This paper uses high-frequency spot price data from fourteen wholesale electricity markets in Europe to analyze asymmetric volatility in European day-ahead power markets with Exponential GARCH (E-GARCH) and TARCH models. Our data set ranges from 1992 to 2015 and consists of approximately 926,000 observations. As such, this paper constitutes the most extensive and comprehensive work conducted so far on European power markets, to the best of our knowledge. Unlike most of the literature that treats price as a continuous variable and attempts to model its trajectory, this paper adopts a unique approach and regards each hour in a day a separate market. The results show, in post-2008 period, the most expensive electricity is consumed in Turkey, Ireland, and UK while the cheapest power is in Russia, Nordic countries, and Czech Republic. Russia, Poland, and Czech Republic have the least volatile markets while France, Ireland, and Portugal have the most volatile ones. Volatility has decreased in many European countries in post-2008 period. Besides, we find magnitude effect is usually larger than the leverage effect, meaning that the absolute value of price change is relatively more important than the sign of the change (whether it is an increase or a decrease) to explain volatility in European day-ahead power markets. Moreover, the results imply there is not a uniform inverse leverage effect in electricity prices; that is, price increases are more destabilizing in some European markets (e.g. Poland, Slovenia, Ireland, Netherlands) than comparable price decreases but vice versa also holds true in some other countries (e.g. Portugal and France). Leverage (or inverse leverage) effect in post-2008 period is relatively stronger in Portugal, France, and Ireland, but its impact is quite limited in Turkey and Germany. Furthermore, although the impact of seasonality on prices is obvious, a specific pattern cannot be identified. Finally, large changes in the volatility will affect future volatilities for a relatively longer period of time in Nordic countries, Ireland, and the UK while changes in current volatility will have less effect on future volatilities in Czech Republic, Russia, and Turkey.
Energy Economics, 56(4), 398-409, http://dx.doi.org/10.1016/j.eneco.2016.04.002