Effectiveness of new policy and incentive mechanisms

1 Effectiveness of new policy and incentive mechanisms Project NOVASOIL Project title INNOVATIVE BUSINESS MODELS FOR SOIL HEALTH Work Package WP4. Envisioning incentives & policies Solutions Deliverable 4.3 Period covered M18-32 Publication date 27/08/2025 Dissemination level PU Organisation name of lead beneficiary for this report WU Authors Insa Thiermann, Lucrezia Abruzzo, Mohammed Hussen Alemu, Vanya Bankova, Martin Banov, Fabio Bartolini, Ivan Boevsky, Cheng Chen, Liesbeth Dries, Fabian Frick, Krasimir Kostenarov, Ferdinand Lang, Thomas Lundhede, Steven van der Maarel, Bettina Matzdorf, Dimitre Nikolov, Carina Ober, Søren Bøye Olsen, Maria Raimondo, Riccardo Scarparo, Kristina Todorova, Ekatherina Tzvetanova-Georgieva Contributors WU, NBU, ZALF, TUM, UNIFE NOVASOIL INNOVATIVE BUSINESS MODELS FOR SOIL HEALTH Grant agreement ID: 101091268 Ref. Ares(2025)8236476 - 30/09/2025 2 QUALITY ASSURANCE PROCEDURES This document has been shared to the consortium in order to ensure their quality and to include a multidisciplinary point of view. Following a description of the different reviews can be found. TABLE REVISION HISTORY DELIVERABLE Row Version Date Reviewers Description 1 0.5 05-03-2025 WAGENINGEN UNIVERSITY Draft version 2 0.6 13-08-2025 NBU, UNIFE, TUM, ZALF Review 3 0.8 14-08-2025 WAGENINGEN UNIVERSITY Final version for review 4 1.0 25-08-2025 EVENOR Final version reviewed 3 Project Consortium Nº Participant organisation name Country 1 EVENOR TECH SLU ES 2 LEIBNIZ-ZENTRUM FUER AGRARLANDSCHAFTSFORSCHUNG DE 3 ZEMNIEKU SAEIMA LV 4 NEW BULGARIAN UNIVERSITY BU 5 CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE CNRS FR 6 KOBENHAVNS UNIVERSITET DK 7 TECHNISCHE UNIVERSITAET MUENCHEN DE 8 ASSEMBLEE DES REGIONS EUROPEENNES FRUITIERES LEGUMIERES ET HORTICOLES FR 9 ISTITUTO DELTA ECOLOGIA APPLICATA SRL IT 10 UNIVERSITA DEGLI STUDI DI FERRARA IT 11 WAGENINGEN UNIVERSITY NL 12 CENTRE OF ESTONIAN RURAL RESEARCH AND KNOWLEDGE EE 13 UNIVERSIDAD POLITECNICA DE MADRID ES 14 UNIVERSITA DI PISA IT 15 ASOCIACION AGRARIA JOVENES AGRICULTORES DE SEVILLA ES 16 UNIVERSITY OF LEEDS GB 4 Table of Contents Summary ................................................................................................................................................... 5 1 Objectives .............................................................................................................................................. 7 2 Overview of the case studies & the case study incentives ............................................. 9 2.1 WU – The Netherlands .............................................................................................................. 9 2.2 TUM - Germany ........................................................................................................................... 9 2.3 ZALF – Germany ........................................................................................................................ 10 2.4 UNIFE - Italy ................................................................................................................................ 11 2.5 NBU - Bulgaria ........................................................................................................................... 11 3 Results of the Discrete Choice Experiments ....................................................................... 12 3.1 DCE in a Nutshell ....................................................................................................................... 12 3.3 Germany (TUM).......................................................................................................................... 35 3.4 Germany (ZALF) ........................................................................................................................ 51 3.5 Italy ................................................................................................................................................. 61 3.6 Bulgaria ........................................................................................................................................ 71 4 Joint Discussion ............................................................................................................................... 84 6. References ..................................................................................................................................... 93 7. Acknowledgments ..................................................................................................................... 96 5 Summary Improved soil health is important for multiple reasons, such as to achieve climate neutrality, or to stop desertification and biodiversity loss. The project NOVASOIL aims to promote soil health measures by identifying potential business models for farmers that allow remuneration for their implementation. The requirements that farmers must fulfil to participate in these business models are shaped by the underlying policies and regulations. This report provides results of tests that were conducted of the selected policy options and incentive mechanisms using Discrete Choice Experiments (DCEs). DCEs are widely used to analyse how farmers choose among contracts that demand more soil health-friendly production practices. The DCEs were conducted in five case-studies looking at different business models and different policies or incentives. The Dutch case study, Rabo Carbon Bank, and one of the German case studies focus on the carbon credits trade for certified carbon removals. The Bulgarian case includes remuneration through the food value chain for wine produced with a focus on soil health. The second German case study focuses on an online marketplace (AgoraNatura) where farmers can finance the implementation of sustainable practices by selling certificates to other value chain actors (e.g., private investors and companies). The Italian case investigates the combination of incentives to stimulate carbon farming, for instance, through agri-environmental schemes and payments from carbon markets. The incentives and policies included in each study were determined through interviews and by conducting policy innovation labs (Task 4.2). Key results per case-study are as follows: - The Dutch case finds that farmers are unlikely to choose the presented hypothetical contracts, especially when no additional incentives besides the renumeration of the carbon credits are included (e.g., fixed cost shares, termination options for successors, annual training, or discounts on loans). The need for long-running contracts is evaluated as a crucial barrier. Including the option for farm successors to terminate the contract and discontinue carbon farming can help to motivate farmers to opt for longer contracts. - The German case (TUM) shows that inflexible contracts with long duration and without guaranteed cost recovery are mostly rejected by farmers. Contracts should be more flexible but also more reliable. Farmers should have the option to terminate contracts when handing over to a successor. In addition, ancillary investment subsidies or free advice can increase the willingness to participate, especially for farmers who have not yet invested much in soil health. Considering the fluctuating price of carbon credits, a guarantee for farmers that costs of the carbon farming practices are covered is particularly relevant. 6 - The second German case (ZALF) finds that there is general interest of farmers to supply biodiversity and ecosystem services credits via an online marketplace. In addition, public seed-funding covering fixed costs for credits holds substantial value for farmers and may boost engagement by financially derisking participation. Finally, the use of artificial intelligence-based monitoring applications are valued by farmers. - The Italian study reveals that farmers are hesitant to join soil conservation programs, especially when long-term programs involve significant income risks. Participation increases with higher per-hectare payments, while longer durations and additional effort reduce the likelihood of acceptance. To improve uptake, programs should offer greater flexibility, minimise perceived risks, and align better with farmers' economic priorities and expectations. - This Bulgarian case shows that producers of traditional products (grapes, wine) can be motivated to implement soil-improving practices (especially cover crops) through a business model that targets improved quality of wine, and resulting increased opportunities and demand from wine tourism in Bulgaria. General conclusions and recommendations can be derived from the results for each of the three main business models under consideration. First, carbon credit trade seems to be perceived as too risky, and especially the long durations of the contracts curb farmers’ enthusiasm. Allowing hybrid payments that combine result-based and action-based payments, and including termination options to successors may help to overcome these barriers. Second, value chain initiatives are perceived positively by farmers in the experiments, and the literature. More flexibility and less strict requirements may help these favourable perceptions. It remains to be seen whether the upcoming EU regulation regarding sustainability labelling for the EU food sector may influence this. The third business model, public payments through agrienvironmental schemes (AES) or eco-schemes, was not the primary focus of any of the case studies. However, public payments offer substantial funding, and their uptake can be improved, especially in countries with high opportunity costs. Furthermore, options of blended financing, combining public and private sources of remuneration, should be further investigated. 7 1 Objectives The EU aims to improve the status of soil health, for instance, through the Green Deal or the Soil Monitoring law. Improved soil health is important for multiple reasons, such as to achieve climate neutrality, or to stop desertification and biodiversity loss. Currently, 60% of soils are in a poor condition. This is due, among others, to high management intensities of soils (European Commission, 2023). Research shows that farmers can be financially incentivised to farm less intensively and apply soil health measures (e.g., additional cover crops, grazing, agroforestry) (COWI et al., 2021; McDonald et al., 2021). The project NOVASOIL aims to promote soil health measures by identifying potential business models for farmers that allow remuneration for their implementation. Since soil health measures can potentially lead to carbon sequestration, carbon credit trade can offer a new business model for farmers. Besides this, soil health measures can be funded by the EU through eco-schemes or agri-environmental schemes (AES). Furthermore, private incentives such as payments over the food value chain for more environmentally friendly production can hold potential (McDonald et al., 2021). The requirements that farmers must fulfil to participate in these business models are shaped by the underlying policies and regulations. EU regulation 2024/3012 sets the requirements for certified carbon removals on arable land. Certified carbon removals will demand quantifiable, reliable long-term storage of carbon that results from measures above statutory requirements (European Commission, 2024). Considering value chain payments, the regulation on sustainability labelling for the EU food sector will be of interest. This regulation will aim for clear and trustworthy information on the environmental impact of products (Mengual et al., 2024). This deliverable will provide information on the effects of new incentives and policies related to these business models. A detailed overview of the business models, and their underlying regulations is provided in deliverable 2.2 of the project (Thiermann & Dries, 2024). The case studies of the NOVASOIL project consider real-life setups of the business models. The Dutch case study Rabo Carbon Bank, and one of the German case studies focus on carbon credits trade for certified carbon removals, while the Bulgarian case includes remuneration through the food value chain for wine produced with a focus on soil health. The second German case study focuses on an online marketplace (AgoraNatura) where farmers can finance the implementation of sustainable practices by selling certificates. The Italian case focuses on incentives to stimulate carbon farming, for instance, through agri-environmental schemes. The incentives and policies included in each study were determined through interviews and by conducting policy innovation labs (Task 4.2). The task underlying this report (Task 4.3) aimed to test the selected policy options and incentive mechanisms using Discrete Choice Experiments (DCE). DCEs are widely applied in micro-econometrics (Greene, 2018). They can be used, for instance, to 8 analyse how farmers choose among contracts that demand more soil health-friendly production practices. DCEs are considered particularly useful as participants can be presented with alternatives that are hypothetical and do not yet exist (Train, 2009). This was considered useful as Task 4.3 seeks to analyse the potential of new policy options and incentive mechanisms. The report presents the results of the DCEs for the Dutch, German, Italian, and Bulgarian cases and derives joint conclusions relevant for policy. The remainder of the report presents the key features of the case studies in section 2. The main features of DCE and the results of the performed DCEs per case study are presented in section 3. Section 4 discusses the results, before deriving joint conclusions on the effectiveness of the new policy and incentive mechanisms. 9 2 Overview of the case studies & the case study incentives This section summaries the key features of each case study, and the relevant policies and incentives that were identified through interviews with key stakeholders and during the policy innovation labs. 2.1 WU – The Netherlands Rabo Carbon Bank is a private initiative by Rabobank that remunerates farmers for generating carbon credits through sustainable soil management practices. Sustainable soil management practices enhance soil health while capturing carbon. The sequestered carbon amounts are then quantified and traded as carbon credits. The bank advises farmers on sustainable soil management practices and connects them with large companies (e.g., agribusiness firms) that seek to buy carbon credits to offset their CO2 emissions. The results from the interviews and the policy innovation lab suggest that financial incentives generated by selling carbon credits may not be sufficient to encourage widespread participation of farmers. This is particularly due to the relatively low carbon amount that is expected to be captured. In addition, the ambitious requirements set in EU regulation 2024/3012 for the certification of measures that lead to carbon sequestration in agriculture (carbon farming), were viewed as barriers and the participants in the interviews raised doubts about farmers’ willingness and ability to meet the requirements. These requirements for achieving certified carbon credits are referred to as the quality criteria. They demand a robust quantification process, additionality (measures must exceed statutory requirements), long-term storage of CO2, and the applied measures must have a positive (or at the minimum a neutral) impact on other sustainability objectives. The DCE conducted for the case study explores farmers’ willingness to opt for carbon farming contracts that aim at generating carbon credits that meet the quality criteria. The attributes of the hypothetical contracts did not only consider the quality criteria. Some of the attributes also referred to other promising incentives (e.g., training on sustainable soil management, discounts on loans, termination options for successors). These are hypothesised to positively influence farmer participation rates, following the outcomes of the interviews with key stakeholders. The results of WU’s DCE are detailed in section 3.2 of this report. 2.2 TUM - Germany The Technical University of Munich (TUM) conducted a DCE in connection to the case study CO2-Land. CO2-Land is a carbon farming initiative in southwestern Germany that generates carbon credit certificates through soil carbon sequestration on partner farms. The initiative creates CO₂ sinks in arable soils, with farmers committing to 16 The report is structured as follows. First, the quality criteria of the CRCF regulation are summarised in section 2. Next, the design and the analysis of the DCE are described (section 3). Section 4 presents the results. Section 5 provides a discussion. 3.2.3 Requirements for EU certified carbon credits First, quantification requires that carbon removals should be quantified in a relevant, accurate, complete, consistent, transparent and comparable manner. During quantification, the net carbon removal benefit should be computed following two steps: in the first step, the gross amount of additional carbon removals that a carbon removal activity has generated in comparison to a baseline is calculated. Standardised baselines should be used that reflect the standard performance of comparable farms in similar social, economic, environmental, and technological circumstances and geographical locations. (European Commission, 2024). Next, to quantify the net carbon removal, any increase in greenhouse gas emissions related to the implementation of the carbon farming measure should be deducted (European Commission, 2024). Second, to ensure beneficial effects for the overall emissions, carbon farming measures should be additional. Therefore, activities should go beyond EU and national statutory requirements, and the incentive effect of the certification is necessary to make the activity financially viable (European Commission, 2024). Third, carbon storage should be permanent or is aimed at storing carbon over the long-term. This also implies monitoring rules and liability in case of reversals. (European Commission, 2024). Finally, carbon farming is assumed to have the potential to not only mitigate climate change but also to deliver co-benefits for sustainability. Therefore, minimum sustainability requirements are introduced that ensure that carbon farming measures have at least a neutral impact or generate co-benefits for other sustainability objectives (e.g., biodiversity, protection of water and marine resources, pollution prevention). (European Commission, 2024). 3.2.4 Methodology 3.2.4.1 Selection of measures The selected carbon farming measures in the hypothetical contracts ensure the fulfilment of the CRCF quality criteria (European Commission, 2024). The measures were chosen based on the existing literature. Kik et al. (2024) developed the programme Farm Analytics, which identifies sustainable soil management strategies that maximise the expected profit of a crop farm in the long term, while meeting certain soil quality goals using Mixed-Integer-Linear Programming. In addition, the authors provide information about the optimal strategies for Dutch farms. The farms either worked extensively (farms with standard arable crops) or intensively (farms with potatoes, carrots or onions) on clay or sandy soils and were representative of the Netherlands. Building on this, Matis (2023) used the RothC model by Rothamsted 17 Research (2024) to calculate the carbon removals that result from these soil health measures. Both Kik et al. (2024) and Matis (2023) present results for farms that would continue standard farm management, and for farms implementing soil quality-oriented management strategies. Kik et al. (2024) find the following optimal strategies: for intensive farms with high-value arable crops (onions, carrots, and potatoes), a minimum share of 25 % of the crop rotation needs to include grains (incl. corn), farms are required to at least use 100 kg N/ha from organic sources, i.e. cattle manure or compost, and farmers must plant cover crops whenever possible, meaning that there are at least eight weeks between harvesting one crop and seeding another. For extensive farms with standard arable crops, the income maximising strategy that allows to reach soil health is a minimum share of 50% of the crop rotation needs to be grains, farms are also required to at least use 100 kg N/ha from organic sources and farmers must also plant cover crops whenever possible. The carbon removal amounts resulting from these requirements are relatively low: for intensive farms on clay soils with soil-quality-oriented management, Matis (2023) calculates an additional removal in comparison to farms applying standard management practices of 0.27 tons of carbon equivalents per hectare per year. For farms farming intensively on sandy soils, the amount is 0.7 tons of carbon equivalents per hectare per year. Overall, the indicated values appear to be in line with other estimations (McDonald et al., 2021. Lessmann et al., 2022). To adequately inform farmers in our survey, the amounts found by Matis (2023) were provided to both intensive and extensive farmers, as Matis does not provide amounts for extensive farms separately. When farmers stated to operate on loam or loess soils, they were provided with an average of the sequestered amounts. The management requirements depended on their level of intensity. The quantification criterion foresees that the carbon amount sequestered should be determined with respect to a standardised baseline (European Commission, 2024). The comparison of soil organic carbon removals from adjusted farm management practices with carbon removals from standard management practices provides such a baseline. Using the comparison to standard Dutch farms also fulfils the additionality criterion. Neither the CAP nor the Dutch legislation hold a minimum requirement for grain shares in crop rotations. The enhanced conditionality of the CAP foresees crop rotation and diversification (GAEC (good agricultural and environmental conditions) 7), but this only prescribes a change in primary crops and limits the share of the main crop. The requirement of a minimum amount of soil cover (GAEC 6) is less strict than in Kik et al. (2024). Finally, only a maximum amount is defined for the usage of organic fertilisers. In 2024, this maximum amount became 170 kg N per hectare in the Netherlands (European Commission, 2023a). Furthermore, Kik et al (2024) developed 18 strategies that meet soil quality indicators and hence ensure that the sustainability criterion is fulfilled. Besides arable farmers, our survey also includes farmers that mainly have grassland and work on mineral soils. Günther et al. (2024) evaluate the effects of carbon farming measures and explain that grasslands are potential carbon sinks but turn into carbon sources when managed too intensively. The requirements presented to these farmers were derived from programs for meadow bird protection (more organic fertiliser, extensive grazing, resting periods, and no ploughing), which is important for the Netherlands (Boerennatuur, 2023; Tanis et al., 2020). Farmers with grassland on mineral soils were also presented with the removal potential calculated by Kik et al. (2024) and Matis (2023), and received the information that precise amounts for grassland are not available yet. 3.2.4.2 Attribute selection, design & conduct of the DCE The attributes of the DCE and their levels were chosen based on articles determining farmer preferences for agri-environmental schemes (AES) focused on soil health measures or carbon farming contracts. The first two attributes of the DCE refer to the monetary aspect. One is the carbon credit price on offer. It varied between € 10 and € 50 per carbon credit (per ton of CO2). € 25 per carbon credit represents current carbon prices. The upper bound of € 50 per carbon credit was offered in a trial of the Interreg carbon farming project (Demeyer et al., 2022). The second monetary attribute is a fixed cost share. In some hypothetical contracts, farmers received a fixed compensation of up to 100% of the cost of applying the measures. In other contracts, they received no fixed compensation, these display contracts where compensation is only foreseen through carbon markets. A fixed cost share for the applied measures was included because farmers can oppose riskier, purely result-based payments (Herzon et al., 2018). The next attribute is contract duration. When deciding to choose a hypothetical carbon farming contract, farmers needed to commit for at least 10 years (long-term storage). (Rochecouste et al., 2017) point out that some farmers may be particularly critical toward long contracts because they limit flexibility for their successors. To grant successors more flexibility, special termination options are commonly included in contracts when children take over the business (e.g., insurance contracts or contract farming agreements) (UNIDROIT et al., 2015) and were added to the DCE’s attributes. The last attribute is ‘support for sustainable soil management’. This support is defined as an annual training on carbon farming practices or being granted cheaper credit rates for all loans of the company. The first additional support incentive was included as Schaub et al. (2023) show that training opportunities can effectively motive farmers. Furthermore, Demeyer et al. (2022) identify a lack of finance as a hindrance for farmers to choose carbon farming contracts. 19 The attributes and their corresponding levels are presented in Table 3.2.1. To create the experimental design, the dcreate package in Stata, as described by Hole (2016) was utilized. Designs for DCEs are statistically derived. This process ensures model identification while minimising collinearity among attributes and levels (Reed Johnson et al., 2013). The resulting design achieved a D-efficiency of 96 % and consisted of nineteen choice cards. Table 3.2.1. Attributes chosen for the discrete choice experiment. Attribute Levelsa Compensation of costs (%) 0%, 25%, 50%, 75%, 100% Carbon credit price €0 per tona, €10 per ton, €20 per ton, €30 per ton, €40 per ton, €50 per ton Contract duration 0 yearsa, 10 years, 20 years, 30 years Termination option (for successors) Yes, No Support for sustainable soil management none, annual training on sustainable management, 0.5% discount on new farm loans a The levels used for the none-option are in bold. For carbon credit prices and contract durations the none-option levels are only expressed in the third alternative (no contract). The survey was set up online on the platform Qualtrics. The data collection for the experiment started in April 2024 and ended in January 2025. Seven choice cards were randomly assigned to each participating farmer. On the choice cards, farmers were always offered two hypothetical contracts with varying attributes and a third, nocontract option. An example of one of the choice cards is provided in Figure 3.2.1 (in Dutch). Following the choice cards, farmers were asked to answer questions about themselves and their farms. 20 Figure 3.2.1. Example of one of the choice cards used for the survey. 3.2.4.3 Estimation For the estimation, a mixed logit model and a latent class model (LCM) were estimated. The mixed logit model assumes that respondents (i) select the most attractive alternative from a set of J alternatives. The utility of choosing an alternative (Uij) depends on the contract’s attributes and their observed levels, represented by a vector xij, and an independently distributed random term (εij). Individual-specific coefficients βi determine the effect of the attributes on the utility (Eq. 1): 𝑈𝑖𝑗 =⁡𝛽𝑖𝑥𝑖𝑗 + 𝜀𝑖𝑗 (1) The mixed logit model assumes that estimators are continuously distributed with density f(β│θ), where θ represents the distribution parameters. The probability of choosing a contract, as shown in Equation (2), becomes an integral over β values and is determined by simulation. In the end, the coefficients display the weighted average of the participant’s preferences: 𝑃𝑖𝑗 =∫𝑒𝑥𝑝(𝛽 𝑥𝑛𝑖) ∑𝐽 𝐽=1 𝑒𝑥𝑝(𝛽 𝑥𝑖𝑗)(𝛽)𝑓(𝛽|𝜃)𝑑𝛽 (2) While the mixed logit accounts for preference heterogeneity using individual-specific coefficients, it lacks detailed information on differences in attribute perception (Train, 2009). Therefore, an LCM was also estimated. In LCM, the same coefficients are estimated per class. Besides this, the LCM presents estimates for the farmer and farm characteristics and displays how they affect class membership (Greene & Hensher, 2003). Willingness-to-accept (WTA) values were computed for both the mixed logit and LCM. WTA measures the rate of substitution between an attribute and monetary value (Train, 2009). In models with random effects, calculating the ratio directly would 21 lead to unreasonably high values (Train & Weeks, 2015), hence the price coefficient was considered as fixed when calculating WTA for the mixed logit model. 22 3.2.5 Results 3.2.5.1 Descriptive statistics Table 3.2.2 shows the descriptive statistics. Table 3.2.2. Descriptive statistics of the analysed sample. Variable mean S.D. Description Hectares 90.13 124.47 Total farm area in hectares (sum of arable and grassland) Hectares arable land 71.19 126.87 Arable land in hectares Hectares grassland 18.94 32.21 Grassland in hectares Sand 0.29 0.46 The predominant soil type is sand Clay 0.43 0.50 The predominant soil type is clay Loam 0.02 0.14 The predominant soil type is loam Loess 0.01 0.10 The predominant soil type is loess Grassland 0.24 0.43 The farm has grassland only Intensive management 0.37 0.48 Farm is classified as intensivea Organic 0.06 0.24 The farm is certified as organic, or in transition Fulltime 0.82 0.39 The farm is a full-time operation Arable farms 0.48 0.50 The farm is an arable farm only, and has no other branches of production, e.g. livestock husbandry AES – soil health 0.17 0.38 The farmer participates in AES to improve soil health on grassland or arable land Soil-health-friendly 0.51 0.50 The farm applies soil health-friendly managementb Share of rented farmland 30.86 30.31 Share of rented farmland Year of birth 1969.38 13.47 Year the farmer was born Higher education 0.44 0.50 Share of farmers with higher education (HBO (Hoger beroepsonderwijs) or university degree) Succession 0.77 0.42 The farm will stay in business for at least the next ten years (run by the farmer or a successor) Statement – Climate 3.52 0.91 Farmer’s level of agreement with ‘Climate change impacts my farm negatively’c Statement - Productivity 4.35 0.77 Farmer’s level of agreement with ‘Soil health is important for my farm’s productivity’c Statement - Risks 2.87 1.02 Farmer’s level of agreement with ‘To achieve profit maximation - I am more willing to take risks than others’c Participants 95 a For grassland(dairy)-only farms this included a self-classification. For arable and mixed farms this evaluation depends on the crops included in crop rotations. Following Kik et al. (2024) intensive farms have onions, carrots, and/or potatoes in their crop rotation. b Soil health-friendly management is assumed when arable farmers already display grain shares exceeding the hypothetical contract requirements, or when farmers with grassland offer grazing. c Farmers’ opinions on these statements were evaluated using Likert scales. On these scales, 1 indicated full disagreement and 5 indicated full agreement. 23 3.2.5.2 Results from the Mixed Logit Estimation Table 3.2.3 contains the results from the mixed logit estimation. The coefficients display which attributes significantly impact farmers’ choices for a hypothetical carbon farming contract. The pseudo-R² of the model is 0.438. The predicted probability for farmers to choose a hypothetical contract is estimated to be 40 %. The coefficients in the model were chosen based on a likelihood ratio test or a robust Waldtest. Table 3.2.3. Mixed logit model to explain farmer’s decision for a carbon farming contract. Coef. SD. Error S.D. SD. Error Interaction effects ASC * Sand 6.983** (2.859) ASC * Grassland -9.638** (4.692) ASC * Hectares -0.051** (0.024) ASC * Intensive mgmt. -14.32** (5.638) ASC * Fulltime -5.409** (2.560) ASC * AES – soil health 7.888** (3.536) ASC * Soil health-friendly 10.74** (4.541) ASC* Share rented farmland -0.038 (0.024) ASC *Higher education 6.186** (2.425) ASC * Succession -2.461 (1.686) ASC * Climate 3.900** (1.587) Contract attributes Compensation of costs (%) 0.199** (0.0815) 0.138*** (0.0537) Carbon credit price 0.188*** (0.0708) 0.159*** (0.0574) Contract duration -0.671*** (0.250) 0.480*** (0.179) Termination option 8.058** (3.165) -8.523*** (3.257) Support – annual training 4.041** (1.773) 7.676*** (2.678) Support– discount loans 1.974* (1.138) 5.967** (2.584) ASC -28.63** (11.67) 15.28*** (5.757) Observations 1,995 Pseudo-R2 0.438 The more productive soil types clay, loam and loess serve as the base category. Standard errors in parentheses. * p<0.10, ** p<0.05, *** p<0.01 24 Farmers display a higher predicted probability to choose a contract when the fixed compensation of costs and when the carbon credit price is expected to be higher. They oppose longer contract durations but value the termination option for successors. Furthermore, the coefficients show that the additional support for sustainable soil management (cheaper loans and training) has an effect: even though they are significant at a low level only (* p<0.10, ** p<0.05). The farm and farmer characteristics were included as interaction effects with an alternative specific constant (ASC). When two alternatives are very similar, the inclusion of one instead of two ASCs is commonly recommended (Auspurg & Liebe, 2011). The ASC used in the model is one for hypothetical carbon farming contracts and zero for alternatives representing the no-contract option. Such an ASC can also provide information about the perception of the hypothetical contracts that may not be captured by the attributes (Train, 2009). The ASC in the model is significantly negative, this indicates that farmers perceive the offer of carbon farming contracts as rather unattractive for reasons not explained by the attributes. The interaction effects provide information on which farmers are more or less likely to choose a carbon farming contract. Unsurprisingly, the interaction effects with the variables ‘Soil health friendly’ or ‘AES - soil health’ are significantly positive, showing the relevance of how close farmers already are to fulfilling the requirements. In addition, farmers who classified themselves as intensive, larger, and fulltime farmers are less likely than extensive, smaller, parttime farms to participate. Furthermore, farmers on sandy soils, that agree more strongly with the statement ‘Climate change impacts my farm negatively’ and farmers with higher education are more likely to accept a contract. 3.2.5.3 Results from the LCM The standard deviations of the attributes in Table 3.2.3 are all significantly positive. This indicates a heterogeneous perception of the attributes, which is explored further in an LCM (Table 3.2.4). Based on the BIC, a model with three classes is preferred. The first preference class has a high predicted probability of 86 % to choose a contract. The second preference class displays a predicted probability to choose a contract of only 4 % (uninterested in the offer). The third preference class is less certain about their perception of the contracts and has a predicted probability of 32 %. Considering the class shares, most farmers (42 %) are in class 2 (‘uninterested’), 38 % are in class 1 (‘interested’), and 20 % are in class 3 (‘uncertain’). Comparing the perception of the contract attributes. Class 1 farmers display preferences like those already described for the mixed logit. They are motivated by higher carbon credit prices, higher fixed payments, the offer of termination options for successors, and the support through annual training. However, they dislike long 25 contract durations. Interestingly, long-running contracts would be accepted by the class 3 farmers. As the other classes, farmers in class 3 also value higher prices for carbon credits and a higher fixed cost share. They also value the option to receive a discount on loans and annual training. The uninterested class 2 farmers strongly oppose long contract durations, and only some positive effects are found for the carbon credit price and the fixed compensation of costs. The second part of the table holds the class membership regression and displays which characteristics make farmers more likely to be part of the classes. They are interpreted with respect to the third preference class. Farmers in class 1 show a tendency to farm on grassland only. In addition, they are less likely to already apply soil health-friendly management practices. They tend to be younger and have a higher level of education than class 3 farmers. They further agree more strongly that soil health is important for the farm’s productivity and consider themselves more willing to take risks. Their lower share of rented farmland might make it easier for these farmers to accept contracts. Class 2 farmers are uninterested in the contracts and hence do not pay much attention to the contract’s attributes. They are also unlikely to already apply soil health-friendly management practices. These farmers are also more likely to be young farmers. Nevertheless, they are less sure that their farm will still be in business in the next ten years. Furthermore, they might have less experience in how to farm soil health-friendly (AES - soil health, Soil health-friendly). Class 3 farmers are older but are more likely to agree that their farms are still in business in 10 years. This could explain why the termination option is more important for them than the initial runtime of the contract. In addition, they are more likely to already apply soil healthfriendly management practices. 32 Hensher, D. A., Rose, J. M., & Greene, W. H. (2015). Applied choice analysis. In Applied Choice Analysis. https://doi.org/10.1007/9781316136232 Herzon, I., Birge, T., Allen, B., Povellato, A., Vanni, F., Hart, K., Radley, G., Tucker, G., Keenleyside, C., Oppermann, R., Underwood, E., Poux, X., Beaufoy, G., & Pražan, J. (2018). Time to look for evidence: Results-based approach to biodiversity conservation on farmland in Europe. Land Use Policy, 71, 347–354. https://doi.org/10.1016/j.landusepol.2017.12.011 Hole, A. R. (2016). Creating efficient designs for discrete choice experiments. Nordic and Baltic Stata Users Group Meeting. . Kik, M. C., Claassen, G. D. H., Ros, G. H., Meuwissen, M. P. M., Smit, A. B., & Saatkamp, H. W. (2024). FARManalytics – A bio-economic model to optimize the economic value of sustainable soil management on arable farms. European Journal of Agronomy, 157. https://doi.org/10.1016/j.eja.2024.127192 Kragt, M. E., Dumbrell, N. P., & Blackmore, L. (2017). Motivations and barriers for Western Australian broad-acre farmers to adopt carbon farming. Environmental Science and Policy, 73, 115–123. https://doi.org/10.1016/j.envsci.2017.04.009 Lapierre, M., Le Velly, G., Bougherara, D., Préget, R., & Sauquet, A. (2023). Designing agrienvironmental schemes to cope with uncertainty. Ecological Economics, 203. https://doi.org/10.1016/j.ecolecon.2022.107610 Lessmann, M., Ros, G. H., Young, M. D., & de Vries, W. (2022). Global variation in soil carbon sequestration potential through improved cropland management. Global Change Biology, 28(3), 1162–1177. https://doi.org/10.1111/gcb.15954 LTO Noord. (2023). Verdienen met grasland en Co2 -uitstoot te grazen nemen. 1–1. https://www.ltonoord.nl/actueel/verdienen-met-grasland-en-co2-uitstoot-te-grazennemen Matis, M. (2023). Conceptual Framework and Bioeconomic Modelling Approach, Applied at Dutch Intensive Arable Farms [Master Thesis]. Wageningen University & Research . McDonald, H., Frelih-Larsen, A., Lóránt, A., Duin, L., Andersen, S. P., Costa, G., & Bradley, H. (2021). Carbon farming Making agriculture fit for 2030. https://www.europarl.europa.eu/RegData/etudes/STUD/2021/695482/IPOL_STU(2021) 695482_EN.pdf Niskanen, O., Tienhaara, A., Haltia, E., & Pouta, E. (2021). Farmers’ heterogeneous preferences towards results-based environmental policies. Land Use Policy, 102, 1–10. https://doi.org/10.1016/j.landusepol.2020.105227 Paul, C., Bartkowski, B., Dönmez, C., Don, A., Mayer, S., Steffens, M., Weigl, S., Wiesmeier, M., Wolf, A., & Helming, K. (2023). Carbon farming: Are soil carbon certificates a suitable 33 tool for climate change mitigation? Journal of Environmental Management, 330. https://doi.org/10.1016/j.jenvman.2022.117142 Rabobank. (2022). Rabo Carbon Bank sells first carbon credits to Dutch companies. https://www.rabobank.com/about-us/carbon-bank/011264819/rabo-carbon-banksells-first-carbon-credits-to-dutch-companies Reed Johnson, F., Lancsar, E., Marshall, D., Kilambi, V., Mühlbacher, A., Regier, D. A., Bresnahan, B. W., Kanninen, B., & Bridges, J. F. P. (2013). Constructing Experimental Designs for Discrete-Choice Experiments: Report of the ISPOR Conjoint Analysis Experimental Design Good Research Practices Task Force. Value in Health, 16(1), 3–13. https://doi.org/10.1016/j.jval.2012.08.2223 Rochecouste, J. F., Dargusch, P., & King, C. (2017). Farmer perceptions of the opportunities and constraints to producing carbon offsets from Australian dryland grain cropping farms. Australasian Journal of Environmental Management, 24(4), 441– 452. https://doi.org/10.1080/14486563.2017.1379037 Rothamsted Research. (2024). ROTHAMSTED CARBON MODEL (ROTHC). 2024. Runge, T., Langlais, A., & Cardwell, M. (2022). Legal aspects of contract solutions to foster the provision of agri environment climate goods. https://literatur.thuenen.de/digbib_extern/dn065563.pdf Schaub, S., Ghazoul, J., Huber, R., Zhang, W., Sander, A., Rees, C., Banerjee, S., & Finger, R. (2023). The role of behavioural factors and opportunity costs in farmers’ participation in voluntary agri-environmental schemes: A systematic review. Journal of Agricultural Economics. https://doi.org/10.1111/1477-9552.12538 Smit, P. (2022). Rabo Carbon Bank verkoopt eerste Nederlandse carbon credits. https://www.nieuweoogst.nl/nieuws/2022/04/13/rabo-carbon-bank-verkoopt-eerstenederlandse-carbon-credits Tanis, M. F., Marshall, L., Biesmeijer, J. C., & van Kolfschoten, L. (2020). Grassland management for meadow birds in the Netherlands is unfavourable to pollinators. Basic and Applied Ecology, 43, 52–63. https://doi.org/10.1016/j.baae.2019.12.002 Tiusanen, K., Lampinen, A., Hulkkonen, V., Harjama, N., Rimhanen, K., Ilvesniemi, H., Juuso Joona, L., Farm Kaj Granholm, T., Naukkarinen, V., & Marianne Tikkanen, B. (2022). LIFE Carbon Farming Scheme final report: Guidance for future carbon farming schemes Best practices for expanding carbon sequestration activities. https://content.st1.fi/sites/default/files/202206/LIFE%20Carbon%20Farming%20Scheme%20final%20report%2001062022.pdf Train, K. (2009). Discrete choice methods with simulation. In Computers & Mathematics with Applications (2. Edition). Cambridge University Press. https://doi.org/10.1016/s0898-1221(04)90100-9 34 Train, K., & Weeks, M. (2015). Discrete Choice Models in Preference Space and Willingness-to-Pay Space. In Applications of Simulation Methods in Environmental and Resource Economics (pp. 1–16). Springer-Verlag. https://doi.org/10.1007/1-40203684-1_1 UNIDROIT, FAO, & IFAD. (2015). Legal guide on contract farming (Vol. 1). https://openknowledge.fao.org/server/api/core/bitstreams/02738d2a-ca79-48588d8e-b826acfc50dc/content 35 3.3 Germany (TUM) 3.3.1 Synthesis of the results Title: German Farmers' Acceptance of Carbon Farming Contracts Authors: Carina Ober, Fabian Frick Structured Abstract: Introduction: While carbon farming initiatives are seen positively by many farmers, contract design is key to foster farmer participation. Thus, our research objective was to overcome participation barriers and find ways to increase farmers’ acceptance of carbon farming contracts. Methodology: We conducted a discrete choice experiment with farmers in Germany to elicit preferences for carbon farming contract attributes. A latent class analysis identified distinct groups of farmers, based on carbon farming preferences and farm characteristics. Results and Discussion: Inflexible contracts with long duration and without guaranteed cost recovery proved to be not promising, as they are mostly rejected by farmers. Contracts should be more flexible but also more reliable. Farmers should have the option to terminate contracts when handing over to a successor. In addition, ancillary investment subsidies or free advice can increase the willingness to participate, especially for farmers who have not yet invested much in soil health. Considering the fluctuating price of carbon credits, a guarantee for farmers that costs of the carbon farming practices are covered is particularly relevant. Key Words: carbon farming, carbon credits, farmer acceptance, discrete choice experiment, soil health Case study: CO2-Land Incentives tested: payments for carbon credits, fixed cost shares, termination options for successors, annual training, or discounts on investments Policy considered: EU Carbon Removals and Carbon Farming (CRCF) Regulation (EU/2024/3012) 3.3.2 Introduction The Regulation (EU) 2024/3012, adopted on 27 November 2024, establishes the European Union's first certification framework for permanent carbon removals, carbon farming, and carbon storage in products. This voluntary framework aims to enhance the credibility and transparency of carbon removal activities within the EU, supporting its climate neutrality objectives by 2050. One of the main categories the regulation covers is Carbon Farming. In Carbon Farming, land management practices that enhance carbon sequestration in soils are rewarded based on the amount of sequestered carbon. Common practices are reforestation, peatland restoration, improved fertilizer use, diverse crop rotations, or reduced tillage (Block et al. 2024; 36 McDonald et al. 2021). To be certified under the EU Carbon Removals and Carbon Farming Regulation, activities must meet four core quality criteria. First, they must demonstrate clear and measurable benefits through accurate quantification of carbon removals or soil emission reductions. Second, they need to show additionality by going beyond existing legal requirements and standard practices. Third, certified activities must ensure the long-term storage of carbon, with safeguards in place to minimize the risk of reversal. Lastly, all activities must adhere to sustainability principles, ensuring they do not cause significant harm to the environment and that they contribute positively to broader goals such as biodiversity and soil health. As the participation of farmers in carbon farming initiatives is voluntary, farmers' acceptance of these initiatives is essential for their success. Figueredos’ (2024) review of research on farmers’ perspectives and attitudes towards carbon farming shows an overall positive attitude of farmers about carbon farming initiatives in multiple studies. While the basic attitude of farmers is often positive, there are still barriers that prevent farmers from actually participating in carbon farming initiatives. These obstacles range from administrative and legal challenges to practical difficulties in implementing new practices. Uncertainty around the impact of carbon farming on productivity and profitability, coupled with fluctuating carbon prices, also discourages involvement (Han and Niles 2023; Dumbrell et al. 2016; Kragt et al. 2016). High upfront investment costs and the need for significant changes in farm management further complicate adoption (Dumbrell et al. 2016). Despite these challenges, many farmers recognize the added value of carbon farming through co-benefits such as improved soil quality and reduced erosion, often cited as the most compelling incentives (Dumbrell et al. 2016; Wang et al. 2023). Providing clear information about these advantages can significantly boost willingness to participate (Dumbrell et al. 2016). However, carbon farming initiatives are frequently associated with substantial transaction costs, complex bureaucratic processes, and high opportunity costs (Latawiec et al. 2019; Raina et al. 2024). Long-term contractual commitments, which can severely limit land-use flexibility, pose yet another barrier, especially in cases where land is rented, making participation unfeasible (Figueredo 2024; Potthoff and Dramstad 2023). In summary, the main obstacles farmers face include uncertain price developments that fail to guarantee cost recovery, high upfront investment costs, and long-term commitments that limit flexibility in farm management. Additionally, a lack of awareness about the on-farm benefits of carbon farming practices often hinders participation. This study aims to build on existing knowledge about these barriers by exploring the acceptance of business models designed to address and overcome them. 37 3.3.3 Methodology 3.3.3.1 Attribute selection, design, and implementation of the DCE To determine farmers' preferences regarding the design of different carbon farming alternatives, we conducted a DCE with German farmers. The online data collection was conducted by a commercial farmer panel provider and took place in May 2025. 315 farmers took part in the survey. At the beginning of the survey, general characteristics of the participating farms were collected. These included, for example, information on acreage, production priorities, or the commercial character. In addition, questions relevant to carbon sequestration were asked, such as soil type or which soil organic carbon-promoting practices are already being carried out. This was followed by the DCE, in which farmers were presented choice cards to choose one of two carbon farming initiatives, or not to choose a program at all (opt-out option). After a brief assessment of the attribute attendance, the farmers were lastly asked to answer socio-demographic questions. The choice cards consisted of two contract options and the possibility of not choosing either option, which means staying in the status quo. The contract options were composed of 5 attributes with two to six attribute levels (see Table 3.3.1). Using a Defficient design, 18 choice cards were created and divided into three blocks. Thus, each farmer had to answer six choice cards. Table 3.3.1 Attributes chosen for the discrete choice experiment. Attribute Levelsa Contract duration 0, 10, 20, 30 Termination option (for successors) No, Yes Support for sustainable soil management None, Free training, 20% of one-time investment costs Compensation of costs (%) 0%, 25%, 50%, 75%, 100% Carbon credit price (€) 0€, 10€, 30€, 50€, 70€, 90€, 110€ aThe levels used for the none-option are in bold. For carbon credit prices and contract durations the none-option levels are only expressed in the third alternative (no contract). In the following, we will go into more detail about the five attributes: 38 · Contract duration: In the regulation (EU/2024/3012), a quality criterion to be certified under this framework is long-term storage. The Commitment Duration attribute has the attribute levels of 10, 20, or 30 years. · Termination option: Acceptance studies show that farmers tend to reject longterm contracts because they restrict farm management in the long term. Therefore, this attribute offers the possibility of exiting long-term obligations when the business is handed over to a successor. · Support for sustainable soil management: Previous studies have shown that farmers are often unaware of the benefits of carbon farming initiatives and the impact on farm productivity. Support by providing free training or advice could therefore boost farmer participation. Furthermore, farmers have indicated in the past that some practices require high investment costs (e.g., no-till machines). The second attribute level offers the possibility of a one-time payment, which covers 20% of the investment in soil-friendly farming machinery. Farmers in Germany are familiar with this kind of subsidy scheme because investment subsidies are a typical form of public support (e.g., in recent years subsidies for machines for ground-level slurry application were popular). · Compensation of costs: Soil organic carbon-promoting practices can incur additional costs (e.g., seeds for a catch crop). One risk for farmers is the uncertain price for a CO2 certificate on the carbon market. Due to fluctuating prices, farmers cannot rely on a secure profit, and it is uncertain whether the additional costs will be recovered. This attribute indicates what percentage of the ongoing additional costs are covered when participating in this program. This cost coverage is an additional payment in addition to the carbon credit payment for the CO2 certificate. · Carbon credit price: The attribute shows the current price per ton of CO2. The price is given in euros per tonne of CO2 that could be stored. 3.3.3.2 Model estimation To analyse the results of the discrete choice experiment, we initially estimated a conditional logit model, grounded in Random Utility Theory (RUT). According to RUT, each alternative has an associated utility composed of observable and unobservable components. The conditional logit model estimates the probability of an alternative being chosen based on its attributes, under the assumption that all respondents evaluate these attributes similarly. A key assumption of this model is the Independence of Irrelevant Alternatives (IIA), which implies that the relative odds of choosing between two options are unaffected by the presence of a third. 39 To test the validity of the IIA assumption, we employed the Hausman test. The test yielded statistically significant results, indicating a violation of the IIA assumption and suggesting the presence of heterogeneous preferences in the sample. Consequently, we proceeded to estimate a mixed logit model, which relaxes the IIA restriction and allows for random variation in preference parameters across individuals. This model accounts for unobserved heterogeneity and captures more realistic choice behaviour. Given that participants in our experiment had the option to opt out of carbon farming programs, we included an alternative-specific constant (ASC) to represent the status quo. The ASC takes the value of 1 when the status quo is chosen. A positive coefficient for the ASC indicates a preference for maintaining the current situation rather than adopting a carbon farming scheme. Recognizing that not all individuals value attributes in the same way, we conducted a Latent Class Analysis (LCA) to further explore preference heterogeneity. LCA is a segmentation technique that identifies unobserved subgroups within the sample based on similar choice behaviours. Each latent class represents a group of respondents with relatively homogeneous preferences, while preferences differ substantially between classes. The model estimates attribute-level preference parameters for each class and the probability that a respondent belongs to a given class. This approach provides valuable insights into the diversity of decision-making patterns and helps to explain why different groups make different choices. 3.3.4 Results 3.3.4.1 Descriptive statistics The average farm size among the 315 participating farms is approximately 131 hectares, which is significantly above the national average in Germany. While the average area of permanent grassland at around 23 hectares, aligns closely with the national mean, the average arable land area in the sample is substantially larger. The sample is representative in terms of the proportion of organic farms. Compared to the national average, farms in the sample have a higher share of rented land and a greater prevalence of full-time farms. Approximately 22.2% of the farms are exclusively arable farmers with no grassland, while 7.6% manage only permanent grassland. Table 3.3.2 Descriptive statistics of the analysed sample Variables Sample Germany Description Average farm size (ha) 131.4 66a Total farm area in hectares 40 Average arable area (ha) 115.1 45.8b Arable land in hectares Average Grassland area (ha) 23.3 21.5e Grassland in hectares Share of rented land (%) 32.8 59e Share of rented farmland Part-time farms (%) 39.7 47.2c Share of part-time farms Organic farms (%) 10.5 11.4d Share of organic farms Pure grassland farms (%) 7.6 Share of farms with only grassland Pure arable farms (%) 22.2 Share of farms with only arable land Predominant Soil Type (%) Sand 27.3 Predominant soil type is sand Silt 11.1 Predominant soil type is silt Clay 7.9 Predominant soil type is clay Loam 44.8 Predominant soil type is loam Peat 0.32 Predominant soil type is peat Intensive (%) 68.2 Share of farms with intensive crops (sugar beet, potato, mais) AECS soil health (%) 26.6 Participation in AECS for soil health Age by classes (%) 41 <55 years 48.3 61i Farmers younger than 55 years >=55 years 51.7 39i Farmers older than 55 years Agricultural education (%) 80.6 85i Share of farmers with agricultural training Higher education 50.2 Share of farmers with higher education (“Hochschulreife”) Successor (%) 36.2 Farm will be transferred to a successor within the next 10 years Statement – Climate 3.5 Farmer’s level of agreement with ‘Climate change impacts my farm negatively’ Statement - Productivity 4.5 Farmer’s level of agreement with ‘Soil health is important for my farm’s productivity’ Statement – Risk 2.9 Farmer’s level of agreement with ‘To achieve profit maximisation - I am more willing to take risks than others a) BMEL-Statistik: Betriebsstruktur und Entwicklung landwirtschaftlicher Betriebe b) BMEL-Statistik: Bodennutzung in Deutschland c) Studie zum Nebenerwerb in der Landwirtschaft: So ist die Lage - Bauernzeitung d) BMEL-Statistik: Ökologischer Landbau e) Agrarstrukturerhebung (statistisches Bundesamt) 2023 The predominant soil type is loam, and nearly 30% of the farm managers reported participating in Agri-Environmental and Climate Schemes (AECS) aimed at improving soil health. Most participants agree with the statement that soil health is crucial for productivity, and also the view that climate change has a negative impact on production is widely accepted. Roughly one-third of the farms are expected to be transferred to a successor within the next 10 years. Farm managers in the sample tend to be older than the national average, while the share of those with formal agricultural training is in line with the national average. 48 farmers’ willingness to participate. Additionally, higher carbon credit prices were also associated with increased participation. A literature review by Figueredo (2024) has previously shown that farmers are generally open to carbon credit programs. Our mixed logit model does not support this result: opting to remain in the status quo, i.e., not participating in such programs, provides significantly greater utility for farmers, on average. A payment of nearly 100€ would be necessary to get farmers to participate. However, farm characteristics do influence participation. Farmers operating on pure grassland farms are less likely to take part in carbon credit programs. This may be because targeted increasing of soil organic matter is generally more difficult in grassland cultivation. On the other hand, farmers who believe in the negative impacts of climate change are more likely to participate in carbon credit programs. These programs may be seen as a strategy to adapt to changing production conditions and as a risk management strategy for increasing production risk due to more frequent weather extremes. Following the mixed logit analysis, we conducted a latent class analysis to explore heterogeneity in farmers’ preferences. This analysis revealed three distinct farmer classes. Class 1, comprising about 44% of respondents, is the largest. Like class 2, they show a general willingness to participate in carbon credit programs. Farmers in this group tend to be younger compared to the other two classes. Therefore, they respond particularly positively to subsidies for initial investments, which may help young farmers to adapt farm management to their personal preferences. Class 2, accounting for 23% of farmers, is oftentimes engaged in more intensive practices. They demonstrate a high willingness to participate, provided the programs offer sufficient cost coverage and include an exit option from long-term contracts. Coverage of investment costs seems to be less decisive, as maybe due to already intensive management, they are already well positioned in terms of their machines. Class 3 includes 32% of the respondents and in contrast to classes 1 and 2 shows a low willingness to participate overall. These farmers tend to be less educated and do not expect negative impacts from climate change, both of which likely reduce their interest in joining such programs. Our findings indicate that a substantial segment of farmers (Class 1 and 2) is considering sustainable land management and is open to participating if certain conditions are met. These are the option to terminate contracts when giving the farm to a successor and a high cost compensation. Class 1 can be additionally encouraged to participate through financial support for investments. Farmers who are sceptical of climate change and less educated (Class 3) are significantly less likely to participate. 49 In summary, strict contract models, such as those with long durations, are not conducive to high acceptance of carbon farming practices among the farmers in our sample. When contracts are designed more flexibly, for instance by offering an exit option, or by providing financial support for initial investments, they can attract farmers’ interest. However, it is important to keep in mind that one of the core criteria of EU Regulation (EU/2024/3012) is the additionality of measures. Practices that are already funded by other programs, such as agri-environmental and climate schemes (AECS), for example, cannot be additionally financed by carbon farming initiatives. With the new CAP funding period starting in 2023, farmers in Germany have many opportunities to receive support for soil health–promoting practices through both the first and second pillars of the CAP. For example, eco-schemes under Pillar I support agroforestry systems or extensive grassland management. AECS under Pillar II promote humus-enhancing crop rotations, the avoidance of intensive crops, or conservation tillage methods. Farmers often view AECS as a reliable source of funding, while carbon credit programs are perceived as risky (Figueredo 2024). Introducing guaranteed cost coverage into carbon credit programs could make them more attractive. However, to make definitive conclusions about farmer preferences, more studies are needed in which farmers directly choose between these two types of support programs. 3.3.6. References Block, Julia B.; Danne, Michael; Mußhoff, Oliver (2024): Farmers' Willingness to Participate in a Carbon Sequestration Program - A Discrete Choice Experiment. In Environmental Management, pp. 1–18. DOI: 10.1007/s00267-024-01963-9. Dumbrell, Nikki P.; Kragt, Marit E.; Gibson, Fiona L. (2016): What carbon farming activities are farmers likely to adopt? A best–worst scaling survey. In 0264-8377 54, pp. 29–37. DOI: 10.1016/j.landusepol.2016.02.002. Figueredo, Alejandra (2024): A review of research on farmers' perspectives and attitudes towards carbon farming as a climate change mitigation strategy. number: 2024:5. Urban and rural reports: Department of Urban and Rural Development, Swedish University of Agricultural Sciences. Available online at https://pub.epsilon.slu.se/35795/. Han, Guang; Niles, Meredith T. (2023): Interested but Uncertain: Carbon Markets and Data Sharing among U.S. Crop Farmers. In Land 12 (8), p. 1526. DOI: 10.3390/land12081526. Kragt, M. E.; Gibson, F. L.; Maseyk, F.; Wilson, K. A. (2016): Public willingness to pay for carbon farming and its co-benefits. In Ecological Economics 126, pp. 125–131. DOI: 10.1016/j.ecolecon.2016.02.018. Latawiec, Agnieszka E.; Strassburg, Bernardo B. N.; Junqueira, André B.; Araujo, Ednaldo; D de Moraes, Luiz Fernando; Pinto, Helena A. N. et al. (2019): Biochar amendment improves degraded pasturelands in Brazil: environmental and costbenefit analysis. In Sci Rep 9 (1), p. 11993. DOI: 10.1038/s41598-019-47647-x. 50 McDonald, Hugh; Frelih-Larsen, Ana; Lóránt, Anna, Duin, Laurens; Pyndt Andersen, Sarah; Costa, Giulia; Bradley, Harriet (2021): Carbon farming. Making agriculture fit for 2030. Ober, Carina; Canessa, Carolin; Frick, Fabian; Sauer, Johannes (2025): The role of behavioural factors in accepting agri-environmental contracts – Evidence from a Q-method and thematic analysis in Germany. In 0921-8009 231, p. 108544. DOI: 10.1016/j.ecolecon.2025.108544. Potthoff, Kerstin; Dramstad, Wenche E. (2023): Management of rented farmland in Norway: Factors impacting on tenants’ decisions to make investments. In 02648377 135, p. 106941. DOI: 10.1016/j.landusepol.2023.106941. Raina, Nidhi; Zavalloni, Matteo; Viaggi, Davide (2024): Incentive mechanisms of carbon farming contracts: A systematic mapping study. In Journal of environmental management 352, p. 120126. DOI: 10.1016/j.jenvman.2024.120126. Wang, Tong; Jin, Hailong; Sieverding, Heidi; Cheye, Stephen; Clay, David (2023): Factors affecting farmers’ willingness to accept payment for carbon farming in the U.S. Midwest. In AgEcon Search. 51 3.4 Germany (ZALF) 3.4.1 Synthesis of the results Title: The supply of biodiversity and ecosystem services credits: what matters to farmers in Germany Authors: Ferdinand Lang, Cheng Chen, Mohammed Hussen Alemu, Thomas Lundhede, Søren Bøye Olsen, Bettina Matzdorf Structured Abstract: Introduction: To increase private investments into biodiversity and ecosystem services, market-based business models such as credits are increasingly being seen as a promising opportunity. In this context, the European Commission recently published its ‘Roadmap towards Nature Credits’, formulating initial steps to develop a regulatory framework enabling such a business case in the European Union. Particularly, farmers are envisioned to play a crucial to supply such credits by implementing environmental measures on their land in exchange for remuneration from private investors. However, little is known about farmer preferences for participating in privately financed market-based schemes such as credits. Our study aimed at investigating farmers’ willingness to participate in such credit schemes, addressing some of the entry barriers and technological opportunities currently being observed in real-world marketplaces (e.g., online marketplaces). Methodology: We conducted a discrete choice experiment with farmers in Germany to investigate preferences for supplying biodiversity and ecosystem services credits and analysed our data specifying a random parameter logit model (or mixed logit) in willingness-to-accept-space. Moreover, as our case study we used a real-world and currently operating online marketplace for such credits: AgoraNatura. Results and Discussion: Our results indicate general interest of farmers to supply biodiversity and ecosystem services credits via an online marketplace. In addition, we observe that public seed-funding covering fixed costs for credits holds substantial value for farmers and may boost engagement by financially derisking participation. We also note that artificial intelligence-based monitoring applications are valued by farmers. Finally, our analysis shows that an online marketplace for biodiversity and ecosystem services credits is of interest to farmers and could serve as an appreciated channel to facilitate private sector investment in respective credits. Key Words: biodiversity, ecosystem services, credits, blending, discrete choice experiment, willingness to accept Case study: AgoraNatura Incentives tested: remuneration for biodiversity and ecosystem services credits, public funding for fixed costs of credits, artificial intelligence-based monitoring technologies, independent vs monitoring through service provider Policy considered: Biodiversity Strategy for 2030, EU Soil Strategy for 2030, EU Roadmap towards nature credits 52 3.4.2 Introduction Aiming to incentivise increased private investment into nature conservation, the European Commission has recently presented its ‘Roadmap towards Nature Credits’, formulating initial steps to develop a regulatory framework that is able to support scalable business models for biodiversity and ecosystem service (ES) credits (European Commission, 2025). For the supply of such credits, farmers are anticipated to play a pivotal role. However, the extent to which farmers are open to participating in such privately financed and voluntary business models remains insufficiently understood. Analysing farmer preferences for biodiversity and ES credits is different to preferences for publicly funded agri-environmental and climate measures (AECMs). Arguably one of the main differences is that AECMs generally offer somewhat guaranteed compensation for opportunity costs (such as foregone economic revenue) (Kuhfuss et al., 2016), while market-based schemes carry risks for farmers such as demand uncertainty for credits (European Commission, 2025; zu Ermgassen et al., 2025). Consequently, farmers face the risk that initial investments made to supply such credits (e.g., expenses for credit development and certification) may turn into sunk costs. Such a risk presents an entry barrier to farmers wanting to participate in credit schemes. To overcome this barrier, blended finance – which combines public funds with private capital – has previously been discussed as a promising opportunity to de-risk participation and leverage stakeholder engagement (Reed et al., 2022; den Heijer & Coppens, 2023). Public funding, for instance, could help reduce risk by covering initial fixed costs related to the conceptualisation and certification of credits (see also European Commission, 2025; Flammer et al., 2025; Rode et al., 2019). On the other hand, revenue made through the selling of credits can remunerate farmers for forgone production income or even generate profits. Although the strategic value of blended finance in mobilising private capital for conservation efforts has been previously recognised (Havemann et al., 2020; Reed et al., 2022; UNEP, 2023; Seidl et al., 2024; zu Ermgassen et al., 2025), its specific influence on farmer engagement within biodiversity and ES credit markets remains insufficiently studied. This represents a timely research gap with direct relevance for current policy initiatives. Another crucial aspect in the context of increasing private investment for biodiversity and ES credits is that investors often hesitate to invest into such markets due to concerns about reputational risks (e.g., the fear of being associated with greenwashing (Chen et al., 2024; Krause & Matzdorf, 2019; zu Ermgassen et al., 2025). Enhancing the integrity and demonstrable impact of biodiversity and ES credits is therefore essential, and robust and transparent monitoring standards will play a central role to attract more investment. Recent innovations in artificial intelligence (AI) applications, especially machine learning (ML), offer promising tools to address this challenge. ML technology applied to visual and acoustic data collected from agricultural landscapes have shown considerable promise for monitoring biodiversity and ES outcomes (Alotaibi & Nassif, 2024; Andermann et al., 2022; IEEP, 2024; Norouzzadeh et al., 2018; Reynolds et al., 2024; Wäldchen & Mäder, 2018). Furthermore, such automated monitoring technology can also substantially lower the costs associated with monitoring (IEEP, 2024; Norouzzadeh et al., 2018) 53 and reduce administrative burdens (IEEP, 2024; Hannus et al., 2020). However, widespread adoption hinges on farmers’ acceptance of these technologies. To date, no research has examined European farmers’ preferences regarding the use of AIbased monitoring in the context of biodiversity and ES credit schemes. Importantly, deploying such technologies on farms may raise new concerns, particularly around data privacy and autonomy. An important question, therefore, is whether farmers would prefer to manage AI-based monitoring themselves or delegate this responsibility to service providers. In summary, this study aims to investigate the supply side of biodiversity and ES credits and what preferences farmers have towards mechanisms and opportunities seen to provide promising pathways for the scalability of such business models. 3.4.3 Methodology To assess farmer preferences for biodiversity and ES credits, characterised by blended finance and AI-based monitoring opportunities we conducted a DCE. Our survey was distributed in April 2025 to farmers across Germany managing arable land. Overall, 301 farmers completed the survey. Descriptive statistics of our sample are provided in Table 3.4.1 . Table 3.4.1: Descriptive statistics of the analysed sample Variables Sample Age Mean: 52.4 Gender (%): Female: 8.64% Male: 91.36% Diverse: 0% Ownership of managed land (%): Leases most of managed land: 40.5% Owns most of land: 54.5% Neither owns nor leases managed land but makes land use and investment decisions: 5% Size of managed land (ha): Less than 1: 1.33% 1-10: 1.33% 10-100: 62.13% 100-500: 23.26% More than 500: 0.66% 54 German federal state where largest share of farmland is located (%): Baden-Württemberg Bavaria Berlin Brandenburg Bremen Hamburg Hesse Mecklenburg-Vorpommern Lower Saxony North Rhine-Westphalia Rhineland-Palatinate Saarland Saxony Saxony-Anhalt Schleswig-Holstein Thuringia 9.30% 52.16% 0% 1% 0% 0% 6.31% 1.33% 18.60% 4.32% 2.33% 0.33% 1.33% 1.66% 1% 0.33% In our DCE we framed biodiversity and ES credits representing flower strip measures on arable land with a duration of five years. We chose a commonly known and general reference scenario to provide farmers with a common starting ground before answering our choice scenarios and to ensure comparability of results. Farmers were informed these flower strip measures would generate credits which they could offer to investors via the online marketplace AgoraNatura. In total farmers were asked to answer six choice scenarios. Each time they were able to choose among two individual flower strip measures and one opt-out alternative. The two individual flower strip measure alternatives where characterised by four varying attributes (a more detailed description can be found in Table 3.4.2): 1. Public funding for fixed costs (0% to 100%); 2. Monitoring type (farmer-led vs service provider); 3. Monitoring technique (manual vs automated monitoring); 4. Remuneration (ha/year) (€500 to €2,500). The funding attribute reflected the blended finance mechanism mentioned in the introduction, where public support was intended to derisk of up-front investment 55 for developing and certifying credits. For simplification – and based on expert input – we told farmers to assume fixed costs €1,000 for a five-year flower strip project. The goal was to explore whether public funding for fixed costs could stimulate farmer participation for otherwise market-based and privately financed (or remunerated) biodiversity and ES credits. Monitoring techniques were inspired by currently applied techniques on AgoraNatura and emerging AI-based monitoring applications. Manual monitoring was framed as traditional species observation, while automated monitoring was framed as automatically uploaded and processed image, video or acoustic data obtained via drones or smartphones. Remuneration levels per credit were inspired by past AgoraNatura projects in combination with expert consultations of the marketplace. Table 3.4.2: Attributes and levels of choice experiment. Attribute Levels (coding) Funding for conceptualisation and certification costs 0% (reference) 33.33% (dummy) 66.66% (dummy) 100% (dummy) Monitoring – type Monitoring through service provider (reference) Independent monitoring (dummy) Monitoring – technique Manual monitoring (reference) Automated monitoring (dummy) Remuneration (ha/year) €500, €750, €1000, €1500, €2000 or €2500 (continuous) 3.4.4 Results Below, we present the results of our random parameter logit (or mixed logit) model (RP – MXL) using the WTA-space specification (see also Hannus et al., 2020). For the final analysis we excluded a total of 9 farmers due to protest behaviours. The final sample used for analysis included 292 farmers. 3.4.4.1 Results Funding for conceptualisation and certification costs (blending): 56 From our results we observe that farmers on average show a positive mean estimate for the opt-out alternative (€590/credit). This estimate represents the lowest average amount of remuneration our farmer sample would want to receive to supply credits. Among public funding levels for fixed costs, full funding (100%) is most valued (€330/credit), followed by 66.66% (€228) and 33.33% (€149). These estimates must be understood as monetary values farmers place on the respective funding levels. Thus, the results imply a strong preference for higher funding, where farmers are willing to accept less remuneration per credit when funding increases. However, we also observe significant standard deviation estimates which point to potential subgroup differences across our sample. Monitoring alternatives: From our results we observe an insignificant estimate for independent monitoring (€10/credit) while farmers on average significantly value automated monitoring (€180/credit). This indicates if farmers have the possibility to use automated monitoring (compared to manual monitoring), they on average are willing to forego €180 of remuneration per credit. Though, we again observe significant standard deviation estimates which point to potential subgroup differences across our sample. 3.4.5 Discussion 3.4.5.1 Willingness to Supply Biodiversity and ES Credits As studies on farmers’ willingness to offer biodiversity and ES credits via marketbased approaches generally remain scarce, we perhaps compare our findings regarding general willingness to supply biodiversity and ES credits to current compensation levels of publicly funded five-year flower strip measures (i.e. AECMs). Since the majority of our sample are farmers in either Bavaria (52.16%), Lower Saxony (18.60%), Baden-Württemberg (9.30%), and Hesse (6.31%), we here focus on these regions. In Bavaria remuneration can vary between approx. €400 – €1100/ha/year (StMELF, 2025) depending on yield index, in Lower Saxony between approx. €910 – €1250/ha/year (ML, 2023), in Baden-Württemberg approx. €730/ha/year (MLR, 2024) and in Hesse approx. €750/ha/year (HMLU, 2025). Notably, the minimum remuneration our sample on average requires to receive in order to consider supplying biodiversity and ES credits (€590/credit) is less than offered remuneration by the above-mentioned federal states except for the minimum in Bavaria. 3.4.5.2 Blended Finance Regarding the use of blended finance for biodiversity and ES credits via marketbased systems, particularly applied to derisking fixed costs, no prior research appears to mirror our supply-side focus or experimental design. Nonetheless, we may compare our findings to existing literature investigating the effect of fixed costs in publicly funded AECMs which also show that lowering the financial risk of fixed costs positively affects farmer participation (Ducos et al., 2009; EspinosaGoded et al., 2013). As we observe that farmers are willing to forego an increasing amount of remuneration per credit as respective seed-funding increases, we contribute to a timely discussion on the development of such business models 57 (European Commission, 2025) as well as the integration of publicly funded with private market-based schemes (Reed et al., 2022). Furthermore, we find support for our blending approach in the European Commission’s recently published ‘Roadmap towards Nature Credits’ which also acknowledges the role of fixed costs potentially presenting a barrier to farmer participation and that participation might be enhanced through “seed finance or by de-risking mechanisms to reward and scale up early biodiversity certification and nature credits initiatives that are essential to attract private capital” (European Commission, 2025:10). Thus, we believe our results may be useful for policymakers to further develop the Roadmap towards Nature Credits. 3.4.5.3 Monitoring The second objective of our study was to investigate farmer preferences for AIbased monitoring technologies and the responsibility on implementing them in order to substantiate biodiversity and ES credits. While technological advancements present great opportunities for monitoring environmental improvements that substantiate biodiversity and ES credits, we are not aware of any studies having examined farmers willingness to adopt such technologies. On average, we do find farmers value the opportunity to apply such technologies. That said, our analysis is not without limits. Future work should expand our sample size to better represent German farmers – especially with farmers from East Germany, which were underrepresented. In this context, it would be especially interesting to see how public funding for fixed costs affects farmers in these regions. With larger average farm sizes in East Germany, these farms might be more flexible in managing fixed costs and thus are perhaps less risk averse. Finally, it would be interesting to replicate the study conducting a revealed preference DCE, which is more realistic in capturing farmer behaviour and to some extent is able to mitigate hypothetical bias inherent in stated preference DCEs (Hensher, 2010; Mariel et al., 2021). 3.4.6 Conclusion This research offers empirical insight into German farmers’ willingness to supply biodiversity and ES credits via an online marketplace. Moreover, we observe that public funding for fixed costs of credits (i.e. seed funding) as well as the opportunity to apply AI-based monitoring technologies may stimulate participation. However, preference heterogeneity across the sample exists. In summary, we hope our findings can contribute to recent policy initiatives driving the development of biodiversity and ES credits such as the European Commission’s ‘Roadmap towards Nature Credits’. Moreover, we emphasise the importance of flexible framework conditions in designing biodiversity and ES credit programmes in order to both account for varying preferences of farmers and maximise farmer participation. 3.4.7 References Alotaibi, E., & Nassif, N. (2024). Artificial intelligence in environmental monitoring: in-depth analysis. Discover Artificial Intelligence, 4(1), 84. https://doi.org/10.1007/s44163-024-00198-1 64 Table 3.5.1: Attribute selection for the DCE Attribute Levels Description Annual payment per hectare 100 €, 200 € Compensation for adopting conservation practices Type of payment 100% fixed; 50% fixed + 50% result-based; 100% result-based Payment modalities: fixed or performance-based Estimated income reduction Low, Medium, High Represents perceived income loss risk Commitment duration 5, 10, 20 years Minimum period of practice adoption These attributes reflect economic aspects (payment, income loss) and operational and risk-related factors (program duration, type of payment), all of which are known to influence farmers’ decision-making. Payments from €100 to €200/ha reflect realistic ranges, consistent with CAP incentives and marginal costs associated with practices such as no-till farming and cover crops. Durations from 5 to 20 years reflect the time required to obtain tangible soil benefits. Payment types test preferences between area-based and result-based mechanisms, a key aspect in the literature, as the latter are expected to provide cost-effectiveness of the measure (D’Alberto et al., 2024; Vergamini et al., 2024) Income reduction and administrative burdens reflect fears of economic loss, a frequent barrier to adoption (Canales et al., 2024). Each survey participant received randomised choice cards containing two program alternatives (A and B) and a third “opt-out” option. An example is shown below: Table 3.5.2: Example of a choice card (in Italian) Attribute Alternative A Alternative B Durata di adozione della pratica 5 anni 20 anni Variazione (riduzione) del reddito agricolo (Reduction of agricultural income) Basso Alto Pagamento per ha (Payment per ha) €100 per ha €200 per ha 65 Tipo di pagamento (Type of payment) 100% in base all'aumento della sostanza organica 50% di pagamento fisso per ha + 50% in base all'aumento della sostanza organica The survey was set up online on the Qualtrics platform. Data collection for the experiment began in October 2024 and ended in March 2025. 3.5.4.3 Estimation For the estimation, a mixed logit model was estimated. The mixed logit model assumes that respondents (i) select the most attractive alternative from a set of J alternatives. The utility of choosing an alternative (Uij) depends on the program’s attributes and their observed levels, represented by a vector xij, and an independently distributed random term (εij). Individual-specific coefficients βi determine the effect of the attributes on the utility (Eq. 1): 𝑈𝑖𝑗 =⁡𝛽𝑖𝑥𝑖𝑗 + 𝜀𝑖𝑗 (1) The mixed logit model assumes that estimators are continuously distributed with density f(β│θ), where θ represents the distribution parameters. As shown in Equation (2), the probability of choosing a program becomes an integral over β values and is determined by simulation. In the end, the coefficients display the weighted average of the participants’ preferences: Pij=∫Lij(β)f(β)dβ (2) Where Lij(β) is the conditional choice probability given a specific value of the preference parameters β; f(β) is the probability density function of the random coefficients β, and ∫⋅dβ integrates over all possible values of β. While the mixed logit accounts for preference heterogeneity using individualspecific coefficients, it lacks detailed information on differences in attribute perception (Train, 2009). WTA measures the substitution rate between an attribute and monetary value (Train, 2009). In models with random effects, calculating the ratio directly would lead to unreasonably high values (Train & Weeks, 2015), hence the price coefficient was considered fixed when calculating WTA for the mixed logit model. The Willingness to Accept (WTA) is a measure used to quantify the monetary compensation that individuals require to accept a less favourable condition in a choice experiment. It is particularly useful in evaluating trade-offs between monetary and non-monetary attributes in program design. The WTA is calculated as follows (3): WTA = – β_attribute / β_payment (3) This formula represents the negative ratio between the coefficient of a nonmonetary attribute (β_attribute) and the coefficient of the payment attribute (β_payment). It indicates the amount of money (in euros per hectare) a farmer 66 would need as compensation to accept an unfavourable change, such as a longer program duration or a higher income reduction risk. 3.5.6 RESULTS 3.5.6.1 Descriptive Statistics The sample includes 109 farmers from Italy. The main socio-demographic characteristics are reported in the table below. Table 3.5.3: Descriptive statistics Variable Mean Std. Dev. Description Age 53 years 11.7 Average age Farm size 42.5 ha 28.3 Total area under management Organic farming 18% – Certification Presence of a successor 64% – Active or expected family successor Higher education 35% – Technical diploma or university degree These data suggest a heterogeneous sample with a good level of professionalisation. The average age is 53 years, indicating a relatively high experience among the sample. The average farm size is substantial, at 42.5 hectares, suggesting significant land management responsibilities. Only 18% of the sample practices organic farming, highlighting that organic certification is still a minority. A relatively high share (35%) of farmers has higher education, indicating good technical or academic training. Furthermore, 64% of respondents report the presence of a successor, either active or expected, pointing to generational continuity in farm management. 3.5.6.2 Mixed Logit Model Estimation The models’ results are presented in tables 3.5.4 and 3.5.5. Table 3.5.4: Results of the MLE Variable Coeff. Std. Error z MWTA (€) Payment per hectare 0.004** 0.001 3.02 – Duration (years) -0.075*** 0.021 -3.56 15.87 67 Effort required (impegno) -0.126 0.121 -1.04 26.59 Payment type (100% resultbased) -0.371* 0.209 -1.78 78.00 Opt-out (no contract option) -2.560** 0.820 -3.12 – (**p < 0.05, ***p < 0.01) The coefficients indicate that farmers are more likely to participate in agrienvironmental programs when higher per-hectare payments are offered (positive and significant coefficient). Conversely, negative coefficients suggest that certain attributes decrease the likelihood of participation. Specifically, farmers show aversion to longer program durations and result-based payment, as these factors significantly reduce the probability of joining the program. Finally, the effort required in terms of income reduction is not statistically significant, suggesting it does not influence farmers' decisions. Willingness-to-accept (MWTA) values provide further insight: farmers are willing to forego €15.87 per hectare to avoid longer program durations. They are also willing to give up €78 per hectare to avoid result-based payment. Overall, the results highlight that financial compensation plays a crucial role, but minimising risk and ensuring flexibility are equally important to encourage farmer participation. Table 3.5.5: Results of the MLE including interaction terms Variable Coeff. Std. Error z MWTA (€) Payment per hectare 0.005** 0.001 2.94 – Duration (years) -0.081*** 0.022 - 3.70 16.32 Effort required (impegno) -0.177 0.143 -1.24 35.35 Payment type (100% resultbased) -0.372 0.233 -1.60 74.13 Opt-out (no contract) -2.151** 0.717 - 3.00 – Soil health perception × optout 3022** 1.051 2.87 – (* p < 0.05, **p < 0.01) The negative coefficient for the duration variable indicates that longer program durations decrease the likelihood of participation. The MWTA of €16.32 suggests that farmers would require an additional €16.32 per hectare to accept an extra year of program duration. 68 The Effort required variable has a negative but not significant coefficient, meaning increased effort requirements reduce participation probability. The MWTA of €35.35 implies that farmers need €35.35 more per hectare to compensate for the additional effort. The coefficient for Soil health perception × opt-out is positive and suggests that farmers who perceive their soil as healthy are more likely to opt out of the program. This indicates a higher disutility associated with participation among these farmers. 3.5.6.3 Willingness to Accept (WTA) The calculated WTA values indicate the compensation required to accept less favourable program conditions. Table 3.5.6: Willingness to accept measurement Attribute MWTA (€) 95% CI Additional years of commitment 15.87 €/ha (6.93 ; 42.15) Increase from low to medium and high-income reduction 26.59 €/ha (-24.9 ; 118.9) Switch to a 100% resultbased payment 78 €/ha (-10.90; 238.08) These results highlight the need for higher financial compensation to convince farmers to accept less attractive program features. Farmers are relatively riskaverse and strongly value flexibility. For instance, they require an additional €15.87 per hectare to commit to longer durations, around €27 per hectare to offset expected income losses and approximately €78 per hectare to accept performance payment structures. Programs that fail to address these concerns may face lower participation rates, even when offering some level of financial support. 3.5.7 Discussion and Conclusions The results of this study offer insights into the preferences and behavioural patterns of Italian farmers regarding the adoption of soil-conservation agricultural practices through program mechanisms. Overall, the findings indicate a cautious attitude among farmers towards engaging in long-term conservation programs, particularly in the absence of flexible conditions and additional incentives. This observation aligns with similar studies conducted in other European contexts, suggesting an increasing trend of hesitancy in economic and structural concerns within the farming community. 69 One of the principal barriers to adoption identified by the analysis is the aversion to long-term commitments. Many farmers, especially those without a clear successor or intergenerational continuity on their farms, perceive extended program durations as limiting their autonomy and future decision-making capacity. This concern increases when programs lack options for early withdrawal or renegotiation. Therefore, introducing greater flexibility in the program’s framework could be a crucial factor in enhancing the attractiveness of these schemes. Another major obstacle is the perceived risk of income loss. Farmers often weigh the uncertainty of potential yield impacts or increased costs associated with conservation practices, even when financial compensation is offered. The estimated willingness to accept (WTA) values, in this study, highlight how much compensation farmers need to take less favourable program conditions, such as longer commitments or more restrictive payment mechanisms. Notably, the aversion to income loss appears to outweigh other considerations, suggesting that economic viability remains a dominant factor in decision-making processes. Mixed payment mechanisms, which combine fixed and performance-based elements, appear to be a promising avenue for encouraging participation, since the result-based payment decreases the probability to participate. While these structures offer a potential balance between security and results-driven incentives, their successful implementation depends on building farmers’ trust in monitoring systems and outcome evaluations. A lack of transparency or perceived complexity in these systems could undermine their effectiveness. Trust in the Institutions and clear communication are crucial in fostering confidence and acceptance. The role of intermediaries, such as cooperatives, farmers' associations, agricultural consortia, and banks, emerges as a crucial element in bridging the gap between policy and practice. These actors can act as trusted facilitators, providing technical support and financial services. Their involvement can help overcome scepticism and build a sense of shared responsibility and community endorsement, particularly in areas where institutional trust is low or fragmented. A set of operational recommendations can be drawn to increase the appeal of conservation programs further. Programs should allow for early withdrawal under specific conditions, reducing the perceived rigidity and risk. Accompanying each program with technical training and ongoing support can empower farmers and improve implementation outcomes. Moreover, transparent communication about the expected environmental and economic benefits is essential to align farmer expectations with policy objectives. Finally, involving local stakeholders in the design and adaptation of programs can ensure that they are sensitive to regional characteristics and farmer needs, enhancing their relevance and legitimacy. In conclusion, the success of soil conservation programs in Italy, much like in other parts of Europe, will depend on the capacity to design flexible, trustworthy, and economically viable instruments. These programs must respond to farmers’ practical realities, address their concerns transparently, and build institutional and social trust. Only by integrating these elements can such programs contribute 70 meaningfully to sustainable agricultural transitions and the broader goals of the European Green Deal. 71 3.6 Bulgaria 3.6.1 Synthesis of the results Title: Stimulating the use of cover crops in traditional wine production in Bulgaria Authors: Dimitre Nikolov, Ivan Boevsky, Martin Banov, Ekatherina Tzvetanova, Krasimir Kostenarov, Kristina Todorova, Vanya Bankova Structured Abstract: Introduction. This case-study targets the production and consumption of a traditional product (grapes and wine) in Bulgaria. The aim is to assess the attitudes of the respondents regarding the production of traditional local products (grape and wine), produced with and without agri-environmental measures focused on the prevention of soil erosion. The business model of interest is the value chain model where soil-improving practices in grape production can be stimulated through improved quality of the wine, and resulting increased opportunities and demand from wine tourism in Bulgaria. The specific agrienvironmental measure that is chosen for the study is the application of cover crops as a way to reduce erosion and improve soil health. Cover crops have multiple benefits for the soil, including decreasing soil erosion, reducing soil compaction, and increasing organic matter. Methods. Interviews were conducted with the farm managers/owners of vineyards in the Plovdiv and Pazardzhik districts. The interviews took place in August, September and October 2024. A DCE is conducted to derive the willingness of farmers to participate in a scheme focused on this measure. Farm and farm characteristics (farm size, education, gender) are assessed to see whether they influence the willingness to participate. Results. Results show that price is an important determinant to stimulate participation. However, no significant results were found connecting the farm and farmer characteristics to the willingness to participate. Key Words: cover crops, wine production Case study: Traditional wine in Bulgaria Incentives tested: Value chain payments for cover crops in wine production Policy considered: Value chain payments (no policy) 72 3.6.2 Introduction This case-study targets the production and consumption of a traditional product (grapes and wine) in the Plovdiv and Pazardzhik districts. The interviews with the farm managers/owners took place in August, September and October 2024. The aim is to assess the attitudes of the respondents regarding the production of traditional local products (grape and wine), produced with and without agrienvironmental measures focused on the prevention of soil erosion. The business model of interest is the value chain model where soil-improving practices in grape production can be stimulated through improved quality of the wine, and resulting increased opportunities and demand from wine tourism in Bulgaria. The specific agri-environmental measure that is chosen for the study is the application of cover crops as a way to reduce erosion and improve soil health. Cover crops are believed to have multiple benefits on the soil, including decreasing soil erosion, reducing soil compaction, and increasing organic matter. A DCE is conducted to derive the willingness of farmers to participate in a scheme focused on this measure. 3.6.3. Methodology A DCE and a farmer survey was conducted. At the start of the interviews, farmers were informed that the aim of the survey is to analyse the production and consumption of traditional products (grape and wine) and rural tourism as a result of a new business model “Integrated Vineyard Production with Wine and Rural Tourism” in the Plovdiv and Pazardzhik districts. The aim was to assess the attitudes of the respondents regarding the production of grapes and wine with and without the use of cover crops. The survey took place in August, September and October 2024. The total number of participants in the survey is 55. The questionnaire was divided into three parts: Part A – General questions; Part B - Choice experiment Part C – Environmental questions. In part B, respondents were confronted with choice cards containing two attributes (Table 3.6.1): (i) area enrolled in the contract; (ii) Levels of compensation for the additional environmental efforts. The DCE methodology involves creating an experimental design in which the different levels of the attributes are systematically combined to generate realistic but varied profiles. Each respondent makes a choice, allowing researchers to collect data on their preferences. Table 3.6.1: Example of a choice card in the DCE Attributes Option 1 Option 2 Option 3 Option 4-Out Area enrolled in the contract 30% 50% 100% None of these. Stick to production without AEM. Compensation for environmental efforts 67 euro per ha 102 euro per ha 152 euro per ha 73 (planting cover crops) I prefer:     3.6.4. Results 3.6.4.1 Descriptive analysis The interviews covered all farms with vineyards and wine production in the two studied regions. Figures 4.6.1 – 4.6.4 show results of the descriptive analysis (55 respondents). Figure 3.6.1: Age structure of the sample Figure 3.6.2: Work experience of the managers in the sample 18-40 14% 41-60 71% 61+ 15% up to 5 11% 5-15 29% more than 15 60% 80 3.6.4.4 Analysis of the choice experiment and respondents’ perceptions of agrienvironmental measures One of the survey questions is targeting the perception of four different types of agrienvironmental measures. The respondents were asked to assess which of the four measures could have a positive impact on their farm – cover crops, buffer strips, intermediate crops and drainage furrows. If we relate these answers with the choice of our four options in the choice experiment, we can observe the following outcomes: - Regarding Option 1 (dedicating 25% of the total cultivated land for AEMs) the predominant part of the respondents accept all of the four AEMs as positive for their farm (the percentage is above 80%). - Regarding Option 2 (dedicating 50% of the total cultivated land for AEMs) the measure cover crops is the one, where all of the respondents find acceptable and positive for their farm. In this option with the exception of cover crops, for the other three measures, nearly half or over the half of the respondents are not sure if these environmental measures are applicable and beneficial for their farm. Regarding the measure buffer strips almost 60% of the respondents are uncertain that the measure is beneficial for them, those results are 46% and 50% for intermediate crops and drainage furrows respectively. - Regarding Option 3 (dedicating 100% of the total cultivated land for AEMs) the results are similar with those in Option 2 regarding cover crops. However, here the percentage of uncertain answers regarding the measures buffer strips and drainage furrows is the highest - 65% and 60% respectively. These are also the two measures, where almost one fourth of the respondents don’t agree about the applicability on their farm. - Those who opted out (the fourth option) are predominantly uncertain about any of the four measures. Overall, we can conclude that the higher the land dedicated for AEMs, the lower the acceptance of buffer strips and drainage furrows as environmental measures. Cover crops appear to be the measure that is most acceptable in all of the options that accept to implement AEMs. 3.6.4.5 Analysis of the choice experiment and respondents’ attitudes to soil quality Another question of the survey that deserves attention is the one referring to the different aspects of soil quality as an ecosystem service. The respondents were asked to state which of the three types of ecosystem services must be achieved through implementation of agri-environmental practices on the vineyards. All of the three ecosystem services are connected to different aspects of soil health - reducing soil 81 erosion, increasing soil nutrients and increasing soil biodiversity. Several things can be observed here: - For Option 1 (dedicating 25% of the total cultivated land for AEMs) where the dedication of land is the smallest percentage, respondents are not sure about the eco service increase of soil nutrients (almost 60%) and increasing soil biodiversity (40%). In contrast, all of the respondents find that reducing soil erosion must be achieved through AEMs on their farm. - For Option 2 (dedicating 50% of the total cultivated land for AEMs) and 3 (dedicating 100% of the total cultivated land for AEMs) the answers are almost similar. Here, most of the respondents agree in highest percentage with the ecosystem services reduction of soil erosion and increase in nutrients (answers vary between 90-95%). However, with the service increase of soil biodiversity the percentage of disagreements is the highest – (15% for Option 2 and 18% for Option 3). - For the opt-out option it is not surprisingly that the answers for all three types of ecosystem services are dominantly “unsure” or “don’t agree”. Where for the ecosystem services decrease of erosion and increase in nutrients the percentage of agreement is 41% and 48% respectively. For soil biodiversity all of the respondents are uncertain that this service should be achieved through AEMs on their farm. This leads to the conclusion that from all three types of ecosystem services biodiversity probably is the most understated and unacceptable from farmers’ perspective. The overall conclusion is that the higher the land dedicated for AEMs, the higher the understanding and acceptance that ecosystem services must be achieved through environmental measures. 3.6.5. Conclusions The analysis is targeting the production and consumption of traditional products (grape and wine). Our goal was to assess the attitudes of the respondents regarding the production of traditional local products (grape and wine), produced with and without agri-environmental measures focused on the prevention of soil erosion. We interviewed 55 managers/owners of farms with vineyards and wine production in Plovdiv and Pazardzhik districts in Bulgaria. The following conclusion can be derived:The predominant part of the respondents find that reduction of soil erosion and increase of nutrients are important ecosystem services and must be achieved via implementing AEMs. However, to a lesser extent they have the same attitude towards increasing soil biodiversity. Probably, the lower level of agreement for this ecosystem service is due to the difficulty to comprehend its importance for soil health. 82 Regarding the positive impact of four agri-environmental measures - cover crops, buffer strips, intermediate crops and drainage furrows – respondents’ answers showed that “cover crops” is the measure with highest acceptability among them. On the other hand, it appears that buffer strips and drainage furrows are seen as the measure of greatest uncertainty. One explanation for this outcome could be the level of technical difficulty for application of these two measures. Regarding respondents’ attitudes towards soil health and climate change all of them agree that climate change has a negative impact on their farms, that soil health is essential for them and they find it an important factor for their farms’ productivity. Regarding the choice experiment outcome, almost 70% of the respondents chose to dedicate half or all of their land to the implementation of AEMs (in our case cover crops). Regarding the attitude of farmers towards several AEMs (cover crops, buffer strips, intermediate crops and drainage furrows) the analysis showed that the higher the land dedicated for AEMs, the lower the acceptance of buffer strips and drainage furrows as environmental measures. Also, the higher the land dedicated for AEMs, the higher the understanding and acceptance that ecosystem services must be achieved through environmental measures. Based on the analysis and the interviews, the following recommendations are made: Increasing the specific knowledge of agricultural producers regarding environmental protection and provision of ecosystem services is of paramount importance. Some ecosystem services are very well recognized by the farmers like reducing soil erosion. The reason for this is not only the direct observation of the physical process, but also the effects it has on the farm productivity. On the other hand, ecosystem services like soil biodiversity is recognized by farmers to a lesser extent. Nonetheless, this service has an equal importance for farm productivity as well as soil erosion. Provision of sufficient and appropriate compensation for implementation of AEMs. As we saw in the analysis almost half of the respondents agree to dedicate 100% of their farmland for introducing cover crops for the compensation level of 152 euro/ha. This is twice the compensation currently offered by the public payments. Technical support for farmers for the accurate implementation of certain environmental measures. From the analysis for some measures, the acceptance level of farmers is relatively low as compared to cover crops for example. This might stem from the fact that adoption of these two measures is more difficult from technical perspective. 83 84 4 Joint Discussion Five DCE experiments were conducted. They had different regional contexts, two DCEs were conducted in Germany, one in the Netherlands, one in Italia, and one in Bulgaria. The DCEs referred to different business models, offered different incentives, and had different underlying policies. The experiments for the Dutch, and one of the German (TUM) cases focused on incentives for carbon credits. The second German experiment (ZALF) and the Bulgarian experiment were on value chain incentives. The Italian case looked at combinations of incentives (agri-environmental schemes and carbon credit trade). Table 4.1. provides an overview of the case studies conducted. A comparison of the case studies is difficult, and the results might not allow for a general conclusion regarding farmers’ preferences for a business model type. For instance, the regional context might play a role. Dutch farmers are described to be difficult to motivate to extensify their practices as shown, for instance, for the participation in AES (Zimmermann & Britz, 2016). This can be explained by the high management intensities and higher opportunity costs when deciding about implementing soil health-friendly practices. These potentially also result in the low predicted probabilities observed in the Dutch carbon credit trade DCE. In addition, the experiments’ attributes were different. In the Dutch DCE only contract features (e.g., ‘result-based payments’ for carbon prices, or contract durations) are included as attributes. The measures that farmers must implement were fixed per farm type and did not vary. In other DCEs contract features and management requirements were included (e.g., cover crop requirements for the Bulgarian case). Table 4.1 displays the incentives included in the DCEs. All approaches (using contract attributes only, management requirements only, or a mix of both) are common in DCE on farmers’ decisions to participate in contracts that seek to promote less intensive agricultural practices, as a review by Schulze et al. (2023) shows. Table 4.1. Overview of the conducted DCE Case Business Model Policy analysed Incentives included Pred. Probabilities Rabo Carbon Bank (NL) Carbon credit trade Carbon Removals and Carbon Farming (CRCF) Regulation (EU/2024/3012) Fixed compensation of costs (%), carbon credit price, termination option for successors, free training, discount loans 40 % CO2-Land (GER) Carbon credit Carbon Removals and Carbon Fixed compensation of costs (%), 46% 85 Farming (CRCF) Regulation (EU/2024/3012) carbon credit price, termination option for successors, free training, subsidies for investments AgoraNatura (GER) Biodiversity & ecosystem service credits through online marketplace Roadmap towards Nature Credits (initial step towards EUlevel regulation) Compensation of costs, credit price, monitoring options - Conservation agriculture (IT) Public payments for sustainable practices Common Agricultural Policy (AES) Price, cost compensation, contract duration, result-based payment - Traditional wine (BG) Value chain payments for soil-improving practices Common Agricultural Policy (AES) Price - 86 Even though the contract requirements and the incentives varied, some joint conclusions might be possible when looking at the effects of specific attributes, and the included covariates (farmer and farm characteristics). The included attributes (incentives, commitments etc.) demanded by the contract and the effects found are summarized in table 4.2. The literature review conducted in Work Package 2 showed that DCEs can include various monetary and non-monetary incentives. The monetary incentives (fixed payments per hectare, result-based payments or premium prices paid to farmers) were significantly positively evaluated in all experiments. Monetary attributes are described as decisive in multiple articles on the topic, e.g. for carbon credit trade (Canales et al., 2024; Shaikh et al., 2007), for AES (Schaub et al., 2023), or for value chain payments (Salazar-Ordóñez et al., 2021; Weituschat et al., 2023). These findings are also in line with the theoretical model, where farmers are expected to choose the option that offers the highest utility (Train, 2009), in a business management context, the utility of a business model strongly depends on the potential profits that are affected by the to-be-expected payments. In some of the experiments result-based, action-based or hybrid payments were featured. These have different advantages and disadvantages (see Burton & Schwarz (2013) or Herzon et al. (2018) for an overview). Carbon credit trade commonly includes result-based payments (meaning that the payment is based on the amount of carbon stored). The experiments from the Netherlands, Italy and Germany (TUM) on carbon credit trade and the experiment on biodiversity credits traded on an online market place in Germany (ZALF) featured the different payment types. It became apparent that the risks entailed in result-based payments is negatively perceived, and that purely result-based payments can only attract few farmers. In the Dutch, both German and the Italian experiments, action-based or hybrid payments were also offered in some choice sets. Such hybrid payments might be possible when allowing combinations of payments from carbon credit markets and public payments. Comparing the DCE' s results to the results of the policy innovation lab, this was to be expected, as the participants commonly highlighted that result-based payments are too risky (Bartolini et al., 2024). Considering the experiment from the food value chain. The Bulgarian experiment found that price is an important determinant to stimulate participation but that farmers’ attitudes and perceptions also matter and may differ between farm and farmer types. Looking at other sources, few articles consider value chain payments for soil health measures so far. One exception is the article is by Weituschat et al. (2023). They evaluated farmers’ willingness to participate in Barilla’s sustainable farming initiative (food value chain payment), where farmers receive premium prices when farming more environmentally friendly. They found that some farmers are interested in such incentives, but that the incentive cannot attract those, for instance, valuing higher flexibility when making farm management decisions. 87 Finally, the NOVASOIL project also identified public payments for AES or eco-schemes as potential business models. The conducted DCEs do not focus on public payments alone, and only combine public action-based payments with other payments (e.g., in the German (ZALF) and Italian cases). This seems reasonable as a large body of literature on farmers’ acceptance of public payments for soil health, or other less intensive practices exists (e.g, Barreiro-Hurlé et al., 2010; Jaeck & Lifran, 2014; Lapierre et al., 2023; Niskanen et al., 2021; Trenholm et al., 2017; Villanueva et al., 2017). Besides the different payment types (action-based, result-based, hybrid) other nonmonetary incentives were commonly featured. These included training opportunities in the Dutch and the German (TUM) experiment. They were found to promote the take up of the measures as well. However, the Dutch and the German cases show that training might not be equally valued across all groups. That training opportunities affect farmers’ acceptance of contracts for public payments (Lapierre et al., 2023; Schaub et al., 2023), or contracts to participate in carbon credit markets (Buck & Palumbo-Compton, 2022) is described in the literature. Buck & Palumbo-Compton (2022) reviewed the literature on farmers’ willingness to adopt practices that lead to soil carbon sequestration. They point out that the literature on soil health practices is too limited to draw general conclusions. Still, they point out that the results indicate that an awareness of co-benefits is a strong driver of adoption and that better education and training opportunities can help. In addition, some articles (e.g., Kragt et al., 2017) describe access to credits as important for farmers' decision to accept contracts. For carbon farming this might be because a payment for carbon credits can only be expected after carbon cumulated in the soil, and because investment costs can be high. Better access to credit helps farmers to finance the period between measure implementation and receiving a payment, or supports the investment in machinery necessary for soil health-friendly management. Access to credit as an incentive was investigated for the Dutch and the German (TUM) case-study but the effects were not found to be significant. An interesting result of the experiments conducted is that the Dutch study and the German (TUM) study included a termination option of contracts for successors. They show that such options also hold potential to motivate farmers to opt for longer contracts. Farmers’ reluctance to accept long contracts (longer than five years) is described as problematic for public payments by Harmanny & Schulp (2023) or Tiusanen et al. (2022). For carbon credit markets, Rochecouste et al. (2017) conducted interviews with farmers and found that they are not willing to limit the flexibility of themselves or their successors. Including termination options for successors could be an interesting option for all contracts in the business models researched (public payments, carbon credit markets, value chain payments). However, offering such options might be challenging for the carbon credit trade business model. For carbon farming, reliable, long-term storage needs to be ensured. Whether contracts between farmers and buyers can include such options is, therefore, questionable. A suitable option might be to set up a contract between intermediaries 88 (e.g. farmer associations that implement carbon removal measures with their members), and the buyers. It might be easier for such intermediaries to build up buffers in case farms terminate contracts. Besides this, a collective implementation with groups is considered to hold further potential to motivate farmers to participate, in nature protection (Barghusen et al., 2021; van Dijk et al., 2015). Schaub et al. (2023) also describe a close relationship with peers as influential for the pick-up of public AES contracts. Table 4.2. Overview of the included attributes in the DCE (and their effect). Rabo Carbon Bank (NL) CO2 Land / Soil Fertility Fund (TUM, GER) AgroNatura (ZALF, GER) Traditional Wines (BG) soil fertili ty (IT) Incentive – Fixed, action-based payment (e.g. to cover the costs of a measure) + + + + + Incentive – Result-based payment (e.g. for carbon stored) + + - Incentive – Information on lower income volatility Incentive – Information on the expected farm income (e.g. prices at farm gate) + Incentive – Premium prices (through eco-labelling) Incentive – Bonus payments for reaching spec. quality or quantity Incentive – Collective bonus (for building networks) Incentive – Sharing of costs (production costs, specific asset costs) + + + Incentive – Supply of inputs Incentive – Training and advice + + Incentive – Access to credit (discounts on loans, easier access to loans) 0 0 Incentive – Co-benefits (soil fertility, less erosion, water holding capacity etc.) Incentive – Public recognition (e.g, labels, signs usable at farm gates) CommitmentsDuration (contract length) - - - Commitments - Min. quantity requirements to enter a contract (e.g., hectares enrolled, soil type) Commitments – Management requirements (e.g. cover crops or red. tillage requirements) 89 Others - Flexibility (termination options, options for renegotiation of measures / contract requirements) + + Others – – Management parties (e.g., monitoring and administration) + Significantly positive effects are indicated with a +, significantly negative effects with a -, insignificant effects display a 0 Besides this, farmer and farm characteristics were included in (most) DCEs. They allow to learn which farmers are more or less likely to implement the business models. For instance, suppliers interested in value chain initiatives could use this information to identify farmers they can approach more easily. The results found for the covariates are summarised in Table 4.3. They refer to the results found for the average farmer. The table is structed following a review by Dessart et al. (2019). The authors distinguish between dispositional factors, social factors, and cognitive factors. Dispositional factors are stable internal variables like personality, motivations, and beliefs. Social factors refer to farmers’ engagement with others but also include social norms. Cognitive factors relate to learning and knowledge and are assumed to impact farmers’ perception of the costs and benefits of measures. In addition to these personal factors, farm characteristics are included following Schaub et al. (2023) who focused on farms’ structural features to provide information on the farmers’ opportunity costs. The result of Dessart et al. (2019) that better cognitive abilities promote farmers’ engagement is confirmed for the Dutch, the Italian and the German (TUM) case. In addition, farmers’ beliefs (e.g. regarding climate change) are commonly described to play a role (confirmed in the Italian and the Dutch case). As expected, some DCEs also showed that farmers who farm more intensively are less likely to contract (e.g., the Dutch case). This was to be expected and is commonly described in review articles (Schaub et al., 2023).However, the finding remains challenging as more intensive farms create more negative effects on the environment. Barreiro-Hurlé et al. (2010) find that farm features and contract features have a stronger importance when very demanding measures are required. When it comes to softer requirements with lower opportunity costs, personal factors play a larger role. The varying contract requirements might explain the heterogeneity across the experiments. 96 Weituschat, C. S., Pascucci, S., Materia, V. C., & Caracciolo, F. (2023). Can contract farming support sustainable intensification in agri-food value chains? Ecological Economics, 211, 107876. https://doi.org/10.1016/j.ecolecon.2023.107876 Zimmermann, A., & Britz, W. (2016). European farms’ participation in agrienvironmental measures. Land Use Policy, 50, 214–228. https://doi.org/10.1016/j.landusepol.2015.09.019 7. Acknowledgments The authors would like to thank the EU for funding, in the frame of the European Union’s Horizon Europe research and innovation programme under Grant Agreement GA 101091268. The document reflects only the author’s view. The Agency is not responsible for any use that may be made of the information it contains.