Research on Factors Influencing and Forecasting Regional Gross Domestic Product—A Case Study of Sichuan Province

Abstract

This article analyzes the Gross Domestic Product (GDP) of Sichuan Province from 2012 to 2022 and its influencing factors, selecting ten relevant indicators. Through preliminary linear analysis, weakly correlated indicators were eliminated, ultimately identifying three independent variables: the proportion of the tertiary industry, urban population, and total number of insured individuals at year-end. A multiple linear regression model was constructed to analyze the impact of various economic indicators on GDP. The resulting regression equation was y = -0.008112 - 0.287037x_3 + 1.144240x_5 + 0.130342x_8, indicating a negative correlation between the proportion of the tertiary industry (%) and GDP, while urban population and year-end insured individuals showed a positive correlation with GDP. A time series autoregressive model was utilized to forecast GDP for the next decade. The results confirmed the negative correlation of the tertiary industry proportion (%) with GDP, and positive correlations of both urban population and year-end insured individuals with GDP. The model fit well, explaining 99.71% of the variability in the dependent variable. Forecasts indicate continued growth in urban population and stability in the number of insured individuals at year-end. During the model establishment and resolution process, preliminary linear relationships among the independent variables were defined based on the multiple linear regression model, with significance analysis conducted using R software to derive the regression equation. Subsequently, a dynamic forecast of GDP for the next decade was carried out using the time series autoregressive model, employing differencing to ensure data stationarity. Hypothesis tests of the model revealed that residuals approached normal distribution, with no significant autocorrelation, although heteroscedasticity was present, indicating the need for further model improvement. Transformations of variables (e.g., logarithmic transformation) may be considered to reduce heteroscedasticity.