Riding the Nordic German Power-Spread: The Einar Aas Experiment.

• Author: Ewald, Christian-Oliver

1. INTRODUCTION

   In 2018 the Norwegian trader Einar Aas has had a fantastic run on simultaneously trading electricity futures on the Nordic and German markets. He seemed to effortlessly exploit differentials in the risk premia in Nordic and German electricity futures, hoping that over time, due to better integration of the two markets, these differentials would diminish. Even though the exact nature of his trades is unknown, at least to the authors, his basic idea was to enter the market with a long-short position in Nordic vs. German electricity futures when the futures price spread, and therefore the spread in the risk premia, was high, and then wait for the price spread to diminish and close out the position at a large profit. (1) In principle electricity can be carried from the Nordic to the German region and vice versa, as the power grids are well connected. As one would expect that the two markets become more integrated in the future, any large spreads should decline naturally. However, Einar Aas crashed out of the market spectacularly in September 2018, as he was unable to cover a margin call when the spread unexpectedly opened up again. This created headlines in the financial press and Norwegian newspapers. Essentially a wet weather forecast for Norway in combination with rising emission prices brought his strategy to a fall. In retrospect, if it would not have been for this dramatic price change, Einar Aas would have walked away some hundreds of millions euros richer. The spread turned out to narrow again.

   The case of Einar Aas and his downfall has been prominently covered in the financial and non-financial press. However it is not the only case where energy traders failed spectacularly. The following table provides a summary of cases within the last 20 years. The case of Einar Aas' has strong resemblance with the case of Amaranth, which at a net loss of over 8 billion USD tops this list. (2)

   Further, as for the trading of electricity contracts, it is also well known that a number of other traders have adopted strategies exploiting price differentials, not only between the Nordic and German markets, but also between French and German Markets, often in combination with the purchase of transmission capacity in auction. (3) The question investigated in this article is whether price spreads on the Nordic vs. German electricity futures markets provide systematic opportunities for excessive profits. More precisely, we examine the Nordic and German electricity futures markets for the presence of any pricing anomalies that could give reason to believe that arbitrage-like strategies as the one adopted by Einar Aas are possible. The literature presents some tests for statistical arbitrage, e.g. Hogan et al. (2004) and Jarrow et al. (2012), however their assumptions are very restrictive and in many cases do not apply. We follow a different path and first assess the two markets as to whether there is in fact a systematic difference in risk premia and/or Sharpe ratios, the first signs of possible arbitrage. We then follow up on this and construct an arbitrage-seeking strategy directly. More precisely, our strategy "beats" the market in the sense of Jensen (1968). (4) In fact we present an explicit strategy and test whether this chosen strategy can produce positive alphas in the context of the Capital Asset Pricing Model (CAPM), when trading simultaneously in the Nordic and German markets. We call this the Einar Aas experiment.

   We look at time series of Nordic and German electricity futures prices between 2010 and 2018 and estimate risk premia and Sharpe values in various ways. The data are daily data for all monthly Nordic and German electricity futures extracted from the Refinitiv database. We conclude that there is a significant difference, which shows more clearly at different periods during the time span investigated.

   There are a number of studies that have previously examined the presence of risk premia in electricity futures prices traded in the Nordic and German financial power markets. For some examples see Pietz (2009), Botterud et al. (2010), Viehman (2011), Haugom et al. (2014), Weron and Zator (2014) and Haugom et al. (2018). The general conclusions in most of these studies are that there exist risk premia which are related to the term structure, seasonality, or both. Additionally, there are also some studies suggesting that risk premia are directly related to fundamental risk factors in the various markets. However, to our knowledge, no previous study has systematically examined how the risk premia in the two markets move together and whether it is possible to exploit this in trading.

   Does a difference in risk premia attached to the same commodity already point towards arbitrage? This question is discussed at the end of section 2 and the answer in general is no. In conclusion, while the intuition of Einar Aas' strategy is very plausible, it is not in contradiction with the efficient market hypotheses and the no-arbitrage principle per se.

   In section 3 we therefore go one step further and perform an experiment in which we test an intelligently chosen dynamic long-short strategy in the spirit of Einar Aas against the Capital Asset Pricing Model (CAPM), and show that this strategy is able to generate positive and significant alphas. The strategy we use originates from the rich literature on pairs trading, and more specifically is an adaption of the strategy derived in Ewald et al. (2019). Under realistic assumptions on the market mechanics, e.g. transaction costs, margins etc., this strategy delivers positive alphas for suitably chosen pairs of futures from the Nordic and German markets. Strategies with positive alphas would generally be considered to be breaking the equilibrium condition, hence this is a serious indicator for a market anomaly and violation of the efficient market hypotheses that should not be ignored. Our paper therefore also makes an important contribution to the question of informational efficiency of energy markets, which has also been discussed in Nick (2016). He looks at price formation and arbitrage efficiency between spot and futures markets in the European natural gas market, rather than two separate futures markets, which is the case in our paper, and does not involve the CAPM in his analysis.

   The remainder of this paper is organized as follows. Section 2 contains a brief review of risk premia in general and the role that futures contracts play in this context. Further it contains an empirical analysis concerned with the risk premia and Sharpe values at the two markets, followed by a discussion on how this connects to no-arbitrage. In section 3 we present our experiment, showing that under realistic assumptions on the trading mechanism it is possible to create positive alphas by engaging in both markets simultaneously. Section 4 summarizes the main conclusions. Additional material relevant to the derivation of the strategy as well as its trading performance is presented in two Appendices A and B.

2. RISK PREMIA IN THE NORDIC VS. GERMAN MARKET

   2.1 Background

   A forward contract is an agreement to deliver or receive an asset at time T for a price specified at time t, where t

   The payoff from a long position in a forward contract (receive) at the time T when the contract matures (5) is given by

   [Please download the PDF to view the mathematical expression] (1)

   For the short position (deliver) it is the negative of (1). In equilibrium the value F(t,T) is determined as such that this contract has a value of zero at time t under a no-arbitrage assumption. This means that it is costless to enter a forward contract and both parties are indifferent about entering into the contract or not. The forward price F(t,T) therefore corresponds to the certainty equivalent and is given by

   [Please download the PDF to view the mathematical expression], (2)

   where r(s) is the instantaneous interest rate at time s, B(t,T) the price of a zero coupon bond at time /with maturity at time 7 and Q the market/pricing measure which reflects a risk neutral representative agent.

   While both parties to a forward contract will eliminate price risk, some risks remain, such as the risk of one party defaulting on their payment obligation at maturity, (1) resp. the negative of (1). Futures contracts eliminate the latter. They are set-up like forward contracts, but cleared on a daily basis, which means that the daily price difference

   [DELTA]F(s,T)=F(s+l,T)-F(s,T) (3)

   is paid to the holder of the long position and its negative to the holder of the short position at the end of each trading day, for any s [member of] [t,T] between entering and maturity or closing out. This difference reflects the change in the value of a forward contract, and the futures price is set in exactly such a way that it is costless to enter/withdraw at any time s [member of] [t,T], for both positions, long or short. In practice daily clearing is facilitated through a margin account, which in addition to facilitating the payments of (2) will require a collateral depending on the number of contracts entered to cover the risk associated with large intra-day movements of the underlying spot (S). The futures price can then be determined as

   [Please download the PDF to view the mathematical expression] (4)

   While futures contracts differ from forward contracts as outlined above, it can be seen from (2) and (4) that under the assumption of deterministic interest rates, futures and forward prices are the same. Looking at (4), the futures price can also be interpreted as a forecast of the asset price S(T), but under the measure Q. As the pricing measure Q reflects a risk neutral representative agent, the difference [Please download the PDF to view the mathematical expression] (5)

   where E,(S(T)) denotes the expectation under the real world measure, generally reflecting risk averse agents, can be identified as the risk-premium. (6)

   Using (4) we...