Strategic Withholding through Production Failures.

• Author: Fogelberg, Sara

1. INTRODUCTION

- We decided the prices were too low... so we shut down.

- Excellent. Excellent.

- We pulled about 2,000 megs of the market.

- That's sweet.

- Everybody thought it was really exciting that we were gonna play some market power. That was fun!

Intercepted exchange between Reliant traders, June 2000, Weaver (2004)

A competitive and well-functioning market is one of the goals of modern, liberalized electricity markets. However, a commonly voiced concern has been that firms strategically reduce their generating capacity in order to increase the electricity price. Strategic withholding of electricity was, for example, observed during the electricity crisis in 2000-2001 in California, and has been determined to be one of the reasons why the crisis became so severe (Kwoka and Sabodash 2011, Weaver 2004). Theoretical studies have also shown how firms benefit from this behavior (Crampes and Creti 2005, Kwoka and Sabodash 2011). The studies of market power investigating the Nordic electricity market Nord Pool have so far been mostly inconclusive (Vassilopoulous 2003, Hjalmarsson 2000, Fridolfsson and Tangeras 2009). However, certain circumstances have been proven to enable exercising of market power like high demand, congestion in the market (Olsen et al. 2006; Mirza and Bergland 2015), and concentration of production (Amundsen and Bergman, 2002).

In this article we look at a previously under-researched method that electricity producers can use to withhold capacity in order to increase prices on the Nordic electricity market. We consider instances when generators shut down part of their production due to a failure, and we verify whether the decision to stop production and provide information about this failure depends on economic incentives rather than being the result of a technical problem. Market participants on Nord Pool are obliged to publicly inform about changes to consumption, generation or transmission that exceed 100MW and last longer than 60 minutes in so-called Urgent Market Messages (UMMs). We investigate whether spot prices on Nord Pool influence the probability of production failures being reported in UMMs. Production failures should not depend directly on prices, as failures should be irregular and difficult to foresee. The detection of a significant relationship between prices and market messages therefore indicates that market participants base decisions concerning reporting a failure not only on technical problems, but also on economic incentives.

We use text scraping methods to create a unique dataset containing information about UMMs released by market participants with information about unplanned reductions of production i.e. production failures. Our dataset permits us to examine how prices affect market participants' decisions about issuing failure messages and how these decisions vary by generator type. Since prices may directly influence how generators are operated an OLS regression would give biased results. In order to estimate a causal effect of prices on production failures we therefore use a linear two-step model and instrument for prices using temperature and precipitation levels. Temperature was chosen as an instrument because of its exogenous nature and because prices on the Nordic electricity market are highly correlated with temperature. Especially during the cold season, when electricity is used for heating, temperature and demand follow each other closely. Additionally, since Nord Pool is largely based on hydro generation we instrument prices with precipitation levels, since during a "wet" ("dry") year prices are in general low (high).

We distinguish messages issued by different types of baseload unit production (nuclear and hydro (1), and marginal unit production (coal, gas and oil). When the demand is high, a small reduction in produced quantity can have a large impact on prices, and this reduction can be achieved by either a marginal or baseload unit. However, a producer with several types of generators primarily has an incentive to decrease production for marginal fuel types, as these production units have higher marginal costs. Hence, we expect larger effects for marginal fuel types which, in the case of Sweden, are oil, gas and coal.

We also separate the effects for new messages regarding failures and follow-up messages concerning already reported failures. This distinction is important as the incentives might differ depending on whether a producer decides to report a new failure or the prolongation of an existing outage. It is possible that a producer decides to report a new failure based on the encountered technical problems, but that the length of the failure depends on economic incentives. An increased number of follow up messages indicates that it takes longer to fix a failure, and the time it takes to fix a failure should not depend on prices in a competitive market.

To our knowledge, this is the first article studying strategic withholding that uses a quasi-experimental set up. The results indicate that there is a significant relationship between day-ahead electricity prices and the number of reported failures. The effect depends on the type of fuel used for generation. We find a positive effect of an increase in price on the number of reported failures in the case of oil and gas--marginal technologies. This is consistent with the hypothesis that it is more profitable to withhold capacity from generators with a high marginal cost. The results also show that prices have a larger effect on follow-up messages as compared to messages reporting an initial failure.

We first describe the economic rationale for withholding capacity in Section 2, followed by a description of the Nordic electricity market in Section 3. The data used in the analysis is presented in Section 4. Section 5 presents the econometric strategy and results are discussed in Section 6. The last section concludes.

(2.) ECONOMIC RATIONALE FOR WITHHOLDING CAPACITY

In Sweden, the general EU transparency rules apply, in particular REMIT (2011) (2) a regulation penalizing market manipulation in wholesale energy markets across Europe. Strategic withholding is regarded as a way of exploiting market power on electricity markets. In economic literature whenever a generator would in expectation profit from the sale of an additional unit of electricity, assuming that the market price would not change, and the generator decides not to sell, it is believed he has exercised market power (Twowey et al. 2005). A multi-unit generator that wants to increase the market price can achieve it in two ways. It can either strategically bid all of its production, asking for high prices (above its marginal costs (3), or it can physically keep some of its capacity away from the market. (4) This article deals with a specific version of physical withholding--the reduction in capacity through production failure. Here, and in the relevant literature, "strategic" is not defined as an interaction between multiple market players, but as a unilateral decision of one player to systematically influence prices (Wolfram 1998, Kwoka and Sabodash 2011).

2.1. Theoretical background

There are some pre-conditions that are thought to ease unilateral withholding of capacity: tight market conditions i.e. high demand, congestion in the market, and concentration of production.

Figure 1 illustrates physical withholding behaviour and its intended impact on prices. The graph depicts characteristics of a liberalized wholesale electric power market with inelastic demand (in the short run) and a hockey-stick shaped supply curve. The special shape of the supply curve is due to the merit order of electricity production, that is, the ordering of electricity production technologies according to their increasing marginal cost of production. Electricity is supplied by either baseload production with large starting costs but low, almost zero, marginal cost (for instance, through nuclear power plants), or by marginal production that starts producing when the baseload cannot fulfil the demand (for instance, coal or gas in Nord Pool). Moreover, different plants have some fixed capacity with steady costs that rise sharply when this capacity is exceeded.

The electricity market of the Nordic Region operates as a uniform auction, resulting in a single equilibrium price for each hour. A reduction of supplied quantity shifts the supply curve to the left, which can result in big price changes, especially if the demand curve is close to the almost vertical part of the supply curve. At this very steep part of the supply curve, where more expensive units are required to match the rising demand, prices become very sensitive to changes in supply (Patton, 2002), so even small output fluctuations might have a substantial impact on prices. Congestion can worsen the situation by helping to keep a high demand market tight.

Congestion can also directly facilitate exertion of market power when a large market due to congestion splits into several regions where only a few generators can provide electricity in each area thus creating conditions where some firms can use their dominant position to influence prices. Congested transmission lines causing a split of the market into zones and creating divergent prices in adjacent areas are indicated as a possible explanation of abnormally high prices in Western Denmark (Olsen et al. 2006) and southern Norway (Mirza and Bergland, 2015).

Another factor impacting the potential of exercising market power is the size of a producing firm. For a larger firm, an incentive to withhold is greater since in order for the strategy to be profitable, a producer needs to own several production units and the increase in profit after a production failure needs to be larger than the lost profit from the reduced output. In a 2002 study Amundsen and Bergman point out that a generating company becoming a minority owner in another generating...