The Symbiotic Production of Food and Green Electricity: Consumer Preferences for Food Produced in Agrivoltaic Systems

Schulze, Maureen; Jürkenbeck, Kristin Article — Published Version The Symbiotic Production of Food and Green Electricity: Consumer Preferences for Food Produced in Agrivoltaic Systems Business Strategy and the Environment Provided in Cooperation with: John Wiley & Sons Suggested Citation: Schulze, Maureen; Jürkenbeck, Kristin (2024) : The Symbiotic Production of Food and Green Electricity: Consumer Preferences for Food Produced in Agrivoltaic Systems, Business Strategy and the Environment, ISSN 1099-0836, Wiley, Hoboken, NJ, Vol. 34, Iss. 2, pp. 2088-2102, https://doi.org/10.1002/bse.4080 This Version is available at: https://hdl.handle.net/10419/319302 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by-nc-nd/4.0/ Business Strategy and the Environment, 2025; 34:2088–2102 https://doi.org/10.1002/bse.4080 Business Strategy and the Environment RESEARCH ARTICLE OPEN ACCESS The Symbiotic Production of Food and Green Electricity: Consumer Preferences for Food Produced in Agrivoltaic Systems MaureenSchulze1 | KristinJürkenbeck2 1Consumer and Behavioural Insights Group, Department of Management, Society and Communication (MSC), Copenhagen Business School, Frederiksberg, Denmark | 2Department for Agricultural Economics and Rural Development, Marketing for Food and Agricultural Products, University of Goettingen, Göttingen, Germany Correspondence: Kristin Jürkenbeck ([email protected]) Received: 29 February 2024 | Revised: 8 October 2024 | Accepted: 14 November 2024 Funding: We acknowledge support by the Open Access Publication Funds of the Göttingen University. Keywords: agrivoltaic systems| consumer preferences| food| renewable energies ABSTRACT In light of the commitment by EU member states to achieve climate neutrality for the European continent by 2050, the expansion of renewable energy sources emerges as a significant challenge of our time. Agrivoltaic systems, which combine the production of renewable energy and food, offer a solution to alleviate the competition for limited land resources. However, scientific insights into whether consumers value food produced in agrivoltaic systems are lacking, so far. Knowledge of consumers' preferences is, however, crucial to successfully commercialize food production within agrivoltaic systems. This study addresses this research gap by conducting a hypothetical choice experiment with a sample of 448 German consumers. It examines consumers' preferences for food produced in agrivoltaic systems, using raspberries as a case study. Results of the random parameter logit (RPL) modeling revealed that consumers are willing to pay a premium for food labeled with information about the production of green electricity. Additionally, information on reduced plastic and water usage of food produced within agrivoltaics systems were also valued, next to domestic and regional production. A latent class segmentation was carried out for the targeted marketing for food produced in agrivoltaic systems. This resulted in four distinct consumer segments that differ according to their preference structure. Targeted recommendations are provided to enhance consumer acceptance and facilitate the diffusion of agrivoltaic systems. 1 | Introduction The shift toward renewable energy sources, which aims to address global energy demands while simultaneously replacing fossil fuels—a major driver of climate change—stands out as a significant sociopolitical challenge in our time (European Commission2019). The expansion of most renewable energy sources requires large areas of land, which are also needed to face the growing challenge of food security due to the impacts of climate change and a growing world population. To alleviate the competition for limited land resources, the implementation of agrivoltaic systems has been suggested. In agrivoltaic systems, photovoltaic panels are mounted above the ground, facilitating food production such as grain, fruit, and vegetable crops underneath while simultaneously generating solar electricity above (ISE 2022), thus offering promising synergy effects (Weselek etal.2019). With a capacity increase from 2.9 gigawatts (GW) in 2018 to over 14 GW in 2020 worldwide, the installation of agrivoltaic systems has also experienced exponential growth but only accounts for a small share This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made. © 2024 The Author(s). Business Strategy and the Environment published by ERP Environment and John Wiley & Sons Ltd. [Correction added on 20 December 2024, after first online publication: The Funding section has been updated in this version.] 2088 of solar energy so far (ISE2022). Although agrivoltaic systems offer several benefits, scientific insights into whether consumers value food produced in agrivoltaic systems is lacking, so far. Knowledge on consumers' preferences is, however, crucial to successfully commercialize food production within agrivoltaic systems. A steadily growing group of consumers is increasingly interested in how the food they consume is produced. For these consumers, a sustainable food production process that neither harms the environment nor human health is of particular importance (Bangsa and Schlegelmilch2020). Meat and dairy with higher animal welfare requirements (Ammann etal.2024) or locally/regionally produced food (Bannor and Abele2021) are just two examples of consumers' increasing interest in how the food they eat is produced. Because consumers cannot experience, neither before nor after consumption, whether the product is produced sustainably or not, additional frontofpackage information is needed (Schulze, Spiller, and Risius2021). The food industry has already been reacting to consumers changed preferences providing consumers with additional information on sustainability or healthrelated information (e.g., labels, health claims). The dual use of agricultural land to simultaneously produce food and solar electricity offers various benefits for crop cultivation that has the potential to serve consumers' increasing interest in environmentally friendly produced food products and healthbenefits. Because the solar modules offer sun protection for plants, which in turn prevents excessive heating (Marrou etal.2013) agrivoltaic systems are characterized by improved water use efficiency (Elamri et al. 2018; Marrou etal.2013). It is widely acknowledged that consumers value information on the environmental impact of the food production process (Bangsa and Schlegelmilch2020) and the few studies that specifically investigated consumers' preferences for water usage in food production suggest a strong consumer interest in information about water usage in food production (Peschel etal.2016). Another benefit of agrivoltaic systems, in terms of sustainability is, that the production systems offer benefits in terms of reduced plastic usage in fruit production (Weselek etal.2019). Especially in fruit production, sheltering of plants is necessary to protect plants from frost and hail (Gandorfer, Hartwich, and Bitsch2016). For this purpose, in traditional farming, plants are covered with antihail nets made of plastic. Cultivation of plants under solar modules makes this redundant. This reduced usage of plastic in the production process of food produced in agrivoltaic systems has the potential to serve consumers with an interest in environmental protection. Research on food packaging has already shown that some consumers prefer environmentally friendly food packaging (Macht, KlinkLehmann, and Venghaus2023). Next to environmental sustainability, health is another important purchase motive for food (Raaijmakers etal.2018). Food produced in agrivoltaic systems can offer consumers healthrelated product characteristics. As such, sheltering plants by solar modules protect plants from fungal diseases following extended rainfall (Toledo and Scognamiglio 2021; Weselek et al. 2019; Sinha etal. 2018). This in turn allows farmers to reduce the usage of pesticides. Studies by Koch etal.(2017) and Simoglou and Roditakis(2022) identified that consumers perceive pesticide residues in food products as major risks to human health. However, although food production in agrivoltaic systems offers various environmental sustainability and healthrelated benefits, the evaluation of agrivoltaic systems presumably confronts consumers with conflicting preferences. It is widely acknowledged that consumers usually tend to initially reject novel food technologies and unfamiliar foods. In contrast to many other domains, novel technologies in the food sector are predominantly perceived as negative by consumers. Novel food production methods often suffer from being perceived as unnatural (Siegrist and Hartmann2020). Food produced in agrivoltaic systems thus confronts consumers with the conflicting preferences of a general interest in sustainable food production and skepticism toward novel food production technologies. How consumers react to this interplay of conflicting preferences and whether they value the simultaneous production of food and solar energy on the same land is unknown so far. This study tries to close this research gap investigating consumers' preferences for raspberries from agrivoltaic systems. Raspberries were chosen as a product example because they are a frequently consumed berry in Germany. In total 64.9 kg of fresh fruit, including 5.1 kg of berries and approximately 1 kg of raspberries per inhabitant was consumed in 2022/2023 (Bundesanstalt für Landwirtschaft und Ernährung 2021). Because raspberries available in German supermarkets are among the fruits that are most frequently contaminated with pesticides (Bundesamt für Verbraucherschutz und Lebensmittelsicherheit 2023), production in agrivoltaic systems offers great potential. Due to the plant's high sensitivity to weather effects, the cultivation of raspberries is highly dependent on sheltering, which again offers great potential for production in agrivoltaic systems (Destatis2023). Additionally, agrivoltaic raspberry production on a larger scale is gaining increasing interest among farmers in Europe (Lohrern.d.), and thus, it can be expected that raspberries cultivated under solar modules will be available in supermarkets in the near future. Therefore, to successfully market raspberries from agrivoltaic systems, consumers' willingness to pay (WTP) a price premium is crucial. An online survey, including a discrete choice experiment was conducted with German consumers. The paper is organized as follows: Section2 provides an overview of the previous scientific results and theoretical background of consumer preferences for sustainably produced food with a special focus on preferences for fruits. Section3 describes the methodological approach by explaining in detail how (1) the overall preference structure for raspberries produced within agrivoltaic systems was assessed and (2) the segmentation approach to account for differences according to consumers' preferences for product attributes associated with raspberries produced in agrivoltaic systems was applied. Section4 provides the reader with the results of the study, followed by Section5 where results are discussed and related to previous research. In Section6, recommendations regarding the market potential, as well as marketing recommendations for food products produced within agrivoltaics systems, are derived. The article ends with Section7 highlighting limitations and future research. 2089 2 | Theoretical Overview Raspberry production in agrivoltaic systems holds great potential to provide food with climate and environmental benefits— food product characteristics that are valued by a growing group of consumers (Schulze, Spiller, and Jürkenbeck2022). However, research on consumers preferences for raspberries in particular is scarce. Previous studies on consumers' buying motives for different kinds of fruits showed that consumers value characteristics such as freshness (Péneau etal.2006), taste and visual appearance (Hueppe and Zander 2024), as well as packaging (Koutsimanis etal.2012) and geographical origin (Segovia and Palma2016). In the decisionmaking process, consumers need to rely on product attributes to evaluate the quality of a product. Usually, consumers rely on search, experience, and credence attributes (Darby and Karni1973). Search attributes, such as price, can be evaluation prepurchase, experience attributes, such as taste, after purchase, and credence attributes, such as the environmental impact of foods, can be assessed neither prepurchase nor postpurchase. Consumer's increasing interest in sustainable food production processes has thus enhanced the importance of credence attributes. Consumers value information on the production method, such as organic certification (Rahman and NguyenViet2023) or fair trade (Fuller and Grebitus2023; Iweala etal.2022). Tait, Saunders, and Guenther(2015) highlighted that UK consumers appreciate increased water use efficiency when evaluating the fruit production process. Another study by Oh, Herrnstadt, and Howard (2015) also found evidence for consumers' interest in sustainable production methods by revealing a higher WTP for local production and natural bird pest management in apple and grape production. More recent studies, for example, from Mazzocchi, Ruggeri, and Corsi(2019) or Di Vita etal.(2021) also pointed to consumers preferences for sustainable fruit production processes. As known from research on other food categories, sustainabilityrelated information on the production process of foods matters to an increasing subgroup of the population, yet not the whole population (Raaijmakers etal.2018). To thoroughly understand these heterogeneous consumer preferences, as well as to estimate the market potential of food produced in agrivoltaic systems, it is beneficial to classify consumers into homogeneous groups based on their preferences, a method that has been widely applied previously in food marketing research (Jürkenbeck, von Steimker, and Spiller2024; Risius, Hamm, and Janssen2019). The same applies to consumers' reluctance toward new food technologies and production methods, as some individuals value sustainable food characteristics and technologies more than others (Piracci etal.2023; Siegrist and Hartmann2020). However, most consumers usually have scarce knowledge about how their food is produced (Connor and Siegrist2010). Consumers' opinions about the production process of food are rarely based on elaborate information processing. To compensate for the lack of profound knowledge consumers tend to rely on heuristics to facilitate the decisionmaking process or process of opinion formation (Siegrist and Hartmann2020). One widely acknowledged heuristic to evaluate the quality of food is that consumers overestimate the value of naturalness. The so called naturalisbetter heuristic leads to consumers' assumption that natural foods (e.g., foods that are produced with [rarely] no human interference) (Siegrist and Hartmann2020) are perceived as healthier and tastier, and the production process as more sustainable (Román, SánchezSiles, and Siegrist2017). Reluctance toward human intervention in the food production process has been shown in various sectors along the food supply chain, such as novel food technologies applied in the production process (e.g., gene technology) (Scott etal.2018) or perception of using novel technologies in housing and handling of livestock (e.g., robots in dairy farming) (Langer and Kühl2024). Whether the beneficial characteristics of food production in agrivoltaicsystems overcome consumer reluctance toward novel food production processes still needs to be investigated. The simultaneous production of green electricity on the same land as food is one core sustainability benefit. It remains, however, an open question whether consumers value frontofpackaging information on the production of green electricity in the food production process. Previous research, though not in the food domain, showed that consumers are generally open to green energy. Rogers etal.(2008), Liu, Wang, and Mol(2013), and Vuichard, Stauch, and Wüstenhagen (2021) showed that wind and solar energy are mostly accepted. Similarly, recent polls from 2023 showed that a large share of European consumers supports the development of renewable energy sources, especially the expansion of solar energy (Tesvolt2023; European Union2023). However, the implementation of large solar parks is often accompanied by landscape changes and therefore has a visual effect on landscape quality—a consequence of solar energy production that is less accepted by the public (Cousse2021). 3 | Materials and Methods 3.1 | Data Collection and Questionnaire Structure Data were collected by means of a structured online questionnaire that was available to German consumers between September 19 and 26, 2023. Data collection was supported by the online access panel provider Bilendi GmbH, Berlin, Germany. The questionnaire was structured as follows: After a short introduction, participants were asked to indicate quotarelevant sociodemographic characteristics, followed by questions regarding their shopping behavior for food in general. Subsequently, participants were presented with an informational text that briefly explained the concept of agrivoltaics systems (AppendixA). Given the early adoption stage of agrivoltaic systems and the limited empirical data on consumer acceptance of food produced in such systems, it is reasonable to assume that consumers require additional information on what food production in agrivoltaic systems entails. This was especially important to ensure participants understood the product characteristics related to the production process. However, this experimental design must be considered when interpreting the results and deriving recommendations (e.g., emphasizing the importance of educating consumers to foster acceptance of sustainably produced food products). The informational text was developed together with an expert from the field of agrivoltaics and subsequently discussed among researchers within the field of food marketing to guarantee Business Strategy and the Environment, 2025 2090 neutrality (see AppendixA). This was followed by the discrete choice experiment, where respondents were confronted with hypothetical choice situations on raspberries. Because discrete choice experiments are of hypothetical nature, a cheap talk script was included directly before participants were confronted with the choice tasks. The cheap talk script is a widely applied approach to reduce biases by making participants aware of the importance to answer as honestly and realistically as possible (Lusk2003). The questionnaire closed with questions regarding participants' awareness of different sustainability dimensions. To check participants' attention, two explicitly instructed response items (e.g., “To check for your continuous attention, please select ‘agree’”) were included to ensure high data quality. Participants who incorrectly responded to those items were directly excluded and not allowed to finish the survey. Another n = 24 respondents were excluded due to rapid response behavior (faster than 1/3 of the overall median). Additionally, we controlled for careless response behavior by excluding participants who chose the same option in more than 90% of questions (n = 20). From the remaining (n = 528) participants, n = 80 participants stated to never buy raspberries and were therefore excluded from the sample. Finally, n = 448 participants were included in the final sample. Table1 provides a detailed overview of the sociodemographic sample characteristics. 3.2 | Choice Experiment Method Choice experiments are based on two influential theories of consumer behavior, namely, Lancasters' theory of consumer behavior (Lancaster1966) and random utility theory (McFadden1986). The former assumes that a consumer obtains utility from a product's different characteristics and not just the product itself (Lancaster1966). The latter assumes that an individual, when provided with different alternatives, would choose the product that provides the highest utility, and that consumers' utility has also a random component to account for unobserved influences (e.g., subjective taste and psychological factors) (McFadden1986; Chinedu etal.2018). Choice experiments have been widely applied to elicit consumer preferences for sustainable food product attributes (e.g., Fuller and Grebitus2023; Sonntag etal.2023). As such, choice experiments provide participants with different choice sets, each consisting of different products with different product attributes. Participants are then asked to choose their preferred product as they would in a supermarket. Including a nobuy alternative, and thus giving participants the option to choose no product at all, is beneficial because it reduces biases (Dhar and Simonson 2003). First, a random parameter logit model was applied to assess an initial overview of the overall preference structure. Second, latent class conditional logit modeling was applied to identify different consumer segments. The approach allows for the integration of a discrete representation of unobserved heterogeneity among individuals (Yoo2020). The choice experiment included six different attributes describing potential product characteristics from raspberries cultivated conventionally and in agrivoltaic systems. Additionally, the product price was included. Table2 provides an overview of the attributes and their corresponding levels considered in the experiment. Because the production of green electricity is a crucial benefit of agrivoltaic systems, the attribute “production of green electricity” was included with three different levels: “with production of green electricity,” “without production of green electricity,” and “no information” regarding energy production. Additionally, as already described earlier, food production in agrivoltaic systems offer benefits in terms of increased water use efficiency, reduced usage of plastic, and pesticides. Those attributes were therefore included in the choice attributes, each with three varying levels (i.e., reduced usage, usual usage, and no information). Given prior research indicating that the origin of a product TABLE 1 | Sample description. Total sample (%) n = 448 German populationa (%) Gender Male 50.4 48.9 Female 49.6 51.1 Age 16–29 years 17.6 18.1 30–39 years 16.5 15.2 40–49 years 15.8 14.5 50–59 years 17.2 18.9 60+ years 32.8 33.5 Education Primary school 33.9 34.3 Secondary or vocational education 31.7 30.8 Higher education 34.4 34.9 aAccording to the German Federal Statistical Office, 2021. 2091 is crucial for consumers (Segovia and Palma2016) and in accordance with EU regulations (Food Information Regulation No 1169/2011), which mandate that food packaging must disclose the country of origin or place of provenance, the attribute of origin—including the following levels: produced regionally, produced in Germany, and no information—was included into the experiment. Despite the legal requirement to provide information in the origin, we decided to include the “no information” attribute level for two key reasons: (1) It serves as a statistical baseline for comparing the impact of different attribute levels. (2) In practice, origin information is often placed on the back of packaging, making the absence of origin information on the front a familiar scenario for consumers. Additionally, the production method (i.e., organic and no information) was included. “No information” (i.e., participants did not receive any information) was the base level of all attributes, except for price. The price attribute ranged from €0.99/125 g to €2.99/125 g. The levels were selected according to average prices of raspberries in German supermarkets in April and May 2023. The Defficient design was generated using Stata 17 with the dcreate command written by Arne Risa Hole, University of Sheffield (Hole2017). Based on the Defficiency criterion, the choice design with 11 choice sets, each consisting of three varying alternatives and a nobuy option, was selected. The nobuy option was included to reduce bias (Dhar and Simonson2003) and to depict the most realistic market conditions (Risius and Hamm 2017). During the experiment, each participant was asked to choose from the set of three varying alternatives and the nobuy option, as if they were purchasing the product in a supermarket. The alternatives in each choice scenario, as well as the choice sets, were randomly presented, while the nobuy option was always presented last. Figure1 depicts an example of a choice set presented to participants (please see AppendixB for the English translation). Subsequent data analysis was performed with Stata 17. First, a random parameter logit (RPL) model was applied to assess the overall preference structure. RPL modeling aligns with random utility theory and builds on the same choiceprobability model as the conditional logit. However, RPL accounts for preference heterogeneity by allowing preference weights to vary randomly across the sample. Unlike other methods, such as hierarchical Bayes estimation, which estimate individualspecific preferences, RPL models these preferences as deviations from the population mean rather than estimating them for each individual (Hauber etal.2016). Given that RPL has been previously TABLE 2 | Attributes and levels used in the discrete choice experiment. Attribute Attribute level Production of green electricity No information, with production of green electricity, without production of green electricity Plastic usage No information, with reduced plastic usage, with usual plastic usage for hail protection nets Pesticide usage No information, with reduced pesticide usage, with usual pesticide usage Water usage No information, with reduced water usage, with usual water usage Production method No information, organic farming Origin No information, produced regionally, produced in Germany, imported from Spain Price €0.99/125 g, €1.99/125 g, €2.99/125 g Note: “No information” served as the base level for all attributes except for price. FIGURE 1 | Example of a choice set presented to participants. Business Strategy and the Environment, 2025 2092 applied in the analysis of discrete choice experiments investigating consumer preferences for sustainable food (Risius and Hamm2017; Sonntag etal.2023) and the focus of this study was not on individuallevel preference estimates, RPL was considered a suitable approach. Effect coding was used for all attributes except for price, which was coded as a categorical variable, to avoid confounding with the nobuy option (Bech and GyrdHansen2005). As such, the attributes were coded with a value of 1 when applicable, a value of −1 for the base level (no information) and zero otherwise. Price was modeled as a random parameter. Subsequently, willingnesstopay measures were calculated to allow for relative comparisons of attribute levels among attributes. The calculation of confidence intervals of WTP values followed Krinsky and Robb(1986). Second, a latent class conditional logit model was applied to account for individuals' preference heterogeneity (McFadden1986; Yoo2020) and the identification of distinct segments that differ according to their preferences. Again, effect coding was used for all attributes except for price. Due to the nonlinear decrease in the price–utility function in most segments, the price attribute was treated as categorical (Risius, Hamm, and Janssen2019) and effect coded, too. To allow for relative comparisons among attributes, relative attribute importance was calculated. This was done by dividing the range of utility values for a specific attribute by the total sum of the ranges of utility values across all attributes (Malhotra2009). 4 | Results The results of the RPL model show the overall preference structure (see Table3). All included attribute levels, except for organic, had a significant effect on consumers' purchasing decisions. Compared with “no information,” “with production of green electricity” had a positive effect on consumers' choice, while a product labeled “without production of green electricity” negatively impacted consumers' choices. The information “reduced usage of plastic/pesticide/water” positively affected consumers' choices, while the “usual usage of plastic/pesticide/water” decreased consumers' marginal utility compared with “no information.” Two attribute levels associated with the product's origin, namely “produced in Germany” and “regionally produced,” had a positive influence on consumers' purchasing decisions, while marginal utility decreased for a product labeled “imported from Spain.” Price had a negative coefficient, indicating that consumers' marginal utility decreased when the product price increased. WTP values allow for comparison among attributes and show that consumers valued the product level “with production of green electricity” the most. Specifically, for €0.27/125 g, consumers showed the highest WTP for raspberries labeled “with production of green electricity,” followed by €0.26/125 g for “regionally produced,” and €0.25/125 g for “produced in Germany.” Further, WTP measures showed that consumers were willing to pay a price premium of €0.19/125 g for raspberries labeled “with reduced water usage,” €0.19/125 g for “with reduced plastic usage,” and €0.15/125 g for “with reduced pesticide usage.” As the coefficient of the remaining attribute levels was negative, consequently, WTP was also negative, indicating that consumers were not willing to pay a price premium for the described change in production. Subsequently, latent class conditional modeling was applied. Based on the model fit criteria (see Table4), four distinct segments within the sample were identified. Relative importance of the product attributes is depicted in Table5. Results of the latent class conditional model are shown in Table6. Segment 1 is the largest group and accounts for 32.4% of the sample. For segment 1, price was most important attribute (29.97%), followed by pesticide usage (19.55%). With 9.85%, the production of green electricity was still important in the purchasing decision, but to a lesser extent. Latent class conditional logit analysis revealed that all attribute levels included in the experiment, except for the lower price level (€1.99/125 g), had a highly significant impact on consumers' buying decision. Product levels “with production of green electricity,” “with reduced plastic usage,” “with reduced pesticide usage,” and “with reduced water usage” had a significant positive impact on consumers' purchasing decision compared with each attribute's base level “no information.” Similarly, “organic,” “produced regionally,” and “produced in Germany” had a positive significant effect on consumers' purchase decision. All other coefficients had a significant negative effect on consumers' choices, indicating that raspberries labeled with those attribute levels led to decreased marginal utility. Accordingly, segment 1 was named “Priceconscious pesticide skeptics, less interested electricity production.” Segment 2 accounts for 22.4% of the sample. Relative product importance showed that for segment 2, with 24.39%, production of green electricity was the most important product attribute, followed by origin (21.78%). Price only accounted for a relative importance of 10.09% in the purchasing decision. Thus, segment 2 is the least price conscious compared with the other segments. The attribute levels “with production of green electricity,” “with reduces water usage,” “organic,” “regionally produced,” “produced in Germany,” and “€1.99/125 g” had a significant positive coefficient, indicating that consumers obtained utility from those characteristics. “Reduced pesticide usage,” “reduced plastic usage,” and “usual plastic usage” had no significant impact on participants' choices. The same applied to the attribute level “€2.99/125 g,” which also had no significant effect on consumers' choices, indicating that consumers associated with this segment were less priceconscious. “Without production of green electricity,” “usual pesticide usage,” “usual water usage,” and “imported from Spain” had a negative significant coefficient, indicating that marginal utility decreased for raspberries labeled with those attributes. Accordingly, segment 2 was named “Less pricesensitive proponents of green electricity production.” Segment 3 is the smallest and accounts for 19.1% of the sample. Regarding the relative importance of attributes, individuals associated with segment 3 highly valued origin (38.05%), followed by price (14.01%) and production of green electricity (12.45%). The attribute levels “with production of green electricity,” “reduced plastic usage,” “usual water usage,” “organic,” “produced regionally,” “produced in Germany,” and “€1.99/125 g” had a significant positive coefficient, indicating that labeling raspberries with those attributes increased the utility for consumers and thus their likelihood to buy the product. The attribute level 2093 TABLE 3 | Results of the random parameter logit model. Attribute Level Coefficient Standard error WTP (€/125 g) 95% confidence intervalb Production of green electricityaWith production of green electricity 0.376*** 0.040 0.27 0.207; 0.325 Without production of green electricity −0.271*** 0.047 −0.19 −0.257; −0.128 Plastic usageaWith reduced plastic usage 0.259*** 0.044 0.19 0.123; 0.249 Usual plastic usage for hail protection nets −0.114** 0.041 −0.08 −0.137; −0.025 Pesticide usageaWith reduced pesticide usage 0.213*** 0.059 0.15 0.073; 0.239 Usual pesticide usage −0.328*** 0.048 −0.23 −0.310; −0.163 Water usageaWith reduced water usage 0.272*** 0.033 0.19 0.147; 0.245 Usual water usage −0.164*** 0.038 −0.12 −0.170; −0.060 Production methodaOrganic production 0.034 n.s. 0.036 n.s. OriginaRegionally produced 0.359*** 0.055 0.26 0.174; 0.340 Produced in Germany 0.354*** 0.042 0.25 0.193; 0.310 Imported from Spain −0.735*** 0.049 −0.52 −0.600; −0.453 Price −1.402** 0.062 Note: Loglikelihood: −4616.8901, Wald χ2 = 1095.39, prob > χ2 = 0.0000. aReference category: “no information.” b95% confidence interval following Krinsky and Robb(1986). ***p ≤ 0.001, **p ≤ 0.01, and *p ≤ 0.05. Business Strategy and the Environment, 2025 2094 “usual pesticide usage” had no significant effect on consumers' purchasing decisions. The remaining attribute levels had a negative coefficient, indicating decreasing marginal utility of products labeled with those characteristics. Based on this, individuals associated with this segment were labeled “import opponents, open to green electricity production.” The remaining segment 4 accounts for 25.3% of the sample. For consumers of segment 4, price was the most important attribute (44.31%), followed by origin (21.12%). “Without production of green electricity,” “usual water usage,” and “produced in Germany” had a positive significant coefficient. “€2.99/125 g” and “imported from Spain” decreased consumers' likelihood of purchasing the product. The remaining levels had no significant effect on consumers' choices. Accordingly, individuals associated with this segment were named “Priceconscious consumers, not interested in green electricity.” 5 | Discussion This study investigated consumers' preferences for food produced in agrivoltaic systems, taking the example of raspberries. Overall, consumers' WTP a price premium was highest for raspberries labeled with additional information regarding the simultaneous production of green electricity in the production process. This provides evidence for consumers valuing frontofpacking information of one of the major sustainability benefits of agrivoltaic systems—the production of green electricity. The results point in the same direction as previous literature highlighting general public acceptance of solar energy production (Vuichard, Stauch, and Wüstenhagen2021). Our results, thus, suggest that the marketing of food produced in agrivoltaic systems could benefit from information regarding the simultaneous production of green electricity and food. In contrast, explicitly highlighting that the production process did not include the production of green electricity yielded negative WTP values indicating a decreased likelihood for the purchase of food labeled with this respective attribute. Origin and the associated product characteristics “produced regionally” and “produced in Germany” were also highly valued by consumers. Geographic origin has already been shown to be a highly valued product characteristic for fruits and vegetables purchases (Segovia and Palma2016). Additionally, due to EU requirements, information regarding the country of origin is legally required. Thus, consumers are presumably used to be confronted with this product attribute when shopping for fruits and vegetables. In contrast to domestic and regional production that has already been shown to be appealing to consumers (Feldmann and Hamm 2015), imported raspberries from Spain yielded a highly negative WTP value. This result might be based on consumers' perception of domestic production being a cue for high product quality (Witzling and Shaw2019; Thøgersen2023). Regarding the wellknown strong consumer preference for the origin of food products, and the presumably high familiarity with information on the origin, it is remarkable that information on the production of green electricity yielded an even higher WTP value. These results might be based on the recently increasing sociopolitical interest in renewable energy TABLE 4 | Model fit criteria of the latent class analysis (LCA). Loglikelihoodmodel df AIC BIC Sample size ≤ 10% Segment 2 −4684.81 29 9427.624 9656.405 0 Segment 3 −4574.55 44 9237.104 9584.22 0 Segment 4 −4481.38 59 9080.761 9546.211 0 Segment 5 −4429.38 74 9006.756 9590.54 1 Segment 6 −4397.309 89 8972.618 9674.737 1 Segment 7 −4358.57 104 8925.139 9745.594 2 Note: The bestfitting model is presented in bold. Pseudo R2: 0.13. Abbreviations: AIC, Akaike information criteria; BIC, Bayesian information criteria. TABLE 5 | Attribute importance of the attributes included in the choice experiment and the respective segments. Attributes Segment 1 Segment 2 Segment 3 Segment 4 Average attribute importance Production of green electricity 9.85% 24.39% 12.45% 12.09% Plastic usage 12.41% 9.18% 8.03% 1.30% Pesticide usage 19.55% 14.85% 11.36% 5.73% Water usage 9.26% 12.60% 11.74% 12.03% Production method 6.84% 7.11% 4.36% 3.42% Origin 14.13% 21.78% 38.05% 21.12% Price 27.97% 10.09% 14.01% 44.31% 2095 Appendix B Example of a choice set presented to participants (originally in German, translated to English) Business Strategy and the Environment, 2025 2102