Forecasting Chinese provincial carbon emissions using a novel grey prediction model considering spatial correlation

Abstract

In response to the errors caused by the uniform background value coefficients in the traditional grey model and the lack of analysis ability of panel data, this study proposes a two-stage background value calculation method and introduces a spatial weight matrix to employ the spatial correlation of variables in the grey model, creating a spatial grey model SGM(1,1,m) and realizing the modeling of spatial data by a grey model. The validity of the SGM(1,1,m) model was tested using carbon emission data from 30 provinces in China, and the carbon emissions of these provinces are predicted from 2020 to 2025. Conclusions are drawn as follows. First, the two-stage background value optimization mode and the addition of spatial overflow term in the model are reasonable, and the SGM(1,1,m) model improves the modeling performance in a certain sense compared with the GM(1,1) model while realizing regional association modeling. Second, the SGM(1,1,m) model has some formal similarities with the grey multivariate model, but while both are similar in terms of modeling requirements, the modeling purposes or economic meanings represented are different and should not be confused. Third, the SGM(1,1,m) model can achieve modeling predictions while providing a simple analysis of the spatial correlation of carbon emissions. Fourth, the prediction results present that the rise of carbon emissions in eastern China will level off, but the rise of carbon emissions in the central and western China will accelerate, which is largely because of the faster rise of carbon emissions in key provinces such as Shaanxi, Gansu, Ningxia, and Inner Mongolia.