How Does Oil Future Price Imply Bunker Price—Cointegration and Prediction Analysis

Yanhui Chen, Jinrong Lu, Mengmeng MaEnergies 2022 · DOI

TL;DR

This paper explores the relationship between oil futures and bunker fuel prices, finding a significant long-run cointegration. A Vector Error Correction Model (VECM) leveraging this relationship outperforms other forecasting models, suggesting oil futures are valuable predictors of bunker prices, crucial for shipping companies managing costs and risks.

What Problem Does This Paper Address?

Marine transportation is a cornerstone of global trade, and bunker fuel costs represent a substantial portion (30-70%) of shipping companies' operating expenses. Fluctuations in bunker prices introduce significant financial risk. While bunker prices are linked to crude oil prices, they exhibit their own volatility due to supply and demand dynamics. Shipping companies need accurate methods to forecast bunker prices to manage these risks, optimize refueling strategies, and control costs. Existing research has largely focused on bunker price management or the relationship between oil futures and spot prices, leaving a gap in understanding how oil futures specifically imply bunker prices and how this relationship can be leveraged for prediction.

Methods

The study employs a multi-faceted econometric approach:

  • Cointegration Analysis: This is the core technique used to establish a long-run equilibrium relationship between bunker fuel prices (specifically MGO - Marine Gas Oil) and crude oil future prices. The Johansen cointegration test, utilizing trace and maximum eigenvalue statistics, is applied to confirm this relationship.
  • Vector Error Correction Model (VECM): Once cointegration is established, a VECM is constructed. This model is specifically designed to capture both the short-run dynamics and the long-run equilibrium adjustments between cointegrated time series. The error correction term in the VECM quantifies the speed at which the series revert to their long-run relationship.
  • Comparative Forecasting Models: To validate the VECM's effectiveness, the authors compare its predictive performance against more standard time-series models:
  • ARMA (Autoregressive Moving Average): A basic time-series model.
  • ARMAX (ARMA with Exogenous Variables): Extends ARMA by incorporating external factors, in this case, lagged crude oil future price returns.
  • VAR (Vector Autoregression): Models the interdependencies between multiple time series by including lagged values of all variables in the system.
  • Prediction Evaluation: Out-of-sample forecasting is performed for one-step-ahead and four-step-ahead horizons. Performance is evaluated using Root Mean Square Error (RMSE) and Theil's Inequality Coefficient (Theil's U), which measure forecast accuracy and compare it against a naive prediction.
  • Impulse Response Analysis: This technique is used to trace the effect of a shock in one variable (e.g., crude oil futures) on another (e.g., bunker prices) over time, providing insights into the dynamic transmission mechanisms.

Datasets

The study utilizes weekly price data for:

  • Bunker Fuel (MGO): From three major ports: Rotterdam, Singapore, and Fujairah.
  • Crude Oil Futures: Two contracts from NYMEX (New York Mercantile Exchange): Contract 1 and Contract 3.

Data spans from December 30, 2005, to November 26, 2021. The dataset is split into an in-sample period (December 30, 2005 – February 22, 2019) for model estimation and an out-of-sample period (March 1, 2019 – November 26, 2021) for forecasting evaluation.

Key Insights and Ideas

  • Oil Futures as a Leading Indicator for Bunker Prices: The core finding is that oil futures prices are not just correlated with bunker prices but exhibit a statistically significant long-run cointegration. This implies that oil futures prices contain valuable information that can predict future bunker price movements.
  • VECM's Superiority in Capturing Long-Run Dynamics: The VECM, by explicitly incorporating the cointegration relationship, consistently outperforms simpler models like ARMA, ARMAX, and VAR in predicting bunker prices. This highlights the importance of accounting for long-term equilibrium in financial time-series forecasting.
  • The Error Correction Mechanism: The negative and significant error correction term in the VECM indicates that deviations from the long-run equilibrium between oil futures and bunker prices are corrected over time, with the error correction term pulling down bunker returns. This suggests a stabilizing influence.
  • Contract Specificity Matters: The predictive power varies slightly between the two crude oil future contracts examined. Contract 1 shows better one-step-ahead forecasting, while Contract 3 offers a marginal improvement in four-step-ahead forecasts. This suggests that the choice of future contract can influence prediction accuracy.

Key Results

  • Cointegration Confirmed: The Johansen cointegration test strongly rejects the null hypothesis of no cointegration, confirming a stable long-run relationship between bunker fuel prices (MGO) and both crude oil future contracts (Contract 1 and 3) across all three ports.
  • VECM Outperforms Other Models: In out-of-sample predictions, the VECM consistently achieved lower RMSE and Theil's U values compared to ARMA, ARMAX, and VAR models for both one-step-ahead and four-step-ahead forecasts.
  • Contract 1 Dominates Short-Term: For one-step-ahead forecasts, the VECM using Crude Oil Future Contract 1 generally showed better prediction performance (lower RMSE/Theil's U) across the ports.
  • Contract 3 Slightly Better for Longer-Term: For four-step-ahead forecasts, Crude Oil Future Contract 3 showed a slight improvement over Contract 1 in most cases, though Contract 1 remained superior for Rotterdam.
  • Robustness: The conclusions regarding VECM's superiority and the general findings were found to be robust when the forecast period was shortened.
  • Impulse Response: Shocks to crude oil futures prices do have a discernible impact on bunker prices, with varying degrees of positive and negative feedback over different time horizons, eventually converging to a steady state.

Limitations and Future Work

  • VLSFO Data Scarcity: The paper notes that data for Very Low Sulfur Fuel Oil (VLSFO) is limited, preventing a similar in-depth cointegration analysis for this increasingly important bunker fuel type. The authors suggest that if sufficient VLSFO data becomes available, a similar cointegration analysis could be performed to forecast its prices.
  • Contract Choice Nuances: While VECM is superior, the slight differences in performance between Contract 1 and Contract 3 for different forecast horizons suggest that a combined approach or further investigation into contract characteristics might yield even better results.
  • Model Simplification: The study focuses on specific models. Future work could explore more complex machine learning or hybrid models to potentially enhance prediction accuracy.
  • Exogenous Factors: While ARMAX included one exogenous variable, future research could incorporate a broader set of macroeconomic or geopolitical factors that influence both oil and bunker prices.

Summary updated

Abstract

This paper investigates how oil’s future price implies the bunker price through cointegration analysis first. A cointegration test confirms the long-run equilibrium condition of bunker and oil future prices. Based on the cointegration relationship, we construct VECM model to forecast bunker prices. In addition, we also consider ARMA, ARMAX, and VAR models for certifying whether considering the long-run equilibrium between bunker and oil future prices is helpful in prediction. One-step-ahead and four-step-ahead forecasting are considered and two out-of-sample datasets are used. The empirical results show that the increase in the value of the error correction term in the VECM model has the effect of pulling down the bunker return. VECM performs better than other models in prediction. The Crude Oil Future Contract 1 has better forecasting performance for bunker prices with VECM in the 1-step-ahead forecast, while Crude Oil Future Contract 3 performs slightly better than Crude Oil Future Contract 1 in the 4-step-ahead forecast.
BibTeX
@inproceedings{chen2022how,
  author = {Yanhui Chen and Jinrong Lu and Mengmeng Ma},
  title = {{How Does Oil Future Price Imply Bunker Price—Cointegration and Prediction Analysis}},
  booktitle = {Energies},
  year = {2022},
  doi = {10.3390/en15103630},
}
Figure