Cassava: Farmers Adoption and Livelihood in Bono Region Performance of Stochastic Frontier Functions

The objective is to analyze the stochastic frontier function in comparing the performance of the Cobb-Douglas and Translog Frontier functions in the cultivation of cassava in the Bono Region of Ghana. The assumption is that given the same level of productive inputs of farmers at any given farming season, across heterogenous farm lands, both functions are likely to produce the same results. Interview was used as the instrument for obtaining plot-specific data from 120 cassava farmers across six districts and data was analyzed using quantitative technique. Direct predictors of output include plot size, labour, hoes, cutlasses and cassava stems. Using half-normal distributional assumption, the study evaluates variance parameters of the composed error terms. The results showed that the estimated functions produced comparable results in terms of magnitude and signs of input variables. While efficiency appeared to be much higher in Cobb-Douglas than the Translog function, the variance parameter score for CD function is significantly different from those of Translog function and the maximum output attainable for the given productive inputs were 40% and 15% respectively. The means that farmers can scale up their current crop yield by 60% and 85% respectively of their frontier functions using the same inputs and technology, if the appropriate interventions are carried out. The limitation of the study is non-inclusion of environmental factors such as rain as productive input and the study is limited to comparing frontier functions. The results underscored the importance of examining the current production behaviour of farmers for reliability and policy inferences