Identifying driving factors of energy poverty and their interaction mechanism based on the BPNN-WINGS method: Household-level evidence from China

Energy poverty is the inability to meet necessary household energy needs and has threatened human well-being and exacerbated climate change. However, there has been a lack of systematic analysis regarding the driving factors of energy poverty and their interaction mechanisms, which is crucial for achieving effective energy poverty reduction. This paper proposes a novel integrated technique composed of the Backpropagation Neural Network (BPNN) and Weighted Influence Non-linear Gauge System (WINGS) method to identify the most important driving factors of energy poverty and understand their interaction mechanism by using householdlevel survey data in China from 2016, 2018, and 2020. The findings reveal that per capita household income is the most important cause-driving factor. Household heads ' education attainment and policy support, such as government subsidy and housing provident fund, are also significant causes of energy poverty. Factors, such as household central heating system usage and house ownership, are in the downstream of the interaction mechanism chain to be considered as the effect-driving factors. Finally, a robustness test is conducted to validate the effectiveness of the proposed method and the obtained findings, which could assist energy-related decisionmakers with the most significant drivers for reducing energy poverty.