Addressing agent specific extreme price risk in the presence of heterogeneous data sources: A food safety perspective

Since the recent food crisis, there has been prominent literature relating to the study of commodity prices volatility and other factors impacting on food security and resilience measures, especially in an African context. Numerous studies have shown that aside production factors, commodities prices instability and the associated price anticipation difficulties constitute a major point of preoccupation when it comes to food insecurity assessment. The multiplicity of historical price data represents one of such difficulties, as it creates uncertainty when making investment decisions. This paper assumes that risk factors based on these historical prices are observed under uncertainty with no unique observable efficient price, and are therefore modeled as intervalvalued random variables. Under such conditions, a linear portfolio of agricultural commodities characterized by a vector of their interval-valued risk factors is considered, just as extreme risk models (VaR and ES) under uncertainty as introduced by Mbairadjim Moussa et al. (2014) are applied for risk computation and portfolio selection. Moreover, our paper proposes specific portfolio selection models under uncertainty, which discriminate between producers, consumers and political decision-makers concerns, in establishing an optimally allocated portfolio. Finally some empirical examples carried out on real data set from Niger, show the effectiveness of the proposed methods in a practical decision making context.