Technical efficiency and its determinants in Vietnamese rice farming: a comparison of DEA and SFA approaches

Understanding efficiency performance and the sources of inefficiency in rice production enables farmers to optimize resource use and mitigate greenhouse gas emissions. This paper measures technical efficiency (TE) and explores its determinants in rice farming through a comparative analysis of Data Envelopment Analysis (DEA) and Stochastic Frontier Analysis (SFA) methods. We use primary farm-level data collected in the Mekong River Delta, Vietnam. The results show that mean TE estimates are substantially different between methods, with values of 0.62 for conventional DEA and 0.73 for SFA models. This gap becomes even larger when comparing bias-corrected Bootstrap DEA and SFA models with explanatory variables, at 0.55 and 0.82, respectively. In this case, the SFA model with exogenous variables is preferred for measuring TE and its determinants. A mean TE of 0.82 indicates significant potential for inefficient farmers to increase output by improving TE. The study finds that being a male household head, attending field training, and owning rice land reduce technical inefficiency, whereas poor soil quality (due to alkalinity and salinity intrusion), rice diseases, and natural disasters increase technical inefficiency. These findings suggest that policies should focus on enhancing farmers’ production skills to help them mitigate the adverse effects of rice diseases and natural disasters, as well as improve soil quality. The research also emphasizes that practitioners and policymakers should be cautious when selecting estimation methods and interpreting efficiency results in rice farming. In particular, the DEA approach may overestimate inefficiency if the data contain significant random noise.