the agri-food sector. Policies are expected to protect public interests, the interests of farmers and other supply chain actors that have weak bargaining power or are less organised.

3.1.1. Justifications for interventions

Risks of imbalanced progress. Digitalisation increases the risk of divisions in the agri-food sector due to differences in the speed of technological adoption, access to the benefits of digitalisation and potentially damaging effects of some technologies. Public policy interventions are necessary to improve access to new technologies (Bellon-Maurel & Huyghe, 2017; Carmela Annosi et al., 2020). For example, access to technologies that have the potential to reduce the environmental impact of the agri-food sector (Von Braun et al., 2017) and tackle the climate crisis (Radulescuu et al., 2019). At the same time, public policies should also prevent potentially negative effects of the new technologies and intervene to change the technology trajectory if necessary (Barrett & Rose, 2020). In the context of implementing a “precautionary approach” to the technology development, the lessons from the overly technology focused approach of Green Revolution are cited (Avaria, 2020). Interventions are also needed to avoid digital divides (Klerkx et al., 2019). In this respect, Barrett and Rose (2020) highlight that farmers are the ones at risk of missing out on the benefits of digitalisation. Shepherd et al. (2020) conclude that more equal sharing of benefits serves as a prerequisite to increase confidence in technologies.

3.1.2. Policy interventions

Support data quality, integrity and availability. A fair level of data accessibility, security and integrability should be treated as a prerequisite in protecting the society and farming sector’s interests in the digitalisation of the agri-food sector. Consequently, public policy interventions are needed to support the bargaining power of farmers and maintain transparency, for example, when agreeing on the terms of the relationship between service providers and farmers (Fielke et al., 2020). The relationship between the expected data privacy of the farming community and the interests of the society needs deliberation, as farm-level data can be used to protect societal interests chiefly when designing and implementing environmental policies (Bronson, 2019).

Encourage sustainable adoption of new technologies. Public policies should support the adoption of new technologies by improving the availability of open-source tools (Rotz et al., 2019), promoting open data solutions (Adamashvili et al., 2020; Von Braun et al., 2017) and encouraging the development of more “simple and practical technologies” (Bucci et al., 2018). Further support for the intensification of R&D activities is necessary to improve the availability of technologies that would improve sustainability and suit the requirements of current agricultural systems (Chuang et al., 2020). Furthermore, excessive power asymmetries in the value chain should be regulated as they usually exist to the detriment of farmers and may cause lagging technology adoption at the farm level (Jakku et al., 2019).

3.1.3. Ways for carrying out the policy interventions

Balanced policies. According to existing literature, considering socio-economic limitations and aiming to provide benefits for all the relevant groups should help ensure that the policies consider the interests of farmers and wider society fairly. For this, ex-ante assessment of the development pathways of the digital transition is needed to avoid potential negative impacts on the ecosystem services (Lajoie-O’Malley et al., 2020) and include less represented groups such as e.g. agroecological farmers (Rijswijk et al., 2019; Rotz et al., 2019). In general, policymakers should ensure that the interests of the “high adopters” are not disproportionally represented in the planning and execution phases of policies (Groher et al., 2020). Different root causes for the adoption and/or non-adoption of the technologies and the inclusiveness and fairness of the digitalisation should be considered in policy development, for example, characteristics of the users and technologies (Bronson, 2018; Giua et al., 2020; Li et al., 2019; Rotz et al., 2019; Yoon et al., 2020). It is also important not to focus only on technical innovation; involved stakeholders also need to invest in social innovation. Responsible Research and Innovation (RRI) framework could be used as a guiding principle that would allow the facilitation of an “equitable digital farming transition” (Bronson, 2019).

Effective dissemination of information and knowledge. Fast, up to date, accessible and applicable agricultural knowledge and information systems should help in establishing a fair digitalisation of the agri-food sector, and public policies have an important role to play in making this goal attainable. It is necessary to support the training and education about new technologies (Yoon et al., 2020). This could be done by integrating the relevant topics into the curriculum of vocational schools and universities (Bellon-Maurel & Huyghe, 2017). Dedicated policies are needed to facilitate the improvement of the skills of farmers that are otherwise at risk of being left behind (Barrett & Rose, 2020), especially regarding data processing and interpretation skills (Jakku et al., 2019). New technologies have the potential to improve the acceptance of agriculture in the eyes of the public, for example, by improving animal health and welfare, but it takes specialised communication skills of the stakeholders to communicate this benefit (Pfeiffer et al., 2020), and public policies should help in improving these skills of the farming sector. Public policies should encourage the implementation of Information and Communications Technology (ICT) solutions for innovation diffusion (Nelson et al., 2019; Norton & Alwang, 2020). This should go together with support for the dedicated efforts of the innovation brokers, for example, private or public extension services that aim to create communication between technology companies and technologically less advanced farms (Lioutas & Charatsari, 2020). Improved access to information and advice should be viewed as a way to increase the adoption of new technologies (Knierim et al., 2019); this would help justify more ambitious policies dedicated to reforming innovation diffusion. While the policy framework should facilitate faster dissemination of information about new technologies (Tiwari, 2008), it is important to adapt to changing circumstances. One of the trends to consider is the likely decrease in the role of the state in the process of knowledge transfer, which is to some extent replaced by the increasing importance of private sector initiatives like the farm media representing ICT actors (Relf-Eckstein et al., 2019).

Dialogue between the stakeholders. In line with the approach of RRI, the reviewed papers stress the need for public policymakers to act as neutral facilitators of dialogue to support cooperation between stakeholders. Such dialogue should be viewed as a prerequisite for establishing a fair process of digitalisation. A major topic that emerges is the need to develop inclusive collaborative platforms that would facilitate dialogue with citizens and other stakeholders around the digitalisation of the agri-food sector that serves as a prerequisite for establishing a balance in viewpoints represented (Fleming et al., 2018), enabling more ethical outcomes (Eastwood et al., 2017), ensuring sustainable transition (Fielke et al., 2019) and identifying sufficient synergies between the new stakeholders and agricultural expertise of the more traditional stakeholders (Ingram & Maye, 2020). More specifically, Devitt et al. (2019) describe the need to implement co-creation methods when designing new bio-surveillance systems. Carmela Annosi et al. (2020) point to the importance of public–private partnerships (PPPs) in overcoming several digitalisation barriers. Additionally, von Braun et al. (2017) refer to the need to enable international collaboration to adapt to the changing environment, and Relf-Eckstein et al. (2019) highlight the need to also involve policymakers from outside the agricultural policy field (e.g. concerning autonomous vehicles).

3.2. Enabling innovation

The other main role of public policies is to be open to the opportunities offered by the development and implementation of digitalisation and enable necessary innovation.

3.2.1. Justifications for policy interventions

Risk of lacking infrastructure and data regulations. Technological development presents a constant challenge to the suitability of infrastructure (e.g. the availability of last-mile connections and state-provided digital services) and regulations (e.g. access to data and fairness of data exchange). It is in the strategic interests of the states not to fall behind on the demands of the innovation process as it would inhibit the competitiveness and sustainability of the digitalisation process and decrease trust in technologies. Public policies are to provide data regulations on access and use of data that are clear and precise, which would allow the trust of the farming sector to be maintained (Relf-Eckstein et al., 2019) and encourage the adoption of technologies that expect data integration (Jakku et al., 2019). The lack of quality of the digital infrastructure is a bottleneck in the adoption of data-based technologies (Galliano & Roux, 2003; Jakku et al., 2019). This hindrance cannot be overcome without dedicated support from public policies and funds (Barrett & Rose, 2020; Bowen & Morris, 2019; Michels et al., 2020; Rotz et al., 2019).

Missing the opportunity to improve the competitiveness of the agri-food sector. Public policy interventions should view digitalisation as a game-changer that enables the improvement of the competitiveness of the agri-food sector by an order of magnitude, that is, a way to step further away from slow improvement or, in some cases, reverse regression. Enabling and promoting the new technologies support the increase in farm productivity and efficiency (Bowen & Morris, 2019; Ratten, 2018) and allow for diversification and internationalisation in the agri-food value chain (Bowen & Morris, 2019). To enhance technology development, regulations should aim to increase the competition between technology providers, which would allow the value of the new technologies for users to be improved (Bronson, 2019).

3.2.2. Policy interventions

Reform the process of interventions. Public policies are not only enforcers and facilitators in the relationship between the agri-food and technology sector, but they are also directly impacted by the developments and are one of the consumers of the new technological solutions. This is illustrated by the increasing presence of technology providers who manage an increasingly larger proportion of the farm data that is needed for the policy design and implementation. One of the impacts of the trend is that the “traditional technology transfer and monitoring/surveillance responsibilities become much broader” (Fielke et al., 2020), demanding respective reforms (Fielke et al., 2019; Relf-Eckstein et al., 2019) that would allow the impacts of digitalisation on topics like data ownership, curation and analysis to be renegotiated (Fielke et al., 2020). Furthermore, policymakers are expected to increasingly implement new technological solutions to streamline surveillance/monitoring functions (Devitt et al., 2019), that is, to assess different agricultural practices from the environmental perspective (Fielke et al., 2020; van der Burg et al., 2019) using satellite data to monitor crops (Negula et al., 2017) or big data solutions for data analysis (Carmela Annosi et al., 2020).

3.2.3. Ways for carrying out the policy interventions

Cross-cutting concepts. Klerkx and Rose (2020) and Klerkx and Begemann (2020) have promoted the use of mission-oriented innovation policies, that is, continually stimulating the innovation policy developments, moving in the desired direction while actively inhibiting those not deemed desirable. Pigford et al. (2018) suggest developing mission-oriented policies through the prism of Innovation Ecosystems thinking. This would allow for more heterogeneous digitalisation due to greater consideration of interdependence of relevant actors, which would allow for the existence of niche fields of innovation.

Policies according to the maturity level of innovation. There are suggestions that the policy approach should differ according to the maturity level of the innovation that is being targeted. On the one hand, some papers stress the need to nurture emerging technologies and innovations (Adamashvili et al., 2020; Lombardo et al., 2018). On the other hand, Barrett and Rose (2020) highlight the potential of picking the “low-hanging fruits first”, that is, support the adoption of mature technologies with immediate effect on food production instead of focusing on solutions with lower levels of technology readiness.

4. Discussion

This study demonstrates how public policies are credited in the academic literature with two interrelated roles in the digitalisation of the agri-food sector – protecting the interests of the weaker parties and enabling innovation. According to the literature analysed, the role of policies to protect the weaker parties is justified by the necessity of avoiding excessive divisions in the access to technologies and the benefits that they offer. To fulfil this role, public policies should aim to promote equitable data exchange and sustainable adoption of new technologies. For example, they should do this by supporting dedicated R&D activities as well as open-source and open-data-based technologies. For this purpose, well-balanced public policies are needed that would fairly accommodate the different interests and characteristics of actors, such as the reasons for the adoption or non-adoption of technologies. Furthermore, policy support for the effective dissemination of information and knowledge and a functioning dialogue between the stakeholders is a way to support a fair digitalisation process. According to existing literature, the “innovation enabling” role of public policy interventions entails the need to reduce the risk of inadequate infrastructure and outdated regulations becoming a bottleneck for digitalisation. Moreover, public policies are essential to fully reap the potential benefits of new technologies in increasing the competitiveness of the agri-food sector. To enable innovation, public policies should aim to implement new technologies in the policy process, for example, for surveillance and monitoring, which will allow for reform in the design and implementation of policy interventions. To enable innovation, policy interventions should be supported by a comprehensive innovation process, such as mission-oriented innovation policies, that allow us to nurture the emergence of early-stage innovations while also delivering already mature technologies with an immediate effect on the agri-food sector. Deriving from the results of this research, several aspects will be highlighted in this section that merit further discussion.

Avoiding excessive power asymmetries in the value chain. The findings from this research provide a basis for answering how public policies should avoid excessive power concentration in the agri-food value chain (Klerkx et al., 2019). Similar to previous research by Van der Burg et al. (2019), the results of this research demonstrate that public policy interventions are required to regulate the division of power in the agri-food value chain. The results of current research emphasise how interventions should do it and describe the approaches that the policies should take in their interventions. This is most prominently highlighted by the aims of the policy interventions that have been identified, that is, the need to encourage sustainable adoption of the new technologies by improving the availability of open-source tools (Rotz et al., 2019), promoting open data solutions (Adamashvili et al., 2020; Von Braun et al., 2017) and supporting the bargaining power of farmers in relations with service providers (Fielke et al., 2020). To fulfil these aims, different characteristics of actors need to be accommodated (Bronson, 2018; Giua et al., 2020; Li et al., 2019; Rotz et al., 2019; Yoon et al., 2020) and public policies are expected to provide a platform for dialogue (Eastwood et al., 2017; Fleming et al., 2018; Ingram & Maye, 2020; Fielke et al., 2019). The identified justifications for policy interventions, particularly the responsibility to avoid the risk of imbalanced progress by granting fair access to new technologies (Bellon-Maurel & Huyghe, 2017; Carmela Annosi et al., 2020), preventing potentially negative effects of the new technologies (Barrett & Rose, 2020), avoiding digital divides (Barrett & Rose, 2020; Klerkx et al., 2019; Shepherd et al., 2020) and regulating the competition between technology providers (Bronson, 2019) further underline the approach that public policies are expected to take towards power imbalances. Thus, considering the future scenarios of digitalisation as developed by Ehlers et al. (2022), where the prominent differences between scenarios lie in the position of power (which could dip or be dipped in the favour of the government, food companies or technology companies), this research argues that it is up to public policies to avoid the realisation of any of these scenarios and focus on finding a middle ground.

End-user services as part of the government-provided digital infrastructure. One of the aims of the public policy interventions should be to introduce digital technologies to the agricultural policy process, for example, for surveillance and monitoring. It is up to the context-specific policy deliberation to choose the exact approach to achieve this aim. However, the extent to which digital technologies are used by the farmers is an important factor that seems to determine the success of these policies (M.-H. Ehlers et al., 2021). It has been demonstrated that access to field or farm-level data is crucial for designing precise interventions (Antle (2019) as referred to by OECD (2019)). One of the ways to increase the use of digital technologies and access the data these technologies produce is for the governments to deliver end-user solutions to farmers, which would become an integral part of the farm management systems (Rose et al., 2016). The developers of these systems could be different. However, they would need to meet certain requirements agreed upon in the policy process (Knuth et al., 2018). Examples of such services could be nutrient and carbon management calculators (Lindblom et al., 2017) and solutions for herd health data recording (Rose et al., 2016). On the one hand, the ability to provide such services would offer policymakers an opportunity for further policy interventions to subsidise the use of these services in exchange for access to parts of the recorded data.

In the context of this study’s findings, this approach would be supported by the expectation of public policies to provide appropriate digital infrastructure and support effective dissemination of information and knowledge. Furthermore, providing such services could be instrumental in avoiding the risk of imbalanced progress, supporting data quality, integrity and availability and encouraging sustainable adoption of new technologies. In contrast, offering such services could cause market distortions and risks of excessive digital path dependency as long-term public investments might be directed to technologies that could become outdated in a few years (M.-H. Ehlers et al., 2021). Furthermore, the OECD (2019) has highlighted how governments lack the ability to develop and implement technologically advanced solutions and bear the investment costs, which further casts doubt about the feasibility of this approach. The lack of actions for delivering end-user services to the farms could, however, promote the trend referred to by Fielke et al. (2020); they refer to the increasing need to monitor the relationship between the technology providers and farmers to ensure “representativeness, accuracy and security of the data that the decisions are based on”, a development that would also demand significant investments from the governments. The relationship between policy interventions and the nature of the digital services that facilitate these interventions needs further deliberation. It is clear, however, that the knowledge and skills of the policymakers that act in this changing environment need constant development, as reflected in previous literature (e.g. Ehlers et al., 2022; OECD, 2019).

The potential of the relationship between digitalisation and public policies to transform food systems. The need to strive towards a sustainable transition of food systems, the perceived ambition of some policies to become part of the food system governance and the development stage of digitalisation merit the need to analyse the results of this research from the food systems perspective. In reflection of Klerkx et al. (2019), this would serve as a prerequisite for assessing the potential that digitalisation as a transformative force has. The existing food systems need to be transformed to become more sustainable, for example, by tackling challenges like excessive food losses and waste, GHG emissions and carbon sequestration, as well as improving food security and nutrition (SEPA, 2020; Lajoie-O’Malley et al., 2020). Several policies highlight the necessity of public policies to participate in the food systems’ governance and to initiate the reforms needed. For example, the EU Farm-to-Fork strategy, with its comprehensive approach that, in addition to farming policies, includes food safety and processing as well as food consumption challenges, is one of such policies (Schebesta & Candel, 2020). The initiative to develop EU Data Spaces is another example, as it includes the vision of extensive improvement of data exchange between different economic sectors (e.g. agriculture and logistics; European Commission, 2020b). This policy trend goes hand in hand with the development stage of digitalisation of the agri-food sector, which, according to Wolfert et al. (2021), has entered the “Twilight Zone” where a paradigm shift is needed to detach from user-centric software design and move towards “targeting the optimal use of data available in the whole system”. According to the authors, this would set the stage for the development of applications covering the entirety of the food systems and making them integral components of the data economy (ibid.). The justifications for, aims of, and ways of carrying out the policy interventions as presented in the results of this research are relevant factors for the sustainable transformation of food systems. For example, factors highlighted in the reviewed literature, such as the need to develop infrastructure and regulations, encourage sustainable adoption of new technologies, the risk of imbalanced progress, and data availability play an important part in the functioning of food systems. Therefore, in the future, it would be useful to adopt a food system perspective to assess the transformational potential that the relationship between public policies and digitalisation possesses. A concept that might be applied in this could be one of the mission-oriented agricultural innovation systems together with the approach of defining “wicked” challenges, which would allow us to consider the diversity of actors and their interactions while enabling us to maintain focus on the desired development pathways (Klerkx & Begemann, 2020).

The existing knowledge gaps. The comprehensive nature of the systematic review provides an opportunity to highlight any possible gaps in the current literature that may benefit from further attention. One of the least studied perspectives is the potential for public policies to increase the competitiveness of the farming sector (Bowen & Morris, 2019; Ratten, 2018) and the agri-food value chain (Bowen & Morris, 2019) via digitalisation. This is likely to be reflected in the priorities of the existing policy instruments as well. The lack of policy focus on improving the competitiveness of the agri-food sector could affect the perceived value offered by new technologies for farmers, which could reduce the speed of technology uptake. Therefore, one of the possible research directions could be to study whether this lack of competitiveness being prioritised by public policies could become a limiting factor for the adoption of new technologies in the agri-food sector.

Although insightful research on how to implement public policy interventions in relation to digitalisation has been published (e.g. M.-H. Ehlers et al., 2021), it appears that further research is still required on the characteristics of governments and policymakers that could have an impact on the success of the policies. More detailed knowledge on this subject could make a significant contribution to the ability of governments to adapt to the changing expectations and more strategically guide the process of digitalisation.

5. Conclusions

Digitalisation is in general perceived as a positive process, but one that also entails several challenges that need to be addressed by the agri-food sector’s internal and external stakeholders. This paper has demonstrated that according to the existing academic knowledge the relationship between the public sector and the agri-food sector is one of the keys for realising the full potential of digital technologies in the agri-food sector. The results of this systematic review include synthesis of the existing academic knowledge about the justifications for, aims of, and ways of carrying out policy interventions in the process of digitalising the agri-food sector. Based on the collected data, the two main roles expected from public policies were outlined – that of protecting weaker parties and enabling innovation. The interrelationship of the roles underlines the importance of public policies in avoiding excessive power asymmetries in the agri-food value chain. It raises further questions about the extent to which public policies should provide end-user services as part of the digital infrastructure.

It remains to be seen which unintended consequences the digitalisation of the agri-food sector brings with it and if the resulting impact is comparable to those that accompanied the Green Revolution of the last century. In an effort to avoid the negative consequences, it is crucial, however, that the potential benefits of digital technologies are not overly compromised. This underlines the importance of establishing a balance between the roles of public policies in enabling innovation and protecting weaker parties. Establishing this balance is especially important to help the agri-food sector overcome the challenges imposed by the climate crisis and the loss of biodiversity. As highlighted by the knowledge gaps presented in the previous section, the search for this balance is dependent on further knowledge about the characteristics that governments and policymakers should have to operationalise the expected roles in developing and implementing their policies.

Acknowledgements

The authors thank the project team of Horizon 2020 ERA-NET Cofund ICT-AGRI-FOOD project “SustainIT – Releasing the Potential of ICT for Sustainable Milk and Beef Cattle Value Chains” for the valuable discussions on the topic of this paper.

We would like to thank the editor and the anonymous reviewers for helpful comments on earlier versions of this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The publication of this paper has been supported by Estonian University of Life Sciences and the Doctoral School of Economics and Innovation (Estonian University of Life Sciences ASTRA project Value-chain based bio-economy).

Notes

[1] Due to extensive interlinkages, this paper understands the term “roles of public policies” to also cover the tasks of governments and their agencies as well as those of the corresponding policy instruments.

ANNEXE

Table 1.

Justifications for policy interventions.

1.1 Need to develop appropriate digital and physical infrastructure and data regulations	1.1.1 Data regulations	Provide clarity in data regulations and policies to maintain trust at the farm level (Relf-Eckstein et al., 2019)
		Provide clarity on the rules on access and use of data to encourage the adoption of the technologies that expect data integration (Jakku et al., 2019)
	1.1.2 Digital infrastructure	Digitalisation brings with it the need for public investment into technical infrastructure in the rural areas (Rotz et al., 2019)
		Low-density rural areas need support for developing ICT infrastructure, whose strength in turn is strongly related to the probability of the adoption of new technologies by individual enterprises (Galliano & Roux, 2003)
		More attention to establishing proper digital infrastructure is needed (Michels et al., 2020)
		The lack of quality of the digital infrastructure is a bottleneck in the adoption of data-based technologies according to the potential target group (Jakku et al., 2019)
		Policies are needed to provide necessary infrastructure (Barrett & Rose, 2020)
		“Reactivity of rural agribusinesses” leads to the need for public policy involvement in ensuring the development of digital infrastructure (Bowen & Morris, 2019)
1.2 Risks of imbalanced progress	1.2.1 Avoidance of digital divides	Avoid digital divides (Klerkx et al., 2019)
		Benefits have to be shared among stakeholders in a way that boosts confidence in technologies (Shepherd et al., 2020)
		Farmers that are under the risk of missing out on the benefits of technological development need additional support (Barrett & Rose, 2020)
	1.2.2 Improved access to new technologies	Appropriate policies have the potential to mitigate challenges to do with access and sustainable business models (Carmela Annosi et al., 2020)
		Specialised farm machinery is a niche market; therefore, support from public sector is needed for acquisition (Bellon-Maurel & Huyghe, 2017)
		Support for deployment of technologies that could help alleviate environmental pressure (Von Braun et al., 2017)
		Promote technologies to tackle the climate crisis (Radulescuu et al., 2019)
	1.2.3 Damage prevention	Policymakers should be able to detect technologies that have potentially negative effect and if necessary, take measures to change the technology trajectory (Barrett & Rose, 2020)
		Lessons from the Green revolution should show public sector that a “precautionary approach” to digitalisation has to be maintained, especially in terms of potentially further aggravating power imbalances and not taking into account the previous experience of stakeholders (Avaria, 2020)
	1.2.4 Inclusive innovation ecosystem	A prerequisite for more inclusive agricultural research is further investment “in social and communications innovation and infrastructure, technology and methods” (Nelson et al., 2019)
		Dedicate specific efforts to make sure that the interests of potentially weaker stakeholders are taken into account in the innovation process (Bronson, 2018)
1.3 Improve the competitiveness of the agri-food sector	1.3.1 Farm productivity and efficiency	Enable the development of more efficient farming practices (Bowen & Morris, 2019)
		Promoting digital entrepreneurship is a way for creating potential for growth of profitability of farms (Ratten, 2018)
	1.3.2 Value chain development	Encourage the adoption of digital technologies to achieve diversification and internationalisation in the agri-food value chain (Bowen & Morris, 2019)
	1.3.3 Enhance technology development	Encourage competition between technology providers (Bronson, 2019)

Table 2.

Policy interventions.

2.1 Encourage sustainable adoption of the new technologies		Provide open data solutions to key stakeholders (Adamashvili et al., 2020)
		Efforts should be directed to supporting open-source tools (Rotz et al., 2019)
		Enable the development of more “simple and practical technologies” (Bucci et al., 2018).
		Regulate power asymmetries in the value chain to encourage the adoption of smart farming technologies at the farm level (Jakku et al., 2019).
		Intensification of R&D to provide solutions that meet the requirements of the current farming systems to support wider adoption of technologies (Chuang et al., 2020)
2.2 Support data quality, integrity and availability		It is likely that governance mechanisms need to be put in place to “monitor the representativeness, accuracy and security of the data that decisions are based on” to support the bargaining power of farmers in the value chain (Fielke et al., 2020)
		Regulate farm-level data collection in the interests of many instead of the few (Bronson, 2019)
2.3 Change the process of public sector’s interventions	2.3.1 Changing the role of public sector	The role of public policies in collecting data and information about the agricultural sector should be re-examined in the light of technological developments (Relf-Eckstein et al., 2019)
		Further privatisation of agricultural advisory systems will leave formal governance structures with the role of market facilitator, which means that “traditional technology transfer and monitoring/surveillance responsibilities become much broader” because of the additional stakeholders (Fielke et al., 2020)
		Provide “policy infrastructure that can cope with socio-technical change“ (Fielke et al., 2019)
	2.3.2 Monitoring and decision-making	Biosurveillance systems should exploit the advantages of the new technologies (Devitt et al., 2019)
		Satellite data could be used to monitor crops, including identifying unauthorised extraction of water for irrigation (Negula et al., 2017)
		Big data solutions could be used for decision-making by the policymakers (Carmela Annosi et al., 2020)
		Farm data can be used extensively to assess the level of sustainability of different practices (Van der Burg et al., 2019)
		The technological development will bring along the increased power to survey the processes of agri-food sector (Fielke et al., 2020)

Table 3.

Ways for carrying out the policy interventions.

3.1 Strive for balanced policies	3.1.1 Ex-ante assessment of development paths	Careful analysis is needed about the trade-offs between the different paths in the development of digital agriculture and impacts on ecosystem services (Lajoie-O’Malley et al., 2020)
		Goals and objectives are needed that would enable the use of new technologies by agroecological farmers (Rotz et al., 2019)
		The emerging changes in the role of AKIS require a carefully considered response, which will need to be based on a digitalisation strategy to “reveal emerging issues, such as data ownership and to reflect and respond to potentially undesirable consequences such as the exclusion of certain groups” (Rijswijk et al., 2019)
	3.1.2 Facilitate social innovation	“Equitably realised digital farming transition demands a high level of social rather than simply technical innovation” (Bronson, 2019)
	3.1.3 Clarify technological, social, economic and R&D needs	It could be beneficial in the policy development to differentiate between the factors affecting the adoption of technologies (characteristics of users) and factors affecting the inclusiveness and fairness of digitalisation (characteristics of technologies; Giua et al., 2020)
		The design of policies should take care that the interests of “high adapters” would not be disproportionally represented (Groher et al., 2020)
		Differentiate between the technological needs of different types of farms (Yoon et al., 2020).
		Socio-economic limitations should be addressed to support the acceptance of technological innovation (Li et al., 2019)
		There is a need for regulations that would support the rights of the consumer in accessing software and its’ generated data. In combination, there is a need for steps for the democratisation of data that would “prioritise the right of society over the interests of shareholders” (Rotz et al., 2019).
		Develop an inclusive innovation ecosystem (Bronson, 2018).
3.2 Subscribe to cross-cutting concepts	3.2.1 Link innovation ecosystem development with missions	Use innovation ecosystems thinking for innovation that supports sustainability (Pigford et al., 2018)
		In order to ensure balanced development, mission-oriented innovation policies should be used (Klerkx & Begemann, 2020)
		Innovation policies should strive towards supporting the development of Mission Oriented Agricultural Innovation Systems (MAIS; Klerkx & Rose, 2020)
3.3 Adjust the policies according to the maturity level of innovation	3.3.1 Nurture emerging technologies and innovations	Support the starting-up phase of innovations (Lombardo et al., 2018)
		Create policies to nurture the emerging technologies (Adamashvili et al., 2020)
	3.3.2 Pick low-hanging fruits	More attention should be put on adoption of mature technologies that could have immediate positive effect instead of focusing on solutions with lower readiness levels (Barrett & Rose, 2020)
3.4 Dialogue between the stakeholders	3.4.1 Develop inclusive collaboration platforms	Dialogue between relevant stakeholders serves as a prerequisite for the creation of a moderate discourse around the future of big data in agriculture (Fleming et al., 2018)
		Involve citizens proactively to enable more ethical outcomes (Eastwood et al., 2017)
		Involve a wide circle of stakeholders to discussions to ensure sustainable transition (Fielke et al., 2019)
		New biosurveillance systems should be co-created and high ethical standards must be ensured (Devitt et al., 2019)
		Effective forms of cooperation like PPP have to be established between public and private stakeholders (Carmela Annosi et al., 2020)
		There is a wide array of new stakeholders from the digital technology sector that are playing increasingly important role in the AKIS. It is important to find sufficient synergies between information and skills of the new stakeholders and the agricultural expertise of the more traditional stakeholders (Ingram & Maye, 2020)
	3.4.2 Break the silos of policy makers	The policymakers also from the outside of the agricultural policy field need to be included (policies of autonomous farm vehicles, for example; Relf-Eckstein et al., 2019)
	3.4.3 International collaboration	Enable international collaboration to adopt to changing environment (Von Braun et al., 2017)
3.5 Disseminate information and knowledge effectively	3.5.1 Curriculum development	Smart Farming related issues should be part of the curriculum in vocational schools and universities (Bellon-Maurel & Huyghe, 2017)
	3.5.2 Develop new skills	Training and education on new technologies should be provided (Yoon et al., 2020)
		Trainings on communication with the public should be provided for explaining the benefits of the new technologies for animal welfare and the environment (Pfeiffer et al., 2020)
		Need new skills for processing and interpreting the data collected (Jakku et al., 2019)
		Policies are needed to improve the skills of farmers that are otherwise under the risk of being left behind (Barrett & Rose, 2020)
	3.5.3 Reform innovation diffusion	Newest ICT solutions must be implemented for innovation diffusion (Norton & Alwang, 2020)
		Improve access to information and advice to increase adoption of new technologies (Knierim et al., 2019)
		Use innovation brokering service to communicate to agricultural technology companies the needs of farmers who are not industrialised and technologically advanced (Lioutas & Charatsari, 2020)
		Investments “in social and communications innovation and infrastructure, technology and methods” are needed (Nelson et al., 2019).
	3.5.4 Facilitate the changing role of public sector through policy	Integrated policy framework should facilitate the faster dissemination of information about the ICT developments (Tiwari, 2008)
		The role of public sector in the knowledge transfer is decreasing and is replaced by the farm media (Relf-Eckstein et al., 2019)