We examine the role of adopting environmentally-friendly agricultural technologies on the technical efficiency (TE) among maize farmers in the mountainous areas of Northern Vietnam. While technology adoption is non-randomly assigned to farmers, we draw on recent advances of machine learning models to identify farm adoption profiles of multiple agricultural technologies. Combining the stochastic production frontier with the inverse probability weights, we compare the TE between adopters and non-adopters. Although, farmers adopting new seed varieties achieve significantly higher TE than others, dissemination of integrated pest management practices by extension agents affect negatively farm performance. Results show that there is a positive relationship between actions to reduce the vulnerability of landscapes to soil erosion and corn productivity. We find heterogeneous effects of adopting multiple environmentally friendly technologies on TE. Farmers who are a member of cooperative and employ organic fertilizer are associated with higher odds to adopt new seed varieties. Maize cultivation in sloping land and farmers with large landholdings tend to be the key drivers in the provision of farm advisory services and the exertion of soil erosion control techniques. Sensitivity analysis suggests that the empirical findings are largely robust against the hidden hazard of unmeasured confounders.

我们研究了采用环保农业技术对越南北部山区玉米农民的技术效率（TE）的作用。尽管技术采用不是随机分配给农民的，但我们利用机器学习模型的最新进展来确定多种农业技术的农场采用情况。将随机生产边界与逆概率权重相结合，我们比较了采用者和非采用者之间的TE。尽管采用新种子品种的农民获得的TE显着高于其他品种，但推广人员对病虫害综合治理方法的传播会对农场经营产生不利影响。结果表明，减少景观对土壤侵蚀和玉米生产力的脆弱性的措施之间存在正相关关系。我们发现在TE上采用多种环保技术会产生不同的影响。作为合作社成员并使用有机肥料的农民采用新种子品种的几率更高。倾斜土地上的玉米种植和拥有大量土地的农民往往是提供农业咨询服务和运用土壤侵蚀控制技术的主要驱动力。敏感性分析表明，实证结果在很大程度上抵御了无法衡量的混杂因素的隐患。倾斜土地上的玉米种植和拥有大量土地的农民往往是提供农业咨询服务和运用土壤侵蚀控制技术的主要驱动力。敏感性分析表明，实证结果在很大程度上抵御了无法衡量的混杂因素的隐患。倾斜土地上的玉米种植和拥有大量土地的农民往往是提供农业咨询服务和运用土壤侵蚀控制技术的主要驱动力。敏感性分析表明，实证结果在很大程度上抵御了无法衡量的混杂因素的隐患。