基于动态时间序列建模的全国居民可支配收入预测效能优化研究——ARIMA乘积季节模型与指数平滑的对比及政策启示

Optimizing Predictive Performance of National Residents’ Disposable Income Using Dynamic Time Series Modeling: A Comparative Analysis of Multiplicative Seasonal—ARIMA and Exponential Smoothing with Policy Implications

摘要: 随着国家经济高质量发展与共同富裕目标的深入推进，全国居民人均可支配收入作为衡量民生福祉的核心指标，其稳定增长不仅是“十四五”规划中“扩大中等收入群体”战略的直观体现，更与当前“双循环”新发展格局下提振内需、激发消费潜力的政策导向密切相关。本文基于2013~2023年全国居民人均可支配收入数据，融合内部参数化方法(ARIMA乘积季节模型)与外部非参数化方法(指数平滑模型)，构建动态预测框架。首先，分别选出两种类型的最佳预测模型，预测2024年3、6、9月的人均可支配收入，并与实际收入进行比较，评估模型的预测性能。其次，通过两模型比较，得出ARIMA乘积季节模型较好。最后，利用ARIMA乘积季节模型，预测未来一年内全国居民人均可支配收入趋势，为数字经济赋能乡村振兴、区域协调发展等国家战略的落地实践提供数据驱动决策范式。

Abstract: With the deepening advancement of China’s high-quality economic development and common prosperity goals, the stable growth of national per capita disposable income of residents—a core indicator for measuring people’s well-being—not only directly reflects the “expanding middle-income group” strategy in the 14th Five-Year Plan, but also closely aligns with current policy orientations to boost domestic demand and unleash consumption potential under the new “dual circulation” development paradigm. Based on national per capita disposable income data from 2013 to 2023, this paper constructs a dynamic forecasting framework by integrating internal parametric methods (ARIMA multiplicative seasonal model) and external non-parametric methods (exponential smoothing model). First, optimal forecasting models from both categories are selected to predict per capita disposable income for March, June, and September 2024, with subsequent comparisons against actual income to evaluate model performance. Second, comparative analysis demonstrates the superior performance of the ARIMA multiplicative seasonal model. Finally, the ARIMA multiplicative seasonal model is employed to forecast national per capita disposable income trends over the next years, providing a data-driven decision-making paradigm for implementing national strategies such as digital economy-enabled rural revitalization and coordinated regional development.