ARDL Model Bounds Test Approach: The Case of Turkey

Abstract

In the econometric literature, ARDL bounds testing approach proposed by Pesaran and Shin [16], improved Pesaran et al. [17], and Toda-Yamamoto causality procedures [12] are widely used in empirical analysis because the outcomes of these tests are more likely to be convincing than their predecessors. The most important distinguishing features of both tests are to necessitate none of prerequisites like stationarity or co-integration analyses. Nevertheless, the number of theoretical studies on these co-integration and causality procedures are not sufficient. The primary purpose of this thesis study is thoroughly to examine the issue in framework of the co-integration analysis and the error correction model within the autoregressive distributed lag model. As a second objective of this thesis, Toda-Yamamoto causality procedure is comprehensively reviewed within a sound theoretical basis. In the empirical part of the thesis, the validity of a level relationship between saving and investment rates for Turkish economy over the period 1970- 2015 is analyzed by using ARDL bounds co-integration testing and Toda-Yamamoto causality testing approaches. The result from ARDL bounds testing procedure confirms that there are both long-run and short-run relationship between domestic saving and domestic investment whilst there is no causal relationship neither from saving to investment nor from investment to saving.