Arbitrage behaviour in the exchange rates of Taiwan and Japan: applying the smooth transition vector error correction model with GJR-GARCH and spillover volatility

Abstract

This study utilizes a Smooth Transition Vector Error Correction Model (STVECM) with Glosten–Jagannathan–Runkle-Generalized Autoregressive Conditional Heteroscedasticity (GJR-GARCH) and spillover volatility to investigate the changes in short-run return dynamics when a deviation from the comovement equilibrium between the prices of New Taiwan Dollar (NTD) and Japan Yen (JPY) exists. Empirical results demonstrated that both exchange rates have a nonlinear relationship with the comovement equilibrium, which was captured by the Logistic Smooth Transition Vector Error Correction Model with GJR-GARCH and Spillover Volatility (LSTVECM-GJR-GARCH-SV). Both exchange rates will quickly reverse to equilibrium when there are large deviations between the two exchange rates, implying that information traders obtain profit in two markets, and then both exchange rates continue to deviate from equilibrium when small deviations exist from comovement, owing to noise trader to expand mispricing. Moreover, small and large deviations may have obvious differences and adjustment speeds when reverting to equilibrium for large negative and large positive deviations. Thus, this study applied LSTVECM-GJR-GARCH-SV to elucidate arbitrage behaviour.