Short-term Forecast and Inflation Trend Analysis of China's CPI Based on ARIMA Model

Abstract

The Consumer Price Index (CPI) is a critical indicator for macroeconomic regulation. While time series analysis is widely used for short-term forecasting, existing studies on China's CPI often overlook significant seasonal fluctuations or lack rigorous out-of-sample validation. This study addresses these gaps by applying the Seasonal Autoregressive Integrated Moving Average (SARIMA) model to China's monthly CPI data from February 1996 to August 2025. Following the Box-Jenkins methodology, the data is split into training and test sets to ensure robust evaluation. Empirical results identify the optimal model as SARIMA(1,0,2)(1,0,1)₁₂. In a 24-month out-of-sample forecast, the SARIMA model significantly outperforms the non-seasonal ARIMA model, reducing the Root Mean Squared Error (RMSE) by approximately 14.1% (0.384 vs. 0.447) and the Mean Absolute Error (MAE) by 15.2%. This confirms the necessity of incorporating seasonal factors for accurate monthly CPI forecasting. Based on the optimal model, the 12-month-ahead forecast indicates that China's CPI will follow a trajectory of low-level, moderate recovery, stabilizing within the 1.0%–1.5% range by mid-2026. The findings suggest no immediate risk of sustained deflation but highlight limited upward pressure, providing a quantitative basis for monetary policy fine-tuning.