SARIMA MODELS: REVIEW AND ITS APPLICATION TO KENYAN’S COMMODITY PRICE INDEX OF FOOD AND BEVERAGE.

Abstract

Attaining price stability is one of the objectives of monetary policy in any economy to protect both consumers' and producers' interest. Unpredictable food and beverages prices make it difficult for consumers to plan for their expenditure in case of unexpected inflation. On the flip side, low prices may hurt producers as they may not be able to protect their profit margins. It is therefore imperative to develop a precise and accurate model to forecast Kenya's commodity prices. Therefore, the current sought to model the commodities price of food and beverage in Kenya using a Seasonal Autoregressive Integrated Moving Average (SARIMA). SARIMA model takes into account the seasonal periodic fluctuations in a series that usually recur with about the same time interval. Secondary data on monthly food price index was obtained from the KNBS website. The data covered the period from January 1991 to June 2017 with a total of 318 monthly observations. Data analysis was carried out using the R-statistical software. Using the Maximum Likelihood Estimation method, the SARIMA (0,1,2) (0,1,1)12 model had better forecasts accuracy than other competing orders based on the Bayesian Information Criterion (BIC=1638.42) criterion with MAE of 2.25 in its forecasting ability. The two-year predictions of food and beverages price index showed an oscillatory behaviour with an increasing trend. The forecasts can help consumers adjust expenditure in preparation for periods of inflation. Policymakers should make priorities to ensure stability of future commodity prices.