Comparative Study of Static and Dynamic ARIMA Models in Forecasting of Seasonally Headline Inflation

Consumer Price Index (CPI) is a common indicator of headline inflation. CPI measures the market value of a fixed basket of goods in order to define the inflation of a country’s economy. Headline inflation is the measure of the whole inflation in an economy, which consists of all goods, such a price of consumables and energy, which are volatile and prone to inflationary spikes. Headline inflation is usually related to the shift of living cost, which provides useful information to market consumers. The current paper aims at modelling and forecasting the headline inflation in the case of Greece using the Box–Jenkins methodology for the period 2009:1–2020:12. For this purpose, the ARIMA (6,1,6) model was applied. We estimated the ARIMA (6,1,6) model following the maximum-likelihood approach. We maximized the likelihood by iterating the Marquardt and Berndt–Hall-Hall–Hausman algorithms while using numeric derivatives, the optimum step scale and a convergence criterion for the change in the norm of the parameter vector from one iteration to the next. Finally, in order to forecast the headline inflation through the ARIMA(6,1,6) model, a dynamic process and a static process have been applied. The results of the forecasting process suggest that the static process provides a better forecast comparing to the dynamic one.