Abstract
Energy commodity prices are inherently volatile, since they are determined by the volatile global demand and supply of fossil fuel extractions, which in the long-run will affect the observed climate patterns. Measuring the risk associated with energy price changes, therefore, ultimately provides us with an important tool to study the economic drivers of climate changes. This study examines the potential use of long-memory estimation methods in capturing such risk. In particular, we are interested in investigating the energy markets’ efficiency at the aggregated level, using a novel wavelet-based maximum likelihood estimator (waveMLE). We first compare the performance of various conventional estimators with this new method. Our simulated results show that waveMLE in general outperforms these previously well-established estimators. Additionally, we document that while energy returns realizations follow a white-noise and are generally independent, volatility processes exhibits a certain degree of long-range dependence.
Highlights
The price quoted for an asset reflects the present value of a future stream of expected earnings
We show later that the parameter H is known as the Hurst exponent, named after the hydrologist Hurst who first analyzed the presence and measurement of long-memory behavior in stochastic processes [1]
The application of the wavelet estimator on actual energy price time series is the topic of Section 5, where we find support for long-memory in the spot returns of major global fossil fuels
Summary
The price quoted for an asset reflects the present value of a future stream of expected earnings. Market imperfections cause prices to reflect information slowly, and sometimes the response to new information is dragged over a long period. This well-established empirical regularity, known as the long-range dependence of price observations, serves as the main theme of this paper. It is widely documented that the evolution of the risk of financial assets’ returns constitutes a long-memory stochastic process. To be specific, this type of process is defined with a real number H and a constant C such that the process’s autocorrelation is ρ(l) = Cl2H−2 as the lag parameter l → ∞.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
