Time Series Forecasting of Producer Price Index, using ARIMA

Nikhil Garg, Aakash Varshney, Adarsh Agrawal

Abstract: Producer Price Index (PPI) is a key indicator of economic stability of a country. This project aims to forecast the quarterly future PPI of USA using ARIMA Model for the years 2003 2007, using a data set with quarterly PPI data for the years 1960 2002. Based on our analysis, it was interpreted that the ARIMA (1, 1, 1) was best suited for modeling the future PPI, with maximum log-likelihood of and the minimum AIC of 393. The Ljung Box test reveals that the residuals are free from heteroscedasticity and serial correlation.

Keywords: Time Series forecasting, ARIMA, econometrics, PPI, MA Process, ACF plot