Sarima Modelling of Inflation : The Case of Liberia

Not being doubtful of the fact that Inflation is one of the most important macroeconomic indicators assessing the economic well-being of a nation, this paper seeks to develop a model to forecastmonthly inflation for Liberia using Time Series Econometric tools and concepts. The paper used a monthly data on inflation from January 2005 through May 2018. Part of the data from January 2005 through May 2017 were used to assess the model's in-sample performance. The rest of the data were also used to assess the out-of-sample performance of the model. It was realized that running the routinely ARIMA model was restrictive and not very informative given the model's inability to capture any seasonality when there exists. Raw plots of log of inflation was stationary with evidence from the ADF, DF-GLS and the PP testsfor unit root. Nonetheless, seasonal peaks were highly discernible from AC and PAC plots which appear to decay at a slow rate. Seasonal difference was taken and the paper proposed 8 candidate models intended to slim the likelihood of missing out difference was taken and the paper proposed 8 candidate models intended to slim the likelihood of missing out 12 emerged as the best model. Gross attention for model selection was given to the Information Criteria (AIC and SBIC) and the Ljung-Box test for serial correlation. The model outperforms all 8 proposed models both in-sample and out-of-sample with a fairly lower forecast error