Competition in the Dutch Electricity Market: An Empirical Analysis over 2006-2011.

• Author: Mulder, Machiel

1. INTRODUCTION

   Electricity markets are widely seen as vulnerable to competition problems because of their characteristics: highly inelastic and volatile demand, entry barriers, economies of scale, transmission constraints, frequent market interaction among suppliers on spot as well as forward markets and the impossibility to store electricity while demand needs to be permanently equal to supply (Holm-berg and Newbery, 2010). These factors not only cause these markets to be tight from time to time, but they also complicate the development of competition. In this respect, the fundamental problem is that firms can have dominant positions which are difficult to change (Green, 2007). Firms defend these positions by means of mergers and acquisitions (Bergman, 2005; Jamasb and Pollitt, 2005; Percebois, 2008). Hence, one can doubt to what extent competition in this market can be really achieved (Twomey, Green, Neuhoff and Newbery, 2005). Indications of abuse of market power in US electricity wholesale markets were found by, amongst others, Sheffrin (2001), Borenstein, Bush-nell and Wolak (2002), Joskow and Kahn (2002) and Mansur (2008). Wolak (2007) concluded that the benefits of introducing competition are small compared to the benefits achieved in other network industries because of the exercise of unilateral market power. For several European power markets, London Economics (2007) found that electricity prices were related to the pivotal position of electricity producers. For the German electricity market, Musgens (2006) estimated that prices were on average about 50% above competitive levels in the period 2001-2003, while in 2011 the German competition authority concluded that the level of competition in the German wholesale market is still unsatisfactory (Bundeskartellamt, 2011).

   Some authors, however, find that wholesale markets have been fairly competitive. Green (2011) concluded for the British market that the evidence for strategic behavior by the power companies is weak and that the supply decisions are mainly set in a competitive manner. For the Nordic countries, Bergman (2005) found that through market integration and the introduction of forward markets, "workable" competition has been realized. Joskow (2006, p. 10) even stated that in the Northeast of the United States market-power mitigation measures have been too successful making it impossible for prices to rise to competitive levels when demand is high, resulting in aggregated profits below the levels needed to recoup fixed costs. Puller (2007) found that pricing behavior in the Californian spot markets during 1998-2000 was consistent with Cournot competition, where the large variation in price-cost margins mainly resulted from changes in the residual demand elasticity. So, despite the common view that market power is a fundamental characteristic of electricity markets, the degree to which prices are actually distorted appears to be an empirical issue.

   In this paper, we test whether the intensity of competition in the Dutch wholesale electricity market changed over the period 2006-2011. A number of years ago, serious concerns existed about the limited liquidity of this market and the ability of the incumbents to exert market power during peak hours (Van Damme, 2005). Since then, the Dutch market has become more integrated with the neighboring countries through both physical and virtual extensions of cross-border capacity. We test whether this has intensified competition.

   We estimate the weighted average Lerner index for each hour in 2006-2011. This index is defined as the day-ahead price minus the marginal costs per firm over this price while the weighing is based on the share of each firm in the total level of generation. The marginal costs per firm are based on actual plant-level data, using engineering-costs estimates. In addition, we assess the contributions of a number of explanatory variables to this Lerner index, in particular the HHI based on production by flexible generation plants and import, the level of domestic demand corrected for the supply from decentralized generation plants and a number of indicators for the influence from the German market. We find that, during super peak hours (defined as 10am to 7pm during working days), the annual average value of this index declined from 0.23 in 2006 to 0.03 in 2011, indicating more competition. Together with the increase in competition and the decrease in net demand, the annual operational profit per plant significantly reduced. We find indications that the increase in competition can be attributed to the extension of cross-border connections, a higher price elasticity of demand and more Bertrand-like competition.

   This paper differs from previous studies as it focuses on the development of competition over a relatively long period of time by using high-frequency data on plant level for a specific market, i.e. the Dutch market, where competitive conditions have changed substantially over the past few years. Our conclusions contribute to the debate about the efficiency of additional measures to improve competition in the electricity wholesale markets. From the Dutch experience, we learn that enhancing the role of decentralized generation as well as fostering integration of markets can be effective measures to promote competition in the electricity market.

   The remaining of this paper is structured as follows. Section 2 presents the theoretical underpinning of the analysis. Section 3 describes the main characteristics of the Dutch electricity market. Section 4 presents the empirical model. The data are described in Section 5, while the results are presented in Section 6. Section 7 concludes.

2. MODELLING MARKET POWER

   In studies that model market power in the electricity wholesale market, Cournot models are widely used (e.g. Borenstein et al., 2002; Joskow and Kahn, 2002; Musgens, 2006; Puller, 2007). Although some papers find that prices are consistent with the Cournot equilibrium (e.g. Puller, 2007), the results vary and standard Cournot models typically show prices that are too high while output is too low for realistic values of the price elasticity (Willems, Rumiantseva and Weigt, 2009). In contrast, Supply Function Equilibrium (SFE) models are considered to be more realistic for electricity markets. With these models, suppliers bid in supply functions for each generator instead of single quantities for a certain price. Although they typically present more realistic results compared to the Cournot models, they require extensive calculations with strong assumptions, while often generating multiple equilibria and giving unstable solutions. Comparing the results from Cournot models and SFE models, Willems et al. (2009) find that Cournot models can be easily calibrated with forward contracts to obtain realistic results. They therefore suggest using Cournot models for short-term analysis and SFE models for long-term analysis, as the latter are less sensitive to calibration and therefore particularly suitable to predict long-term effects.

   As our focus lies on the short-term, using hourly data, we assume Cournot competition in which market power is related to the degree of concentration. In a Cournot market, the relationship between competition and market structure can be described as LI = HHI/[epsilon], where LI is the weighted average Lerner index, HHI is the Herfindahl-Hirschman index and [epsilon] the (absolute) elasticity of market demand with respect to price (Motta, 2004). The Lerner index is broadly used as a measure for market power (Elzinga and Mills, 2011). The Lerner index measures the intensity of competition by the degree to which price exceeds marginal costs: L[I.sub.i] = (p-m[c.sub.i])/p, where m[c.sub.i] is the marginal costs of firm i and p is the market price.

   The relationship between the weighted average Lerner index and HHI is, however, not straightforward (Borenstein, Bushnell and Knittel, 1999; Willems et al., 2009). This holds in particular for electricity markets where market power strongly depends on the magnitude of demand, given the non-storability of electricity and the short-term inflexibility of supply. To control for this, one should measure the HHI on the basis of flexible, not yet contracted production (Holmberg et al., 2010). If the HHI is measured on flexible production, its impact on the Lerner index is directly related to the price elasticity of demand. The importance of the latter for assessing competition is confirmed by Puller (2007) as well as Giabardo, Zugno, Pinson and Madsen (2010), who both found that an increase in the price elasticity of demand decreases the possibility of power firms to exploit their market power.

   Competition in electricity markets is also related to the overall level of demand. The demand level may influence competition through a number of mechanisms. The Cournot model is in particular valid at higher demand levels when technical constraints must be considered by firms, while at lower demand levels competition is more Bertrand-like resulting in more competitive prices (Willems et al., 2009). Another mechanism by which the demand level affects the Lerner index is related to the fact that the marginal firm can raise its price up to the level of the marginal costs of the next plant in the merit order (Borenstein et al., 1999). As supply curves often become steeper at higher load levels, including the demand level in the analysis may capture this effect. Note that the steepness of the merit order may change over time, affecting the degree the electricity price is related to the marginal costs. The flatter the merit order, the smaller the distance between the price which firms require and their own marginal costs, suggesting a more Bertrand-like competition. In addition, the more capacity is used, the weaker the ability of the remaining capacity to respond to further increases in demand, resulting in prices (strongly)...