Time Series Modelling of Oil Price Fluctuations: Applications to Libya and Nigeria

This study aims to analyse the behaviour of crude oil prices and to determine the dynamic relationships between domestic crude oil prices and fundamental macroeconomic variables in Libya and Nigeria. The analysis in this study involves two stages. The first stage is to analyse and model oil price returns of the Libyan, Nigerian and OPEC markets. Unlike previous studies, this study examines the existence of a structural break in crude oil prices data. The empirical analysis uses the AR-GARCH, AR-EGARCH, AR-GJR-GARCH, AR-APARCH, AR-CGARCH and AR-ACGARCH models for modelling the conditional mean and conditional variance of the oil prices returns under three error distributions, namely the normal distribution, student-t distribution and generalized error distribution. The results show that the three return series exhibit no structural break in the mean and variance equations but we find evidence of volatility clustering and leverage effect response to good and bad news in the asymmetric models in the three markets. We also assess the out-of-sample forecasts of the class of GARCH models by using four loss functions. The results indicate that the AR-CGARCH-GED model is the best model for forecasting oil returns in Libya, whilst the best models for Nigeria and OPEC are the AR-GARCH-GED and AR-EGARCH-t models, respectively. The second stage is to examine the dynamic relationship between oil prices and GDP, exchange rate and inflation using annual data for the 1970-2017 periods in Libya and Nigeria. Both short-run and long-run relationships between these variables are explored by applying cointegration tests, the vector autoregressive model (VAR), and vector error correction (VECM) model, Granger causality tests, impulse response functions and forecast variance decompositions. The results show that there is a cointegrating relationship between domestic oil prices and macroeconomic variables in both Libya and Nigeria. Furthermore, the results show that there is a unidirectional Granger-causality relationship running from Libyan oil prices to Libya's GDP. Moreover, the results show a unidirectional causality running from Nigerian oil prices to GDP and exchange rate in Nigeria. The findings of the impulse response functions suggest significant impacts of domestic oil prices shocks on the macroeconomic variables in Libya and Nigeria in the short and long term. The results of the variance decompositions analysis indicate that the changes in Libyan oil prices can impact Libyan GDP. While, Nigerian oil price shocks could affect most of macroeconomic variables in Nigeria. The main policy implications from these findings are that policymakers should monitor and predict future oil prices and take these expectations into account when adopting a particular monetary policy.