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

Abstract

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 heterogeneous farmlands, both functions are likely to produce the same results. The interview was used as the instrument for obtaining plot-specific data from 120 cassava farmers across six districts and data was analyzed using the quantitative technique. Direct predictors of output include plot size, labor, hoes, cutlasses, and cassava stems. Using a 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 in the Translog function, the variance parameter score for the CD function is significantly different from those of the Translog function, and the maximum output attainable for the given productive inputs was 40% and 15% respectively. This 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 the 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 behavior of farmers for reliability and policy inferences