A model on charter rate prediction in container shipping

Abstract

The maritime industry has witnessed numerous challenges in recent years after the global pandemic, primarily characterized by sharp fluctuations in the daily charter rates. Considering an unpredictable business environment, this study aims to suggest a financial forecasting model on charter rates, creating added value for the stakeholders of the maritime trade business. The empirical analysis utilized the data from the Clarksons Research Portal, which encompassed 7,409 charter rental transactions of container ships from 01.01.2018 to 10.03.2023. After examining seven different linear and ensemble regressions, it was revealed that the XGBoost regressor resulted in the least RMSE value of 0.1833 with an R2 of 0.9015. The selected predictors were the TEU, container fixture type, charter time, charter time multiplied by TEU, ship age, laycan year, and laycan month, respectively. In addition to coping with the limitations of linear regression, the model revealed that the laycan years, charter time, and charter time multiplied with TEU were the essential variables in charter rate prediction. As a result, the model developed in the study can be used as an important decision support tool for stakeholders in the container shipping industry.