Traditional benchmarking implicitly assumes that decision making units operate in isolation from their peers. For arable production systems in particular, this assumption is unlikely to hold in reality. This paper quantifies spatial spillovers on input‐specific inefficiency using data envelopment analysis and a second‐stage bootstrap truncated regression model. The bootstrap algorithm is extended to allow for the estimation of the parameter of the spatial weight matrix, which captures the proximity between producers. The empirical application concerns Dutch arable farms for which latitudes and longitudes are available. The average inefficiency across years was 3.87% for productive inputs and 2.98% for damage abatement inputs under variable returns to scale. For productive inputs technical inefficiency, statistically significant spillover effects from neighbours’ age and their degree of specialisation depended on the type of the spatial weight matrix used (inverse distance or k‐nearest neighbours). Statistically significant spillover effects of subsidy payments were adverse while statistically significant spillover effects from insurance payments were beneficial. For damage abatement inputs technical inefficiency, statistically significant adverse effects were found for neighbours’ age and subsidy payments and beneficial effects from neighbours’ insurance payments and their degree of specialisation.

中文翻译：

---

荷兰耕地投入低效的空间溢出效应

传统的基准测试隐含地假设决策部门与其同级机构隔离运行。特别是对于耕种生产系统，这种假设不太可能成立。本文使用数据包络分析和第二阶段自举截短回归模型对输入特定效率的空间溢出进行了量化。自举算法得到扩展，可以估计空间权重矩阵的参数，从而捕获生产者之间的接近度。该经验应用涉及荷兰的耕作农场，这些农场可提供经度和纬度。在可变规模收益下，多年的平均无效率生产投入为3.87％，减少损害投入为2.98％。对于生产性投入的技术效率低下，k个近邻）。补贴支付的统计显着溢出效应是不利的，而保险支付的统计显着溢出效应是有益的。对于减轻损害投入的技术效率低下，发现对邻居的年龄和补贴付款具有统计学上的显着不利影响，而邻居的保险付款及其专业化程度则带来了有利影响。