Digital agriculture technologies, efficiency, and climate resilience: evidence from soybean production in São Paulo State, Brazil

Abstract

Agricultural technologies play an important role in improving farm performance. Precision agriculture technologies (PATs) aim to enhance input-use efficiency, whereas climate-smart agriculture (CSA) encompasses a set of technologies and management practices designed to increase resilience and reduce climate-related risks to agricultural productivity. Stochastic production frontier (SPF) models have been widely used to evaluate the relationship between technology adoption and farm performance using observed farm-level data rather than experimental or simulated settings. However, SPF literature still lacks both comprehensive synthesis and empirical evidence on how agricultural technologies - particularly emerging digital and CSA technologies - are associated with farm performance through distinct efficiency pathways. In addition, there remains a need for updated farm-level evidence, particularly in Brazil, one of the world's leading agricultural producers, where information on the adoption and impacts of these technologies remains limited. This thesis addresses these gaps by investigating how and to which extent agricultural technologies and management practices are associated with the technical efficiency (TE) and climate resilience of soybean production in São Paulo State, Brazil. Specifically, the thesis pursues three complementary aims: (i) to synthesize the existing evidence on the pathways through which agricultural interventions are associated with farm performance; (ii) to assess the association between PATs adoption and farm-level TE; and (iii) to evaluate whether CSA technologies and practices are associated with TE and resilience under adverse weather conditions. The empirical analyses are based on primary farm-level data collected from 152 soybean farms in São Paulo State, Brazil, during the 2023/2024 season. The first empirical analysis applies the SPF model with heteroscedasticity in inefficiency term to evaluate the association between adopting an integrated bundle of PATs (soil maps, variable-rate fertilizer application, yield monitors, and agricultural management software) and TE of soybean farms. In the second analysis, the latent class SPF model was employed to assess the yield sensitivity of soybean production systems to climate shocks under different levels of CSA adoption, given the climatic conditions of the 2023/2024 growing season, which was marked by above‑average temperatures and below‑average rainfall. The empirical results indicate that adopting PATs as an integrated bundle is associated with considerably higher TE than adopting the technologies in isolation, highlighting the importance of complementarities among digital technologies. Similarly, farms characterized by higher levels of CSA adoption exhibited greater TE and lower sensitivity to climatic shocks than low-intensity adopters. Together, the systematic literature review and the empirical analyses contribute to a better understanding of how agricultural technologies and management practices are associated with farm performance. The findings of the thesis provide novel and robust empirical evidence on the relationships among PATs, CSA technologies, technical efficiency, and climate resilience. Overall, the findings highlight the potential of those technologies to support more productive, resilient, and sustainable soybean production systems under increasing climate variability.