A Time Series Anaylsis Of Autoregressive Distributed Lag (ARDL) To Co Integration Technique: The Results Applications

Abstract

This research examines Autoregressive Distributed Lag (ARDL) to co integration technique and its theoretical background, advantages, application, interpretation and problems of ARDL to co integration approach. ARDL approach to co integration technique has flexible behavior about variables, can be applied in the presence of same or different stationary order or mutually integrated. This approach has a compulsory condition that none of variable in the model should be at second difference. This technique is applied on time series dataset while time series dataset has to face unit root problems. In the presence of unit roots problems the forecasting model gives the spurious results.   The bound test (Wald test) or F-statistics reflects the long run relationship of the variables. The value of F-statistics exceeds the lower and upper limits at 90% and 95% confidence interval represents the co integration among variables exists in long run. In case of low value of bound test shows absence of co integration. Error Correction Mechanism is used for the picture of short run results. The negative value of ecm (-1) represents the model is highly significant and convergence towards equilibrium. The Lagrange Multiplier test for serial correlation and Ramsey Reset test for correct functional form is used in diagnostic test scenario. The value of F-version and LM-version is more than 10% or 5% fulfills all the assumptions of Ordinary Least Square. Pesaran and Pesaran (1977) applied the stability test if the plot of CUSUM and CUSUM sum of square lies in between 5% critical bound limits showed model is stable without structural breaks.

Keywords:

ARDL Approach; Bound Test; Error correction Mechanism; Unit Root Tests; Stability