APPLICATION OF LONG-MEMORY PROCESSES TO DESCRIBING INTERNATIONAL INFLATION RATES: SOME FRACTIONALLY INTEGRATED ARFIMA-GARCH EVIDENCE.

Neslihan KAYA (Bilkent University)
Kıvılcım METİN-ÖZCAN(Bilkent University)

The paper considers the application of long-memory processes to describing international inflation rates. For this purpose, long memory in inflation rates based on the Consumer Price Index (CPI) and Wholesale Price Index (WPI) for all countries in the IFS Database is tested for, using ARFIMA models. Additionally, the process is allowed to have GARCH type conditional heteroscedasticity. The interaction of mean and volatility of inflation is also observed in order to find evidence for or against two hypotheses: Friedman hypothesis (1977), which states that volatility or uncertainty increases in high-inflation regimes, and Cukierman&Meltzer hypothesis (1986), which states that increases in the inflation uncertainty raise the optimal average inflation rate. Significant long memory component in the CPI and WPI-based inflation rates exists. Implications and sources of the long memory and persistence are evaluated.