Digital Opportunities for Better Agricultural Policies

Report

23 September 2019

Chapter 3. Realising digital opportunities for better agri-environmental policies

Abstract

This chapter considers how the advances in digital technologies and related institutions can support delivery of better policies for agriculture. Much of the discussion and examples are drawn from the field of agri-environmental policy specifically. This allows an in-depth analysis of how digital technologies can be useful throughout the policy cycle. The approach also serves to highlight that some issues which agri-environmental policy makers need to consider when making use of digital technologies for policy are actually part of broader discussions about digitalisation in the economy. Chapter 5 considers some of these broader issues in more depth.

3.1. Digital opportunities for agricultural and agri-environmental policies: A conceptual framework

Use of digital technologies for agri-environmental policies is analysed based on the conceptual framework described in Figure 3.1. This framework begins with identifying fundamental issues which can constrain the use and development of digital technologies for policy. It then identifies the various components of the policy-making cycle and posits that digital technologies can have a role in all of these components. Finally, it identifies several types of challenges or issues that government organisations may need to overcome to fully realise the benefits offered by digital technologies.

Digital technologies can help address fundamental problems1 that constrain existing agri-environmental policies, caused by information gaps (incomplete information),2 asymmetric information,3 transaction costs,4 and non-alignment of incentives5 of different actors. These problems manifest in several ways: firstly, and perhaps most importantly, they can constrain the set of feasible policy alternatives, potentially limiting the scope of the policy or the policy mechanisms available to choose from. Secondly, within the set of feasible alternatives, they can diminish the effectiveness or efficiency of policy implementation.6 This paper identifies how digital technologies can help address these fundamental problems.

As shown in Figure 3.1, opportunities to mitigate or overcome these problems via the use of digital technologies exist in all components of the policy cycle. The policy cycle shown in the figure is a stylised representation of the broad components undertaken to design, successfully implement, and evaluate an agri-environmental policy. In the figure, the components are set out linearly; it is acknowledged that the particular components and ordering of components for a particular policy will depend on context – the emphasis here is on considering the usefulness of digital technologies for each component. The components, drawn from the literature on agri-environmental policy design (e.g. OECD (2008[1]) and OECD (2010[2]), are: Policy Design; Initial outreach and enrolment for the policy mechanism; Implementing policy mechanism; Monitoring and enforcement (if relevant); Policy evaluation; and Communication with broader public about policy.7

The data infrastructure referred to in Figure 3.1 represents the physical, digital and institutional structures enabling and governing the collection, transfer, storage, analysis of agricultural data to produce knowledge and advice, and enabling a feedback loop to farmers as well as policy makers. This underpinning infrastructure conditions how digital technologies are deployed throughout the policy cycle and influences the way policies can be designed and implemented. The data infrastructure is discussed further in Chapter 5.

Next, the conceptual framework identifies sources of challenges to successfully using digital technologies to solve the problems identified above. The first set of challenges relates to institutional constraints and path dependencies. These can be in the form of rigidities of legislation, regulation and legislative processes or consensus mechanisms which cannot easily be adapted for new technologies. They can also refer to resistance or lack of human capital within the public sector to adopt the new technology (including attitudinal factors or lack of skills), or on the part of other actors (for example, farmers may be reluctant to share data with administrators). The second set of challenges are dynamic challenges which are either caused by new technology itself or by actors’ responses to the adoption of technologies intended to improve agri-environmental policies.

Figure 3.1. Conceptual framework for analysing digital opportunities for agricultural and agri-environmental policies

Applying this conceptual framework, this chapter examines how digital technologies can help address the problems caused by information gaps, information asymmetries, transaction costs and incentive non-alignment. This chapter draws evidence relating to agri-environmental policies rather than agricultural policies in general, for two reasons. First, agri-environmental policies are particularly sensitive to the aforementioned issues. Second, his narrower focus makes the analysis more tractable and comparable across organisations and countries, and allows for deeper consideration of sustainability aspects. Nevertheless, many insights drawn are relevant more broadly.

Section 3.2 first provides an overview of use of digital technologies by organisations responsible for administering existing agri-environmental policies, using evidence obtained from an OECD questionnaire (Box 3.1). It then explicitly considers technology use to improve different components of the policy cycle. Section 3.3 considers how digital technologies can enable new policy approaches, which were previously unfeasible due to factors such as high cost or technical infeasibility.

Box 3.1. Adoption of digital technologies by agri-environmental policy makers and administrators: OECD questionnaire

Information on the actual use of digital technologies by public sector agencies is generally difficult to obtain, and there are very few sources which allow for comparisons across countries. While the OECD collects data on “Digital Government” and open data for its members and the United Nations (2016, p. xvii[3]) collects data on “E-Government” in support of sustainable development, these datasets relate to countries’ government or public sector as a whole; no data is available at the agency level or specifically for the agriculture sector, and as such it is difficult to determine the level of adoption of digital technologies by government organisations (e.g. department or ministry or other government agency) responsible for agri-environmental policies.

To bridge this gap, the OECD conducted a questionnaire targeted specifically at these organisations. The questionnaire focusses on:

which types of data are currently used and how they are gathered;
the extent to which agri-environmental policymakers and programme managers make use of particular digital technologies in carrying out their functions as they relate to the agricultural sector, including for policy design, policy implementation, monitoring and compliance, policy evaluation, and communication (i.e. throughout the “policy cycle”);
the extent to which use of digital technologies differs across agri-environmental policy areas (water quality, water quantity, air quality, biodiversity, soils, climate change adaptation (on-farm), climate change mitigation (on-farm));
strategies or management policies organisations are putting in place to maximise the beneficial use of digital technologies;
organisations’ experiences with digital technologies and future plans.

The Questionnaire received 46 responses covering 67 institutions (some responses consolidated data from several institutions) from 16 OECD member countries, plus the European Commission’ Directorate-General for Agriculture. These responses provided data on 108 policies and programmes, as well as respondents’ experiences with and views on use of digital technologies by their organisation. This dataset provides a wealth of information on how digital technologies are currently being used by reporting organisations.

Note: See Annex B for further information on the design and process for the questionnaire.

Employing digital technologies to address problems (whether to improve existing policies or enable new ones) may not necessarily be simple. Sections 3.2, 3.3 and 3.4 consider different types of challenges or barriers to the successful use of digital technologies: institutional constraints (see the left-side box in Figure 3.1) and new challenges caused by use of digital technologies (including issues arising from the public good characteristics of data and knowledge – see the right-side box in Figure 3.1).

Throughout these sections, insights drawn from ten case studies provide illustrations of the way technologies are being deployed in specific policy contexts, the type of challenges faced and the solutions found to overcome them, and “lessons learned” for others considering undertaking similar initiatives. The full set of case studies is available in Part IV.

3.2. Digital technologies throughout the policy cycle: Insights from agri-environmental policies

3.2.1. Use of agricultural data and digital technologies for agri-environmental policies: OECD questionnaire

The OECD Questionnaire gathered data on the use of digital technologies by organisations implementing agri-environmental policies and programmes, using the technology categories listed in Table 2.1.

Respondents were asked to provide data for up to five agri-environmental policies or programmes, selected on the basis of respondent-assessed importance of the policy or programme for maintaining or improving the sustainability of agriculture in the respondent’s country. Data was provided for 108 policies or programmes in total.

The average number of technologies currently used for policies and programmes or expected to be used within the next three years varies considerably between responding countries (Figure 3.2). The Netherlands (consolidated response of Ministry of Agriculture, Nature and Food Quality and Netherlands Enterprise Agency) was the most frequent user of technology for agri-environmental programmes, reporting significant use of digital technologies in all components of the policy cycle; Korea–Rural Development Agency and Portugal–PRODERAM8 were the next most frequent users.

Use of digital technologies varied considerably not only across individual respondents, but also within countries. Use did not differ systematically according to the level of government of respondents (i.e. national versus regional, province, or watershed-level). However several national respondents did not answer this section, noting that all implementation was administered by other (usually sub-national) organisations.9

The most common policy areas10 making use of digital technologies were water quality and biodiversity (each of these policy areas were selected for 76 out of 108 policies and programmes, noting multiple policy areas can be selected), while the most common policy mechanisms were extension services and information provision and agri-environmental payments or subsidies. Multiple policy areas and policy mechanisms were selected for the majority of policies and programmes reported on.

Figure 3.2. Use of digital technologies in agri-environmental policies and programmes

When asked directly whether use of digital technologies differed across agri-environmental programmes or policy areas, 58% of respondents agreed or strongly agreed that technology use does differ, compared to 28% disagreeing or strongly disagreeing. 53% of respondents also agreed or strongly agreed that decisions about technology use were made at the level of the individual policy or programme (25% disagreed or strongly disagreed).

However, technology use did appear to vary with policy mechanism used: policies using environmental taxes as (at least one of) the policy mechanism(s) use on average almost twice the number of digital technologies as policies using trading schemes (environmental markets), although there were a low number of observations in for of these policy mechanisms. When grouped into mechanism categories, some differences across different technology types emerge. For example, GIS-based analytical tools, digital communication tools and online surveys or censuses are currently more intensively used (i.e. more often and in more components of the policy cycle) for administering economic instruments (environmental property rights, environmental taxes, agri-environmental payments or environmental markets) than for regulatory instruments (activity prohibitions or environmental standards). The converse is true for citizen science and crowdsourcing, which may reflect that use of these technologies is currently low overall, but that there are examples where regulators have invited community participation in monitoring programmes.

Overall, it appears that use of digital technologies for policy purposes is often approached on an ad hoc basis, in that decisions about digital technologies are often made at the level of individual policies or programmes. Government organisations should evaluate opportunities systematically, even if actual use of specific technologies remains only for specific purposes. The case for creating new digital tools also needs to consider whether existing tools can be improved, and also how digital tools work together with other tools (Box 3.2). A coherent approach can:

help ensure that initiatives generate “additional” benefits by using a mix of old and new technologies.
provide for multi-dimensional integration of digital tools (e.g. interoperability between digital tools, integration of digital tools with other tools) to ensure efficiency and effectiveness.

Box 3.2. Case Study lesson: Ensure initiatives generate “additional” benefits by using a mix of old and new technologies

Digital technologies have been used in the New Zealand case study (Case Study 1—see Box 3.4 for an overview) both to improve and enhance the functionality of existing analytical systems (e.g. upgrading the NZ Water Model), and to provide wholly new tools (e.g. LUS classification and Physiographic Environments of New Zealand GIS layers) that support decision-making process that were not previously possible. This enables the Challenge to avoid duplication and “reinventing the wheel”, while still ensuring that the tools are fit for purpose. This requires a thorough understanding of the existing analytical tools.

A mixture of old and new tools was similarly found to be the most cost-effective approach in the Dutch agricultural collectives context (Case Study 2—see Box 3.5 for an overview). Based on the experience of case study participants (see full case study for more detail), it is recommended that countries considering implementing a similar approach should:

form a clear view about the technological requirements, including whether these will appropriately reflect (existing or desired) administrative arrangements;
canvass a variety of options (adapting pre-existing tools, new custom built-tools, or a hybrid of both) at the outset. This could include planning for a staged introduction of new digital tools if this is considered desirable;
plan from the beginning for the tools to be able to be adapted to new policy contexts (e.g. the introduction of more result-oriented or targeted policies).

Digital communications technologies including social media, web-based video conferencing and digital data visualisation technologies were the most commonly-used technology categories, closely followed by GIS-based analytical tools. Perhaps unsurprisingly, the emerging technologies such as Deep Learning or Artificial Intelligence is the least-used technology category, followed by citizen science or crowdsourced data and data from precision agriculture. Overall, digital technologies are currently most-used in the Policy Design and Implementation components of the policy cycle (Figure 3.3). Further information about use of different technologies within different components of the policy cycle is provided in sections 3.2.2 to 3.2.6.

Figure 3.3. Digital technologies currently used in the policy cycle components

Online surveys, aerial photography and satellite data key digital sources of data for agri-environmental policies, but traditional methods are still important

Public organisations have a long history of collecting, using, and providing agricultural data. Such data is critical for policymaking and serves a range of other valuable purposes, not least in providing aggregate information about the agriculture sector which is difficult or impossible to obtain from other sources. The digitalisation of data collection methods for agri-environmental policies is still on-going and organisations are currently using both traditional data collection methods (e.g. postal and in-field surveys) and digital methods (Figure 3.4). Manual collection of field data by government is still the single most commonly-used method. The most commonly-used “high tech” data collection methods are online surveys, aerial photography and satellite or geospatial data.

Relatively few organisations are making use of non-traditional sources of Big Data relevant to the agriculture sector: precision agriculture data (11 respondents) and retail scanner data (eight respondents).11 Canada, Chile and Korea are the only countries for which respondents reported making use of both these data sources. One reason is that accessing those data is not straightforward. While they can provide a very high degree of granularity about agricultural production (precision agriculture) and consumption (scanner data), they are often commercially protected, making them difficult to access. In addition, they may have quality or coverage issues which may make them more difficult to incorporate into existing data analysis frameworks. Interestingly, seven respondents (mostly respondents from Canada and Chile) envisage using precision agriculture data in the next three years, but still none envisage using retail scanner data. These responses indicate that there may be an opportunity to learn from leading countries and organisations about how to make use of new data sources and integrate them with existing sources.

Figure 3.4. Methods used to collect and verify data on farmers’ environmental performance and use of “Big Data” for agri-environmental policies and programmes

Most organisations have a good awareness of the benefits of digital technologies, but also see new risks

Respondents generally considered that digital technologies have a range of benefits for their organisation (Figure 3.5). The most commonly-perceived benefits are that technologies help organisations to improve their communications with other government departments or with farmers; this likely reflects that use of digital communications technologies is one of the technology categories that has the highest current use. Respondents also generally agreed or strongly agreed that digital technologies can facilitate new programmes or services and decrease organisational costs. Only a minority of respondents (18%) considered the lack of understanding of the benefits as a challenge.

There is also broad awareness (75%) that digital technologies introduce new risks. However, none of the organisations opted to provide examples of such risks, and more work is needed to better understand what new risks organisations perceive they are facing. Respondents were most commonly neutral on whether their organisation had a clear understanding of the potential for digital technologies to disrupt or change agricultural supply chains. This likely reflects the fact that there is as yet very little evidence on the magnitude of change that adoption of digital technologies by the agricultural sector will bring about.

Figure 3.5. Perception of the benefits of using digital technologies

Most organisations have adopted digital strategies and data policies, and have appointed a Chief Information Officer

In the Recommendation of the Council on Digital Government Strategies, OECD member countries agreed to the recommendation that countries develop and implement digital strategies, and (among other recommendations) to “[e]stablish effective organisational and governance frameworks to co-ordinate the implementation of the digital strategy within and across levels of government” (OECD, 2014[4]). The questionnaire gathered data on three aspects to evaluate to what extent adoption of digital strategies has occurred for organisations administering agri-environmental policies: appointment of a “Chief Information Officer”,12 adoption of a digital strategy,13 and adoption of a data policy.14

Figure 3.6 shows that most respondents (72%) had already appointed a chief information officer; several others were intending to do so within the next three years. Most (78%) had also adopted a digital strategy, or abided by the broader digital strategy set by another organisation (e.g. as part of a whole-of-government digital strategy). Finally, most (74%) had adopted a data policy or abided by a broader one set by another organisation. In addition to information gained via the questionnaire, Case Study 1 (Box 3.3) also provides a practical example of how having a data strategy can assist organisations when implementing new initiatives that make use of digital tools.

Of respondents who had not adopted data strategies, data policies, or chief information officers, most planned to adopt them within the next three years. Almost all of these respondents belonged to countries who submitted more than one response to the Questionnaire and other respondents from these countries had already adopted these institutions. Thus, there appears to be an opportunity for cross-organisational learning: organisations intending to adopt data strategies or data policies or to appoint a Chief Information Officer in the near term could examine existing institutions in similar organisations. Conversely, organisations which already have these institutions in place could use this as an opportunity to review their own institutions and work towards a cohesive approach across all levels of government.

Figure 3.6. Organisational capacity: Chief information officers, organisational digital strategy and data policies

Box 3.3. A data strategy for New Zealand’s Our Land and Water National Science Challenge

New Zealand’s Our Land and Water National Science Challenge (‘the Challenge’) is a government-funded research and innovation programme aiming to improve the productivity and sustainability of the New Zealand primary production sector. The many and varied research projects under the Challenge are producing a “growing diversity, complexity and volume of data” (Medyckyj-Scott et al., 2016[5]). From the start of the Challenge, it was recognised by the Challenge Chief Scientist and Leadership Team that gathering this data into a shared “data ecosystem” is one of the greatest sources of potential value added for the Challenge as a whole. In 2016, a group of experts from the New Zealand public service and the research sector collaborated to produce a “white paper” on the design of this data ecosystem. The data ecosystem is explained as “a system made up of people, practices, values and technologies designed to support particular communities of practice [in which] data is valued as an enduring and managed asset with known quality” (Medyckyj-Scott et al., 2016, p. v[5]).1

Lesson learned: Having a data strategy for a particular initiative can help ensure digital tools are ‘fit-for-purpose’. The data ecosystem “white paper” actively considered the question of “[w]hat are the best data structures for land and water information to achieve the Challenge Mission?”. It also set out a data strategy for the initiative as a whole. This helped ensure that all proposals, including those for new digital tools, actively considered both existing and recommended data structures and existing data tools.

tools are “fit-for-purpose”. The data ecosystem “white paper” actively considered the question of “[w]hat are the best data structures for land and water information to achieve the Challenge Mission?”. It also set out a data strategy for the initiative as a whole. This helped ensure that all proposals, including those for new digital tools, actively considered both existing and recommended data structures and existing data tools.

Lesson learned: Embrace different levels of Data Management Maturity to fit different contexts. The “white paper” also acknowledged that different actors in the initiative have different levels of Data Management Maturity (DMM).2 It recognised that it may not be necessary to advance all (or any) participants to the highest level of data management in order to achieve programme objectives, and that it will take time to move progressively through different DMM levels. The “white paper” recommended the initiative incorporate strategic planning for transitioning through DMM levels, which can be helpful for: i) identifying the current situation; ii) identifying which level(s) need to be reached; and iii) improving the overall level of maturity while still allowing for flexibility and not imposing too high transition costs.

It is also important to recognise that moving towards more advanced levels of DMM may require attitudinal change. The “white paper” identified that “experience shows that one of the major obstacles in the cultural change is the view that data belongs to “me” and that it is not treated as an asset”, and concluded that “it is unlikely that maturity in handling data will emerge if in other ways participants lack a strong sense of community.” (Medyckyj-Scott et al., 2016, pp. 16, 29[5]).

Notes: See Box 3.4 for a brief overview of the Challenge.

1. The data ecosystem is defined to encompass: Policies regarding data management planning, data custodianship and curation, legal frameworks, and the use of externally sourced data; Procedures and processes to execute those policies and manage data; A data governance framework and organisational structures; Engagement with data consumers and stakeholders; and Technology platforms that will support data collection, storage, description, analysis, linking, delivery and curation.

2. Data Management Maturity (DMM) is a concept and framework for analysing institutional capacity to manage and make beneficial use of data assets. The DMM framework assesses data management practices in six key categories that helps organisations benchmark their capabilities, identify strengths and gaps, and leverage their data assets to improve business performance. See Medyckyj-Scott et al. (2016[5]) and https://cmmiinstitute.com/data-management-maturity.

Source: Case Study 1.

3.2.2. Improving inputs into agri-environmental policy-making

In the design phase of agri-environmental policies, one of the most fundamental challenges is understanding complex physical relationships in order to understand how policies translate into environmental impacts (Gholizadeh, Melesse and Reddi, 2016[6]). A second key challenge is to plausibly assess the likely costs and impacts of different policy options, with a view to choosing the best mechanism to achieve the policy objectives. Third, policy design generally needs to take into account input from a variety of stakeholders, which poses a communication challenge. Policy designers need to consider how they can best engage with these stakeholders, many of whom may be in different physical locations and have limited time to contribute (see section 2.2.3).

Information gaps, information asymmetries, administrative costs (transaction costs) and incentive non-alignment can each significantly constrain efforts to obtain the understanding of physical relationships needed for policy design, the preferences of individuals and groups over different policy mechanisms and outcomes, and the intentions of actors in responding to the selected policy mechanism (anticipation of which should be factored into mechanism selection).

GIS-based analytical tools, online surveys and censuses and digital communications technologies are currently the most-often used digital tools for the design component of agri-environmental policies (Figure 3.7). Of the agri-environmental programmes included in the OECD dataset, two-thirds used GIS-based tools during the design of agri-environmental policy mechanisms.

Figure 3.7. Current and expected future use of digital technologies for policy design

Taken together, over the next three years the main area of expected expansion is in use of technologies—remote and in-situ (proximal and ground) sensing and GIS-based applications—which will allow data to be collected at a higher level of spatial disaggregation and with greater frequency (including the possibility of continuous monitoring). Substantially increasing the spatial and temporal data resolution allows for a more precise and nuanced understanding of the impacts of agriculture on the environment and vice versa, and for highly detailed monitoring of the actions of individuals (e.g. farmers and other land managers) and the outcomes of those actions. This enables better policy-making in several dimensions:

improved definition of agri-environmental policy objectives
ability to implement spatially-differentiated policy mechanisms (section 3.2.4)
ability to implement results-based policy mechanisms (section 3.3.2).

Several levels of improvements to agri-environmental policy objectives can be envisaged. First, improved data resolution can lead to a refinement of existing environmental objectives accounting better for spatial heterogeneity. Many existing environmental objectives for agriculture are characterised by significant scientific uncertainty, due to factors such as the complexity of physical processes, the difficulty to project environmental impacts, especially over long timescales, and the significant spatial heterogeneity of environmental impacts and of conservation measures (Rissman and Carpenter, 2015[7]). While improved data resolution cannot fully remove scientific uncertainty, it can allow for more precise estimates and for consideration of uncertainties and risks at finer scales. For example, environmental outcomes may be more certain in one area than another, which may not be evident if data spatial resolution is relatively coarse.

Second, improved data resolution can lead to redefining objectives to better account for complex environmental interactions. Many environmental objectives currently rely on relatively simple indicators to represent highly complex environmental phenomenon. For example, water quality objectives may be set with reference to a specific pollutant (e.g. nitrogen or phosphorous) or with reference to a specific population of interest (e.g. macroinvertebrates or a key fish species in receiving water bodies). Improved spatial and temporal data on variables of interest can improve the ability to understand how different variables interact, and could allow for setting objectives which take into account more complexity and more holistically represent environmental goals.15

Third, improved data resolution can lead to redefining goals to include consideration of attenuation capacity of ecosystems and to integrate environmental goals with other goals (e.g. economic and social goals). Environmental objectives are often defined in terms of reducing environmental pressures from agriculture. While policies are sometimes coarsely spatially differentiated according to different levels of environmental risk (e.g. different policies on highly environmentally sensitive land such as land at higher risk of erosion, land in close proximity to water bodies) or according to economic considerations (e.g. different policies in marginal areas), they are rarely based on a holistic understanding of how environmental pressures from agriculture and other land uses differ across landscapes (in particular, due to the different attenuation capacity of land and water bodies). Nor are they usually based on a holistic understanding of how policies will affect both the productivity and sustainability of the agriculture sector. The data required to underpin such holistic approaches is considerable and requires a high degree of spatial and temporal disaggregation, and moreover for data to be collected on a wide range of physical and economic variables. Case Study 1 (Box 3.4) provides an example from New Zealand’s Our Land and Water National Science Challenge.

Box 3.4. Case Study 1: Digital tools for New Zealand’s Our Land and Water National Science Challenge

The case study provides a practical example of how digital tools can be used to improve understanding of agriculture’s impacts on water quality outcomes and policy options for management of water quality impacts.

New Zealand’s Our Land and Water National Science Challenge (“the Challenge”) is a mission-oriented government-funded, research and innovation programme, which aims to “enhance primary sector production and productivity while maintaining and improving our land and water quality for future generations”. The Challenge as a whole envisages a new approach to fostering a primary agriculture sector that is both productive and sustainable. The Challenge aims to enable New Zealand to move from considering land use capability (generally driven by production potential and other factors such as off-site environmental impact) to land use suitability where economic, environmental, social and cultural factors are considered together.

The Challenge, which commenced in January 2016 and is ongoing, is comprised of three Research Themes; the second Research Theme (RT)—Innovative and resilient land and water use—is the primary focus of this case study. This RT is comprised of a number of research programmes (>NZD 1 million investment) and smaller projects (refer to full case study in Part IV for details).

Existing efforts to manage land for (environmental) sustainability are based on land-use capability (LUC) classifications.1 Data requirements for LUC classification relate to on-site physical and environmental characteristics. In contrast, the Land Use Suitability (LUS) classification which the Challenge aims to produce integrates “information about the economic, environmental, social and cultural consequences of land use choices” (McDowell et al., 2018[8]), and thus requires substantially more, and different, data. Thus, a number of different information gaps need to be filled. Key gaps include: information about natural processes (e.g. nutrient and other contaminant pathways); and information about how producers and other land managers respond to incentives (both policy and other incentives).

These information gaps also prevent the targeting of existing policies to take into account local contexts. Further, the existing research landscape is characterised by fragmented and asymmetric information: often, data sets and digital modelling tools are accessible only by the researchers who work with them directly. This leads to duplication, confusion over the role of different models and research efforts, and impedes effective translation of research efforts into change “on the ground” (McDowell et al., 2017[9]). In addition, licensing issues with some of the datasets mean data sharing between researchers could be difficult. In a collaborative setting, the researchers can settle for a common minimum data that is accessible to all, but which may not be the most up-to-date dataset.

The Challenge is making use of a number of digital tools to address the information gaps and asymmetries identified above (see full case study in Part IV for an overview of specific tools). In some cases, pre-existing tools are being repurposed to help achieve Challenge objectives; in other cases, Challenge funding is being used to enhance pre-existing tools or build new ones. These tools constitute an important part of Challenge activities, but it is important to recognise that they are being developed and used alongside other (non-digital) activities.

1. LUC classification defined as “a systematic arrangement of different kinds of lands according to those properties that determine its capacity for long-term sustained production” (Lynn et al., 2009, p. 8[10]).

Source: Part IV, Case Study 1.

3.2.3. Connecting administrators with programme participants (farmers) and the general public

In the outreach and enrolment component of policy implementation, and also when communicating with the broader community about policies and programmes, policy makers and administrators need to identify who to communicate with, convey information in a meaningful and easily accessible way, and allow others to communicate with the policy-maker or administrator (and perhaps with each-other). There is a broad literature on the public policy benefits of digital communications technologies such as smartphone apps, social media, web-conferencing, online polls, etc. (Picazo-Vela and Gutiérrez-Martínez, 2012[11]). In the context of agri-environmental policies, these technologies can assist in:

lowering information search costs (both for the administrator and for stakeholders) and increasing participation in voluntary programmes
allowing for multi-directional communication between entities (policy makers, administrators, farmers, NGOs, private third parties, general public
facilitating adaptive management
improving public awareness of, and participation in, agri-environmental programmes (and broader environmental initiatives).

These benefits generally arise via use of web-based technologies which lower the transaction costs for the activities listed above. Another newer avenue for reducing transactions costs is to make use of algorithms, machine learning and natural language generation16 (NLG) technologies to automate (at least partially) some kinds of communications. These technologies are particularly useful when policy makers or administrators need to communicate with a range of stakeholders, who may be interested in receiving different information or receiving information in different formats or in different styles. Examples include:

use of web-based submissions and online dialogues to allow comment and discussion on new policies or policy reforms (Brandon and Carlitz, 2002[12])
use of NLG to automate differentiated communications about river heights and flood risk for different stakeholders (Arts et al., 2015[13]; Han et al., 2014[14]; Macleod et al., 2012[15]; Molina, Sanchez-Soriano and Corcho, 2015[16])
use of social media to advertise policy initiatives or opportunities to participate in policy-making
use of teleconferencing or web-based video conferencing to allow participatory policy-making, particularly to include participants in remote and rural areas.

Apart from lowering transactions costs, use of web-based technologies may also allow for increased participation in policy-making simply by fostering greater awareness of policies and opportunities to become involved.17

Further, by increasing transparency about policy administration and encouraging multi-directional communication, digital communication technologies can also help overcome issues arising from a lack of trust between parties, often resulting from information asymmetries. In the context of agriculture, which is characterised by many actors dispersed across often vast landscapes, video conferencing and live-streaming are particularly useful in building trust between physically separated parties.

However, use of digital communication technologies also involves challenges. For example, attempts to make use of digital communication tools can be hampered by insufficient connectivity between actors—particularly in a context where farmers are located in remote areas—and by a lack of digital literacy of some stakeholders (e.g. older farmers). Also, use of social media and other online communication tools can potentially be manipulated or subject to misinformation campaigns.18 Finally, public consultation processes can be very costly and may stymie policy implementation progress if not done well (Crase, O’Keefe and Dollery, 2013[17]).

Use of social media was by far the most commonly used category of digital technology for public communication, and currently used in 77 out of the 108 agri-environmental policies or programmes included in the dataset (Figure 3.8). Data visualisation technologies and web-based video conferencing were also used but to a lesser extent.

Figure 3.8. Current and expected future use of digital technologies for public policy communication

3.2.4. Digital technologies for policy implementation

Practical implementation of agri-environmental policies can involve a range of different activities and processes, depending on policy mechanism choice (Annex A provides an overview of agri-environmental policy mechanisms). This could involve, for example, administering payments provided to eligible farmers; executing contracts; administering tradeable permit programmes.

The OECD questionnaire provides information on use of digital technologies for initial outreach and enrolment in agri-environmental policies, as well as for policy implementation more generally (Figure 3.9). Digital communication technologies were by far the most-used technology for initial outreach and enrolment, followed by GIS-based analytical tools and online surveys or censuses. For implementation in general, digital communications and GIS-based analytical tools were the most-used technologies. Over the next three years, the most significant area of expansion is the use of secure and accessible data storage for policy implementation.

Figure 3.9. Current and expected future use of digital technologies for implementation of agri-environmental policies

The following sub-sections consider three specific aspects of how digital technologies can facilitate improved policy implementation. While monitoring and compliance can also be considered as part of implementation, section 3.2.5 considers use of technology for monitoring and compliance separately.

Facilitating collective governance mechanisms for landscape approaches to agricultural sustainability

Collective governance can provide an alternative to traditional mechanisms in which federal or national governments deal directly with individual farmers. They can also be useful to achieve more flexible policies. Such policy options may be desirable because they may i) foster participation and compliance by reducing the potential for inadvertent individual non-compliance due to uncontrollable natural events; ii) increase benefits of compliance (e.g. by creating conservation-focussed communities); or iii) decrease the cost of compliance by taking into account natural fluctuations (e.g. via regulatory requirements which “follow nature” by adapting to seasonal patterns) and lower transaction costs.

Digital technologies can help in all of dimensions identified by OECD (2013[18]) and others, e.g. (Prager, 2015[19]; Prager, Reed and Scott, 2012[20]) as fostering successful collectives, including: the importance of providing effective, accessible technical assistance; “intensive, transparent communication”; and collaboration between landowners, intermediaries, collective institutions and central governments. Such collective mechanisms are being implemented in the Netherlands (Box 3.5), to achieve environment-climate-biodiversity objectives in agriculture. These technologies and their accompanying administrative and legislative arrangements enable to consider the landscape as a whole while providing spatial and temporal flexibility for participating farmers and other stakeholders.

Box 3.5. Case Study 2: Digital technologies for Dutch agricultural collectives

In 2016, the Dutch government introduced a new scheme such that individual applications under the EU Common Agricultural Policy (CAP) are no longer possible in the Netherlands; all applications must be lodged by an agricultural cooperative (The Netherlands Ministry of Economic Affairs, 2016[21]). The government considered that the collective approach would: foster a “cross-farm approach”; provide greater flexibility in terms of the content, location and financial compensation of conservation activities; be simpler and less error-prone than administration based on individual applications; reduce costs and improve compliance; and be consistent with the existing social structure in the Dutch agriculture sector. In order to achieve this vision, a number of technical and administrative challenges needed to be solved. Conceptually, these challenges relate to addressing information gaps and creating co-ordination and risk management mechanisms between different actors, different scales and different legal frameworks:

To achieve flexibility, the administrative system and the payment rules must be able to “follow nature” which requires high resolution data on where and when the relevant natural events occur, as well as the ability to track individual actions (e.g. on-farm practices) accurately in space and time.
Achieving flexibility at the local level requires recognising that EU rules may not be similarly flexible, and therefore designing a system which allows local flexibility while still “fitting in” with EU requirements. This introduces the risk that local flexibility will not “fit in”, which needs to be mitigated.
To achieve the desired “cross-farm” or landscape approach, the system needs to be able to track all individual efforts and assess the aggregate effect, and enable an interactive regional planning process whereby regional objectives are set taking into account individual actions, as well as vice versa.

To address these challenges, the Dutch collectives are developing a system of digital technologies: SCAN-ICT, Mijnboerennatuur.nl, and Schouwtool.1 SCAN-ICT interfaces with the digital platforms of the Dutch paying agency, for example the Dutch Land Parcel Identification System (LPIS). This direct link makes it possible to change parcels and management activities on a short notice, without losing controllability requirements stemming from EU-legislation. Further, it ensures that the plans, claims and justifications officially submitted to the paying agency fit with the digital information the paying agency obtains from other sources. This helps improve the quality of these products, and reduces paying agency time to make a decision.

The system also includes “Quality Indicators” (QIs), which are system constraints to help prevent errors and to cross-check different elements. The QIs help demonstrate that the system is robust, and help to automate checks and reduce the risk of errors. The system was built by a “Building Team”, comprising information communication technology (ICT) suppliers, the Netherlands Enterprise Agency, Dutch Provinces and BoerenNatuur. Team members worked together in an open, transparent and cooperative approach. The Building Team organises user groups and regularly consults them on their experiences using the system, collects suggestions on improvement and tests new proposals.

1. Mijnboerennatuur is an online platform which will “digitalise” communications between collectives and their participants and allow farmers to view their own data in real time as well as key documents such as contracts. Schouwtool will allow collectives to manage their inspections through SCAN-ICT. See full case study in Part IV for details.

Source: Case Study 2, Part IV.

Facilitating improved spatial targeting of agri-environmental policies

Micro-level agricultural data (e.g. farm level or field level data) is crucial for design and implementation of targeted agricultural and agri-environmental policies, in addition to improving agricultural research more generally (Antle, 2019[22]). First, it allows understanding of how policy impacts differ across dimensions such as location, production practices, industry or sector, socio-economic status. Second, it allows administrators to actually implement policies on a targeted basis: for example, targeting a policy to areas of most environmental concern or where potential benefits are greatest requires a highly-disaggreagted understanding of how farmers’ decisions affect the environment. Another valuable feature for policy analysis is having data which enables tracking policy impacts through time, i.e. panel or longitudinal data.

Many studies assessing the environmental effectiveness and cost effectiveness of agri-environmental policies have recommended that a greater degree of spatial targeting could substantially improve these (Engel, 2016[23]; Lankoski, 2016[24]; Meyer et al., 2015[25]; Moxey and White, 2014[26]; Savage and Ribaudo, 2016[27]; Weersink and Pannell, 2017[28]; OECD, 2008[1]; OECD, 2012[29]). For example, Wardropper, Chang and Rissman (2015[30]) studied constraints to spatial targeting of water quality policies in the US Midwest. They found that the ability to target is constrained by lack of data, both in terms of data gaps and inability to access data on private lands or use it without making it identifiable. This conclusion is not specific to the US context, nor to the policy challenge of nonpoint source water pollution.

Digital technologies can help with both of these problems, and in fact several researchers have already designed data rich computer-based models and algorithms that could be used to implement spatially-targeted policies (Klimek, Lohss and Gabriel, 2014[31]; Rabotyagov et al., 2014[32]; Whittaker et al., 2017[33]). Governments need to recognise that access to agricultural micro data, including the ability to link different agricultural micro datasets (as well as other relevant data such as environmental data) is a crucial to produce more robust and targeted policy analysis, advice and administration.

Reed et al. (2014, p. 47[34]) note that “[d]espite significant advances in recent years, scientific understanding of the complex relationships between biophysical processes and service provision remains limited, and more is known about some services than others. Without adequate scientific understanding of causal relationships between management actions and service delivery, it may be difficult to assign payments to providers, or to demonstrate additionality i.e. not paying for something that has already been provided.” It appears that the technical ability to implement spatially targeted programmes has much improved, but that achieving a very high degree of spatial resolution (or implementing results-based policies) may yet face some difficulties in certain contexts. Thus, policy makers can begin to implement targeted policies now, even if targeting is necessarily coarse due to data limitations, and can work towards improving the degree of granularity over time. Case Study 3 (Box 3.6) provides an example of how science is currently advancing in the field of gully erosion monitoring, which will allow for improved targeting of voluntary erosion remediation efforts in the Great Barrier Reef catchment, Australia.

Even as technological innovations and associated improvements in data are solving many of the previous technical challenges to implementing spatial targeting, there may be other kinds of challenges. One challenge identified by Wardropper, Chang and Rissman (2015[30]) is that institutional factors such as lack of funding and programme requirements constraint the ability to design targeted programmes (an example of a programming constraint is that the programme be voluntary or the targeting take into account national or regional goals in addition to local goals).

Another potential challenge is that farmers themselves might be resistant to targeted policies and therefore decline to participate (a problem particularly relevant for voluntary policies). However, Arbuckle (2013[38]) found evidence that, contrary to assumptions of resistance, some farmers support greater spatial targeting of agri-environmental policies. Endorsement of targeting was found to be associated with certain factors, such as “awareness of agriculture’s environmental impacts, belief that farmers should address water quality problems, having experienced significant soil erosion, belief that extreme weather will become more common, participation in the Conservation Reserve Program, and belief that farmers who have natural resource issues are less likely to seek conservation assistance. Concerns about government intrusion were negative predictors of support for targeted approaches.”

Use of digital communication technologies to better engage farmers and provision of digital services to farmers may help cultivate a positive attitude towards targeted agri-environmental policies. (See section 3.3.1, which discusses social impacts of monitoring and how results-based programmes can foster positive stewardship narratives for agriculture.) In particular, the potential challenge of farmer resistance may actually be a factor that digital technologies can directly help to mitigate, for example by using communications technologies and high resolution agricultural data to improve farmers’ awareness of environmental issues and their contribution to them. Digitally-enabled results-based policies could therefore be an opportunity to improve programme participation and foster a community approach to improving the environmental performance of agriculture.

Box 3.6. Case Study 3: Remote sensing for targeted erosion and sediment control

The Australian and Queensland governments, in collaboration with the relevant local partners, have funded a number of related initiatives to develop remote sensing applications to assist in targeting key areas to improve the effectiveness and efficiency of efforts to control erosion and sediment loadings in agricultural catchments of the Great Barrier Reef (GBR). Advances in remote sensing technologies offer the opportunity to improve information on gully erosion at lower cost than existing methods. While remote sensing—particularly aerial images—has long been used to supplement in-field measurement, it is only recently that a range of newer remote sensing technologies have been deployed in GBR catchments or elsewhere. The Queensland and Australian governments have funded several projects aiming to assess the suitability of remote sensing technologies in this context. Key projects are:

Gully mapping and drivers in the grazing lands of the Burdekin catchment —this project mapped and quantified gully extent and rates of change at a range of scales in the Burdekin catchment using airborne and terrestrial LiDAR1 data.
Monitoring Gullying Processes in the Great Barrier Reef Catchments (Photogrammetry project)—this project assessed the suitability of “digital photogrammetry2 applied to aerial images routinely collected by state land survey agencies [for addressing] the challenges posed by gully erosion and sedimentation” (Poulton et al., 2018, p. i[35]).

Lesson learned: Use of advanced remote sensing techniques to map erosion processes over large spatial scales is technically feasible and yields improved results but is still relatively costly and challenging to undertake. Large knowledge gaps remain, and a combination of tools may be necessary to enable cost-effective mapping techniques and erosion management strategies. Further, knowledge of where and when gullies occur is not the only information gap needing to be filled. Thus, it is important to place the use of technology for a specific purpose (monitoring gullies) in the broader context of the overall policy objective (reducing negative impacts of erosion on the GBR.

Lesson learned: Improved understanding of physical processes must be balanced by economic considerations. The techniques investigated in the projects covered by this case study have the ability to significantly reduce information gaps about where and when gully erosion is occurring. This knowledge is fundamental to efforts to address the negative impacts of erosion, both for the Great Barrier Reef and more broadly for livestock producers and rural communities who rely on the productivity of land at risk from gullying.

However, there is still “very limited information about the cost-benefit of gully prevention and remediation approaches” (Tindall, 2014[36]). Targeting remediation and prevention efforts based only on the information provided by gully mapping ignores spatial differences in management costs and transactions costs, which may be substantial.3 Information on both the benefits and costs of alternative erosion management activities is still needed to ensure efforts are targeted cost-effectively.

Lesson learned: Benefits and challenges of collaboration across organisations and across projects. Both the projects were highly collaborative and brought together researchers from a range of state and national government agencies. These projects are also part of a broader portfolio of research activities that are continuing to contribute to identifying, defining, characterising, measuring and modelling change in gully systems in key Great Barrier Reef catchments. Increasing costs associated with this this type of research and the rapid on-going technological development in the collection of ground based, remoted sensed and large spatial data requires adaption, innovation and successful collaboration of the research community. For the photogrammetry project, having access to a wider research network currently undertaking project activities within in the GBR region enabled transfer of localised knowledge which helped identify suitable case study areas. Collaborative exchange delivered cost savings in data collection for individual projects as well as useful calibration and validation data made available between different project groups. While there was a willingness for collaboration between projects, in reality researchers share their time between a number of competing activities.

1. LiDAR (Light Detection And Ranging) is an active remote sensing sampling tool which uses the length of time a laser beam takes to return to the sensor to calculate distance.

2. As explained by Poulton et al. (2018, p. 16[35]): “Digital photogrammetry is the science of making, among other things, geometric measurements from images”.

3. For example, Wilkinson et al. (2015[37]) report that the direct management cost (i.e. excluding any programme-related transactions costs) of the recommended combination of management techniques for GBR grazing lands varies between AUD 4 500 and AUD 9 000 per km of gully. Variation in cost-effectiveness per tonne of reduction in mean-annual sediment load is largely dependent on the efficiency of fencing.

Source: Case Study 3, Part IV.

Digital platforms for effective market-based agri‑environmental instruments

Digital platforms can support better agricultural and agri-environmental policies by streamlining administration of agri-environmental payments (Case Studies 2 and 3) and facilitating farmer access to services (Case Study 10). They can also be used to facilitate the implementation of agri-environmental market-based instruments (MBIs).

In general, environmental MBIs are less used in agriculture, both relative to other sectors and to other types of instruments (OECD, 2013[39]) Most existing schemes involving agriculture relate to water quality (e.g. nutrient, sediment or temperature trading programmes), water quantity (i.e. cap-and-trade water rights instruments) or greenhouse gas emissions (OECD, 2013[39]; Shortle, 2012[40]).

Most agri-environmental MBIs in OECD countries make use of digital platforms in one way or another (Table 3.1).19 These platforms provide a variety of functions, including:

Registry functions—tracking the creation or registration of property rights and subsequent changes in ownership and location as trades take place.
Compliance functions—for example, in a system which imposes conditions for participating such as baselines, providing a secure system for tracking buyer and seller eligibility to participate in the market.
Exchange functions—digital marketplaces, online clearinghouses, online brokerage services.
Information and oversight functions— for example, historical market information, market-relevant outlooks, public reporting on market outcomes.
Trading support tools— for example, online simulation tools that help buyers such as regulated point sources estimate how many credits they need to purchase to fulfil their obligations, or that help a farmer estimate how many credits could be produced under alternative land management scenarios and taking into account location-specific factors.

Country	Programme	Environmental property right	Platform administrator
Australia	Hunter River Salinity Trading Scheme	Salinity credits	New South Wales Environmental Protection Agency
New Zealand	Lake Taupo Nitrogen Trading Program	Nitrogen allowances	Environment Waikato
Australia	State water registers e.g. Victorian Water Register ; NSW Water Register	Water-related entitlements including water access rights and water delivery rights	Victorian Water Registrar; WaterNSW
Australia	Private water exchanges e.g. Waterfind; h2Ox;	Water-related entitlements including water access rights and water delivery rights	Private entities
United States
United States	Virginia Nonpoint Source Nutrient Credit Registry	Nitrogen and Phosphorous credits	Virginia Department of Environmental Quality
United States	Ohio River Basin Trading Project online registry	Water quality credits	Electric Power Research Institute
Australia	Emissions Reduction Fund	Carbon offsets	Clean Energy Regulator
Canada	Alberta Water Tool	Water availability
United Kingdom	NFU Water Bank web-based noticeboard	Water available under abstraction licences	NFU Water
United States	Maryland Nutrient Trading Tool	Nitrogen and Phosphorous credits	Maryland Nutrient Trading Program, Maryland Dept. of Agriculture
United States	Pennsylvania Nutrient Credit Trading Program	Nutrient credits	Pennsylvania Infrastructure Investment Authority
United States	Nutrient Net	Various	World Resources Institute
United States	Virginia Nonpoint Source Nutrient Credit Registry	Nitrogen and Phosphorous credits	Virginia Dept. of Environmental Quality

Note: Hyperlinks to online platforms were accessed in September 2018. Dept. = Department.

Source: OECD Questionnaire, (Shortle, 2012[40]; Willamette Partnership, World Resources Institute and the National Network on Water Quality Trading, 2015[41]), authors.

There is little evidence available to quantify the specific benefits gained from using digital platforms for agri-environmental MBIs rather than non-digital counterparts such as paper-based registries or information and reporting regimes. Many agri-environmental MBIs are relatively new and have made use of digital platforms since their instigation; this makes it more difficult to isolate the specific benefits of using digital tools. Moreover, economics literature considering such MBIs generally focus on institutional design factors and do not describe implementation tools (digital or otherwise) in much detail. Nonetheless, benefits of using digital platforms to underpin agri-environmental MBIs appear to include:

Integration between digital tools used to quantify property rights and property right registries, leading to increased robustness of the MBI as a whole. For example integration of watershed models and registries to estimate nutrient or carbon emissions reductions which form the basis of nutrient or carbon credits or to estimate water availability in tradeable water allocation regimes.
Increased participation of buyers and sellers leading to increased market liquidity and greater “reach” for the MBI.
Improved pricing transparency as digital platforms can be automated to provide both aggregated and detailed information on trade volumes and prices.
Reduced administrative costs of processing trades.
Improved ability to report (e.g. publicly or to the relevant oversight) on market outcomes.
Improved ability to provide training to potential market participants to enable them to participate.

Table 3.1 suggests that for the majority of cases, existing digital platforms for agri-environmental MBIs are administrated by government agencies. Thus, a relevant question for governments is whether these agencies are suitably equipped (expertise, funding, etc.) to operate these platforms efficiently and effectively, or whether alternatives such as partnering with a third party or privatising the platform could improve them while maintaining their public policy objective. A first necessary step towards answering this question is to ensure that review and evaluation mechanisms for agri-environmental MBIs explicitly include an assessment of the efficiency and effectiveness of existing digital platforms.

3.2.5. Digital technologies for monitoring and compliance

Agri-environmental policies which actively seek to alter farmer behaviour (whether through regulatory or market-based mechanisms—see Annex A) fundamentally work by realigning farmers’ incentives through the introduction of conditional penalties or rewards which would not exist in the absence of the policy and which are intended to be dependent on famers’ own actions. The presence of these conditional rewards and penalties creates the potential for non-compliance to be a farmer's preferred response – if, for example, a farmer can receive a conservation payment without actually undertaking (costly) conservation actions. This problem is one form of “moral hazard” and arises because of information asymmetries between farmers and administrators – if administrators had full information about farmers’ actions they would never incorrectly apply a payment or penalty where it was not warranted.

The potential for moral hazard creates the need for the monitoring and compliance component in the policy cycle. However, these are costly activities, and therefore policy makers often opt to incompletely monitor programme participants. Digital technologies offer the potential to dramatically reduce costs of monitoring, for administrators but also for farmers. They can also facilitate different kinds of monitoring and compliance regimes (section 3.2.1 discusses this latter point).

GIS-based analytical tools are the most commonly-used digital technology currently used for monitoring and compliance of agri-environmental policies (Figure 3.10). Of the policy cycle components included in the OECD questionnaire, this component also showed the highest expected increase in use of digital technologies within the next three years, with increased use expected for all of the included technology categories except citizen science and crowdsourcing.

Figure 3.10. Current and expected future use of digital technologies for monitoring and compliance

Data from remote sensing, digital data from precision agriculture, and automation algorithms are some of the most promising technologies for improving the efficiency of monitoring and compliance in agriculture (Nikkilä et al., 2012[42]). Nash et al. (2011[43]) show that automation of compliance assessments for crop production or management standards (e.g. EU organic standards) is technically feasible in most cases (up to 90% of agricultural production rules). However, the authors note that, as of 2011, “it would be nearly impossible to realise this potential immediately due to the lack of availability of the required data in digital form”. Case Study 4 below shows how far this field has advanced in the intervening years: the initiative detailed in the case study successfully carried out automated compliance inspections based on remote sensing for several EU CAP.

Empirical evidence on the administrative savings that can be made by use of digital technologies for monitoring and compliance is limited. However, available studies show that savings can be considerable. For example, DeBoe and Stephenson (2016[44]) studied the administrative costs of water quality trading programmes in the United States, and found that using satellite data to monitor land conversion (tree planting) required on average a quarter of an hour of administrator time, compared to around ten hours for an on-site visit. Evidence from Case Study 4 (below) shows that use of a remote-sensing based digital platform for performing on-the-spot-checks required under the EU CAP can reduce administrator costs by around 25%.

While using remote sensing data as a basis for more cost-effective monitoring and compliance appears to be extremely promising, it is worth noting that remote sensing is not necessarily suitable for monitoring all types of practices (for practice-based policies), particularly certain management practices (e.g. timing of fertiliser applications). As administrators move to update policies or programmes and monitoring and compliance strategies in light of new possibilities offered by digital technologies, the focus of agri-environmental programmes or production requirements should not be confined to only practices or results which are able to be monitored remotely. Rather, administrators should make use of several digital tools in combination (e.g. both remote and in situ sensing, as well as digital analytical tools such as models) to achieve the greatest overall improvement (of which reductions in compliance costs is just one factor) in cost-effectiveness (see Case Study 1 for an example of how a range of digital tools can be combined).

Beyond making existing monitoring and compliance functions more cost-effective, digital technologies also offer the possibility of transforming compliance approaches. These possibilities are discussed in Case Study 4.

Box 3.7. Case Study 4: Earth Observation initiatives for administration of the EU CAP

Context

The European Union Common Agricultural Policy (CAP) is the overarching system of subsidies and payment programmes for agriculture and rural areas in the European Union. The CAP is fundamentally an eligibility-based system: farmers must meet certain criteria in order to receive payments. There are three main monitoring and control tools used by the relevant competent public authorities (“National Control and Paying Agencies”, NCPA): administrative checks of paperwork submitted by claimants (farmers), visual on-the-spot checks (OTSC) and Control with Remote Sensing (CwRS).

Due to the high complexity and diversity of the obligations that need to be monitored, each method has its limitations. According to DG-AGRI (Borchmann, n.d.[45]), the cost of controls to Member States in 2015 was EUR 1 125 million. The challenge for CAP administrators is therefore to minimise administrative transactions costs (both public and private) while maintaining effective standards of compliance. One crucial aspect of this is to reduce costs of obtaining information on farmers’ activities.

Digital solutions: The RECAP initiative

One initiative from the European context, RECAP—Personalised Public Services in Support for the implementation of the Common Agriculture Policy, provides evidence on the potential benefits of earth observation technologies and online digital platforms. RECAP is a commercial platform (cloud-based Software as a Service - SaaS) that integrates satellite remote sensing and user-generated data into added value services for public authorities (administrators and inspectors), farmers and agricultural consultants, to improve the processes for implementing and monitoring the Basic Payment Scheme (BPS). RECAP makes use of various digital tools comprising six “components”:

The Remote Sensing (RS) component provides automated earth observation (EO) processing workflows (including algorithms) to assist paying agency inspections with respect to farmers’ compliance to their CAP obligations.
The Spatial component depicts the information generated by the RS component as well as external spatial data in digital maps, enabling users to visualise valuable information for an effective and efficient inspection process. For example, it can be used by the NCPA as auxiliary information in their risk analysis process, to help target inspections.
The Business intelligence component is a data mining tool enabling public officers analyse large datasets stored in RECAP platform.
The Workflow component is the core system of the platform. It provides farmers, consultants and inspectors with checklists of Cross Compliance rules applicable to the farm; guides farmers and inspectors on compliance procedures; generates notifications to farmers based on calendar of key dates; etc.
The Software Development Kit (SDK) allows agricultural consultants and developers to develop their own “added value” services in an open approach within the RECAP platform, and deliver improved advisory services to farmers.
The RECAP Digital Platform: Web and mobile applications interconnect the different system components in order to deliver the deliverables earlier described. They cover 5 categories; the general system requirements, the Basic Payment Scheme (BPS) eligibility/applications, farmer record keeping, the inspection process and farmer education/information. Each farmer has their own personal account on the RECAP platform where they are able to store data, records, and documents. The platform also enables NCPAs to increase the effectiveness of risk-based analysis for the selection of farms to be inspected.

Overall, the RECAP pilot demonstrated high accuracy in identifying crop types, but also showed that the suitability of remote sensing for compliance decision-making depend in part on the nature of requirements. Further, the pilot showed that RECAP tools can reduce administrators’ costs for performing required on-the-spot-checks, increase the transparency of inspections and improve the accountability of public organisations. Further specific results indicators are available in Part IV.

Source: Case Study 4, Part IV.

3.2.6. Digital technologies for policy evaluation

The policy evaluation component of the policy cycle entails an overall assessment of the policy mechanism, both on effectiveness and efficiency aspects (as well as other aspects such as synergies and trade-offs with other policies, unintended effects etc.). There is an extensive literature on optimal evaluation design for agri-environmental policy, which notes that in general policy evaluations are not done well, and that maintaining an adequate knowledge base for policy evaluation is a central challenge impeding development of robust, comprehensive assessments (Baylis et al., 2016[46]).

Digital technologies can assist in the creation and maintenance of knowledge bases for policy evaluation, and also help foster collaboration among relevant actors to ensure that evaluations appropriately take into account both qualitative and quantitative aspects. Insofar as digital technologies gather new knowledge or allow for improved analysis (e.g. analysis of big data), they may also facilitate calculation of a wider range of policy impacts and therefore improve the robustness of policy evaluations.

Results from the OECD questionnaire show that two digital technologies were most commonly-used for evaluation of agri-environmental policies (Figure 3.11): online surveys or censuses, followed by GIS-based analytical tools. Over the next few years, more organisations are expecting to make use of online platforms, remote and in situ sensing, online surveys or censuses and data from precision agricultural machinery for policy evaluation. Together with the implementation component, use of technologies for the evaluation component showed the greatest expected expansion in use of technologies in the next three years.

Figure 3.11. Current and expected future use of digital technologies for policy evaluation

3.3. Digital technologies can open new options for agri-environmental instruments

The design of existing agri-environmental policies is constrained by information gaps, information asymmetries, transaction costs and non-aligned incentives. These challenges can be so significant as to render certain kinds of policies (or policy design elements) infeasible. With the advent of digital technologies that can overcome or drastically reduce these challenges, a re-assessment of the “feasible set” of policy options is warranted; that is, digital technologies can open up new options for agri-environmental instruments.

This section focusses on three frontiers where digital technologies can facilitate expansion of the current feasible set of policy options:

Enabling policies based on monitoring all participants rather than relying on controlling a sample of participants and strong negative incentives to compel compliance (section 3.3.1).
Moving towards robust, cost-effective outcomes-based agri-environmental policies (section 3.3.2).
Augmenting agri-environmental extension models with distributed knowledge networks and machine learning (section 3.3.3).

3.3.1. Technologies to enable new monitoring and compliance approaches

In the calculus of designing compliance and enforcement mechanisms, the policy-maker’s objective is to ensure participating actors (farmers, in the case of most agri-environmental measures) find it more profitable to comply than not to comply.20 When it is costly to detect non-compliance (as is generally the case), a common strategy for the administrator is to monitor only a small subset of total participants, and design penalties such that the expected profit from compliance exceeds that from non-compliance, according to the following formula (Becker, 1968[47]; Stefani and Giudicissi, 2011[48]):

$p \times n o n c o m p l i a n c e p e n a l t y \geq c o s t o f c o m p l i a n c e$

In this formula, “p” refers to the probability of detection, which (assuming that the administrator always correctly detects non-compliance for those agents it monitors) is equal to the proportion of the total population of participants who are monitored. Generally, this proportion is quite low – for example, under the EU Common Agricultural Policy (CAP) for the 2014-2020 period, national paying agencies are required to perform yearly on-the-spot-checks for at least 5% of beneficiaries (IACS21 measures) or 5% of expenditure (non-IACS) measures (in addition to 100% administrative checks) (Borchman, date NA). When the probability of undergoing on-site monitoring is low, the regulator is “information poor”, and the penalty for non-compliance needs to be high enough to serve as a deterrent (Macey, 2013[49]).

As discussed in section 3.2.5, digital technologies offer the opportunity to improve monitoring within existing compliance frameworks; that is, without changing policy settings such as minimum requirements about who and when to monitor, and penalties for non-compliance. However, as the above formula shows, increasing the probability of detection allows for a corresponding decrease in the penalty for non-compliance (other things equal). Therefore, digital technologies offer administrators the opportunity to radically recalibrate their enforcement approach.

In fact, administrators may be enabled by technology to move towards new “data intensive” compliance approaches based on high (near-100%) rates of remote monitoring. A discussion paper by the European Union’s Joint Research Centre (JRC) (Devos et al., 2017[50]), considers the possibility for substituting on-the-spot-checks (OTSC) for administration of the EU Common Agricultural Policy (CAP) by a system of monitoring for checking the fulfilment of land use and land cover related CAP requirements. The envisaged system would cover 100% of relevant territory. The discussion paper describes the main components of such an approach and considers the technical and regulatory requirements needed to make it operational (see also case study in section 3.2.5).

Low-cost web-based mechanisms in which participants self-assess compliance and report on their performance, using a range of digital technologies to substantiate compliance claims is also a promising digitally-enabled compliance tool. Such portals can be linked to independent verification mechanisms such as sensor data or online registers to reduce incentives to misreport data.

Another digital technology which may enable new compliance approaches is distributed ledger technology (DLT) (e.g. blockchains). DLT allows for centralisation of information at the same time as decentralisation of access to and ownership of databases, as well as providing low-cost tools for auditing, authentication and traceability. As such, DLT, especially when used in combination with other technologies such as sensors and precision agriculture machinery, may allow more holistic compliance approaches in which intangible or credence attributes of agricultural products (e.g. social and sustainability aspects) are verified and tracked throughout agricultural supply chains. DLT could also be a tool for encouraging collective approaches in which individuals (e.g. farmers, environmentalists and even the general public) can contribute data to demonstrate compliance or achievement of programme objectives. Further, insofar as Blockchain applications can facilitate price premiums for differentiated goods (via providing robust and verifiable information and certifications to consumers), Blockchains may enable a shift from government- to market-provided incentives, which may reduce or remove the need for regulation, or change the nature of regulatory requirements. However, there are to date very few examples of such initiatives, and further work is required to investigate how distributed ledgers could enable new policy designs.

Finally, digital technologies can enable a holistic compliance approach in contexts where (environmental) regulation applies to several different kinds of entities. Sometimes regulators may have adopted compliance approaches which monitor different entities differently, for example because of legacy issues or because of heterogeneous costs of obtaining data across entities. While such a differentiated approach may continue to be warranted, digital technologies may introduce new lines of evidence for regulators for which costs are more homogeneous across different entities, or which allow a better view of the “whole picture”. Box 3.8 provides an example from the context of the Murray-Darling Basin, Australia.

Box 3.8. Satellite imagery to improve compliance with water allocation frameworks

Digital technologies can help administrators to take a systematic approach to monitoring and enforcing environmental policies applying to agriculture but also involving other actors. This box provides an example from Australia, demonstrating the usefulness of satellite imagery as one tool to monitor compliance with water allocation frameworks and monitor ecological outcomes over large spatial scales.

In the Australian Murray-Darling Basin (MDB), agriculture is the largest water user (BoM, NA). However, as a result of government policy recognising the need to return more water for the environment, a new class of water users has emerged: “environmental water holders” (EWHs).1 EWHs provide environmental flows and undertake watering activities throughout the MDB. In order for these environmental flows to fully achieve their objectives, they require protection from extraction by consumptive users, including agriculture. Therefore, the compliance regime needs to monitor different kinds of water “use” (i.e. consumptive and non-consumptive uses), over very large spatial scales, and in an environment characterised by highly varied water availability from year to year.

In recent years, several compliance incidents, including those connected with a limited number of agricultural water users, have highlighted the need for a more proactive approach to environmental water protection. While on-the-ground monitoring already exists in many areas (e.g. water meters, river gauging stations), it is difficult to gain a holistic compliance picture from these data sources alone. Recognising these limitations, in early-to-mid 2018 the Murray-Darling Basin Authority (MDBA) undertook a trial of the use of free, open and publicly available “Sentinel 2” satellite imagery to track and measure a specific environmental flow event. The trial represents the first large-scale use of satellite imagery in this context within the Murray-Darling Basin. The trial aimed to test the ability of satellite imagery to: 1) successfully track the watering event as it progressed through the river system; and 2) measure the degree to which water was present in farm dams and storages during the event, and how they changed over time, as significant changes could indicate compliance issues (since a restriction was placed on consumptive use extractions while this flow event took place).

The MDBA concluded that the trial successfully showed that remote sensing using satellite imagery provides a very important line of evidence for supporting compliance activities. These tools and their associated data offer the opportunity to observe the behaviour of water moving through the landscape, as well as water present in farm dams and storages and crop presence at a range of scales – from farm to catchment level and in close to real-time. These observations provide MDB water agencies with a new line of information, which may be used to trigger further investigation or other compliance responses as appropriate. Further work is underway to test the methods and build the systems required for application more broadly across the MDB.

Digital technologies other than satellite imagery and related data processing software also played a role in making the trial a success. In particular, community events were organised to inform stakeholders about the trial, with associated media releases and (in both digital and non-digital formats) factsheets, as well as social media posts to provide information. The trial also involved a range of actors and relied on freely available data provided by international institutions, highlighting the benefits of sharing data across borders (international data exchange) and between actors in a co-operative effort to improve compliance outcomes.

Notes: Technical details of the trial are publicly available in the MDBA report.

1. The most significant EWH is the Commonwealth Environmental Water Holder. See https://www.environment.gov.au/water/cewo, accessed February 2019.

Sources: Adapted from MDBA (2018) A case study for compliance monitoring using satellite imagery: the Northern Connectivity Event; Australian Bureau of Meteorology (BoM) National Water Account 2016.

Case Study 5 (Box 3.9) provides a practical example of how a range of digital technologies and data transparency tools can be used as part of a broader strategy to improve the flexibility and robustness of regulatory environmental programmes.

Box 3.9. Case Study 5: Digital technologies applied by USEPA to achieve Innovative Compliance

This case study provides a practical example of how digital technologies and data transparency tools can be used as part of a broader strategy to improve the flexibility and robustness of regulatory environmental programmes.

The United States Environmental Protection Agency (USEPA) implements national environmental law by writing and enforcing regulations, setting national standards that US states and tribes enforce, and assisting regulated entities to understand the requirements (USEPA, date NA). USEPA administers a range of US federal legislation relevant to agriculture, which has both regulatory and non-regulatory components (see additional description below).

During 2010-2013, USEPA self-identified a range of issues or areas for applying innovative compliance approaches, including: “gaps in information about the compliance status of regulated entities, unacceptably high rates of noncompliance, deficiencies in state enforcement of delegated programs, and substantial shortcomings in managing (collecting and transmitting) compliance-related information” (Markell and Glicksman, 2015[51]).

To address these gaps and improve the cost-effectiveness of USEPA’s compliance programme, USEPA began exploring innovative compliance tools in 2012 and activities such as the use of optical gas imaging cameras, electronic reporting, and working to improve the effectiveness of regulations and permits, have become more commonplace. Types of tools are:

Advanced monitoring and information technologies: Real-time continuous monitoring generates actual measurements (as opposed to estimates), which reduces information gaps and information asymmetries between the regulator and the regulated entities
Electronic reporting: E-reporting saves time, reduces error, enables automatic checks and triaging to help target monitoring and enforcement activities, reducing transaction costs of compliance and enforcement activities.
Transparency—public disclosure requirements: Increased public disclosure provides incentive for entities to improve their environmental performance via reputation effects.
Rule and permit design—“Compliance-ready” technology and rules with “compliance built in”.1

Use of innovative compliance tools for agri-environmental policy implementation can broadly be separated into applications in regulatory (permit) and non-regulatory (voluntary) contexts.

In the US agriculture sector, some concentrated animal feeding operations (CAFOs) such as certain feedlots, dairies and poultry houses are “regulated by EPA under the Clean Water Act in both the 2003 and 2008 versions of the “CAFO rule” (40-CFR) (NRCS, date NA). These regulations underpin a permitting system known as the National Pollutant Discharge Elimination System (NPDES). Innovative compliance tools are applied for NPDES permittees via several avenues (see full case study for details), including electronic reporting; innovative compliance components in permitting; and innovative compliance tools used in rule-making.

Agricultural enterprises that are not required to obtain a permit may nevertheless choose to participate in a range of voluntary agri-environmental programmes. In this voluntary context, when a producer enters into a voluntary contract for provision of environmental goods, innovative compliance tools can be used in much the same manner as in a regulatory context, with compliance with the terms of the contract taking the place of compliance with a permit. Certification programmes (e.g. organic labelling) can also implement innovative compliance tools as part of the certification process. Programme administrators and producers can also make use of some of the innovative compliance tools—particularly electronic reporting in conjunction with transparency—to foster public support for entirely voluntary environmental efforts (i.e. even those which do not use contracts or any other form of legal enforcement mechanism).

This case study showed that technological change offers new possibilities for improved monitoring and compliance. However, there needs to be clear and fit-for-purpose processes for demonstrating suitability of advanced monitoring tools for regulatory purposes, which may differ from existing processes. However, this may be challenging to achieve in a context of fast-paced technological change. Given the rapid pace of technological change of relevant technologies, existing processes for vetting new technologies for regulatory purposes, particularly ones which take several years to complete, need to become more agile.

1. Recognising that enforcement action alone will not produce full compliance in every instance, this component entails promoting the use of technology, transparency, and other tools. Similarly, rules can be designed to require use of certain technology or processes by upstream manufacturing rather than attempting to regulate the use of technology by end users, e.g. auto manufacturers are required to install pollution control devices rather than requiring automobile buyers to do so (Giles, 2013[52]).

Source: Case Study 5, Part IV.

3.3.2. Result-based agri-environmental policies and modelling versus measuring

Numerous studies (e.g. (Burton and Schwarz, 2013[53]; Savage and Ribaudo, 2016[27]; Shortle et al., 2012[54]) have pointed out various flaws in agri-environmental policies which pay farmers to implement practices linked to the production of environmental services. These authors typically contrast such policies which “pay-for-practices” unfavourably with “result-oriented” policies, which reward producers directly for achieving specific environmental outcomes. However framework based a myriad of policy options are in fact available: e.g. uniform-pay-for-practice, spatially-differentiated-pay-for-practice, pay-for-modelled-edge-of-field-performance, pay-for-modelled-results, pay-for-measured-results. While a full elaboration of this spectrum is beyond the scope of the current project, even the brief description laid out above clearly suggests that the role of digital technologies differs across the spectrum.

Table 3.2 compares how different technology categories (refer to Table 2.1) could be used in two stylised representations of result-based programmes, and also a practice-based programme. This table is not intended to comprehensively list all the ways digital technologies could be used, but rather to compare key points of difference in technology use.

	Action-based programme Cost-share for installation of water quality (WQ) best management practices (BMP)	Modelled result-based programme: Water quality (WQ) trading	Measured result-based programme Farmland birds
Programme description	Payments to individual farmers based on contracts to install and maintain WQ structures or to implement WQ management actions. Payments based on cost.	Market-based mechanism based on capping total amount of nutrient (N, P) emissions in a watershed and allowing trading of nutrient emission reduction credits.	Collective payment based on reaching objectives for bird populations at the landscape level; allocated to land managers based on individual habitat results
Data collection technologies	Remote and in situ sensing to collect data to assess WQ state & impact pathways Online surveys & data submission portals for collecting BMP cost data Remote sensing, geo-referenced digital photographs and data from precision agriculture as model input and to monitor BMP implementation and maintenance	Remote and in situ sensing to collect data to assess WQ state and impact changes, model input Data from precision agriculture (e.g. on fertiliser applications and tillage practices) as model input	In situ sensing to collect field-level data on bird populations (e.g. nesting sites, number of birds) Crowdsourcing / citizen science apps to collect data on birds (e.g. in public spaces)
Data analysis technologies	Software for processing sensor data Watershed model (e.g. SWAT ^a) to estimate BMP effectiveness GIS-based analysis to support planning for BMP installation	Software for processing sensor data Watershed model (e.g. SWAT ^a) to estimate nutrient reductions and credit generation, and to estimate overall WQ impacts	Software for processing sensor data GIS-based analytical tools for tracking bird populations at a landscape level and collective planning of actions to maintain / improve bird habitats Machine learning / algorithms to analyse bird migration patterns
Data storage technologies	Secure storage of programme participants’ personal details Cloud storage of shared data assets (e.g. maps, model data)	Secure storage of programme participants’ personal details Cloud storage of shared data assets (e.g. maps, model data)	Secure storage of programme participants’ personal details Cloud storage of shared data assets (e.g. maps, model data)
Data management technologies	Interoperability programs	Interoperability programs Secure credit register	Interoperability programs
Data transfer and sharing: Digital communications	Online extension services / technical assistance Web-conferencing (e.g. to support processes for agreeing average BMP effectiveness parameters)	Online extension services / technical assistance	Online extension services / technical assistance Social media / apps for communication between farmers
Trading, payment and service delivery platforms	Secure e-payments	Secure e-payments Secure online access to credit register Online marketplace / auction platform	Secure e-payments Online platform tracking bird observations

Note: The stylised policies represented in this table are not intended to represent any specific existing policy.

a. SWAT = Soil and Water Assessment Tool, a public domain model jointly developed by USDA Agricultural Research Service (USDA-ARS) and Texas A&M AgriLife Research. See https://swat.tamu.edu/, accessed September 2018.

Source: Authors.

Transitioning towards more result-based policies would appear to suggest an increasing role for sensor technologies to directly measure results. However, while this may be the case in some contexts, it may not always be: policies which pay based on some kind of outcome or performance measure are in most cases still based on modelled results rather than direct measurements. Water quality programmes are typical examples of policies which implement results-based payments using modelled outcomes (see third column of Table 3.2 as an example).

Even policies which are based on measured outcomes are likely to still make use of models in some way. For example, a policy paying for measured incremental nutrient loadings reductions in a downstream water body may still make some use of models to undertake initial validation of sensor technologies for use as approved measurement technologies in the programme, to establish overall baselines or regulatory targets, etc.

Thus, the most significant factors determining the types of technologies and their relative contributions may not necessarily be whether payments are made based on actions versus results, but rather:

The extent to which policies are based on modelled outcomes versus measured outcomes (whether BMP performance, edge-of-field performance, or environmental outcomes which may be downstream or at the landscape level).
The nature of the policy mechanism: e.g. trading programmes can make use of digital trading (e.g. online marketplaces), whereas a programme where the administrator pays participants directly may have more need for secure online payment mechanisms.
The institutional setting of the policy mechanism, particularly whether the programme requires collective or coordinated action (e.g. policy takes a landscape approach) rather than only individual actions: policies taking a collective or co-ordinated approach are likely to have greater demand for digital technologies which facilitate communication between participants and which provide landscape-level analysis.

Results-based programmes generally have the feature that provision of incentives to programme participants is linked to measured or modelled results. Another important dimension of such policies is the possibility of setting programme objectives or requirements which are adaptable to environmental results. Examples could include:

policies governing water access and use which are linked to river, aquifer or storage levels;
habitat maintenance requirements (e.g. restrictions on mowing) directly linked to monitoring of bird populations;
livestock management policies directly linked to monitoring of large carnivore populations (i.e. in a context where a policy goal is to restore large carnivore populations in or adjacent to a livestock area).

Such policies are promising for fostering an agricultural sector are “nature inclusive” and resilient to frequent changes in environmental conditions. However, they rely on technology to provide robust information on highly variable environmental phenomena in (near-)real-time, potentially over large areas. This will likely require establishment of wireless sensor networks or other in situ data collection technologies, as remote sensing, despite recent advances, does not yet provide continuous monitoring.

3.3.3. Digital networks, platforms, AI and machine learning for policy communication and extension

Extension services play a vital role in increasing farmers’ agronomic skills and their understanding of the productivity and sustainability of impacts of their actions. Depending on the particular policy mechanism selected, extension services can be the key policy focus or can complement other aspects of policy implementation. Further, extension services can also be used to educate the broader group of stakeholders (e.g. environmental groups, the general public) about efforts undertaken to improve agricultural sustainability and about how other stakeholders can participate.

A key challenge for extension services is how to maximise effectiveness given limited resources. In many instances, extension officers may not be accessible when the farmer needs them (Nguyen and Thai-Nghe, 2016[55]), whether due to high travel costs, because ratios of extension officers to farmers are too low to service all demand or other factors.

Extension providers (both public and private) are establishing distributed digital communication networks to improve access to extension services, and also to facilitate peer-to-peer learning. Digital communication networks can reduce costs of communication, provide improved human-human interaction, especially over large distances, and improve educational outcomes (e.g. by providing interactive learning environments) (Kelly, McLean Bennett and Starasts, 2017[56]).

Machine learning goes a step further by making use of artificial neural networks to analyse specific problems faced by farmers and provide semi-automated recommendations. Recent initiatives to develop neural networks to supplement human-delivered extension programmes—exemplified by the work of Mohanty et al. (2016[57]) and Sladojevic et al. (2016[58])—are developing software for the automated classification of crop diseases using deep neural networks.

Another technology for reducing the cost of communicating with diverse audiences is the use of algorithms (e.g. natural language generation (NLG) algorithms) to automate translation of data into easily digestible communications and automated notifications of alerts (e.g. notifying farmers of pest or disease outbreaks, weather-related risks, but also notifications of extension opportunities such as webinars, field days, or training dates). Section 3.2.3 provided examples of using NLG-based automation to facilitate communication between administrators and programme participants, but automated communication can be equally useful for extension services and communicating about agri-environmental policy more generally, including with the general public.

Interactive digital platforms are another useful tool for extension services and wider communication about agri-environmental policies. Examples are shown in Table 3.3. Many of these examples are not specifically focussed on agriculture; rather, they provide for agri-environmental impacts or initiatives to be considered as part of a more holistic picture. They can therefore have the added benefit of forging common ground between agriculture, other sectors of the economy and the general public.

Platform	Country	Description	Website
Atlas of Living Australia	AUS	A collaborative, national project that aggregates biodiversity data from multiple sources and makes it freely available and usable online.	https://www.ala.org.au/
Agroclimatic Observatory	CHL	An interactive collection of maps and other figures that monitor drought at present, near future and in the recent past. The maps and figures can be manipulated and are linked to the original data.	http://www.climatedatalibrary.cl/IMP-DGIR/maproom/?Set-Language=en
Online suite of geospatial products	CAN	A suite of interactive agriculture-related maps, geospatial data and tools to help users make better decisions for environmentally responsible yet competitive agriculture.	http://www.agr.gc.ca/eng/?id=1343066456961
MAGIC	GBR	An interactive online map providing authoritative geographic information about the natural environment from across government. The information covers rural, urban, coastal and marine environments across the United Kingdom.	https://magic.defra.gov.uk/
Korean Soil Information System (KSIS)	KOR	KSIS is an online portal that provides soil information and recommends agricultural crops and fertiliser application amounts based on soil characteristics. The system is based on soil information and agri-environmental data collected by Rural Development Administration of Korea.	http://soil.rda.go.kr
Akkerweb	NLD	An open platform for digital services for precision farming. See Case Study 9 for further information.	https://akkerweb.eu/en-gb/
Interactive Emissions Tracker	NZL	An interactive tool summarising New Zealand’s Greenhouse Gas Inventory.	https://emissionstracker.mfe.govt.nz/
EnviroAtlas	USA	A set of online, interactive tools allowing users to discover, analyse, and download data and maps related to ecosystem services.	https://www.epa.gov/enviroatlas
Sound Impacts	USA	An online portal for all of practitioners and implementers of Green Infrastructure (which could include a range of actors such as government, industry, farmers, or even a member of the public) as well as for “anyone curious” to see what efforts and investments are being made to protect and improve the natural assets of the Puget Sound area.	http://www.soundimpacts.org/projects/list/tracts/all/

Note: Web links accessed September 2018.

Source: OECD Questionnaire and authors’ own work.

References

[10] AgResearch; Landcare Research; GNS Science (ed.) (2009), Land Use Capability Survey Handbook - a New Zealand handbook for the classification of land, http://www.landcareresearch.co.nz (accessed on 7 August 2018).

[22] Antle, J. (2019), “Data, Economics and Computational Agricultural Science”, American Journal of Agricultural Economics, Vol. 101/2, pp. 365-382, http://dx.doi.org/10.1093/ajae/aay103.

[38] Arbuckle, J. (2013), “Farmer Attitudes toward Proactive Targeting of Agricultural Conservation Programs”, Society and Natural Resources, http://dx.doi.org/10.1080/08941920.2012.671450.

[13] Arts, K. et al. (2015), “Supply of Online Environmental Information to Unknown Demand: The Importance of Interpretation and Liability Related to a National Network of River Level Data”, Scottish Geographical Journal, Vol. 131/3-4, pp. 245-252, http://dx.doi.org/10.1080/14702541.2014.978809.

[73] Arts, K., R. van der Wal and W. Adams (2015), “Digital technology and the conservation of nature”, Ambio, http://dx.doi.org/10.1007/s13280-015-0705-1.

[66] Balla, S. and B. Daniels (2007), “Information technology and public commenting on agency regulations”, Regulation & Governance, Vol. 1, pp. 46–67, http://dx.doi.org/10.1111/j.1748-5991.2007.00005.x.

[46] Baylis, K. et al. (2016), “Mainstreaming Impact Evaluation in Nature Conservation”, Conservation Letters, Vol. 9/1, pp. 58–64, http://dx.doi.org/10.1111/conl.12180.

[47] Becker, G. (1968), “Crime and Punishment: An Economic Approach”, Journal of Political Economy, Vol. 76, pp. 169-217, https://www.journals.uchicago.edu/doi/pdfplus/10.1086/259394 (accessed on 7 March 2018).

[45] Borchmann, C. (n.d.), CAP audit system 2014-2020, European Commission, DG-AGRI, https://ec.europa.eu/futurium/en/system/files/ged/c._borchmann_dg_agri_eafrd_audits_-_facts_and_figures.pdf (accessed on 20 August 2018).

[59] Börner, J. et al. (2017), “The Effectiveness of Payments for Environmental Services”, World Development, Vol. 96, pp. 359-374, http://dx.doi.org/10.1016/J.WORLDDEV.2017.03.020.

[12] Brandon, B. and R. Carlitz (2002), “Online rulemaking and other tools for strengthening our civil infrastructure”, Administrative Law Review, Vol. 54/4, pp. 1421-1478, https://www.jstor.org/stable/40712129?newaccount=true&read-now=1&seq=3#metadata_info_tab_contents (accessed on 13 September 2018).

[53] Burton, R. and G. Schwarz (2013), Result-oriented agri-environmental schemes in Europe and their potential for promoting behavioural change, http://dx.doi.org/10.1016/j.landusepol.2012.05.002.

[68] Claassen, R., A. Cattaneo and R. Johansson (2008), “Cost-effective design of agri-environmental payment programs: U.S. experience in theory and practice”, Ecological Economics, Vol. 65/4, pp. 737-752, http://dx.doi.org/10.1016/J.ECOLECON.2007.07.032.

[64] Coggan, A., S. Whitten and J. Bennett (2010), “Influences of transaction costs in environmental policy”, Ecological Economics, Vol. 69/9, pp. 1777-1784, http://dx.doi.org/10.1016/J.ECOLECON.2010.04.015.

[17] Crase, L., S. O’Keefe and B. Dollery (2013), “Talk is cheap, or is it? The cost of consulting about uncertain reallocation of water in the Murray–Darling Basin, Australia”, http://dx.doi.org/10.1016/j.ecolecon.2012.12.015.

[44] DeBoe, G. and K. Stephenson (2016), “Transactions costs of expanding nutrient trading to agricultural working lands: A Virginia case study”, Ecological Economics, http://dx.doi.org/10.1016/j.ecolecon.2016.06.027.

[50] Devos, W. et al. (2017), Discussion document on the introduction of monitoring to substitute OTSC supporting non-paper DS/CDP/2017/03 revising R2014/809., Publications Office of the European Union.

[23] Engel, S. (2016), “The Devil in the Detail: A Practical Guide on Designing Payments for Environmental Services”, International Review of Environmental and Resource Economics, http://dx.doi.org/10.1561/101.00000076.

[6] Gholizadeh, M., A. Melesse and L. Reddi (2016), “A Comprehensive Review on Water Quality Parameters Estimation Using Remote Sensing Techniques”, Sensors, Vol. 16, http://dx.doi.org/10.3390/s16081298.

[52] Giles, C. (2013), Next Generation Compliance, The Environmental Law Institute, Washington DC, http://www.eli.org. (accessed on 16 August 2018).

[14] Han, X. et al. (2014), Latent User Models for Online River Information Tailoring, Association for Computational Linguistics, Philadelphia, Pennsylvania,, http://sepa.org.uk/water/river_levels/river_level_data.aspx (accessed on 13 September 2018).

[71] Heimlich, R. (2005), The policy-related transactions costs of land conservation in the United States: evolution over time and comparison between programmes, http://www.marketvolume.com/glossary/t0282.asp (accessed on 17 September 2018).

[56] Kelly, N., J. McLean Bennett and A. Starasts (2017), “Networked learning for agricultural extension: a framework for analysis and two cases”, The Journal of Agricultural Education and Extension; Competence for Rural Innovation and Transformation, Vol. 23/5, http://dx.doi.org/10.1080/1389224X.2017.1331173.

[31] Klimek, S., G. Lohss and D. Gabriel (2014), “Modelling the spatial distribution of species-rich farmland to identify priority areas for conservation actions”, Biological Conservation, Vol. 174, pp. 65-74, http://dx.doi.org/10.1016/J.BIOCON.2014.03.019.

[72] Krutilla, K. and R. Krause (2011), “Transaction Costs and Environmental Policy: An Assessment Framework and Literature Review”, International Review of Environmental and Resource Economics, Vol. 4/3-4, pp. 261-354, http://dx.doi.org/10.1561/101.00000035.

[24] Lankoski, J. (2016), “Alternative Payment Approaches for Biodiversity Conservation in Agriculture”, OECD Food, Agriculture and Fisheries Papers, No. 93, OECD Publishing, Paris, http://dx.doi.org/10.1787/5jm22p4ptg33-en.

[60] Lawrence, E. and S. Estow (2017), “Responding to misinformation about climate change”, Applied Environmental Education & Communication, Vol. 16/2, pp. 117-128, http://dx.doi.org/10.1080/1533015X.2017.1305920.

[49] Macey, G. (2013), “The Architecture of Ignorance”, Utah L. Rev, Vol. 21/1627.

[15] Macleod, K. et al. (2012), “Communicating river level data and information to stakeholders with different interests: the participative development of an interactive online service”, International Congress on Environmental Modelling and Software, https://scholarsarchive.byu.edu/iemssconference/2012/Stream-B/327 (accessed on 13 September 2018).

[51] Markell, D. and R. Glicksman (2015), “Next Generation Compliance”, Natural Resources & Environment, Vol. 30, p. 22, https://heinonline.org/HOL/Page?handle=hein.journals/nre30&id=160&div=&collection= (accessed on 16 August 2018).

[69] Marshall, G. (2013), “Transaction costs, collective action and adaptation in managing complex social–ecological systems”, Ecological Economics, Vol. 88, pp. 185-194, http://dx.doi.org/10.1016/J.ECOLECON.2012.12.030.

[62] McCann, L. (2013), “Transaction costs and environmental policy design”, Ecological Economics, Vol. 88, pp. 253-262, http://dx.doi.org/10.1016/J.ECOLECON.2012.12.012.

[67] McCann, L. et al. (2005), “Transaction cost measurement for evaluating environmental policies”, Ecological Economics, http://dx.doi.org/10.1016/j.ecolecon.2004.08.002.

[70] McCann, L. and K. Easter (2000), “Estimates of Public Sector Transaction Costs in NRCS Programs”, Journal of Agricultural and Applied Economics, Vol. 32/03, pp. 555-563, http://dx.doi.org/10.1017/S1074070800020642.

[9] McDowell, R. et al. (2017), Research landscape map for the Our Land and Water National Science Challenge (2nd Edition), http://www.ourlandandwater.nz (accessed on 17 August 2018).

[8] McDowell, R. et al. (2018), “The land use suitability concept: Introduction and an application of the concept to inform sustainable productivity within environmental constraints”, Ecological Indicators, Vol. 91, pp. 212-219, http://dx.doi.org/10.1016/J.ECOLIND.2018.03.067.

[5] Medyckyj-Scott, D. et al. (2016), Our Land and Water National Science Challenge: A Data Ecosystem for Land and Water Data to Achieve the Challenge Mission, AgResearch, Hamilton, New Zealand, http://www.ourlandandwater.nz/assets/Uploads/Our-Land-and-Water-Data-Ecosystem-White-Paper.pdf (accessed on 7 August 2018).

[25] Meyer, C. et al. (2015), “Design rules for successful governmental payments for ecosystem services: Taking agri-environmental measures in Germany as an example”, Journal of Environmental Management, http://dx.doi.org/10.1016/j.jenvman.2015.03.053.

[57] Mohanty, S., D. Hughes and M. Salathé (2016), “Using Deep Learning for Image-Based Plant Disease Detection”, Frontiers in Plant Science, Vol. 7, p. 1419, http://dx.doi.org/10.3389/fpls.2016.01419.

[16] Molina, M., J. Sanchez-Soriano and O. Corcho (2015), “Using Open Geographic Data to Generate Natural Language Descriptions for Hydrological Sensor Networks.”, Sensors (Basel, Switzerland), Vol. 15/7, pp. 16009-26, http://dx.doi.org/10.3390/s150716009.

[26] Moxey, A. and B. White (2014), “Result-oriented agri-environmental schemes in Europe: A comment”, Land Use Policy, http://dx.doi.org/10.1016/j.landusepol.2014.04.008.

[43] Nash, E. et al. (2011), “Towards automated compliance checking based on a formal representation of agricultural production standards”, Computers and Electronics in Agriculture, Vol. 78/1, pp. 28-37, http://dx.doi.org/10.1016/J.COMPAG.2011.05.009.

[55] Nguyen, C. and N. Thai-Nghe (2016), “An Agricultural Extension Support System on Mobile Communication Networks”, http://dx.doi.org/10.1109/ATC.2015.7388386.

[63] Nguyen, N. et al. (2013), “Water quality trading with asymmetric information, uncertainty and transaction costs: A stochastic agent-based simulation”, Resource and Energy Economics, Vol. 35/1, pp. 60-90, http://dx.doi.org/10.1016/J.RESENEECO.2012.09.002.

[42] Nikkilä, R. et al. (2012), “A service infrastructure for the representation, discovery, distribution and evaluation of agricultural production standards for automated compliance control”, Computers and Electronics in Agriculture, Vol. 80, pp. 80-88, http://dx.doi.org/10.1016/j.compag.2011.10.011.

[74] OECD (2014), “National strategies for science, technology and innovation”, in OECD Science, Technology and Industry Outlook 2014, OECD Publishing, Paris, http://dx.doi.org/10.1787/sti_outlook-2014-6-en.

[4] OECD (2014), Recommendation on Digital Government Strategies - OECD, http://www.oecd.org/gov/digital-government/recommendation-on-digital-government-strategies.htm (accessed on 11 September 2018).

[39] OECD (2013), Policy Instruments to Support Green Growth in Agriculture, OECD Green Growth Studies, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264203525-en.

[18] OECD (2013), Providing Agri-environmental Public Goods through Collective Action, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264197213-en.

[29] OECD (2012), Evaluation of Agri-environmental Policies: Selected Methodological Issues and Case Studies, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264179332-en.

[2] OECD (2010), Guidelines for Cost-effective Agri-environmental Policy Measures, OECD Publishing, Paris, http://dx.doi.org/10.1787/9789264086845-en.

[65] OECD (2010), Paying for biodiversity: enhancing the cost-effectiveness of payments for ecosystem services, OECD Publishing, http://dx.doi.org/10.1787/9789264090279-en (accessed on 19 February 2018).

[1] OECD (2008), Agricultural policy design and implementation: a synthesis, OECD Publishing, https://www.oecd.org/tad/agricultural-policies/40477848.pdf.

[11] Picazo-Vela, S. and I. Gutiérrez-Martínez (2012), “Understanding risks, benefits, and strategic alternatives of social media applications in the public sector”, Government Information Quarterly, Vol. 29/4, pp. 504-511, http://dx.doi.org/10.1016/J.GIQ.2012.07.002.

[35] Poulton, P. et al. (2018), Evaluating the utility of photogrammetry to identify and map regions at risk from gully erosion. Report to Department of Agriculture and Water Resources., CSIRO Agriculture and Food, https://publications.csiro.au/rpr/pub?list=SEL&pid=csiro:EP18475&sb=RECENT&expert=false&n=1&rpp=550&page=1&tr=1&q=PID%3A%22csiro%3AEP18475%22&dr=all (accessed on 9 August 2018).

[19] Prager, K. (2015), “Agri-environmental collaboratives for landscape management in Europe”, Current Opinion in Environmental Sustainability, Vol. 12, http://dx.doi.org/10.1016/j.cosust.2014.10.009.

[20] Prager, K., M. Reed and A. Scott (2012), “Encouraging collaboration for the provision of ecosystem services at a landscape scale-Rethinking agri-environmental payments”, Land Use Policy, http://dx.doi.org/10.1016/j.landusepol.2011.06.012.

[32] Rabotyagov, S. et al. (2014), “Cost-effective targeting of conservation investments to reduce the northern Gulf of Mexico hypoxic zone”, Proceedings of the National Academy of Sciences, http://dx.doi.org/10.1073/pnas.1405837111.

[34] Reed, M. et al. (2014), Improving the link between payments and the provision of ecosystem services in agri-environment schemes, http://dx.doi.org/10.1016/j.ecoser.2014.06.008.

[7] Rissman, A. and S. Carpenter (2015), “Progress on Nonpoint Pollution: Barriers & Opportunities”, Dædalus, the Journal of the American Academy of Arts & Sciences, Vol. 143/3, pp. 35-47, http://dx.doi.org/10.1162/DAED_a_00340.

[27] Savage, J. and M. Ribaudo (2016), “Improving the Efficiency of Voluntary Water Quality Conservation Programs”, Land Economics, Vol. 92/1, pp. 148–166, E-ISSN 1543-8325.

[40] Shortle, J. (2012), Water Quality Trading in Agriculture, OECD Publishing, Paris, https://agsci.psu.edu/enri/news/spotlight/water-quality-trading-in-agriculture (accessed on 11 September 2018).

[54] Shortle, J. et al. (2012), “Reforming agricultural nonpoint pollution policy in an increasingly budget-constrained environment”, Environmental Science and Technology, http://dx.doi.org/10.1021/es2020499.

[58] Sladojevic, S. et al. (2016), “Deep Neural Networks Based Recognition of Plant Diseases by Leaf Image Classification”, Computational Intelligence and Neuroscience, Vol. 2016, pp. 1-11, http://dx.doi.org/10.1155/2016/3289801.

[61] Stavins, R. (1995), “Transaction Costs and Tradeable Permits”, Journal of Environmental Economics and Management, Vol. 29/2, pp. 133-148, http://dx.doi.org/10.1006/JEEM.1995.1036.

[48] Stefani, G. and E. Giudicissi (2011), Cross Compliance: what about compliance?, http://agris.fao.org/agris-search/search.do?recordID=US2016216307 (accessed on 7 March 2018).

[21] The Netherlands Ministry of Economic Affairs (2016), The cooperative approach under the new Dutch agri-environment- climate scheme Background, procedures and legal and institutional implications, https://enrd.ec.europa.eu/sites/enrd/files/w12_collective-approach_nl.pdf.

[36] Tindall, D. (2014), Gully mapping and drivers in the grazing lands of the Burdekin catchment: RP66G Summary Report, Queensland Department of Science, Information Technology, Innovation and the Arts Remote Sensing Centre, https://publications.qld.gov.au/dataset/gully-mapping-burdekin/resource/40a7bb78-8092-4056-a7bf-93deb3ef5af3 (accessed on 8 August 2018).

[3] United Nations (2016), United Nations E-Government Survey 2016: E-Government in support of sustainable development, United Nations, https://publicadministration.un.org/egovkb/Portals/egovkb/Documents/un/2016-Survey/Executive%20Summary.pdf (accessed on 19 February 2018).

[30] Wardropper, C., C. Chang and A. Rissman (2015), “Fragmented water quality governance: Constraints to spatial targeting for nutrient reduction in a Midwestern USA watershed”, Landscape and Urban Planning, http://dx.doi.org/10.1016/j.landurbplan.2014.12.011.

[28] Weersink, A. and D. Pannell (2017), Payments versus Direct Controls for Environmental Externalities in Agriculture, Oxford University Press, http://dx.doi.org/10.1093/acrefore/9780199389414.013.520.

[33] Whittaker, G. et al. (2017), “Spatial targeting of agri-environmental policy using bilevel evolutionary optimization”, Omega (United Kingdom), http://dx.doi.org/10.1016/j.omega.2016.01.007.

[37] Wilkinson, S. et al. (2015), Managing gully erosion as an efficient approach to improving water quality in the Great Barrier Reef lagoon. Report to the Department of the Environment, CSIRO Land and Water, https://publications.csiro.au/rpr/pub?pid=csiro:EP1410201 (accessed on 8 August 2018).

[41] Willamette Partnership, World Resources Institute and the National Network on Water Quality Trading (2015), Building a Water Quality Trading Program: Options and Considerations, http://www.edf.org/ (accessed on 11 September 2018).

Notes

← 1. Information gaps (imperfect information), information asymmetries and the presence of transaction costs constitute three key factors explaining why outcomes observed in the “real world” can systematically differ from predicted outcomes based on standard neoclassical economic theory, which assumes full information and (often) zero transaction costs. The first two are the subject of the strand of microeconomic theory known as information economics (pioneered by economists such as Hayek, Akerloff and Stiglitz), while transaction cost economics constitutes a separate-but-related branch stemming from seminal contributions by Coase. An extensive literature explores the implications of these three “market failures” in relation to agricultural and agri-environmental policies. See, for example, Coggan, Whitten, & Bennett (2010[64]), McCann (2013[62]), Nguyen (2013[63]), Shortle, Reed & Nguyen and Stavins (1995[61]). In addition, the conceptual framework identifies incentive non-alignments separately as an important factor in explaining why information gaps, information asymmetries and transactions costs persist.

← 2. Information gaps (imperfect information) refer to the absence of relevant information: for example, the environmental impacts of agriculture (particularly nonpoint sources) have historically been prohibitively costly or technically infeasible to monitor.

← 3. Information asymmetries occur when some information is known by some but not all relevant parties: for example, farmers know their own costs and intentions but might not have incentive to reveal this to the regulator or policymaker. Because it is costly for the agri-environmental policy administrators to obtain this information, farmers can extract “information rents” (Lankoski, 2016[24]). Information asymmetries cause problems of moral hazard and adverse selection, manifesting in agri-environmental policies as problems such as non-additionality and leakage. (OECD, 2012[29]); (OECD, 2010[65]; Börner et al., 2017[59]).

← 4. There are many definitions and classifications of transaction costs in the literature (for example, McCann and Easter, (2000[70]); McCann et al. (2005[67]); Krutilla and Krause (2011[72])). Following Lankoski (2016[24]), this paper uses the broad definition offered by McCann et al. (2005[67]): “transaction costs are the resources used to define, establish, maintain, and transfer property rights”, which includes costs arising from the design, implementation, control and evaluation of agri-environmental policy measures (Claassen, Cattaneo and Johansson, 2008[68]); (Heimlich, 2005[71]) (Marshall, 2013[69]); (McCann and Easter, 2000[70]). Transaction costs can erode the direct benefits (total welfare gain) from a policy and their distribution can affect how policies are designed. For example, if the policy maker is more concerned about costs faced by farmers than costs borne by government, they may face a trade-off between minimising total costs (transaction costs plus direct costs) and minimising costs borne by farmers.

← 5. Non-alignment of incentives occurs naturally because different actors have different preferences. Also, in some cases non-aligned incentives occur because incentives created by policies or regulations can (perhaps unintentionally) be mis‑aligned (i.e. policies create different incentives for different actors which conflict) or competing (i.e. several policies create conflicting incentives for one specific actor). Incentive non-alignments are one key reason why information gaps or asymmetries persist, and why they are costly to overcome. Incentive non-alignments can also cause different actors to work against each-other, rather than working together to achieve a common objective.

← 6. This analysis does not assume that the solution to problems caused by information gaps and information asymmetry is to maximise information available to policy makers or administrators.

← 7. See Annex B for an overview of policy components and agri-environmental policy mechanisms.

← 8. Managing Authority of the Rural Development Programme for the Autonomous Region of Madeira (Autoridade de Gestão do Plano de Desenvolvimento Rural da Região Autónoma da Madeira) (Portugal).

← 9. For information on respondents, including respondents’ level of government, see Annex B. For example, in Canada, agri-environment is a shared jurisdiction between federal and provincial and territorial governments. Canada’s provinces and territories design, deliver, and consult with their producers on agri-environmental on-farm programs. Seven provinces of Canada, and Agriculture and Agri-Food Canada, provided responses to the OECD questionnaire on the use of digital technologies for agri-environmental policies, as an assessment for the country as a whole is not possible.

← 10. Policy areas specified were: water quality, water quantity, air quality, biodiversity, soils, climate change (on-farm adaptation) and climate change (on-farm mitigation). Policy mechanisms specified were: environmental taxes; agri-environmental payments or subsidies; extension services and information provision; trading schemes (environmental markets); activity prohibitions; environmental standards; environmental property rights; and environmental stewardship programmes (not elsewhere specified). Respondents were able to select multiple policy areas and policy mechanisms.

← 11. These respondents were: for precision agriculture data: Canada-Prince Edward Island, Canada-Ontario, Chile-INIA, Chile-SAG, Korea-RDA, Denmark, Estonia-Ministry of Environment, Korea-MAFRA, Korea-Gyeong-gi provincial government, the Netherlands and USA-USDA NRCS. For retail scanner data: Austria, Canada-Quebec, Canada-Nova Scotia, Chile-CONAF, Chile-INFOR, Chile-SAG, Korea-MAFRA, and Switzerland.

← 12. Defined as “[a] senior executive position, formally responsible for setting strategic direction for the use and management of information technology systems, including digital records management.”

← 13. The Recommendation does not define the term “digital strategy”. The OECD used the following definition: “A ‘digital strategy’ is an organisational strategy which serves several functions. First, it articulates the organisation's vision regarding the contribution of digital technologies towards achieving the organisation's strategic objectives. Second, it sets the organisation's priorities for digital technology procurement and related investments (e.g. investment in staff training). Third, it sets out organisational reforms which are needed to ensure the organisation effectively and efficiently harnesses opportunities offered by digital technologies, while also appropriately addressing challenges. The development of a digital strategy can engage stakeholders ranging from the research community, other government entities, business, and civil society to regional and local governments. In some cases, organisational digital strategies may reflect or be based on broader national or government-wide strategies.” Adapted from articulation of OECD (2014[74]).

← 14. Defined as “[a] written document specifying a set of broad, high level principles which form the guiding framework in which data management can operate.” (Source: OECD Glossary of Statistical Terms).

← 15. See, for example, Robisch (2014[133]), who discussed the interaction between pollutants in the setting of water quality objectives in the context of setting Total Maximum Daily Loads (TMDLs) in the United States.

← 16. Natural language generation is “the automated generation of language based on digital data processing” (Arts, van der Wal and Adams, 2015, p. S633[73]).

← 17. However, Balla and Daniels (2007[66]) tested the hypothesis that introduction of web-based systems for providing comment on rule-making by the US Department of Transportation would dramatically increase the number of public comments filed and found that, contrary to expectations, overall participation by stakeholders remained approximately the same as before web-based submissions were introduced.

← 18. For example, Lawrence and Estow (2017[60]) examine options to address misinformation about climate change circulated via social media.

← 19. For example, all examples covered in OECD (2013[39]) use a digital platform of some kind.

← 20. This is known in economic theory as satisfying the incentive compatibility constraint. In the case of voluntary mechanisms, there is a little more complexity as the administrator needs to ensure both the participation and incentive compatibility constraints are satisfied (i.e. that (enough) farmers participate and comply).

← 21. Integrated Administration and Control System.

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Countries

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Publications

Publications

Featured publications

Data

Data

Featured data

News & Events

News & Events

Featured Events

About OECD

About

Engage with us

Work with us

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment