Box rates unveiled: Predictive analytics for ocean freight rates with system dynamics and text mining under supply chain disruptions

Ocean freight rates are key drivers of supply chain costs, global inflation, trade and economic growth, and their volatile nature, particularly manifest during supply chain disruptions, causes significant supply chain and economic turbulence. This study presents a parameterized system dynamics model to predict the Shanghai Containerized Freight Index (SCFI), incorporating both linear and nonlinear interrelationships among transport supply, demand, market sentiment, and freight rates. The model comprises four integrated components: (1) a transport demand estimator based on past container shipping volumes, Gross Domestic Product (GDP), and the Purchasing Managers' Index (PMI); (2) an actual containership capacity model that considers shipping liners’ strategic and tactical capacity decisions; (3) a novel Market Sentiment Index (MSI) using lexicon-based text mining of news articles, integrated with PMI to quantify market sentiment; and (4) an SCFI prediction model that captures the bidirectional feedback between these components. Results indicate that our system dynamics model outperforms established methods such as XGBoost and ARIMA. Furthermore, we find that freight rates are sensitive to the balance between supply, demand, and market sentiment. Specifically, an oversupply of capacity and declining demand reduce freight rates, whereas capacity constraints caused by supply chain disruptions increase rates. Sensitivity analysis further demonstrates that strategic capacity adjustments, particularly through blank (cancelled) sailings, can effectively increase freight rates. These insights have important implications for strategic, tactical, and operational decision-making within global supply chains. • Predictive analytics for ocean freight rates under supply chain disruptions. • Nonlinear and bidirectional relationships among transport demand, actual capacity supply, market sentiment, and freight rates are modelled using system dynamics. • Market sentiment in container shipping is modelled with text mining and market data. • Shipping liners adjust capacity mainly through blank (cancelled) sailings to influence freight rates. • System dynamics-based prediction method outperforms XGBoost and ARIMA.