Behavioural Analysis in Management and Commercialisation of Agro-Food Waste among Urban Agro-Producer Households in Kenya

Abstract

In Kenya, most of the waste generated in the urban supply chains end up in dumpsites. Agro-food waste is no different although there are indications that a portion of it is utilised as input in urban agriculture. Despite this, there is limited empirical evidence on agro-food waste management and commercial utilisation behaviour among urban agro-producer households. The study (i) assessed the drivers of agro-food waste typologies generation, (ii) assessed the role of contextual factors in the choice of agro-food waste management practices, (iii) determined factors influencing agro-food waste commercial utilisation behaviour and (iv) evaluated the determinants of demand for agro-food waste among urban agro-producer households. The study sample constituted 456 agro-producer households and 48 key informant interviews drawn from Nairobi City County. Fractional response, multivariate probit, structural equation and linear almost ideal demand system models were used to assess waste generation, choice of management practices, commercial utilization behaviour and waste demand respectively. Findings indicated that the daily per capita agro-food waste generated was 11.42kg (food waste 0.67kg and agricultural waste 10.75kg). Household agricultural inputs budget showed a higher allocation towards conventional inputs (0.73) compared to agro-food waste (0.27). Furthermore, other households constituted the most preferred source (30.85%) of agro-food waste utilised whereas waste emanating from processors had the highest budget allocation of 26.60% in agro-producer households. Fractional Response model output indicated that age, number of enterprises, perceived behavioural control and attitude were among the most important drivers of agro-food waste generation. Knowledge variables (waste sorting and urban agriculture knowledge) largely had influence on agro-food waste management options and safety practices in Multivariate probit model. Structural equation model indicated that commercial utilisation intention, risk perceptions and perceived behavioural control had significant influence on the commercial utilisation behaviour which implied that intention largely transitioned to behaviour. The model could explain 79.1% of the commercial utilisation behaviour variance. Based on the Linear Approximation of Almost Ideal Demand System, expenditure elasticities for agro-food waste sourced from restaurants (0.6998), agro-markets (1.1988), processors (1.7481), own (0.8640) and other (0.8970) households were normal goods whereas dumpsite (-0.0769) waste was an inferior commodity. Although uncoordinated, the demand aspects implied that waste market was in existence and developing. Overall, participatory tailor-made education programs would likely enhance agro-food waste management and supply chain coordination as such streamlining the waste market in Kenya