Carbon Emission Prediction in Shandong Province: A Novel Extended STIRPAT Based on MCP Regression Carbon Emission Prediction in Shandong: An Extended STIRPAT Model with MCP Regression

Against the backdrop of China's "Dual Carbon" strategy, the identification of regional carbon emission drivers and the projection of future scenarios have emerged as pivotal components in promoting high-quality development and green transformation. This study focuses on Shandong Province, a representative high-energy-consumption region located in eastern coastal China, and employs an extended STIRPAT modeling framework. A dataset encompassing thirteen variables across demographic, economic, energy, structural, and openness dimensions is constructed to systematically characterize the mechanisms underlying regional carbon emissions and their evolutionary dynamics. To address the issue of multicollinearity inherent in multivariable modeling, this study introduces the Minimax Concave Penalty (MCP) regression approach for robust estimation of the extended STIRPAT model. The results demonstrate that MCP regression not only enhances model fitting accuracy but also effectively isolates key explanatory variables, mitigating the over-shrinkage problem associated with Lasso. The principal drivers of carbon emissions in Shandong Province are identified as the level of economic development, energy intensity, population size, educational attainment, and degree of openness to external economies. The integration of MCP regression with the extended STIRPAT model offers a feasible and efficient approach for elucidating the mechanisms of regional carbon emissions and conducting scenario-based projections.