Detecting Structural breakpoints in natural gas and electricity wholesale prices via Bayesian ensemble approach, in the era of energy prices turmoil of 2022 period: the cases of ten European markets

TL;DR

This study analyzes how critical geopolitical events, specifically the 2022 Russo-Ukrainian war, caused structural breakpoints in European natural gas, electricity, and financial markets using a Bayesian ensemble approach, revealing diverse market reactions.

Contribution

It introduces a Bayesian ensemble method, BEAST, for detecting structural breakpoints and analyzing market reactions to geopolitical shocks in energy and financial markets.

Findings

Markets reacted differently in timing and nature of breakpoints.

Some markets showed leading or lagging responses to critical events.

Market trends exhibited significant changes post-events.

Abstract

We investigate the impact of several critical events associated with the Russo Ukrainian war, started officially on 24 February 2022 with the Russian invasion of Ukraine, on ten European electricity markets, two natural gas markets (the European reference trading hub TTF and N.Y. NGNMX market) and how these markets interact to each other and with USDRUB exchange rate, a financial market. We analyze the reactions of these markets, manifested as breakpoints attributed to these critical events, and their interaction, by using a set of three tools that can shed light on different aspects of this complex situation. We combine the concepts of market efficiency, measured by quantifying the Efficient market hypothesis (EMH) via rolling Hurst exponent, with structural breakpoints occurred in the time series of gas, electricity and financial markets, the detection of which is possible by using a Bayesian ensemble approach, the Bayesian Estimator of Abrupt change, Seasonal change and Trend (BEAST), a powerful tool that can effectively detect structural breakpoints, trends, seasonalities and sudden abrupt changes in time series. The results show that the analyzed markets have exhibited different modes of reactions to the critical events, both in respect of number, nature, and time of occurrence (leading, lagging, concurrent with dates of critical events) of breakpoints as well as of the dynamic behavior of their trend components.