Dynamics and Short-Term Forecasting of the U.S. Consumer Price Index: An Empirical Analysis

Abstract

This article presents an empirical analysis of the dynamics and short-term forecasting of the U.S. Consumer Price Index (CPI) using econometric time-series modeling techniques. Official macroeconomic data from the Federal Reserve Bank of St. Louis (FRED) serve as the empirical basis for the study. Preliminary analysis reveals the non-stationary nature of the original time series, which necessitates the application of ARIMA-class models incorporating differencing. The optimal model specification is selected based on the Akaike and Bayesian information criteria, while model adequacy is confirmed through residual diagnostics. Based on the selected specification, short-term forecasts are generated over a 12-period horizon with corresponding confidence intervals. The forecasting results indicate the persistence of a moderately upward trend in the Consumer Price Index over the projected period. Forecast accuracy is further assessed using the Mincer–Zarnowitz test, which detects no statistically significant bias and confirms both the unbiasedness and informational efficiency of the forecasts. Overall, the findings demonstrate the high predictive performance of ARIMA models in analyzing CPI dynamics and their practical relevance for assessing short-term monetary conditions.