Time-Varying Model for Non-Oil Export Volatility in Nigeria

Abstract

The study of export volatility is important because it plays important roles in the growth of an economy. Most previous studies on export had concentrated on investigating its dynamics with classical econometric models which have static parameters that are incapable of capturing its associated time-varying dynamics and volatility. This paper proposes a Bayesian time-varying parameter dynamic linear model to investigate major non-oil export predictors in the Nigerian economy. The Kalman filter and Markov chain Monte Carlo (MCMC) algorithm are used to perform posterior Bayesian inference on time-varying parameters which implicitly describes the fluctuating relationships between the key drivers of export in an economy. In particular, we investigate the predictive performance of relevant macroeconomic variables on non-oil export using a Bayesian time-varying parameter model. Empirical results show that Gross Domestic Product (GDP) and Lending Rate predict the level of fluctuation in non-oil export in Nigeria for the period under consideration. Some policy implications and change point analyses of these results are also discussed.