Ersin, Ozgur Omer2024-03-132024-03-1320121300-610X1308-4658https://doi.org/10.3484/iif.2012.319.3320https://hdl.handle.net/20.500.12662/3857Analysis of Real Exchange Rate in Turkey with Nonlinear Models: Band-TAR and STAR Models In this study, we aimed to analyze the Purchasing Power Parity (PPP) relation in Turkey for the Post-February 2001 Crisis period. In order to test PPP, in addition to the traditional approach that utilizes the linear unit root tests, the methodology focuses on the nonlinear econometric real exchange rate models. According to the empirical results obtained, the real exchange rate is accepted to follow unit root both in light of linear ADF and PP tests, which suggest rejection of PPP The KSS nonlinear unit root test shows that nonlinear unit root behavior cannot be rejected. Additionally, linearity in real exchange rates is rejected by the Hansen, Tray, Mcleod-Li and Luukkonnen et.al linearity tests. As a result, the study aimed at conducting the analysis with nonlinear Obsfeld and Taylor Band-TAR, Taylor et.al. ESTAR and the LSTAR2 model which is selected by the LM type STAR model selection tests. If models to investigate PPP are compared with Diebold Mariano tests and MSE, MAE and RMSE error criteria for forecast accuracy the success increased accordingly once we moved from linear to the nonlinear models - from AR towards Band-TAR, ESTAR, and LSTAR2. The results suggest that, for the flexible exchange rate period adopted after the Feb. 2001 Crisis in Turkey, PPP holds only for the outer regimes that occur if the real exchange rate appreciation or depreciation is larger than the upper or lower thresholds.trinfo:eu-repo/semantics/closedAccessPurchasing Power ParityNonlinear Time SeriesTARSTARArticle10.3484/iif.2012.319.33201223198927WOS:000310183300005Q4