Revisiting the J-Curve Hypothesis in Turkey: A Nonlinear Autoregressive Distributed Lag Approach with Asymmetric Exchange Rate Effects

This study empirically investigates the validity of the J-curve hypothesis for Turkey over the period 1975-2024 by employing the Nonlinear Autoregressive Distributed Lag (NARDL) framework. The model incorporates trade balance (TB), Gross Domestic Product (GDP), foreign direct investment (FDI), and the official exchange rate to capture both short- and long-run asymmetries in Turkey’s external adjustment process. Unit root and Brock-Dechert-Scheinkman tests confirm the presence of mixed integration orders and nonlinear dependence among the variables, justifying the use of a nonlinear model. The results indicate a stable long-run cointegration relationship, with the Wald tests validating significant asymmetries in both the short and long run. Specifically, exchange rate depreciations improve Turkey’s TB, whereas appreciations have a stronger deteriorating impact, supporting the asymmetric J-curve hypothesis. GDP and FDI exert negative long-run effects on TB, reflecting import-led growth dynamics and investment-driven import demand. Diagnostic and stability tests confirm the robustness of the estimated NARDL model.