Differentiated Responses of Major Korean Ports to Macroeconomic Conditions: Container Throughput Forecasting and Management Strategies

Abstract

Maritime transport carried 99.7% of Korea's international cargo in 2023, and container throughput at Korean ports moves with economic conditions abroad. This study identifies the macroeconomic indicators that move throughput at Busan Port, Incheon Port, and Gwangyang Port, and carries those indicators into forecasting models built for each port. A common candidate set and selection procedure were applied to the three ports and to their combined total over 120 monthly observations from January 2013 to December 2022, so that differences in the estimates follow from the ports rather than from model specification. Significant variables entered ARIMAX, SARIMAX, LSTM, and Bi-LSTM models as exogenous regressors, trained on 84 months and tested on the 36 months from January 2020. Interaction and squared terms were estimated for the combined total. The base interest rate, the economic growth rate of China, and the SCFI carried negative coefficients for the combined total, Busan Port, and Incheon Port. At Gwangyang Port both Chinese growth and the SCFI carried positive coefficients, and the interest rate did not enter the final model. LSTM produced the lowest RMSE and MAPE at every port, with MAPE between 4.67% and 7.91%. The interaction between the interest rate and Chinese growth was negative, so throughput falls furthest when both run high. Because response directions differ across ports, a forecast aggregated across ports gives weak support for port-level planning. The findings were consolidated through a SWOT analysis of port conditions, and four scenarios over the two interacting indicators are proposed as a basis for differentiated response by port.