TIME SERIES ANALYSIS OF INFLATION IN ETHIOPIA: UNIVARIATE CASE

Abstract

Inflation refers to a situation in which the economy’s overall price level is rising where as Inflation rate is the percentage change in the price level from the previous period. The aim of this research paper focused to fit a Univariate time series model which can be used to forecast the overall inflation behavior in Ethiopia. The secondary data based on the Consumer Price Index (CPI) was used on monthly observations from January 2001 to December 2018 reported from Central Statistical Agency(CSA). The Univariate time series model was employed for modeling inflation. We apply ADF test to detect Stationarity of the series. Model selection is one of the fundamental tasks of scientific inquiry and the partial autocorrelation, autocorrelation functions, AIC and BIC were used to identify the appropriate model. Thus, the estimated conditional mean and variance for in-sample series have been achieved by combination of AR (1) with GARCH (1, 1) model with the minimum AIC and BIC rank sum value for the log return series. In all the cases the coefficients of the ARCH terms are significant at 5% level of significance implying that there is clustering of volatility of overall inflation. That is, large changes in log returns of prices of goods and services are likely to be followed by further large changes. Similarly, the significance of the coefficients of the GARCH terms in each case at the 5% level of significance indicates that the present conditional variance is dependent on its past variances. The forecasting accuracy of the model was checked using like MZ-R 2 , RMSE, MAE and MAPE. The general level of price of goods and services is rising over time at the country level. It creates uncertainty when the average price level of goods and services changes significantly and becomes unstable. Therefore, focus should be given on policies that will achieve price stability in the country level. STATA12 software has been used to analyze the data.