1 Intelligent Well Technology: Status and Opportunities for Developing Marginal Reserves SPE
SELECTING STATIC OLIGOPOLISTIC MODELS IN THE ITALIAN
WHOLESALE ELECTRICITY MARKET
Daniela Floro
Department of Economics, University of Warwick,
Overview
The liberalization process in the electricity market aims to promote competition in the potentially competitive activities(generation and wholesale market) and to enhance transparency and efficiency in the management of natural monopoly(dispatching). The establishment of the spot market for the purchase and sale of electricity aims to increase competition in the generation activity.
The Italian electricity market has been organized according to a zone market given limited transmission capacity across the country. Whenever electricity flows exceed the maximum inter-zone capacity, the national market is split into several zones until the exhaustion of the transmission limits. As a result different market clearing prices are defined.
In order to identify the degree of the competition and the behaviour of the former monopolist (ENEL), this paper proposes a study based on the comparison of two competing models and the selection of the one which fits the market better. The analysis takes into account the zone market structure analyzing the main macro-zones: North and South, which together count for around 90% of total electricity consumption and 75% of total production.
In each market we choose two competing models according to firm production and generation capacity. In the North market, we study the Stackelberg and the Cournot model. In the South, we analyze the Stackelberg and the Dominant model. The comparison of two oligopolistic models is due to the role of ENEL. Although the former monopolist holds a significant proportion of the generation capacity, 50% in the North and 63% in the South, its supply is not as high as expected. In the sample analyzed, ENEL production in the day-ahead market is on average 54% (2005) and 30% (2006) in the North market; and 67% (2005) and 63% (2006) in the South market. Further, in each market, another three firms supply more than 30% of the spot market demand.
The study is applied to hourly firm level-data collected in the winter and summer months of 2005 and 2006, taking into account two seasonal patterns according to weekday and weekends.
We find that the Northern market is better represented by the Stackelberg model in 2005 and by the Cournot model in 2006. By contrast, the Southern market is better described by the Stackelberg model in both periods.
Methods
For each model we first define the strategic player set and estimate the competitive fringe supply via the general method of moments. Assuming a perfect inelastic spot market demand and subtracting the estimated fringe supply, we define the residual demand of the strategic players.
Knowing the residual demand of the strategic players, the next step is the evaluation of the market equilibrium outcome according to the oligopolistic model chosen. In analyzing the Cournot outcome, we apply a grid search approach as in Borestein and Bushnell (1999). In the Stackelberg model, the equilibrium output is determined solving the game by backward induction and considering for which marginal power plant a firm does not have incentive to deviate from the equilibrium. We first derive the best reply functions of the followers and the leader and determine leader output taking into account both leader and follower marginal costs. Knowing the leader optimal output we then evaluate the output of a follower according to its and the other follower marginal costs. In the Dominant firm model, the equilibrium output is defined maximizing ENEL profit according to its residual demand.
In all the models, the marginal cost function is assumed constant up to firm zone capacity for each power plant type. In our model, the strategic supply is based on hydro and thermal sources. The marginal cost of the former is evaluated as the opportunity cost of the thermal plant that they replaces marginally in each period, whereas the one of the latter is evaluated taking into account the fuel costs, the variable operation and maintenance cost (O&M). Finally, the available capacity is decreased by the forced outage factor indicating the probability of the unit being unavailable at any given time.
The oligopolistic model simulations allow the determination of the equilibrium price, the strategic player total output and each firm output in every model.The selection of the model which fits the zone market better is based on a variation of the traditional R2, which is defined as one minus the ratio of the sum of squared errors over the sum of the squared actual values according to the both weekdays and weekends, for each zone market.
The analysis is applied to the hour 7 p.m. for two main reasons. On one hand, 7 p.m. constitutes a peak hour, but not very high, thus firm capacity constraints are never bounded. On the other hand, at 7 p.m. the North market is split from the rest of continental Italy most of the times justifying a zone market analysis.
Results
Competitive fringe supply estimation
In the North, we estimate a unique competitive fringe supply because in the simulated models the fringe does not change, what changes is the interaction between the strategic players according to the model analysed. In the South, we estimate two different competitive fringe supplies according to the definition of the simulated models. In the first case, we consider a Stackelberg model where the former monopolist is the leader and there are three main followers, whereas all the other firms have a marginal contribution to the zone production. In the second case, we consider a Dominant firm model with competitive fringe in which all the firms except the former monopolist are involved in the competitive fringe.
In general the results are consistent with economic intuition. In all the estimated models, the competitive fringe supply is more responsive to price changes during weekends than weekdays. The comparison of the slope coefficients shows that these are higher in the North than in the South models. Specifically, the slope coefficients are higher in the Dominant firm model than in the Stackelberg one.
Oligopolistic model selection results
The model selection test shows interesting results according to the period analysed. As regard to the Northern market we find that in 2005 the Stackelberg model is preferred to the competing one, whereas in 2006 the Cournot model fits the data better. In particular, in 2005the R2 evaluatedon price is 0.81 and 0.51 in the Stackelberg and Cournot model respectively. The comparison of the R2 on total strategic player supply shows very close results in the two models, 0.99 in the Stackelberg and 0.98 in the Cournot, to decide which model can be preferred. However, the test shows a clear preference for the Stackelberg model when applied to single firm supply. For example, the R2 evaluated on ENEL supply is 0.95 (Stackelberg) and only 0.69 (Cournot). Moreover, the R2 evaluated on the other strategic player output indicates that the Stackelberg model is better than the competing one. In 2006, the R2 evaluated on price, total strategic player supply and single firm supply shows a noticeable preference for the Cournot model. Specifically, the R2 evaluated on price is 0.92 (Cournot) and 0.82(Stackelberg); and on total strategic player supply is 0.98 (Cournot) and 0.96 (Stackelberg). However, the gap between the two model fits becomes higher applying the test to single firm supply. In particular, the R2 evaluated comparing ENEL actual and simulated values suggests a higher preference for the Cournot model (0.85) than for the Stackelberg (0.48). The same result holds taking into account the other strategic player supply. Results are slightly stronger during weekdays than weekends.
As regard to the South, the Stackelberg model is strictly preferred to the Dominant firm model in both years. In this case there are no significant differences in the periods analysed. The R2evaluated on the leader supply is around 0.89 in the former model, whereas in the latter is around 0.77.
Conclusions
The study hightlights a difference in the oligopolistic structure characterizing the two markets according to the period analysed. The Northern market records a change in the oligopolistic structure: in 2005 the market is better described by a Stackelberg model, whereas in 2006 there is a clear preference for the Cournot model. This change is due to the increase of both other strategic player and the competitive fringe supply. In particular, the analysis of firm market shares shows a decrease in the Northern concentration structure. Considering this result in term of efficiency we should point out that the Cournot model is less efficient than the Stackelberg model, because the price is higher and the quantity is lower in the former than in the latter model. However, in a liberalization point of view the significant decrease of the former monopolist supply can be considered as first step toward a more competitive spot market.
As regard to the South market the analysis shows that in both years the market is better described by the Stackelberg model The competition in the market could be increased improving the transmission capacity across the country. According to Borestein et al. (2000) limited transmission capacity has two main consequences. On one hand, most efficient generators can not serve the geographically distant consumer due to insufficient transmission capacity. On the other hand, congestions reduce the competitive pressure reinforcing the position of the former monopolist.
References
[1] Boffa F. and Pingali V. (2008), “Zonal Pricing in the Italian Electricity Market”. Working Paper n. 17. DiSSEUniversity of Macerata. Available at
[2] Borenstein S. and Bushnell J.B., (1999)"An Empirical Analysis of the Potential for Market Power in California's Electricity Industry." Journal of Industrial Economics, ,Vol. 47 (3):285-323
[3] Borenstein S., Bushnell J.B and Stoft S. (2000), “The Competitive Effects of Transmission Capacity in a Deregulated Electricity Industry.” The Rand Journal of Economics 31 (Summer): 294-325.
[4] Borestein S., Bushnell J.B. and F. Wolak (2002), “Measuring Market Inefficiences in California’s Restructured Wholesale Electricity Market”, American Economic Review, Vol. 92 (5): 1376-1405
[5] Bushnell J.B, Mansur E. T. and Saravia C. (2008), “Vertical Arrangements, Market Structure, and Competition: An Analysis of the Restructered U.S. Electricity Markets”, American Economic Review, vol. 98 (1): 237-266.
[6] GME (2004), “GME’s Electricity Market: Purposes, Organization and operation”, available at