Short-term forecasting of seafood exports: a hybrid approach for strategic trade planning

Abstract

In this study, a short-term forecasting model for seafood exports is developed by integrating econometric and deep-learning methods. Using Korea's monthly data from January 2000 to December 2023, we identified five key predictors-export price, won-yen exchange rate, Brent oil price, real gross domestic product (GDP) per capita, and seafood production-through a systematic feature selection process. Dynamic regression confirmed their significant effects on export volumes, while long short-term memory (LSTM) and gated recurrent unit (GRU) models produced accurate forecasts for January 2022 through to December 2023. The results highlight product-specific dynamics: seaweed snack exports are highly sensitive to global income and demand, reflecting their income-elastic nature, whereas tuna exports are mainly shaped by production capacity and relative price competitiveness. By simultaneously identifying key export determinants and generating forward-looking forecasts, this framework combines interpretability with predictive accuracy, offering practical implications for tailored trade strategies, proactive risk management, and sustainable policy planning in volatile global seafood markets.