Bilateral J-Curve Between Türkiye and Its Major Non-EU Trading Partners: Evidence from Both Linear and Non-Linear Approach

Abstract

This research analysed the bilateral J-curve phenomenon in the Turkish economy. For this purpose, we applied both the linear and non-linear Autoregressive Distributed Lag (NARDL) co-integration methods, in addition to the asymmetric Toda-Yamamoto causality test, to examine whether the impact of Turkish lira appreciations differs from that of lira depreciation. The findings from the linear model indicate statistically significant coefficients for the long term, and J-curve effects were observed in the case of two countries: Russia and the UAE. This suggests that when the Turkish lira depreciates, it positively affects Turkey's trade balance with these partners; however, lira appreciations have a negative impact. In contrast, the non-linear model provides more evidence, with the results revealing that asymmetry cannot be ignored, as the positive and negative variables exhibit differences in signs, magnitudes, and levels of significance. We found the J-curve effect for only three countries (India, USA, and UAE) out of seven partners in this model. Third, the lira evaluation between the short and long run affected the external balance. Furthermore, the long-run error correction mechanisms converge to steady-state equilibrium faster. Lastly, there is a unidirectional or bilateral linkage between the FX rate and the external deficit for these four partners. Therefore, exchange rate policies are a determinant that should be considered in relationships with certain trading partners.