Comparison Of Autoregressive Distributed Lag Model And Vector Error Correction Model Analysis On The Effect Of Some Macroeconomic Variables On Gdp In Nigeria

Abstract

This research work employed Autoregressive distributed lag (ARDL) and Bound Test for co-integration with Vector Error Correction (VEC) Models for estimation of long run and short run effect on Gross Domestic Product (GDP) in Nigeria. In order to achieve this, annual data on GDP, Unemployment, Exchange, and Interest rate from 1980-2017 The Augmented Dickey Fuller (ADF) Test revealed that the variables are stationary at first difference .ARDL and bound test for co-integration revealed that the decision whether to accept or reject either of the hypotheses is inconclusive. Johansen confirmed the existence of long run relationship between the variables. VECM long run estimate indicated that there is positive and significant effect of Exchange rate and Unemployment rate on GDP while Interest rate had negative significance effect on GDP.ARDL long run estimate indicated that only Unemployment rate had significant effect on GDP. ARDL short run estimate among the variables revealed that the coefficient of the ECM(-1) had a correct sign and statistically significant at 5% level which also indicated that the system corrects its previous period at the speed of adjustment by 46% per annum, it was also revealed that interest rate was positive and statistically significant in the short run . While VEC short run estimate revealed that no any variable has significant effect on GDP and that the coefficient of the ECM(-1) had a correct sign and was statistically not significant at 5% level which also indicated that the system had its previous period correct at the speed of adjustment by 12% per annum,. The models were stable for forecasting, no serial correlation, multicollinearity, heteroskedasticity and the residuals are normally distributed