Each week, we select a recently published Open Access article to feature. This week’s article comes from the Journal of the Royal Statistical Society Series C (Applied Statistics) and presents a methodology for modelling weather derivatives with application to pricing wind power futures.
The article’s abstract is given below, with the full article available to read here.
Härdle, W.K., López Cabrera, B. and Melzer, A. (2021), Pricing wind power futures. J R Stat Soc Series C. https://doi.org/10.1111/rssc.12499
With increasing wind power (WP) penetration an extensive amount of volatile and weather dependent energy is fed into the German electricity system. To manage the volume risk of windless days and the transfer of revenue risk from wind turbine owners to investors, WP derivatives were introduced. These insurance-like securities allow the hedging of the volume risk of unstable WP production on exchanges such as NASDAQ and EEX. We present a modern and powerful methodology to model weather derivatives, with very skewed underlying assets, incorporating techniques from extreme event modelling to tune seasonal volatility. We compare transformed Gaussian and non-Gaussian CARMA(p, q) models. Our results indicate that the Gaussian CARMA(p, q) model is preferred over the non-Gaussian alternative. Out-of-sample backtesting results show good performance, with respect to benchmarks, employing smooth market price of risk (MPR) estimates based on NASDAQ weekly and monthly German WP futures prices. A seasonal MPR of a smile shape is observed, with slightly positive values in times of high volatility, for example, winter months, and negative values, in times of low volatility and production, for example, in summer months. We conclude that producers pay premiums to insure stable revenue steams, while investors pay premiums when weather risk is high.