Autoregressive distributed lag error correction modeling of some macroeconomic indices in Nigeria

Abstract

This study models a relationship between gross domestic product, motor vehicle assembly and oil refining in Nigeria between 1981 to 2022. The autoregressive distributed lag error correction model (ARDL-ECM) was introduced to evaluate the relationship existing among the variables under consideration. Annual data from 1981 to 2022 (41 observations) were extracted from the National Bureau of Statistics Bulletin 2021. Pre-estimation tests such as time plots and unit root tests were conducted to determine the presence or absence of unit roots in the series. The result shows a mixed order of integration as motor vehicle assembly was stationary at a level while oil refining and gross domestic product were integrated at first difference. This prompted the application of the Autoregressive Distributed Lag (ARDL) bounds test of cointegration and the analysis shows the presence of a long-run relationship among the variables. Lag 3 was chosen using Akaike Information Criteria (34.02331*). The estimates from the error correction model analysis infer that oil refining had a strong effect on gross domestic product and the combined lags of oil refining and motor vehicle assembly significantly caused own shocks in the short-run. Post estimation test of stability (CUSUM test) shows that the model was stable and the serial correlation test also confirms this assertion that the residuals obtained were serially uncorrelated. Some recommendations were made to reduce macroeconomic instability in Nigeria.