Modeling the effect of exchange rate on rice trade in China using finite mixture model with Bayesian method

With the rapid expansion of data availability, statistical modeling has become essential for analyzing intricate economic and trade relationships. Despite the increasing application of advanced statistical methods in international trade research, this study employing the Finite Mixture Model (FMM) with Bayesian estimation for time series data remain scarce. Given the heterogeneity and structural complexity inherent in trade data, FMM is particularly well-suited for capturing latent patterns and underlying market dynamics. Meanwhile, Bayesian method provides a robust framework for parameter inference, offering flexibility in handling model uncertainty and improving the reliability of estimates. This research investigates the impact of exchange rate fluctuations on China_s rice import and export by applying FMM fitted with the Bayesian method. Specifically, it identifies the presence of two components representing stable and volatile periods in the time series data, models their structural characteristics, and examines the relationship between exchange rate movements and rice trade. The findings indicate that rice export prices and import quantities exhibit a statistically significant negative correlation with the exchange rate, whereas import prices and export quantities show a statistically significant positive correlation with the exchange rate. These results highlight the asymmetric impact of exchange rate fluctuations on different aspects of rice trade, emphasizing the importance of accounting for multiple underlying structures within the data. By integrating FMM with Bayesian estimation, this research provides a nuanced understanding of the exchange rate-trade nexus, offering valuable insights for policymakers in designing trade and foreign exchange policies. The findings underscore the significance of adopting advanced statistical techniques to analyze heterogeneous economic data. Specifically, this study extends the literature by demonstrating the effectiveness of Bayesian FMM for heterogeneous agricultural trade data, contributing to both methodological advancements and empirical research on agricultural trade.