Benchmarking farms, in order to advise farmers to cure inefficiency, may be biased if heterogeneity is not accounted for. Technological variability in agriculture indeed happens, but productive efficiency analysis with frontier methods usually assumes homogeneity. Heterogeneity influences investment motives and production strategies, but is not always clear-cut, for example, when gradation in external inputs use occurs. Unfortunately, these indistinct (no clear-cut) differences in technologies are very common within farming communities, but have often been ignored by the advisors focusing on the discrete ones such as organic versus conventional farming. This paper explores indistinct heterogeneity in efficiency analysis, aiming at identifying peers/reference farms while reflecting on their significance for benchmarking. The gradual differentiation between low and high input dairy farms in Flanders is used as a case, based on a five-year balanced panel data for 58 farms. A data envelopment analysis (DEA) version of the meta-frontier approach is used to account for heterogeneity. The research revealed that, although stemming from a continuous distribution, low and high input farming can be considered as different strategies but none can be said to be superior to the other. Coupling the efficiency scores with peer information allows distinguishing good and bad performing efficient farms within each strategy, and thus improves benchmarking using frontier analysis.

如果不考虑异质性，为了建议农民解决低效率问题，基准农场可能会有偏差。农业中的技术可变性确实发生了，但是采用前沿方法进行的生产效率分析通常假定是同质的。异质性会影响投资动机和生产策略，但并非总是很明确，例如，当外部投入使用发生分级时。不幸的是，这些模糊的技术差异（没有明确的差异）在农业社区中非常普遍，但是专注于离散技术（例如有机农业与常规农业）的顾问们常常忽略了这些技术差异。本文探讨了效率分析中不明显的异质性，旨在确定同龄人/参考农场，同时反思其对基准的重要性。基于58个农场的五年均衡面板数据，以法兰德斯地区低投入和高投入的奶牛场之间的逐渐差异为例。元边界方法的数据包络分析（DEA）版本用于说明异构性。研究表明，尽管源于持续的分布，低投入和高投入耕作可被视为不同的策略，但不能说任何一种都优于另一种。将效率得分与对等信息相结合，可以区分每种策略中表现良好的农场和不良的高效农场，从而使用前沿分析提高基准。元边界方法的数据包络分析（DEA）版本用于说明异构性。研究表明，尽管源于持续的分布，低投入和高投入耕作可被视为不同的策略，但不能说任何一种都优于另一种。将效率得分与对等信息相结合，可以区分每种策略中表现良好的农场和不良的高效农场，从而使用前沿分析提高基准。元边界方法的数据包络分析（DEA）版本用于说明异构性。研究表明，尽管源于持续的分布，低投入和高投入耕作可被视为不同的策略，但不能说任何一种都优于另一种。将效率得分与对等信息相结合，可以区分每种策略中表现良好的农场和不良的高效农场，从而使用前沿分析提高基准。