Autoregressive Distributed Lag (ARDL) cointegration technique: application and interpretation

摘要

Economicanalysis suggests that there is a long run relationship between variables underconsideration as stipulated by theory. This means that the long runrelationship properties are intact. In other words, the means and variances areconstant and not depending on time. However, most empirical researches haveshown that the constancy of the means and variances are not satisfied inanalyzing time series variables. In the event of resolving this problem mostcointegration techniques are wrongly applied, estimated, and interpreted. Oneof these techniques is the Autoregressive Distributed Lag (ARDL) cointegrationtechnique or bound cointegration technique. Hence, this study reviews theissues surrounding the way cointegration techniques are applied, estimated andinterpreted within the context of ARDL cointegration framework. The study showsthat the adoption of the ARDL cointegration technique does not require pretestsfor unit roots unlike other techniques. Consequently, ARDL cointegrationtechnique is preferable when dealing with variables that are integrated ofdifferent order, I(0), I(1) or combination of the both and, robust when thereis a single long run relationship between the underlying variables in a smallsample size. The long run relationship of the underlying variables is detectedthrough the F-statistic (Wald test). In this approach, long run relationship ofthe series is said to be established when the Fstatistic exceeds the criticalvalue band. The major advantage of this approach lies in its identification ofthe cointegrating vectors where there are multiple cointegrating vectors.However, this technique will crash in the presence of integrated stochastictrend of I(2). To forestall effort in futility, it may be advisable to test forunit roots, though not as a necessary condition. Based on forecast and policystance, there is need to explore the necessary conditions that give rise to ARDLcointegration technique in order to avoid its wrongful application, estimation,and interpretation. If the conditions are not followed, it may lead to modelmisspecification and inconsistent and unrealistic estimates with its implicationon forecast and policy. However, this paper cannot claim to have treated theunderlying issues in their greatest details, but have endeavoured to providesufficient insight into the issues surrounding ARDL cointegration technique toyoung practitioners to enable them to properly apply, estimate, and interpret;in addition to following discussions of the issues in some more advanced texts.