eng
Rouf,Abdur
Evaluating Interventions in Developing Country Agriculture: The Productive Efficiency and Related Analytical Issues
http://ageconsearch.umn.edu/record/258303/files/Research_4.pdf
Productive efficiency, a widely used estimate for evaluating interventions in agriculture, is sensitive to specifications of econometric models and analytical approaches. These methodological issues likely to vary from one production environment to another. Hence it is important to select an appropriate combination of the specifications of the econometric model and analytical approaches in order to get unbiased and efficient estimates. This study, however, looks into some of the crucial specifications of the stochastic frontier model and analytical approaches that apply to developing country agriculture. The empirical evaluation of two competing interventions, namely, the Silt-dredging and Regulated-drainage Management (SRM) and the Tidal River-basin Management (TRM), estimates stochastic frontier model for rice production with different specifications, (e.g., normal half-normal, normal truncated-normal, error component etc.) involving alternative analytical approaches (e.g., pooled modeling vs. individual modeling etc.). The mean productive efficiency (technical) scores obtained from the pooled model are 0.769 and 0.707 respectively for the SRM and the TRM, whereas these scores from the individual models are 0.854 and 0.976 in order. However, both of the model diagnostics (e.g., σ2 and γ) are found statistically highly significant with the pooled model, unlike the individual models. Meanwhile, a translog functional form of the frontier pooled model came up with biased estimates, unlike the Cobb-Douglas form. In fact, the pooled model with normal truncated-normal distributional having Cobb-Douglas functional form came up with consistent and statistically significant outcomes.
2017-07-02T08:10:33Z
http://ageconsearch.umn.edu/record/258303