Stable Time Series Prediction of Enterprise Carbon Emissions Based on Causal Inference
Zitao Hong, Zhen Peng, Xueping Liu
TL;DR
This work addresses cross-environmental non-stationarity in enterprise carbon emissions by introducing Stable-CarbonNet, a temporal prediction framework that fuses causal-invariance (stable learning) with temporal adaptation. It formalizes a problem of cross-environment generalization, decomposes emissions into environment-invariant causal drivers $f^*(X^c)$ and environment-specific perturbations $h^{(e)}(X^s)$, and imposes gradient-consistency constraints to pursue stable representations across multiple environments. The model uses adaptive normalization and sample reweighting to handle temporal non-stationarity, and its loss combines environment-specific risks with a stability penalty that enforces consistent optimality across environments. Empirical results on multi-environment enterprise-panel data show that Stable-CarbonNet achieves superior cross-environment generalization and interpretable mechanism stability, with ablations confirming the central role of stability constraints; findings indicate that causally stable features such as energy input structure and capital equipment provide durable guidance for policy and industrial decarbonization strategies.
Abstract
Against the backdrop of ongoing carbon peaking and carbon neutrality goals, accurate prediction of enterprise carbon emission trends constitutes an essential foundation for energy structure optimization and low-carbon transformation decision-making. Nevertheless, significant heterogeneity persists across regions, industries and individual enterprises regarding energy structure, production scale, policy intensity and governance efficacy, resulting in pronounced distribution shifts and non-stationarity in carbon emission data across both temporal and spatial dimensions. Such cross-regional and cross-enterprise data drift not only compromises the accuracy of carbon emission reporting but substantially undermines the guidance value of predictive models for production planning and carbon quota trading decisions. To address this critical challenge, we integrate causal inference perspectives with stable learning methodologies and time-series modelling, proposing a stable temporal prediction mechanism tailored to distribution shift environments. This mechanism incorporates enterprise-level energy inputs, capital investment, labour deployment, carbon pricing, governmental interventions and policy implementation intensity, constructing a risk consistency-constrained stable learning framework that extracts causal stable features (robust against external perturbations yet demonstrating long-term stable effects on carbon dioxide emissions) from multi-environment samples across diverse policies, regions and industrial sectors. Furthermore, through adaptive normalization and sample reweighting strategies, the approach dynamically rectifies temporal non-stationarity induced by economic fluctuations and policy transitions, ultimately enhancing model generalization capability and explainability in complex environments.