Analysis of the farming behavior of Sri Lankan peasants producing multiple outputs

Typescript (photocopy).This study investigates resource allocation in Sri Lankan peasant agriculture and focuses on dry zone multi-output/multi-input farming systems (Vavuniya district). Investigation is carried out at both farm household and district levels using alternative models. A household production model based on utility maximization is developed at the farm household level while a profit maximization model is used to examine behavior at the district level. The production activity of the farm household is expressed with a primal system. A dual specification of production is developed at the district level. Both primal and dual specifications of production activity are defined using a quadratic functional form. The consumption activity of the household is developed with a direct demand specification using the Rotterdam model. System (SUR) and single equation (OLS) estimators are used to accomplish the estimation of the models. Due to simultaneity in some production decisions, estimates are also obtained using instrumental variable procedures. Output categories include paddy, chili, vegetables, and pulse. Variable input categories include seed, chemicals, fertilizer, farm power and farm labor while land is treated as an allocatable fixed input. Based on proportion of the farm household production parameters with estimated signs consistent with theoretical expectations, the SUR estimates without instruments rank above both the OLS and instrumental variable estimates. By the same criterion, normalization of the district level profit function by farm power price ranks higher than normalization by paddy price. At both farm household and district levels, the estimates are not fully consistent with the profit maximizing hypothesis. However, when the own-price parameter estimates at the district level are raised by one standard deviation, the adjusted profit function exhibits convexity in prices. Input demand and output supply responses differ substantially between the farm household and district level analyses. At the farm household, all crops appear to be competitive to each other. Except for pulse, they are complementary at the district level..