Fat Tails due to Variable Renewables and Insufficient Flexibility: Evidence from Germany.

AuthorHuisman, Ronald
  1. INTRODUCTION

    Electricity markets have experienced radical structural changes over the past few decades. During this period of time, many countries liberalised their electricity sector and set the path to the creation of competitive power markets. Besides that, most of these markets experienced drastic reforms during the energy transition, with the most prominent being the increasing penetration of energy supply from renewable energy sources (we refer to those with RES hereafter). The rapid and large-scale integration of intermittent RES, however, induces significant impact on power prices and substantially increases the demand for power system flexibility, as intermittent energy supply comprises non-controllable variability and partial predictability (Perez-Arriaga and Batlle, 2012; Kyritsis et al., 2017).

    Partial predictability is predominantly driven by the fact that weather-dependent RES do not perfectly adapt their output as a reaction to economic incentives, and therefore to the flexibility demand from the energy system. A better understanding of the impact that intermittent RES have on electricity prices are of great concern to managers, who must take better long-and short-term decisions in the operating on electricity markets, but also to policy makers who endeavour to adjust the electricity market design in order to increase power system flexibility, and thereby accelerate the reduction of emissions in the power sector.

    Focusing on the German electricity market, a prominent example of a market integrating variable energy supply from RES, Kyritsis et al. (2017) show that both solar and wind power generation have an impact on the probability distribution function of electricity prices by decreasing the average price which is--in other words--the a merit-order effect. According to Kyritsis et al. (2017), electricity prices decline when the share of RES in the power system increases. Wurzburg et al. (2013) discuss several studies, of which nine focus on the German electricity market, and all provide evidence for the merit-order effect. More recent studies that yield the same conclusion are, among others, Tveten et al. (2013), Ketterer (2014), Paraschiv et al. (2014), and Dillig et al. (2016).

    Tveten et al. (2013), Ketterer (2014), and Kyritsis et al. (2017) go one step further and examine how changes in intermittent renewable energy supply affect the volatility of electricity prices. In fact, Kyritsis et al. (2017) study both solar and wind power generation technologies and, considering the recent period of high renewable penetration, they show that solar and wind have a different impact on the volatility of electricity prices; while solar power generation reduces the volatility of electricity prices and the probability of electricity price spikes, wind power volatility increases electricity prices volatility and introduces electricity price spikes. The same relation between solar and the volatility of electricity prices manifests in Tveten et al. (2013) and between wind and electricity prices volatility in Ketterer (2014), during the first period of RES integration. A more recent study by Johnson and Oliver (2019) analyzes wind and solar supply together and also find that RES is increasing power price variance. Increased price variance induced by RES calls for more knowledge about managing energy price risk and valuing real options such as the option to store power in batteries and alternative power storage systems and the option to flexibly adjust consumption and supply to respond to changes in RES.

    The view from the literature is that power prices decline (ceteris paribus) is a result of an increase in RES and that the volatility of power prices changes (ceteris paribus) is a result of solar and wind energy supply variations. This view motivates us to further examine the impact of intermittent energy supply on the probability distribution of power prices. We question whether the increasing share of variable solar and wind power generation also affects the tails of power price distribution. The motivation for our research question becomes clear from Figure 1 which shows the development of the day-ahead average daily prices in [euro]/MWh in Germany from January 2010 to June 2015. In this figure, the typical characteristics of day-ahead power prices, such as mean reversion and extremely high and low prices, become apparent. Kyritsis et al. (2017) focus on how power price variation (volatility), being the second moment of a probability distribution function, is affected by changes in RES supply. As extreme prices, which are kurtosis events, influence the fourth moment of a probability distribution function, we think that these observations cannot be captured only by variation (or the second moment as it were). Therefore, the present study examines extreme power prices. (1)

    The expansion of variable or intermittent RES requires an increasing effort from the non-intermittent suppliers to counterbalance abrupt changes in production volumes. This may result in increased supply frictions, which become more prominent during periods of limited power system flexibility, in terms of adjusting the production volumes by the non-intermittent suppliers. Thus, the lower the flexibility of the power system, the higher the probability of extreme prices to occur. Hence, beyond the mean and variance of the electricity price distribution, the shape of the probability distribution in the tails is also driven by the penetration of RES into the power system.

    This reasoning relates the tail structure of electricity price distribution closely to power system flexibility, which is the key challenge towards the large-scale integration of RES. However, there is not a consensus view in the literature on the relation between intermittent wind and solar energy supply and the tails of the power price probability distribution. Limited evidence comes from studies that marginally touch on the link between extreme electricity prices and intermittent supply. For instance, Paraschiv et al. (2014) do not find conclusive evidence for the case of solar supply, but their results show that upward price spikes occur mostly when wind energy supply is low. By comparing the tail fatness of the empirical power price distributions between emerging and developed economies, LeBaron and Samanta (2005) show that one of the factors influencing the distribution of electricity prices is the different penetration level of intermittent renewable generators. From a similar point of view, Lindstrom and Regland (2012) study six European electricity markets through the employment of a regime switching model, and find a positive relation between the frequency of extreme price events and the penetration of renewable energy sources in the power system; hence, they provide evidence of renewable energy supply increasing the tail fatness of the electricity price distribution. In contrast, Keles et al. (2016) apply an AR-GARCH model on EPEX day-ahead market data and indicate that the tail fatness of the power price distribution is reduced over the period from 2008 to 2014. Although the authors do not make a strong claim, they suggest that their results are possibly driven by the increasing share of RES, and particularly wind, in the power generation mix.

    Kyritsis et al. (2017) demonstrate the different impact of wind and solar energy supply on power price variation and provide some main distributional properties of electricity prices related to price spikes, for different solar and wind power penetration levels. Those price spikes (being both extreme high and low prices) are not studied in particular. Extreme Value Theory (EVT) is a field within statistics that focuses on the probability structure of extreme observations only. As extreme high and low prices occur due to unexpected changes in supply from RES and the inflexibility of the power system to cope with these changes, prices behave different than when such changes do not occur. This motivates us to apply EVT as we believe that the probability distribution of extreme prices could not necessarily be caused by higher variance or volatility only. In this study, we therefore proceed a step further and investigate whether the results Kyritsis et al. (2017) found for volatility also hold with regard to the tail fatness of the power price distribution. Not only we look at the effect of solar and wind on the tails, but the main advantage is that we disentangle the effects of each of them on the left and right tail of the power price distribution. This paper contributes to the literature by extending Kyritsis et al. (2017). Using their data and methodology, we examine the impact that the penetration of intermittent RES in the German power supply mix has on both tails of the power price probability distribution, but also separately on the left and right tail.

    The distribution of electricity prices can significantly deviate from the normal distribution, and one needs to incorporate information about the tails to correctly model the shape of the distribution. The tail fatness of the electricity price distribution has direct implications for risk management, energy policy making in the sense that supply from RES in combination with insufficient flexible storage and / or production capacity and inelastic demand lead to extreme electricity prices, and for the real options valuation of flexible power suppliers for which price variation is a key-input variable.

    The remainder of the paper is structured as follows. Section 2 introduces our methodology. Section 3 discusses the data, and section 4 presents the empirical findings. Section 5 concludes.

  2. METHODOLOGY

    Motivated by the aforementioned discussion, we investigate the impact of energy supply from RES on the tail fatness of the empirical power price distribution. Due to price inelastic short-term demand and non-flexible storage and / or production capacity...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT