Analysis of the impact of resource misallocation and socialized services on low-carbon agricultural production with DML based on random forest

The substantial increase in agricultural carbon emissions (ACEs) contributes significantly to global warming, and agricultural resource misallocation (ARM) plays a major role in the rise of ACEs. Therefore, the aim of developing modern agriculture is to reduce ARM and promote low-carbon agriculture. To achieve this, the double machine learning model (DML) and moderating effect model are adopted in the present study to explore the correlations between agricultural socialized services (ASS), ARM, and ACEs. Also, the ARM index is calculated using the panel data collected from 31 provinces during the 2004–2020 period in China. The research results are as follows. Firstly, ARM has a significant positive impact on ACEs at the 1% significance level. Secondly, ASS exerts significant negative regulatory effects on ARM and ACEs at the 1% significance level. Lastly, the goodness of fit (R2) reaches 95% and 40% for DML and the fixed effect model (FEM), respectively. Relatively, DML performs better in the goodness of fit according to this study. Furthermore, it is found out in this study that land misallocation, labor misallocation and capital misallocation increase ACEs and impede the development of low-carbon agriculture. Also, as a regulatory variable, ASS mitigates the ARM-induced increase of ACEs. Therefore, these findings provide guidance on the development of low-carbon agriculture for ASS.