Program officers are faced with difficult modeling choices when monitoring the state of diverse yet related dimensions of household food security. We compare results of single- and multi-output frameworks, applying both econometric and machine learning models, to jointly predict three interrelated outcomes of food security: undernourishment (access), dietary diversity (utilization), and food consumption stability (stability). Using the 2013–2022 Kyrgyz Integrated Household Survey, we evaluate generalized linear models (GLM), a multi-output generalized structural equation model (GSEM), and single- and multi-output artificial neural networks (ANNs) under both cross-sectional and temporal validation settings. The results show that accounting for non-linearity of relationships in our specifications yields greater predictive gains than modeling outcomes jointly. Among the models evaluated, the simplest model (single outcome GLM) often performs as well as others, and even better in some specifications, while also providing interpretable parameter estimates. However, accounting for non-linearity with our single-outcome ANN often achieves superior performance in low-outcome groups, which is especially relevant for policies aimed at targeting vulnerable households. Adding complexity through joint prediction of the related outcomes improves precision in some cases, but reduces recall, increasing exclusion risk. This trade-off makes such approach less desirable from both equity and implementation perspectives. The findings emphasize that complex patterns tie household socio-economic and demographic characteristics with food security dimensions. These results can guide program officers navigating the trade-offs between prediction performance, interpretability, and implementation; showing that linear models can deliver robust predictions, but specific non-linear, data-driven models might allow better targeting of vulnerable households.
Global food governance is increasingly described as being in crisis, as multilateral cooperation weakens amid rising nationalist and populist politics. However, less attention has been paid to how dominant multilateral modes of governance within the broader field of global governance shape the political conditions in which such tensions emerge. This paper examines how contemporary global food governance is organized primarily around coordination as a governing grammar—a mode of governing oriented toward aligning action across fragmented authority. Drawing on an interpretive analysis of a purposively selected body of authoritative global food governance documents produced between 2012 and 2025, the paper examines how coordination structures action, responsibility, evaluation, and contestation across multilateral institutions. The analysis identifies recurring patterns through which coordination diffuses responsibility, manages failure procedurally, and reframes distributive trade-offs as technical problems, enabling sustained engagement and institutional durability in the absence of centralized authority. These features constrain multilateral institutions’ capacity to assume responsibility, adjudicate conflicts, or respond politically to uneven consequences. Grievances associated with food system transformation often struggle to gain traction within multilateral arenas; this may contribute to their articulation in other political contexts, including national and subnational arenas where responsibility is more readily attributed. Rather than treating such mobilizations as external threats to multilateralism, these dynamics are interpreted as being consistent with the institutional features of coordination-based governance. The paper concludes by outlining implications for global food governance, emphasizing the need to complement coordination with institutional arrangements that clarify responsibility, link measurement to material response, and enable political contestation over distributive trade-offs.
Substantive agri-environmental policy change has not materialized within EU agri-food governance despite emerging scientific consensus on the need for food system transformation. Agri-environmental legislative proposals often face strong resistance, resulting in their weakening or reversal. Although the challenges to adopting agri-environmental legislations are clear, existing analytical tools fall short in explaining why legislative rollbacks occur. This study addresses this gap by drawing on Rational Choice Theory to examine how individual-level motivations linked to the agri-food sector influence members of the European Parliament (MEP) voting behaviour on agri-environmental legislation. Roll-call vote data and a binomial generalized linear mixed model are used to test office-seeking, policy-seeking and vote-seeking motivations. Our analysis shows that MEPs with private financial interests in the agri-food sector were more likely to vote against environmental protection across the studied agri-environmental legislations, while MEPs with stronger environmental ideological affinity were more likely to support such measures. Institutional variables, mainly political group affiliation and COMAGRI leadership of legislative files, are also significant determinants of voting behaviour. These findings suggest that individual-level motivations and institutional structures help explain the lack of consistent legislative support for agri-environmental legislations between February 2022 and July 2024. Our findings raise normative questions about the need for stronger transparency and conflict-of-interest measures in EU agricultural governance.
This paper investigates how the agricultural cycle influences protest behavior in Kenya from 2000 to 2019. Using satellite data to identify the crop cycle and local harvest periods at the lowest available administrative level (wards) and geocoded protest records, we find that protest activity does not decline during the harvest period but increases with a temporal lag. A dynamic difference-in-differences specification shows that protest activity gradually rises in the two to three months following harvest, peaking around the onset of the subsequent agricultural cycle. These patterns are consistent with theories of opportunity costs and seasonal scarcity. While the opportunity costs of protest participation are highest during harvest periods, they decline thereafter and coincide with the lean season, when household food stocks become exhausted. These findings are robust across empirical specifications and contribute to a deeper understanding of the temporal dynamics of unrest in agrarian settings. They also highlight the importance of timing in the design of rural development and conflict prevention policies.
Spatial price differences in agricultural markets, after accounting for transport costs, are often interpreted as evidence of market inefficiency. However, price differences may also reflect product quality variation, especially where formal grading systems are absent. This study uses a novel transaction-level dataset on paddy rice sales from rural Madagascar in 2022–2023, including laboratory-assessed grain quality indicators. By employing hedonic regressions, we construct a composite quality index to quantify the extent to which observed price variation across locations can be explained by quality differences. We find that varietal identity commands substantial price premiums, but measured physical and post-harvest quality attributes beyond variety explain only a small share of the observed price dispersion. Controlling for the quality attributes has little effect on conventional measures of market integration. This result suggests that buyers face difficulties in accurately assessing physical quality at the point of sale. In such contexts, buyers may rely on location-specific reputation as a substitute. Consistent with this, transactions originating from villages with higher average quality are associated with higher prices even after controlling for observable quality attributes. We interpret this as descriptive evidence consistent with reputation or relationship-based sourcing in settings with limited quality observability.
Increasing attention on “ultra-processed” foods (UPFs) and their health effects has sparked policy debate. News media influences public perceptions yet understanding of UPF framing and industry influence remains limited. This study analysed UPF framing in UK news media and trade press, exploring interests of quoted individuals and organisations.We systematically searched UK media for UPF articles in 2022–23, identifying 188 articles for mixed-methods content analysis. Articles were examined for their position towards classifying food by processing-level (accepted/contested/mixed). Framing Theory guided identification of UPF frames describing problem definitions, causes, and solutions. Actors’ interests were classified using academic journal guidelines.Half the articles (52%) accepted the UPF concept, 19% contested it, and 29% were mixed. Five frames were identified. The ‘Social Justice’, ‘Lifestyle Drift’, and ‘Science Report’ frames accepted the UPF concept, with the ‘Social Justice’ frame positioning UPF consumption as a structural public health problem requiring policy action such as taxes or front-of-package labelling. The ‘UPF Critique’ and ‘Market Justice’ frames contested the scientific basis of the UPF concept, proposing nutrient-based approaches instead. Among 116 named actors, 33% had UPF industry interests, with academics comprising 45% and industry spokespeople 11%. Actors with UPF industry interests were more commonly cited in contesting frames (e.g. ‘Market Justice’: 73%) than accepting frames (e.g. ‘Social Justice’: 10%).These findings highlight that industry influence on UPF media discourse is often obscured behind academic voices. Greater transparency around actor interests is urgently needed to ensure emerging UPF policies are shaped by independent evidence.
Demand-side food policies are increasingly recognized as important tools for aligning consumption patterns with climate change mitigation targets, and carbon food labels have attracted growing attention as a potential policy instrument. However, evidence on their effectiveness remains mixed, and causal evaluation results from assessing the effects of carbon food labels in real-world supermarket settings is scarce. This study presents evidence from pre-registered randomized field-embedded survey experiment with behavioral outcomes, combining a representative sample of 2,372 Swiss residents with transaction-level purchasing data covering more than five million daily food purchase decisions made by 1,601 consumers over several years. In collaboration with one of Switzerland’s largest grocery retailers, we evaluate the causal impact of an information treatment designed to increase awareness of one of the world’s first carbon food labels implemented in grocery stores on both stated preferences and revealed purchasing behavior. The results show that the information treatment significantly increases label awareness and leads to small but positive effects on attitudes, perceived behavioral control, and intentions to consider the climate compatibility of food purchases. When tracing these effects into revealed purchasing behavior, the treatment also affects actual food purchases, but effects are modest and heterogeneous across food categories. Rather than inducing broad shifts away from meat consumption, the information treatment primarily leads to a higher overall share of labeled products purchased, increased purchases of labeled convenience products, and a partial reallocation within meat consumption (e.g., lower beef and higher pork purchases). We find no evidence of substantial reductions in overall meat purchases or large cross-category substitutions toward plant-based foods. While heavy meat eaters do not react negatively to the treatment, men and more right-wing individuals show a modest behavioral backlash. Overall, the results suggest that carbon food labels capacity to substantially alter habitual food consumption patterns in stimulus-intensive retail environments is limited and should be thus complemented by additional demand-side measures.
Surface ozone pollution has emerged as a major threat to agricultural productivity, yet evidence on its impacts on high-value cash crops and farmers’ adaptive responses remains limited. Combining a household-level panel from the Chinese National Fixed-Point Survey with high-resolution pollution and weather data from 2003 to 2013, we find that rising ozone concentrations significantly reduced farm income of fruit growers, primarily driven by ozone-induced reductions in fruit yields, total production, and labor productivity. Sugar crops also experienced significant yield losses, while other cash crops exhibited negligible responses. Due to limited adaptive capacity, fruit growers responded by abandoning ozone-damaged plots, reducing input use, and reallocating labor towards off-farm employment, resulting in persistent income losses with little evidence of longer-run adjustment. In contrast, sugar producers offset yield losses by expanding cultivated area and increasing material inputs, maintaining output and income. These findings highlight the critical importance of ozone-control policies and targeted support to strengthen farmers’ adaptive capacity, particularly for fruit growers who are most vulnerable to rising ozone exposure.
There is a need to assess the benefits of government policy to improve the welfare of food animals. We present a holistic, animal-centred welfare assessment method using an expert panel of independent animal welfare scientists linked to a survey of food shoppers’ willingness to pay (WTP) to improve welfare levels, using a 0–100 animal welfare scale. We apply this method to six food animal types in the UK and 15 different topical policy scenarios. We find that several current production systems have welfare scores below 50 but that some of the policy initiatives analysed could increase welfare scores substantially if implemented. We provide valuation tables of people’s WTP to increase welfare scores for each of the six animal types. Marginal WTP was found to diminish as welfare scores increased. WTP values are fully transferable across any policies affecting the animals involved and are not tied to specific production practices or contexts. The UK government has been using the method to evaluate a range of policy options to improve welfare. The method can be applied in other countries. The method enables government policy makers to better make the case for policies to improve the welfare of food animals and its application should lead to an increased likelihood of such policies being implemented. The outputs can also be used to inform consumers regarding the animal welfare provenance of the food they purchase.
The Supplemental Nutrition Assistance Program (SNAP) is a safety net for 41 million low-income people in the United States. Although 80% of all eligible individuals participate, SNAP eligible agricultural workers likely participate in SNAP at about half that rate. Agricultural workers may be less attached to the safety net because seasonal employment makes administrative burdens more salient and can induce variation in eligibility. In this paper we examine how seasonality affects farmworker attachment to SNAP. We measure households’ attachment to SNAP with churn rates. Churn occurs when a household exits SNAP and subsequently re-enters the program within 3 months. With household-level administrative data from Fresno County as a case study, we provide descriptive evidence that farmworkers are more likely to churn than other SNAP participants. SNAP churn among farmworkers increases significantly when agricultural employment falls after seasonal peaks, particularly for crop production and support activities. We estimate a seasonal pattern in churn for farmworkers that does not exist for non-farmworkers, indicating that the patterns we observe cannot be explained by baseline seasonality alone. Understanding the relationship between churn and the seasonality of agricultural employment has the potential to improve access to the safety net for a vulnerable population, as well as reduce SNAP related costs for both households and the state.
The concepts of local food and food heritage have attracted increasing attention in academic and policy debates. Yet, both terms remain ambiguous, contested, and open to multiple interpretations. Within the EU, these notions carry particular importance: food has long been central to EU identity and governance, and current strategies for food system transformation and rural development frequently invoke related terms. While prior work has examined some aspects of the EU food policy framing, no systematic text analysis has specifically examined how the two concepts are being constructed and mobilised within the EUThis study aims to fulfil this gap by exploring how local food and food heritage are conceptualised and operationalised in EU policy documents. It adopts a qualitative research design grounded in a constructionist paradigm. Twelve EU policy documents published between 2011 and 2025 were selected from a broader corpus and analysed using Qualitative Document Analysis (QDA). The codebook was derived from the literature and refined through pilot testing. The findings reveal a shift in the treatment of local food: while earlier and less formal documents refer to it explicitly, more recent, strategic, and legally binding texts rely on indirect framings aligned with food producers’ economic priorities. Food heritage is more consistently present, primarily linked to traditional knowledge and cultural identity. The two concepts intersect most clearly in geographical indications system, and in discourses of local branding and rural development. The analysis demonstrates that EU policies frame local food in line with broader economic objectives, and that food heritage is recognised as part of the EU’s cultural identity and worthy of protection as such. We also conclude that a lack of coherent vision of what is the place of both concepts within the EU food governance risks them being primarily determined by producer competitiveness, thereby causing the relational and cultural dimensions that originally justified these concepts to become marginalised in policy.
Technology adoption is an important farm management decision and is frequently made using information that is unavailable to the researcher. Aside from their own experiences, producers often rely on extension services to inform their adoption decisions. Here, we propose a novel instrumental variable (IV) approach to identify the effect that a new technology, genetically modified (GM) corn, has on farm yield using instruments constructed from extension information. The proposed approach allows us to explore the degree to which endogeneity bias is present in the GM adoption effect on farm productivity. Candidate instruments are constructed from variety trial data to resemble information acquired by producers from university extension variety trial reports and are spatially merged with on-farm production data. We find that failure to account for the endogeneity of the adoption decision does indeed result in an upward bias in the estimated productivity gains. After correcting for this bias through our IV approach, GM corn results in large on-farm yield gains of 19.6 bushels per acre. We demonstrate that variety trial data can be used to estimate the on-farm impact of new technology, and this approach is general enough to be applied to additional contexts (i.e., technologies, crops, and/or locations).
Integrated staple food markets are essential for food security in sub-Saharan Africa, where national production is often sufficient but spatial disparities limit access for households in deficit regions. This study examines price connectedness in maize and yam markets across ten major regions in Ghana and assesses how transmission patterns respond to domestic and global shocks. Using monthly wholesale prices from January 2006 to December 2024, we employ a dynamic Quantile Vector Autoregression framework to estimate spillovers across the full distribution of price movements. The results show strong integration in maize markets, with total dynamic connectedness averaging 92 %. Major assembly and transit markets, including Kumasi, Techiman and Bolgatanga, consistently transmit price shocks, while coastal consumption markets such as Accra and Takoradi receive them. Yam markets exhibit weaker and more variable integration, with an average connectedness of 87 % and a more dispersed structure of price leadership across regions. For both commodities, connectedness strengthened during the 2008 food price hikes, whereas connectedness during the COVID-19 period was uneven and volatile. In contrast, political regime changes in 2009 and 2017 do not generate substantial shifts in market connectedness, highlighting the dominant role of structural supply chains and trader coordination over short-term political events. The findings identify regional markets where households face greater exposure to price volatility and highlight the importance of spatially targeted investments in transport, storage and market coordination alongside measures to manage shock transmission and reduce vulnerability to price volatility and unequal food access.
Mitigating antimicrobial resistance (AMR) in veterinary medicine is critical to achieving the UN 2030 Sustainable Development Goals (SDGs). Most countries have adopted tailored antimicrobial reduction policies in animal production, yet rigorous cost-benefit evaluations remain scarce. We leverage China's large-scale 2021 veterinary antimicrobial use reduction (AUR) policies as a quasi-natural experiment, employing a Fuzzy Regression Discontinuity (FRD) design to estimate its causal effects on farmers' net income and disentangle the transmission mechanisms using three-wave panel data from hog farmers (2022-2024). The empirical results demonstrate that the AUR policies exhibited a dynamic pattern of initial income suppression followed by income enhancement, with the estimated effect turning positive in the second year of implementation (2023). Based on the baseline FRD estimates, the average effect over the full panel period is an increase of 186.504-198.708 CNY per head, representing 34.09 %-36.32 % of the three-year average net income. This effect is primarily driven by changes in cost-benefit indicators such as slaughter weight, market premiums, biosecurity expenditures, antimicrobial input costs, and antimicrobial alternative costs. We elucidate the underlying mechanisms through which AUR policy interventions operate, specifically via antimicrobial use regulation, compulsory immunization, antimicrobialreduction compliance certification, and antimicrobial-reduction technical training. During the AUR's initial rollout (2022), antimicrobial use regulation was the primary income suppressor, whereas market premiums became the dominant income driver in the maturation phase (2023-2024). Moreover, heterogeneity analysis shows that the income-enhancing effect of the AUR policies was more pronounced among farmers who possessed high digital literacy and had organizational support, compared to those without these advantages. Based on China's AUR experience, we propose five policy insights that are particularly relevant for developing-country contexts: optimized policy mixes, tiered biosecurity systems, scalable antimicrobial alternatives, targeted market incentives, and inclusive global governance frameworks.
Fish consumers are often challenged by tradeoffs between nutritional benefits and contaminant risks, which increase due to environmental pollution. Health campaigns and labeling initiatives can guide decision-making by providing information both on contaminant risk and nutritional value of a product, but it is not well understood how consumers react to such complex dual labels. We use data from a stated choice experiment in Kenya’s Lake Victoria region to study how consumers respond to dual labels on fish products, and how their responses to each label interact. We focus on the tradeoff between polyunsaturated fatty acids and contamination with microcystins, a class of toxins produced by harmful algae blooms that can accumulate in fish tissue. Our findings suggest that, faced with a dual information policy, consumers react rationally to dual health attribute labeling, and that nutrient labels and contaminant warnings can function concurrently, indeed even be mutually reinforcing, but pose a risk of inadvertently concentrating unhealthful consumption in less responsive subpopulations.
