Cointegration with nonlinear ARDL approach and an application

Linear unit root and cointegration tests are widely used in the literature of econometrics. However, linear models are not always sufficient in modelling economic variable. It is more appropriate to use nonlinear methods to investigate the asymmetric behavior of variables. In this thesis the aim and theme of the research topic are adressed in the introduction chapter. The linear unit root and cointegration tests are discussed in chapter two, right after introduction. In chapter three, the concepts of nonlinearity and linearity are introduced firstly, and preliminary tests are explained for nonlinearity. Next, nonlinear unit root and cointegration tests commonly used in the literature are discussed. The NARDL approach, which is classified as nonlinear cointegration tests, takes into account the short and long term asymmetric effects in addition to the advantages of the linear ARDL model. In the fourth chapter of the thesis, the validity of Fisher Hypothesis in Turkey is tested with the NARDL approach which has been newly introduced in literature. In the study, both the linear and nonlinear ARDL models are analyzed for comparison purposes. Both approaches find a long-run relationship between nominal interest rates and inflation rates. The linear ARDL approach assumes that response of nominal interest rates with positive and negative changes in the inflation rate will be the same in the short and long run. But, the analysis with the NARDL approach shows that nominal interest rates react differently to the positive and negative changes in inflation.