ANALYSIS OF INFLATION VOLATILITY FOR FISH AND MEAT MARKETS IN NIGERIA USING ARMA AND GARCH TYPES

Abstract

This paper analyzed Consumer Price Index data (2009M9 -2016M10) on fish and meat in Nigeria. Findings reveal that the values of coefficient of variation of Consumer Price Index for fish and meat are 26% and 22% respectively. The experiment to model the absolute and log returns series of fish and meat inflation was carried out using 80 out of the selected 85 observations. This enabled the paper to access the adequacy of the models to forecast the remaining 5 observations. Using Akaike Information Criteria (AIC) and Schwarz information Criterion (SIC) to select model’s goodness of fit, Autoregressive Moving Average - ARMA (4,3) is selected for Fish Absolute Returns (FAR) and Fish Log Returns (FLR) while Autoregressive Exponential Generalized Autoregressive Conditional Heteroscedasticity - AR-EGARCH (1,1) is selected for Meat Log Returns (MLR) series. Similarly, Autoregressive Generalized Autoregressive Conditional Heteroscedasticity - AR-GARCH (1, 1) and AR-EGARCH (1,1) are both selected for Meat Absolute Returns (MAR) series. In forecasting the inflation returns, the dynamic forecast selects ARMA (4,2) and AR (1) as equations that best forecast the out of samples for MLR and MAR series respectively. The static forecast chooses AR-EGARCH (1,1) model for the MLR and MAR series. The returns from the models selected by the dynamic forecast yield near accurate CPI. The better performance of dynamic forecast attest to the marginal effect of seasonality in fish and meat production in Nigeria. The paper concludes that the presence of high volatility persistence indicates volatility today will significantly have a long term influence on meat price.