Volatility in shipping markets: a multi-perspective analysis using time series and deep learning models

Shipping markets are volatile in nature and surrounded by many uncertainties. Understanding such volatility helps stakeholders mitigate risks. This research aims to study the volatility in the shipping industry based on the patterns of changes in the three major cargo shipping sectors—container, dry bulk, and tanker. Variance analysis, time series models as well as long short-term memory (LSTM) deep learning models are applied to investigate the long-term trend of volatility over the 30 years from 1995 to 2024. The Port of Singapore is our theatre and volatility is investigated by means of cargo volumes, vessel arrival statistics and freight and charter rates. The results quantitatively reveal the volatile patterns of shipping markets and give a reference ranking in terms of the degree of volatility. The dry bulk shipping sector is the most volatile one, largely because of the highly volatile commodity and energy markets, followed by the tanker and container shipping sectors. Methodology-wise, the LSTM model is more efficient in predicting market change, compared to time series models, with lower error metrics value and higher adjusted R-squared. The implications of the research offer applicable solutions when facing risks and uncertainties, and address the necessity of developing automation and digitalisation in the industry.