Probabilistic forecast-based procurement in seaborne forward freight markets under demand and price uncertainty

Abstract

Volatility in freight rates and shipping demand poses financial risks for charterers and ship owners. Freight forward agreements (FFAs) are popular hedging tools for fixing freight rates in advance by specifying the amount of cargo to be transported at the maturity period of the agreement. Procurement decisions with FFAs require assessing future freight rates and shipping demand. Accepting an FFA price offer higher than future FFA and spot prices or procuring a larger amount than the actual demand constitutes risks for charterers. We consider the freight procurement problem of a charterer, minimizing the total expected cost using FFA and spot markets under price and demand uncertainty. We show that a state-dependent base-stock policy is optimal with non-decreasing base-stock levels as the demand period approaches when price and demand forecasts are not updated. To determine base-stock levels, we propose probabilistic forecast-based policies with updated forecasts and an increasing base-stock level policy (IBP) adjusting base-stock levels based on the number of periods left until the demand period. The proposed methods are compared with benchmark methods using synthetic data covering different market conditions and real data from 14 bulk and tanker routes. The evaluation period covers pre-crisis (2016–2019) and during-crisis periods (2020–2023), considering major events after 2019, such as the COVID-19 pandemic and the Russia–Ukraine conflict, which led to high market volatility. Numerical evaluations show that policies based on probabilistic forecasts outperform those based on point forecasts. Utilizing probabilistic demand forecasts results in lower costs than probabilistic price forecasts. Experiments on the market data show that IBP results in the lowest cost on average while avoiding excessive procurement due to being in line with the optimal procurement policy. IBP outperforms probabilistic forecast-based policies due to forecast biases in the volatile freight market.