A bio-economic model of small-ruminant production in the semi-arid tropics of the Northeast region of Brazil: Part 2—Linear programming applications and results

Abstract

Survey data covering production systems for mixed farms in the Northeast region of Brazil has been synthesized within a linear programming (LP) framework. The resulting model contains activities covering the production of cattle, sheep and goats, and a vector of alternative cropping activities. Farm resources include two categories of grazing land, planted forages, family labour, two categories of hired labour, and working capital. The major livestock activities represented in the region were included as production options. Initial results did not discriminate between categories of available grazing resources. Therefore, cattle, by virtue of their higher dressing percentages and higher price per kilogram, were the optimal livestock species. A series of adjustments was then carried out to reflect types of feed resources and patterns of animal species selectivity. Optimal farm solutions for a representative traditional-production unit found objective function levels close to those found by farm surveys, but discrepancies between model results and the actual farm situations for sheep and goat activities. Model results excluded small-ruminant breeding activities because of the low net offtake at weaning levels assumed in the model. Data that became available after these initial model runs showed a higher net offtake level, and these revised coefficients resulted in optimal LP results very close to those actually found on farms. The model was then used to simulate the response of activities and farm economic performance to ‘good’ and ‘bad’ years defined by ± half standard deviation from mean annual levels of precipitation. Model results indicated much higher variability of farm income in response to weather than that found with changes in levels of technical efficiency of sheep and goat production.