Forced labour risk is pervasive in the US land-based food supply

The transformation of food systems is crucial for achieving the UN Sustainable Development Goals (SDGs), particularly the elimination of forced labor (SDG 8.7). Previous research on labor conditions in food production has largely focused on specific subpopulations, regions, and commodities, lacking a comprehensive, systematic assessment across diverse food products. This study aims to fill this gap by providing a quantitative assessment of forced labor risk embedded within the US land-based food supply. The complexity of globalized supply chains and the illicit nature of forced labor pose significant challenges for detection and elimination. However, the evolving regulatory landscape, including international trade sanctions and human rights due diligence requirements, necessitates new approaches, data, and indicators to inform effective policy and business decisions. This research builds upon previous work that identified high forced labor risks in US fruit and vegetable production, expanding the scope to encompass a wider range of food products and supply chain stages. The objectives are threefold: (1) to expand the forced labor risk scoring method to include new data sources and processing stages; (2) to estimate the risk of forced labor embedded in the diverse foods consumed in the US; and (3) to identify forced labor risk hotspots within and across food categories.

Existing literature on social risk assessments and case studies of labor conditions in food production primarily focuses on specific subpopulations, regions, and commodities. There is a lack of systematic assessment of labor conditions against international standards across diverse, complex food products. Some research has used a country-level lens to analyze aspects of the social performance of food systems, but none has attempted to link social performance to particular food commodities at scale. This gap is significant for policymakers designing targeted interventions to achieve SDGs while remaining within planetary boundaries. The agriculture, fishing, and forestry sectors have some of the highest incidences of forced labor globally due to reliance on manual labor, often by migrant workers vulnerable to exploitation. However, existing analyses of sustainable food systems lack data on forced labor in supply chain stages beyond agricultural production, creating potential unintended consequences for policy and practice. While instances of forced labor are documented, the incidence in processing stages is not quantified, highlighting the need for better data and indicators to inform business and policy decision-making.

To compute forced labor risk, the researchers compiled origin data for the US land-based food supply (excluding seafood). They then qualitatively coded the forced labor risk in agricultural production and processing for each country-commodity combination using a three-tiered approach (Table 1), prioritizing the most granular data available. This tiered approach involved assessing risk at the commodity-country level (Step 1), sector-country level (Step 2), and country level (Step 3). Qualitative risk levels were translated into quantitative scores using medium risk hours equivalent (mrh-eq) conversion factors, following the Social Hotspots Database (SHDB) approach. The risk of forced labor was calculated as a function of characterized risk and worker hours, and data quality was assessed using a pedigree matrix approach. Data sources for known occurrences included government and NGO reports, investigative journalism, and assessments of hand-harvest risk. Government response data were obtained from the Trafficking in Persons Report. The final dataset included 212 food products and 1,312 product-country combinations. Processed products were mapped to estimated origin countries using FAO's Supply Utilization Accounts (SUA) and Detailed Trade Matrix. The total number of activity-country combinations scored for risk was 2,661. A hotspot analysis, adapting a grouping schema by Kim et al., was used to analyze the distribution of forced labor risk across the US food supply. The methodology also involved estimating labor intensity (worker hours per tonne) for each food product, supply chain stage, and country combination using price data and working hours per unit value from the SHDB. A data quality assessment framework, adapting the S-LCA pedigree matrix, was employed to assess the reliability, temporal, geographical, and technical aspects of the risk coding, working hours, and price data. The final data quality score for each observation was calculated by averaging the scores across these four indicators for each data source and then averaging the scores across all data sources.

The analysis revealed that 41% of product-country combinations were unprocessed, 48% had one processing stage, and the remainder had two or three stages. 18% of activity-country combinations were scored using commodity-country-specific data, 49% using sector-country data, and 33% using country data. Step 1 risk data (commodity-country specific) were available for 27% of combinations spanning 81 countries, with significant variation across countries. Much less Step 1 data were available for processing stages (4%). The top three product categories contributing to forced labor risk were meat, poultry, and eggs (28%); other products (23%); and processed fruits and vegetables (18%). Over half (62%) of the forced labor risk was attributable to domestic production or processing. The second and third highest contributing countries were China (13%, primarily due to apple juice concentrate) and Mexico (8%, mainly from unprocessed fruits and vegetables). While agricultural production contributed 85% of the total risk, processing's contribution varied substantially across products, with significant contributions for products like maize starch, frozen potatoes, beer, and shelled cashews. A hotspot analysis within food categories revealed that a small number of products contributed large shares of forced labor risk. For instance, among fruits, avocados, lemons and limes, and pineapples had disproportionately high risks; for vegetables, tomatoes and chillies and peppers; and for processed products, apple juice concentrate. Shelled cashews showed disproportionately high risk among pulses and nuts, while boneless beef showed disproportionately high risk among meat products. Cocoa powder/cake and refined sugar were also high-risk products. Data quality assessment showed that the average quality of data on working hours was medium.

The findings demonstrate that forced labor risk is prevalent across diverse food products and originates not only from imports but also from domestic production and processing in the US. This contradicts the common assumption that high-income countries' risk is primarily embedded in importing practices. The high domestic risk is attributable to the US's significant domestic production and processing, coupled with long-standing systemic risks in food-related labor. The study highlights the need to address both domestic and international dimensions of forced labor through a multifaceted approach, including harmonizing import controls with national and local regulations and developing human rights due diligence frameworks. The study also reveals potential overlaps and tensions between assessments of the environmental impacts, health outcomes, and social sustainability of US food consumption, suggesting the need for future research that integrates all four pillars of sustainability. The method used in this research represents an advance in social life cycle assessment (S-LCA) of food systems and labor risk assessment, offering a more dynamic approach compared to intermittent social audits. The incorporation of investigative journalism reports enhanced the dataset by filling gaps and revealing risks in overlooked sectors and geographies. The limitations associated with using media reports as data sources were partially addressed through a rigorous coding approach and data quality assessment.

This study identified widespread forced labor risk across diverse food products in the US food supply, emphasizing the need for collaborative, worker-centered approaches. Future research should explore novel data sources and expand the assessment to other supply chain stages, including animal feed production, transport, retailing, and waste management. While this study has limitations in data completeness and potential reporting bias, its contribution lies in utilizing a standardized methodology across diverse products and supply chain stages, providing a broader and more nuanced understanding of forced labor risks in the US food system than previously available.

The study acknowledges limitations in data availability, particularly regarding precise working hours data and the potential biases inherent in using investigative journalism reports as a data source. The exclusion of seafood due to data limitations represents another constraint. Furthermore, while the methodology addresses some data quality issues, inherent biases in reporting data and the illicit nature of forced labor may lead to underreporting of risk in certain sectors or geographies. The reliance on existing databases and the use of average values for some parameters may also introduce uncertainties into the estimations.