Machine learning approaches for explaining determinants of the debt financing in heavy-polluting enterprises

Under the background of green credit policy, more and more attention has been paid to the debt financing of high-polluting enterprises. This paper collects 224 financial and non-financial indicators in 40 listed enterprises in the mining, steel, and power industries to investigate their relationship with those measurement indicators. This paper selects the XGBoost method for feature selection to sort out the top six indicators of the combination of subsets under the condition of high dimension. The screened indicators have a good effect in predicting long-term debt. Further explanation is given based on the Shapley additive explanation value.

Automated summary

This paper investigates the relationship between financial and non-financial indicators of 40 listed enterprises in the mining, steel, and power industries and debt financing. Using the XGBoost method, the top six indicators are identified for predicting long-term debt. Further explanation is provided based on the Shapley additive explanation value.