Regional heterogeneity and industrial system efficiency in China: A two-stage dynamic network DEA analysis based on the nonconvex metafrontier

In recent years, the growing volume in industrial solid waste generation in China has made its management a critical policy issue. Efficiency evaluation helps assess policy effectiveness and provides decision-making support for industrial development. Given the pronounced disparities across different areas of China in terms of economic development levels, industrial structures, and technological capabilities, regional heterogeneity has emerged as an indispensable concern in research on industrial system efficiency. A provincial industrial system in China contains production and abatement stages. Within this structure, the output of general industrial solid waste can be treated as a linking variable, while waste treatment capacity can be treated as a carry-over variable. On this basis, a two-stage dynamic network data envelopment analysis (DEA) model integrated with a nonconvex metafrontier is constructed. By applying this model, this study evaluates the operational efficiency of industrial systems across China’s four major regions—eastern, central, western, and northeastern China—from 2015 to 2023. It further reveals in depth their spatiotemporal evolution patterns and regional heterogeneity characteristics. The study yields several key findings. First, when regional heterogeneity is incorporated, the efficiency evaluation results of various regions become more impartial and can more accurately reflect the actual status of each region’s industrial system. Second, the overall efficiency of the industrial system is primarily influenced by abatement efficiency, and improvements in the latter effectively promote the former. Third, the disparities in overall efficiency are primarily driven by transvariation density, whereas the disparities in production stage and abatement stage efficiency are mainly caused by inter-area differences. Fourth, from 2015 to 2023, the overall efficiency, production efficiency, and abatement efficiency of all the regions generally tended to converge. Finally, in the absence of external interference, regions with lower overall efficiency tended to converge toward those with higher efficiency, indicating a trend of collective progress and balanced development in abatement efforts across regions.