Relation Between Wind and Electricity Prices in a Deregulated

Relation between wind and electricity prices in a deregulated

market: the case of Ireland

Laura Malaguzzi Valeri; Economic and Social Research Institute, Dublin and Trinity College Dublin

Valeria Di Cosmo; Economic and Social Research Institute, Dublin and Trinity College Dublin

Overview

We are interested in determining the effect of increasing wind generation (and investment in wind capacity) on the costs of the system as a whole. A fundamental part of this analysis requires the identification of the effect of wind on the wholesale price of electricity.

We start by analyzing historic data for the Irish Single Electricity Market (SEM) between 2008 and 2011. The SEM encompasses the electricity systems of both the Republic of Ireland and Northern Ireland, making it a unique cross-jurisdiction, cross-currency system. The SEM is a compulsory pool system, where plants bid their short-run marginal costs and are called to generate on the basis of the subsequent merit order.

The system marginal price (SMP) is the sum of the shadow price, determined by the marginal plant that is called to produce, and uplift. The uplift is the payment made to plants to avoid short-run losses and covers the cost of turning the plant on.

In this paper we study the effect of wind generation on the shadow price.

As wind increases, traditional thermal plants (operated by natural gas, coal or oil) are forced to change their generation more often to accommodate the fluctuating wind. This issue is known (Troy et al, 2010; Perez-Arriaga, 2012). Most analyses of the effect of wind generation on prices to date approach the problem by simulating the effects of increased generation on a given system. This paper differs from previous work by providing an econometric analysis of the historic effect of wind.

The Irish dataset is particularly well-suited to our analysis for several reasons: first, the island has limited interconnection with other systems allowing us to identify the effect of wind more easily. Second, it has experienced a large increase in installed wind capacity, more than doubling from about 900MW at the end of 2007 to more than 2000MW at the end of 2011. Third, it is a compulsory pool system and therefore the published data refer to almost all of the electricity traded in the SEM.

Methods

We create a dataset of hourly information on electricity generation, demand, plant availability and on daily costs for the last 4 years. Following Hardle and Truck (2010) and Huisman (2007) we estimate a system of seemingly unrelated regressions, one for each hour of the day:

Pi,d=αi+βihLi,dh+γiWi,d+jφij Fi,d-1j+μiCOd-1+θimari,d+jζs Dis+εi,d

The shadow price Pi,d in hour i of day d depends on: the load Li,d, which is allowed to have different effects depending on the intensity of demand h; the previous day's fuel prices F, where j indexes the fuel; the previous day's carbon dioxide permit prices CO, the generation margin mar; wind generation W; and finally a set of dummies D to account for the weekend and the months. There are N=24 equations in the system, one for every hour of the day.

Results

Preliminary results on hourly data suggest that there is a statistically significant negative effect of wind on the shadow price. The following table presents results for select hours: 4 p.m. to 7 p.m. in the evening, and midnight.

Wind generation has a consistently negative effect on the shadow price. In order to compare our results to papers that do not analyse the relation separately for each hour, we calculate the weighted average of the wind coefficient, with weights given by the loads of the corresponding hour. Our coefficient is equal to -0.004, which can be interpreted as saying that every 100MW increase in wind generation (equal to about 25% of the average wind generation in our sample) will lead to a decrease of the shadow price equal to 0.4 €/MWh, or about 0.8% of its average value in our sample.

Conclusions

In this paper we have analysed how wind generation influences the electricity shadow price in the Irish Single Electricity Market.

We control for other variables that can theoretically affect shadow price levels, including the level of demand, the prices of the main fuels (coal, gas, oil), the price of carbon dioxide permits, and the generation margin, that measures how much extra generation is available in each period, in addition to dummy variables for weekends and the different months of the year.

We estimate a system of hourly equations by FGLS, correcting for the presence of serial correlation and find a small but significant negative effect of wind generation on shadow prices. We also find positive and statistically significant effects of the main fuel prices on loads on the shadow price, as expected. The coefficient on the wind variable is equal to -0.004, which can be interpreted as saying that every 100MW increase in wind generation (equal to about 25% of the average wind generation in our sample) will lead to a decrease of the shadow price equal to 0.4 €/MWh, or about 0.8% of its average value in our sample.

At times when generation capacity is abundant, this dampening effect is likely to be captured by consumers. At times when there is need for more generation investment, however, the lower expected profits might act as a deterrent to potential investors.

References

Hardle, W. and S. Truck (2010) The dynamics of hourly electricity prices, SFB 649 Discussion Paper

Huisman, R.,C. Huurman and R. Mahieu (2007) Hourly electricity prices in day-ahead markets, Energy Economics, vol.29(2), 240-248

Perez-Arriaga, I. and C. Batlle (2012) Impacts of intermittent renewables on electricity generation system operation, Economics of Energy and Environmental Policy, vol. 1 (2), 3-17

Troy, N., E. Denny and M. O'Malley (2010) Base-load cycling on a system with significant wind penetration, IEEE Transactions on Power Systems, vol. 25 (2), 1088-1097