Shipping news sentiment as a predictor of iron ore freight rates: Hybrid evidence from lexicon-based analysis and threshold autoregression modelling

Accurate and timely predictions of shipping freight rates provide a crucial basis for market participants to enhance their decision making. To this end, this study proposes a hybrid approach that integrates language models (LMs) and lexicon-based methods to construct an innovative shipping sentiment index (SSI) for the iron ore shipping market. By mining 9034 iron ore-related shipping news headlines and their accompanying summaries from two major shipping websites covering the period from January 2011 to December 2023, the following main results are obtained: First, the SSI closely tracks the dynamics of iron ore freight rates on major routes, effectively capturing uncertainties triggered by unforeseen events such as the COVID-19 pandemic. Second, the threshold autoregression model reveals a nonlinear relationship between the SSI and iron ore freight rates, indicating that the positive correlation is more nuanced during relatively optimistic market conditions. Moreover, iron ore prices positively influence freight rates, with this effect being stronger in optimistic market conditions than in periods of pessimistic sentiment. Additionally, economic policy uncertainty in China has a positive impact on iron ore freight rates on the selected routes, regardless of the prevailing shipping sentiment. Third, when analyzed within the nonlinear threshold framework, the SSI effectively predicts iron ore freight rates in the out-of-sample forecasting analysis. These results have important implications for iron ore importers, enhancing their ability to manage production costs, ensuring financial stability, and increasing supply chain resilience. They also provide a sound basis for shipping market participants (e.g., shipowners, brokers, investors, and shipbuilders) to make informed decisions that can improve profitability.