Soil health practice adoption and productive efficiency: evidence from Swiss wheat growers

Summary

In this paper we assess the productivity and efficiency implications of adopting different soil health promoting agricultural production strategies. We do so by considering both the extent of individual practice adoption (i.e. area covered) and the bundles of jointly practices adopted over the arable farmland. The analysis makes use of novel agricultural survey data collected in 2024 representing 2’714 arable farms in Switzerland. We utilise a two-staged approach, combining Gaussian Mixture Modelling and Stochastic Frontier Analysis to address our research question. Gaussian Finite Mixture Modelling, an unsupervised machine learning method, was used to cluster the adoption extent of 12 arable production practices into 6 different soil health promotion strategies predominant in the sample, the results of which we verified with soil health experts. Clustering in such a way was performed, in order to reduce the high dimensionality of practice adoption to a manageable level for conducting the analysis. We then analyse the productivity and efficiency effects across each soil health practice adoption cluster – representing heterogenous soil health management strategies - using a Meta-Frontier approach. Whilst the analysis is still ongoing and final results are not presented in this abstract, we can already draw findings of both policy and methodological relevance with regard to assessing and supporting highly heterogenous practice adoption.