Modelling agricultural 'performance': Smallholder weed management in Southern Malawi

The concept of agricultural 'performance' is explored through a case study of smallholder weed management in Southern Malawi. Weeding performance was analysed at the aggregate, farm and field level using survey data for three crop seasons. Farmer decision-making for weeding was modelled using a hierarchical decision-tree. Rainfall, termites, fertilizer use and labour supply were identified as the main determinants of timely weeding. Farmers had developed specific weeding techniques to cope with these contingencies. Understanding weeding as a farmer 'performance' highlights the importance of diagnostic research to codify farmer knowledge and exposes the limitations of prescriptive research recommendations based on a fixed, predetermined plan. We argue that in the short- to medium-term, the most effective strategy for improving the timeliness of weeding is to increase the effective demand for hired labour. This will require crop management interventions that improve the ability of the farming system to generate cash income for reinvestment.