Dynamic Connectedness of International Crude Oil Prices: The Diebold–Yilmaz Approach

Xiaoyong Xiao, Jing‐Zhi HuangSustainability 2018 · DOI

TL;DR

This paper uses a novel network analysis framework to quantify how interconnected international crude oil prices are. It reveals that while WTI and Brent act as "price makers," influencing others, the overall market structure remains remarkably stable even through significant price shocks.

What Problem Does This Paper Address?

The international crude oil market is often debated: is it a unified "great pool" where events in one region quickly affect others, or is it fragmented? Understanding this interconnectedness is crucial for risk management, forecasting, and policy decisions. Existing studies often look at correlations or simple price movements, but lack a comprehensive framework to measure the dynamic and directional nature of these connections across multiple oil markets. This paper aims to fill that gap by quantifying how shocks in one oil market propagate to others and how these relationships evolve over time.

Methods

The core of this study is the application of the Diebold-Yilmaz connectedness framework. This approach, rooted in vector autoregression (VAR) models, quantifies connectedness by analyzing the decomposition of forecast error variances. Specifically, it measures:

  • Total Connectedness: An overall measure of how much shocks to one oil price explain the forecast error variance of others in the system.
  • Total Directional Connectedness: This breaks down into two parts:
  • "From Others": The proportion of a price's forecast error variance due to shocks from all other prices in the system.
  • "To Others": The proportion of other prices' forecast error variance that is due to shocks from a specific price.
  • Pairwise Directional Connectedness: The specific influence of one oil price on another.

To capture the dynamic aspect, the authors employ a rolling sample window (200 days) to analyze how these connectedness measures change over time, rather than just a single static snapshot. They also use network structure diagrams to visualize these relationships, making them more intuitive.

Datasets

The study utilizes daily international crude oil price data for six key benchmarks:

  • WTI (West Texas Intermediate)
  • Brent
  • Dubai
  • Daqing (China)
  • Minas (Indonesia)
  • Tapes (Malaysia)

The data spans from January 2, 2003, to August 17, 2018, resulting in 3351 daily observations for each price. The data was sourced from the Wind database.

Key Insights and Ideas

  • The "Great Pool" Hypothesis Holds: The study provides strong evidence that the international crude oil market functions as a highly interconnected system, supporting the "great pool" theory.
  • WTI and Brent as Market Leaders: These two benchmarks consistently emerge as "price makers," exerting a significant influence on other oil prices. Their shocks tend to propagate widely.
  • Price Takers Identified: Conversely, prices like Tapes, Daqing, Dubai, and Minas are identified as "price takers," meaning they are more influenced by external shocks than they are drivers of them.
  • Remarkable Structural Stability: Despite significant price volatility and market shocks (like the 2008 financial crisis), the fundamental structure of connectedness among these oil prices remains surprisingly consistent. The "network" of influence doesn't drastically change.
  • Visualizing Complexity: The use of network diagrams is a powerful tool for understanding complex interdependencies, moving beyond simple correlation matrices.

Key Results

  • Total Connectedness: The overall connectedness of the six international crude oil prices is found to be 67.3% using the full sample.
  • Directional Influence:
    • Tapes, Daqing, Dubai, and Minas are highly affected by other prices (receiving significant volatility shocks).
  • WTI and Brent have a greater impact on other prices, contributing more to their forecast error variances.
  • Pairwise Dominance: The pairwise directional connectedness between Brent and WTI is particularly high, indicating a strong, direct link between these two major benchmarks.
  • Dynamic Fluctuations: While the overall structure is stable, the degree of connectedness fluctuates over time, with four distinct peaks observed during the sample period, often coinciding with major global economic events.

Limitations and Future Work

  • Economic Mechanisms: The paper acknowledges that while it identifies how prices are connected, it doesn't deeply explore the economic mechanisms driving these connections (e.g., supply chain dynamics, geopolitical factors). This is identified as an area for future research.
  • Data Granularity: The study uses daily data, which is considered high-frequency for this type of analysis. However, the authors note that even higher frequency data (if available) could potentially reveal more nuanced relationships.
  • Limited Sample: While the six chosen prices are representative, they are still a subset of the global oil market. Future work could expand the analysis to include more regional or specialized crude oil prices.

Summary updated

Abstract

Connectedness is the key to modern risk measurement and management. This study investigates the international connectedness of crude oil prices and explores its time-varying characteristics based on a connectedness measurement framework using daily international crude oil prices. The international connectedness of crude oil prices is investigated from three perspectives: total connectedness, total directional connectedness, and pairwise directional connectedness. We find that the total connectedness of crude oil prices is 67.3%. We also find that the crude oil prices of Tapes, Daqing, Dubai and Minas are highly affected by Brent and WTI (West Texas Intermediate) crude oil prices. Furthermore, WTI and Brent are the price makers of international crude oil prices, while Tapes, Daqing, Dubai and Minas are price takers. From the perspective of pairwise directional connectedness, we find that the degree of pairwise directional connectedness between Brent and WTI are high. Finally, the structure of international crude oil markets stays the same even after market shocks. The main contributions of this study are identification of dynamic connectedness and presentation of the network connectedness of international crude oil prices.
BibTeX
@inproceedings{xiao2018dynamic,
  author = {Xiaoyong Xiao and Jing‐Zhi Huang},
  title = {{Dynamic Connectedness of International Crude Oil Prices: The Diebold–Yilmaz Approach}},
  booktitle = {Sustainability},
  year = {2018},
  doi = {10.3390/su10093298},
}
Figure