Market Power and Transmission Congestion in the Italian Electricity Market.

• Author: Bigerna, Simona

1. INTRODUCTION

   Electricity market reform is nearly two decades old in most industrialized countries. These reforms followed the pioneering revolutions that occurred at the beginning of the 1990s in the UK (1990), Norway (1991) and the USA (1992). In particular, there has been a progressive fading of monopolies and administered prices in Europe since Directive 96/92/EC. Market mechanisms have been introduced progressively, but there is still doubt that electricity markets have become competitive. The prior studies have analyzed the exercise of unilateral market power by generators by utilizing empirical evidence that demonstrates that markets are oligopolistic and primarily employs bid data (Borenstein et al. 2002, Wolak 2003, 2009, Sweeting 2007, Bollino and Polinori 2008, McRae and Wolak 2009, Bosco et al. 2012). However, the analysis in this literature has been limited to those cases in which the transmission network is similar to a single bus bar, i.e., when there is only one hourly market. The relationship between congestion and market power in electricity spot markets has been analyzed primarily on a theoretical basis (Cho and Kim 2007). Recently, Lee et al. (2011) have proposed a new approach to measure market power in presence of transmission constraint, while Karthikeyan et al. (2013) provided a comprehensive review of market power. The benefits of transmission expansion are well-described in Wolak (2012), but to date, there is no empirical analysis of the effect of line congestion on zonal price differentials using bid data. (1)

   The aim of this paper is to measure market power when different zonal prices arise in an electricity market as a result of congestion. To this end, we propose a novel approach to measuring market power empirically by explicitly disentangling this from the impact of network congestion on market structure, equilibrium prices, and operators behavior. In detail, we provide the correct formula for calculating the zonal Lerner Index (LI), by explicitly considering the transmission congestion problem. We propose to modify the residual demand computation (as originally proposed in Wolak 2003, 2009), considering explicitly line congestion and consequently, two different types of offer bids. The first are offer bids refused by the Transmission System Operator (TSO), (2) even if they are more competitive than the marginal bid. The second are the offer bids accepted by the TSO, even if they are more expensive than the marginal bid. This approach enables us to compute a correct measure of zonal market power.

   We investigate the initial period (April 2004 - December 2007) of the day-ahead market of the Italian Power Exchange (IPEX). This period is an interesting one to study, characterized by two main features: (i) the transmission network structure was relatively stable; (ii) the renewable energy sources (RES) share was negligible. Indeed the RES boom in Italy started in 2008 to 2009. Therefore, we study a period of stable generation and transmission network structure, which can be interesting also for developing electricity markets in other regions of the world. In fact, in countries such as Argentina, Colombia, Venezuela, Thailand, the Philippines, and Taiwan and most oil producers, the share of RES in electricity generation is currently approximately 2 or 3% (excluding hydro). In other words, analysis of the Italian market at an early stage of the liberalization process can shed light on the critical issues that those countries could face in the future, if they overlook the congestion issue.

   This paper is organized as follows. Section 2 describes the main features of the IPEX and market segmentation. Section 3 describes the new methodological approach to measure market power at the zonal level with a transmission congestion and data construction procedure. Section 4 discusses the empirical analysis and results. Section 5 summarizes our primary findings. The method details and statistical information are provided in the Appendices.

2. THE ITALIAN ELECTRICITY MARKET

   The IPEX is one of the last pool markets created in Europe after Directive 96/92/EC for energy sector liberalization. The IPEX is organized as a day-ahead, adjustment and dispatching resource market, similar to that of other countries (Newbery, 2005), and started operation in April 2004 (Law 79/99, Law 240/04 and other Decrees). (3)

   We focus on the hourly spot market, (4) which yields a System Marginal Price (SMP). When there is congestion, market segmentation arises, and the consequence is that different zonal SMP are determined. To appreciate the importance of the congestion effect on the market in the period analyzed, note that there are a total of 32,880 hours and a total of 77,004 market zones and consequently 77,004 different market equilibrium prices to be considered. The TSO determines congestion as intended flows in excess of the physical transmission capacity based on supply and demand bids. Consumers pay a unique SMP on the demand side, which is computed as a weighted average of the zonal SMP on the supply side.

   In the period analyzed, the Italian transmission network characteristics remained unchanged compared to the pre-liberalization period. Investment in new lines started with a period of delay, and the network management began dispatching intermittent renewable energy sources in 2008. Therefore, in this period, market liberalization developed under the constraint of the old network arrangement, generating line congestion due to the physical line capacity limits coupled with an approximately stable generation capacity. The main features of the market in this period can be summarized as follows. (i) In 2004, only suppliers participated in the market, and demand was inelastically represented by the TSO. (5) (ii) In January 2005, active demand bids entered the market, and the TSO was able to overrule bids if the total market demand was "too different" from the independent day-ahead forecasts used for security management. (iii) In January 2005 the Single Buyer, in Italian Acquirente Unico (AU), was empowered to aggregate non-eligible and poor customer demand (6) and was instructed to utilize contracts for differences extensively and to buy into the market; as a result, market liquidity rose over 60%. (iv) The Italian Energy Authority enacted several pro-competitive market surveillance mechanisms aimed at discouraging producer quantity withholding strategies. The Energy Authority also regulates market data disclosure. (v) The RES has had dispatching priority since 2008, and it started to inject relevant flows into the network with the new feed-in tariff mechanism. In this period, the generation capacity in Italy has been quite concentrated and over two-thirds of this capacity is fueled by oil and natural gas (Table 1). ENEL, the former State-owned monopolist, had a market share of 48-49% in 2004 and of approximately 30% in 2007. The other main generators (newcomers, which have bought capacity from ENEL according to the liberalization rules) initially had a share of another 25% of production in 2004, which gradually rose to 33% in 2007.

   The hourly demand pattern is typically bimodal, indicating the increasing similarity between summer and winter peak demand (above 50,000 MW in 2007), but the primary feature has been that the household and industrial user prices were higher in Italy than in the rest of Europe. Thus, the IPEX differs from other European electricity markets in terms of prices, market liquidity, fuel mix, incentive mechanisms and market segmentation or configuration because of congestion problems (as discussed in detail below). All things considered, the German market appears to be the most similar to the IPEX. In particular, the prices paid for low voltage usage (e.g., contracts in the 2.5-5 MW range) are similar in Italy and Germany (Bigerna and Polinori 2014)

   The Italian market is divided into seven physical national zones (PNZ) (Table 2, Panel 1): North Italy (N), Center-North Italy (Cn), Center-South Italy (Cs), South Italy (S), Calabria (Cal), Sicilia (Si, Sicily in English) and Sardegna (Sa, Sardinia in English). (7)

   When there is congestion, the IPEX is segmented in a variable number of zones (Nz), (1[less than or equal to]Nz[less than or equal to]7), and we name this configuration the "Nz-market". Each configuration includes Nz "markets," each of which is generally characterized by a different SMP. Obviously, these markets can be constituted by a single PNZ or by many interconnected PNZs. In the second case, we label "elementary zone" each PNZ belonging to that market (Table 2, Panel 2). For example, if Nz is equal to three, then there are several three-market configurations. In one case, the three markets can be ML, Si and Sa. In this configuration, Si and Sa are both markets and PNZ, but ML is a market made up by the interconnected elementary zones of N, Cn, Cs and S. In another case, the three-market configuration can be N, IeNSi and Si, and so on. IPEX segmentation due to congestion resulted in 77,004 hourly markets in the period considered (Table 3), which yielded an average of 2.34 markets per hour (as there were 32,880 hours in the period).

   The hourly number of markets varies throughout the period. One-market (ITA) occurred for 5,989 hours (18.2% of the hours in the period). Five or more markets occurred for only 110 hours (0.33% of the total). The most common configurations are two- and three-market, which together occurred for 23,540 hours (71.59% of the total), and the two most frequent configurations are Si, MLSa, which occurred for 7,765 hours (23.65% of the total), and Si, Sa, ML, which occurred for 4,260 hours (12.96% of the total). Not surprisingly, the islands are often separated from the rest of the network: Sa was a separate market for 11,618 hours (35.3% of the total), and Si was a separate market for 18,889 hours (57.4% of the total). At other times, one of the...