Consumer Price Index Prediction by ARIMA And ETS

Abstract

Consumer Price Index (CPI) is one of the most widely used indicators to measure inflation and reflect changes in the cost of living. Accurate forecasting of CPI has become increasingly important as it provides essential guidance for economic decision-making and financial planning. This paper applies two classical time series approaches, the ARIMA model and the ETS model, to monthly CPI data, with the aim of comparing their forecasting performance. The results show that both models capture the upward trend of CPI, while the ARIMA (2,1,2) with drift produces slightly lower RMSE, MAE, and MAPE values than ETS (A, A, N), suggesting that ARIMA has marginally better predictive accuracy. These findings confirm the effectiveness of both methods in modeling CPI but also indicate that ARIMA provides a more precise fit. The study contributes to the literature by offering an updated evaluation of CPI forecasting methods, which may support future empirical work and provide references for policy and market analysis.