Market Makers and Liquidity Premium in Electricity Futures Markets.

AuthorPena, Juan Ignacio

    Electricity futures trading offers benefits to electricity producers, retailers, and consumers, such as price discovery (Steinert and Ziel, 2019), a hedge against spot price market risk (Dupuis et al., 2016), favors competitive strategies (Holmberg, 2011), and mitigates market power (Newbery, 1995). But these benefits come at a cost when the futures price diverges from expected spot prices, giving rise to a forward premium, defined as the difference between futures prices and expected spot prices (Bessembinder and Lemmon, 2002; Longstaff and Wang, 2004; Douglas and Popova, 2008; Cartea and Villaplana, 2008; Redl et al., 2009; Redl and Bunn, 2013; Bunn and Chen, 2013). If the futures price is higher (lower) than expected spot prices during the delivery period, this cost is borne by consumers (producers).

    Current literature has investigated the existence, or lack thereof, of the forward premium in electricity futures markets. Economic theory (e.g., Hirshleifer, 1990, Bessembinder and Lemmon, 2002) suggests that the forward premium should compensate risk-averse market participants for bearing systematic risk. Another source of the forward premium may be producers' market power (Ito and Reguant, 2016), which materializes even when agents are risk-neutral.

    The forward premium has been studied by comparing futures prices against expected spot prices. If we use realized (ex-post) spot prices, futures prices contain forecast errors that may induce bias in the estimated forward premium. Besides, random shocks to spot electricity prices are large compared to any premium in the futures price, so huge samples are needed to get tests with statistical power (Fama and French, 1987). If we use the estimated (model-based, ex-ante) spot prices, the forward premium becomes dependent on the spot price model used. There are many spot price models, and none enjoys general acceptance (1) (see the comparisons in Benth et al., 2012). Weron and Zator (2014) also point out that some divergences in the results found in the literature may arise from the varying terminology regarding the variable under study. The variables risk premium, forward premium, forward risk premium, and market price of risk are not uniquely defined and sometimes used as synonymous. (2) As an alternative to dealing with these problems, Fleten et al. (2015) argue that the forward premium got from daily returns of electricity futures prices is more informative than the forward premium based on ex-ante or ex-post spot prices.

    In this paper, we make several contributions to the existing literature. First, we study the extent to which the insights in Grossman and Miller's (1988) model may help understand whether the intuition in Fleten et al. (2015) makes sense from a formal economic model's vantage point. We conclude that the daily forward premium, as computed by Fleten et al. (2015), is mostly a liquidity premium. Second, the empirical application to futures electricity contracts traded in the French, German, Spanish, and Nordic electricity markets, suggest a negative premium in the first three markets and a null premium in the last one. The liquidity premium is negative (positive) when market makers sell (buy) futures contracts because an imbalance between producers' and retailers' hedging choices creates a demand for immediacy. The negative premium appears because retailers wanted to offload a higher amount of price risk than the producers. The premium decreases when the number of market makers increases. The size of this liquidity premium depends on the number of market makers, a proxy of the level of competition in the electricity futures market. Third, in the empirical application, this paper considers a more recent period from 2008 to 2017 and uses a panel data model with many controls, including fuel and carbon prices, stock index returns, the level of water reservoirs, and several moments of the distribution of spot prices.

    We organize the rest of this paper as follows. Section 2 reviews the literature. After describing the formal model in Section 3, we present data in Section 4. Section 5 discusses the empirical results. Section 6 concludes.


    As far as we know, there is no conclusive evidence on the sign, size, variability, and determinants of the forward premium in futures electricity markets, Weron (2014). Current literature has documented positive, negative, and zero premium. The empirical evidence suggests that the forward premium may vary throughout the hour of the day, among days of the week, between months or seasons, or over the year. Results differ from one market to another, within the same market over different periods, and whether the researchers use ex-ante or ex-post measures.

    The European market receiving more attention so far is the Nordic market, analyzed in Cartea and Villaplana (2008), who report that during the months of high (low) volatility of demand, forward prices are above (equal or below) expected spot prices. Therefore, they exhibit a positive (zero or negative) forward premium. Redl et al. (2009) report a positive forward premium, as do Botterud et al. (2010), which stresses the importance of using the water level in the Norwegian reservoirs as a fraction of total capacity as an explanatory variable of the premium. Lucia and Torro (2011) also report a positive forward premium, but Huisman and Kilic (2012) register a time-varying forward premium. Weron and Zator (2014) discuss the impact of the bias coming from the simultaneity problem in regressions with futures and spot prices and study the ex-post (realized) risk premium, defined as the negative of the forward premium. They report a negative average risk premium, and therefore a positive forward premium. However, its size and significance vary over time, depending on the reservoir levels.

    The second market attracting more attention is the German market, studied in Benth et al. (2008), Redl et al. (2009), Viehmann (2011), Benth et al. (2012), Fleten et al. (2015), and Valitov (2019) among others and the results suggest a positive, negative or even null forward premium, depending on the period and the methodology. Bunn and Chen (2013) and Cartea and Villaplana (2008) study the British market and document seasonal sign reversals. Huisman and Kilic (2012) focus on the Dutch market and report a time-varying risk premium, similar to the results found in the Spanish market in Capitan Herraiz and Rodriguez Monroy (2009) and Furio and Meneu (2010). On the US markets, evidence on time-dependent risk premium is provided using NYMEX contracts for delivery at the California-Oregon border in Shawky et al. (2003). Similar evidence on the PJM market can be found in Longstaff, and Wang (2004), Douglas and Popova (2008), Cartea and Villaplana (2008), and Haugom and Ullrich (2012); Hadsell and Shawky (2006) study the NYSO market and Borenstein et al. (2008) the California market. Bevin-McCrimmon et al. (2018) analyze the impact of liquidity on the forward premium in New Zealand and conclude that liquidity affects the premium premia only with long-dated futures. Some papers examine several markets jointly; for instance, Daskalakis et al. (2015) study the Nordic, French, and British markets and find that electricity risk premia are significantly related to the volatility of electricity spot prices, demand, and revenues, and the volatility of carbon price volatility.

    Perhaps one common problem with studies focusing on the forward premium using ex-post measures, making their results hard to interpret, is that, besides the forecasting errors included in futures prices, the ex-post forward premium measures compensation for risks that market makers need not bear. On studies using ex-ante measures, the need for assuming a model for computing expected spot prices complicates the interpretation of results because of a lack of consensus over such a model. As an alternative to dealing with these problems, Fleten et al. (2015) argue that the forward premium got from daily returns of electricity futures prices is more informative than the forward premium based on ex-ante or ex-post spot prices. They define the overnight return as the daily log return of daily futures closing prices at day t, PC(t) and t-1, PC(t-1) (i.e. log(PC(t) /PC(t-1)). But this is the standard definition of daily return. The financial literature calculates overnight return as the log of the ratio between the opening price at day t, PO (t) and the closing price at day t-1, PC (t-1) (i.e., log(PO (t) /PC (t-1)); Ronen 1998). The two definitions are equivalent only if PC (t) = PO (t), which is not generally the case. We hasten to add here that Fleten et al. (2015) call the forward premium "the risk premium." Still, their definitions suggest that the variable they are interested in is the difference between futures prices and expected spot prices, i.e., the forward premium (3). Describing a futures market populated by hedgers and market makers, Fleten et al. (2015) posit that a market maker commands a premium from hedgers on exposures she must carry and cannot hedge. The only exposure a market maker must hold is daily (close-to-close) exposure because longer-term exposures can be offset using suitable contracts. (4)


    This section summarizes the key elements of the model in Grossman and Miller (1988). (5) The model focuses on a world with three dates: 1, 2, and 3. At date, t = 1, an exogenous liquidity event happens, causing an imbalance of size i between the supply and demand of futures contracts. Market makers compensate for the imbalance by trading futures contracts at the current market price. They satisfy the demand for immediacy that the model assumes is the critical driver of market makers' trading activity. The market makers hold their positions until date t =2, in which another liquidity event happens, causing an imbalance of the same size but opposite sign -i. Again...

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