Modelling Rates of Inflation in Kenya: An Application of Garch and Egarch Models

Abstract

The purpose of this study was to determine an effective Arch-type model for forecasting Kenya’s inflation. Using Kenya monthly inflation data from January 1990 to December 2015, the performance of GARCH and EGARCH type models was analyzed to come up with the best model for forecasting Kenyan inflation data. Since the inflation series is non-stationary, the Consumer Price Index (CPI) was first transformed to return series by logarithmic transformation. Afterwards, the data was tested for the presence of ARCH effects and serial correlation using both Ljung Box Pierce Q test and Engle Arch test. The test showed presence of heteroscedasticity and correlation in the inflation return series which is a key feature of a financial time series data. The project adopted AIC and BIC in selecting the the best model. From the fitted models EGARCH (1,1) had the smallest AIC and BIC values followed by the GARCH(1,1) model. Model diagnostic test was conducted on the selected model EGARCH (1,1) model to determine its adequacy and goodness of fit. QQ plot was fitted to the residuals of the model and fairly straight line was produced looking roughly linear. Furthermore weighted Ljung Box Test on standard squared residuals showed the absence of correlation in the model. In conclusion, EGARCH(1,1) model is the best model for forecasting Kenyan inflation data.

Keywords: AIC, BIC, Model adequacy and heteroscedasticity