A Federated Agricultural Market Intelligence Framework for Distributed Price Discovery and Inflation Monitoring

Abstract

Agricultural markets often experience considerable differences between the prices received by farmers and the prices paid by consumers. These disparities arise from fragmented supply chains, limited market transparency, transportation costs and the involvement of multiple intermediaries. As a result, farmers may receive low returns for their produce while consumers face significantly higher retail prices. This paper presents a privacy-preserving framework based on Federated Learning (FL) to support transparent and fair agricultural price discovery across geographically distributed markets. Instead of sharing sensitive personal or farm-level information, farmers or field agents contribute local crop price data that remain on their devices. The federated model aggregates insights from decentralized data sources to estimate fair wholesale prices, transportation and storage expenses and expected retail prices. To improve reliability, the framework includes anomaly detection mechanisms that identify unusual or potentially manipulated price submissions, thereby boosting data integrity and trust. The proposed system can assist government agencies and market regulators in monitoring price fluctuations, detecting emerging inflationary trends and implementing timely interventions when necessary. By providing accurate and privacy-aware market intelligence, the framework empowers farmers to make informed decisions regarding storage, transportation and direct sales opportunities. The proposed architecture is scalable, secure and adaptable to diverse agricultural ecosystems, offering a practical approach to reduce information asymmetry and promoting fair pricing for both producers and consumers.