Testing Energy Market Manipulation in Great Britain
In June 2014 the Gas and Electricity Markets Authority made a reference to the Competition and Markets Authority (CMA) for an investigation into the energy market in Great Britain. The reference was made in response to rising energy prices, which, combined with a lack of clear understanding of the factors driving the movement in price, renewed public scrutiny of the energy market. Concerns were raised in the public mind, and expressed on political platforms, in relation to the affordability of energy, levels of service and perceived levels of profitability of the six large energy firms that are all partially integrated with respect to electricity.
The CMA’s Energy Market Investigation, which was the most comprehensive investigation into the energy market since privatisation, set out to identify and address discrete problems within the energy market, as well as to clear the air of speculations. As far as energy market manipulation was concerned, there were two areas that we considered may lead to some harm in the market, namely the exercise of unilateral market power, and the manipulation of contract for difference (CfD) reference price to increase CfD payments – both in the generation and wholesale market for electricity.
In both areas we found that while there were some initial grounds for concern, we considered it very unlikely that manipulation would occur to the detriment of consumers. In the rest of this paper we set out for both wholesale market power and CfD manipulation, how manipulation might occur, how we set about assessing it and our findings. Before that, we first provide a brief overview of the relevant regulation for identifying and penalising energy market manipulation in the GB energy markets.
Britain’s energy regulator Ofgem has several tools that it can use to investigate and, where merited, enforce against market manipulation in the GB energy markets, including the EU Regulation on wholesale energy market integrity and transparency (REMIT), and the relevant provisions of competition law notably in relation to excessive pricing. Ofgem can undertake a wide range of investigative activities as part of both REMIT and competition law investigations, as well as to take measures against those who break the rules, including imposing severe financial penalties that can more than offset any gains from manipulation.
These regulations can have a powerful effect on incentives to engage in market manipulation. If manipulation is obvious it is likely to be observed by Ofgem, and therefore the regulation is likely to constrain generator behaviour. However, where the manipulation is less obvious, then the degree to which regulatory enforcement is a constraint might be questioned. It is with that in mind that we decided to examine whether or not manipulation took place in the past, and what this may tell us about the likelihood of this happening in the future.
Market power in electricity generation
How might manipulation occur?
Our main hypothesis relating to the exercise of unilateral market power (UMP) was that certain generators may have market power in the GB wholesale electricity market at particular times. Although market shares in generation as a whole are relatively low, the nature of demand and supply, including the need to balance demand and generation in real time, means it is possible that at certain times, one generator (or more) may unilaterally be able to influence the price of electricity in spot markets.
The main way in which a generator can exercise market power is to withhold some capacity (either physically or economically) to force a shift in the supply curve and hence the price.
The concern was that price increases from market power would be passed through to final consumers. In addition, there were also concerns that if vertically integrated generators had market power, they could use this to partially foreclose the retail market to other non-integrated retailers, leading to harm to final consumers.
How did we assess UMP?
Modelling market power in electricity markets has a long history. The liberalisation of wholesale markets in 1991 in Great Britain led to two seminal papers: Harbord and von der Fehr (1993) and Green and Newberry (1992). Both papers sought to establish the Nash equilibria of a bidding game that was meant to be similar to the GB market, which at that time was dominated by two privatised generating companies.
Another notable contribution to the literature was what became a very influential article by Allaz and Vila (1993), which argued that forward markets for electricity changed the logic of market power completely. Effectively, forward sales competed with spot sales from the same supplier in a way somewhat analogous to results about durable good monopolists. Allez and Vila’s result suggested that market power would be much less of a problem than either Green and Newberry or Harbord and von der Fehr approaches would suggest.
Our work was very much in the tradition of these pioneering models, although we did not need to establish an equilibrium bidding strategy but only the existence or not of an ability and incentive to move spot prices. A necessary condition of the exercise of market power in all the early theoretical models was profitable unilateral deviations from competitive bidding conditional on competitive bids by others. This is the condition that we decided to test.
In order to assess whether electricity generators had the ability and incentive to exploit UMP in the GB wholesale electricity market, we built a model to compare the competitive market price to an optimal price for each firm for each half-hourly period in 2012 and 2013. We took the firm’s optimal withholding strategy as the profit-maximising response to other firms’ competitive offerings (i.e., assuming that rival firms offer their output as if the market were competitive and do not withhold capacity.) The best response of other firms to withholding by one firm is likely, in the specific circumstances of the GB electricity market, to be to maintain competitive levels of output.
This modelling exercise provided something close to a necessary condition to a finding of UMP by any generator. However, it did not provide a sufficient condition. Any modelling exercise of this type needs to make a number of simplifying assumptions. These simplifications limit the ability to strongly rely on positive results, and therefore would require further investigation and processing. However, if harm could not be found in the simple model, we could almost certainly rule out there being harm in a more complex model that better represents the real world.
The main simplifications included (1) assuming all output was sold through the spot market at a single price; (2) assuming no uncertainty about the market (including both demand, other generator’s capacities and strategies); and (3) assuming no dynamic constraints on generators.
Single price in the spot market
While our model assumed a single price with all output being sold on the spot markets, the GB wholesale electricity market can be better characterised by both bilateral trading between parties and forward trading of output. This means that generators may necessarily receive different prices and may have sold much of their output before the spot market. Therefore, withholding capacity will not lead to the withholding generator benefiting fully from the increase in prices. However, while there is not a single market price that all generators receive for their output, we would nevertheless expect the prices agreed to reflect underlying fundamentals of the market, and thus the single price assumption is reasonable in the circumstances.
In contrast, with respect to forward trading, the model may not perform so well because either a generator would have to be completely unhedged to benefit from capacity withholding, or the withholding has to be anticipated by other participants to affect forward prices. We considered the former unlikely as it would be a highly risky strategy to pursue, thus for a generator to benefit from a withholding strategy they would have to affect expectations of the forward prices.
To affect expectations of forward prices, other market participants would need to believe that this strategy was being consistently pursued. This might be difficult for a generator to achieve because the incentives for withholding in the spot market are affected by prior forward market behaviour. If a generator is expected to withhold, and therefore manages to sell forward energy at a high price, it could be tempted not to withhold when the time comes to withhold. A generator would therefore have to develop a reputation for consistently withholding even if immediate incentives did not warrant it. It would be very difficult for a generator to achieve this without attracting regulatory scrutiny.
Uncertainty and dynamic constraints
Withholding capacity to affect future expectations involves a degree of risk that reduces the profitability of the strategy. This is because market circumstances may change between the period in which a generator withholds capacity and the period in which the expectations are affected (e.g., the relative position of coal and gas plants in the merit order may be reversed between one year and the next).
Uncertainty around key aspects of the market can affect the optimal withholding strategy in the short term. While we expect that generators, through repeated interactions, will be able to estimate the position of their power plants in the stack relative to other plants, the ability to forecast actual demand and output by other plants can lead to a suboptimal withholding strategy being pursued.
This means the benefits of withholding may be overstated but the costs from withholding will not be. In addition, as mentioned in the section above, the relative prices of gas and coal will affect the strength of signal about current withholding to other participants. For instance, one generator may withhold some of its CCGT plants when they are peaking plants, but if the price of coal and gas were to change such that coal plants were the peaking plants, the ability and incentives to withhold may change. In this scenario, the cost of making a signal to the rest of the market would still be present, but the gain from withholding may not be.
In addition, the lack of dynamic constraints meant that the model predicted higher off-peak prices and much lower evening peak prices. To account for these factors, we filtered the results to ensure we found results more consistent with reality.
What did we find?
For 2012 and 2013, in our initial filtered results we found that there were some opportunities for generators to exploit UMP. However, the number of periods was limited. Furthermore, once we considered filters allowing for uncertainty and dynamic constraints, there were minimal opportunities to increase profits by withdrawing capacity.
In addition, as part of the EMR programme, a number of fossil-fuelled power stations will be in receipt of capacity payments following the capacity market auction. Generators in receipt of capacity payments need to be available to generate at times of system stress. Failure to generate at these times can lead to penalties, leading to a reduction in the incentives for a generator to exploit UMP.
Overall, we concluded that it did not appear likely that generators had the ability and incentive to increase profits by withdrawing capacity through the exercise of UMP.
How might manipulation occur?
A CfD is a private contract between the holder and the CfD counterparty in which the holder receives from (or pays to) the counterparty the difference between a previously agreed strike price and a CfD reference price. The CfD counterparty makes (or receives) a payment per megawatt hour generated, meaning the level of support is based on actual output of low carbon generation (rather than capacity).
The CfD payment is the difference between the strike price and the CfD reference price. The reference price is the average price on the reference market, weighted by volumes traded. In general, the reference price should reflect the prevailing market price, especially in liquid markets.
If the reference price can be manipulated, there is a theoretical possibility that firms in receipt of a CfD could manipulate the reference price to receive higher overall revenues.
How did we assess it?
To assess whether the energy market could be manipulated to increase CfD payments we considered both the ability and incentive for generators to manipulate the reference price.
a With respect to ability, we identified the two key factors as the extent of liquidity in the reference market and whether there were any opportunities for arbitrage. If the market was illiquid and there were few opportunities for arbitrage, there would likely be greater opportunity for manipulation.
b With respect to incentives, we assessed the profitability condition for manipulation. This allowed us to identify that a generator in receipt of CfD would need to sell more CfD receiving generation outside the reference market than is being sold in the reference market by all other generators. Given a generator would need to sell some output into the reference market in order to manipulate it downwards, this would require a very large generator or a small reference market.
What did we find?
In terms of ability, we were unable to rule out the possibility that a large generator in receipt of a CfD could sell sufficient output on the reference market to manipulate the price. However, we noted that selling output in the reference market below the prevailing market price might create arbitrage opportunities that could increase the volume of trading in that market, thereby making it harder to manipulate the price.
In terms of incentives, our analysis indicates that for a manipulation strategy to be profitable, a generator would have to sell at least as much output outside the reference market as the total volume traded by other parties in the reference market. The capacity traded in the reference market was approximately 18.5GW in 2014, and the generator that has been awarded the most baseload CfDs to date will have 3.2GW of capacity in receipt of CfDs when it comes online. As a result, absent significant changes in the amount of trading in the reference market, we concluded that generators in receipt of CfD were unlikely to have incentives to manipulate the CfD reference price.
In the CMA’s energy market investigation, we assessed two potential incidences of market manipulation. In both cases we assessed the ability and incentives of parties to manipulate the market. When investigating market manipulation, it is not sufficient to find that a party has the ability to manipulate the market if you find they have no incentive to follow through on the action. In addition, we noted that the incentives to manipulate the market may be reduced through the design of the market, regulation or the monitoring of behaviour by regulators.
 James Jamieson is an economic adviser and Nenad Njegovan is an economics director at the Competition and Markets Authority.
 The electricity grid in Great Britain is separate to that in Northern Ireland, which is regulated by the NI Utility Regulator. In addition, the terms of reference for the CMA’s investigation only covered the Great British energy market rather than the whole of the United Kingdom
 We did not look at the activities of energy market traders as this was outside the scope of our reference.
 The REMIT regulation (Regulation (EU) No. 1227/2011 of the European Parliament and of the Council of 25 October 2011 on wholesale energy market integrity and transparency) explicitly prohibits market abuse, requires effective and timely public disclosure of inside information by market participants, and obliges firms professionally arranging transactions to report suspicious transactions. If GEMA is satisfied that a regulated person has failed to comply with a REMIT requirement, it may impose a penalty of such amount as it considers appropriate. While the primary responsibility rests with the regulated person, GEMA may also take action against individuals where there is evidence of personal culpability on the part of that individual (see The Electricity and Gas (Market Integrity and Transparency) (Enforcement etc.) Regulations 2013).
 See Ofgem, National Report 2017, Figure 4.
 In addition, our theory of harm requires that affecting the spot market is likely to lead to increased price expectations in forward markets. This is not a naïve assumption, and the issues are discussed further in CMA’s Energy Market Investigation (EMI) Final report, Appendix 4.1, Paragraph 20. See https://www.gov.uk/cma-cases/energy-market-investigation#final-report.
 Economic withholding would involve a party contracting its power plant as if it were higher in the merit order.
 NHM von der Fehr, D Harbord (1993) ‘Spot market competition in the UK electricity industry,’ The Economic Journal, 103 (418), 531-546.
 Green, RJ and Newberry, DM, (1992), ‘Competition in the British Electricity Spot Market,’ Journal of Political Economy, Vol:100, Pages: 929-953.
 Allaz, Blaise, and Vila, Jean-Luc. (1993). ‘Cournot Competition, futures markets and efficiency.’ Journal of Economic Theory, 59 (1), 1-16.
 The competitive market price is the marginal cost of the marginal plant when all plants are stacked up in order of their marginal cost.
 The optimal price is the price that maximises profits for the firm in question. If the price increases from the competitive strategy, this optimal price is achieved by a firm by withholding capacity. If the withholding strategy is inferior to the competitive strategy, the optimal strategy is the competitive strategy for that period.
 A full explanation of the methodology is outlined in EMI Final report, Appendix 4.1, Annex A.
 The reason for this is that market power, when it is exercised, involves making another technology the price-setting technology – for example, shifting this from coal to gas. Once this has been done, there is no further opportunity to raise prices by small additional capacity reductions. Therefore, we believe that the strategies we identified as optimal for each firm would also be stable for the market as a whole.
 Unilateral market power could conceivably be exercised by specific exploitation of the constraints that we have excluded from this model. For example, a particular generator might find that it has market power in the supply of the very flexible, rapid response capacity that is required at peak times or in times of system stress. We have not explored this possibility further in this paper. We considered the arrangements and incentives for the supply of flexibility in EMI Final report, Appendix 5.1.
 In addition, under EMR, new generating capacity in receipt of CfDs will be at least indifferent, if not against increases in benchmark prices as this reduces the payments a generator would receive from the CfD (see section on CfD below).
 The CfD counterparty is the Low Carbon Contracts Company (LCCC) – a company wholly owned by the government. Its duties include acting as the counterparty for CfDs issued to low carbon generators. See DECC (August 2014), Low Carbon Contracts Company Ltd: framework document.
 For baseload generation, the CfD reference price is the volume weighted average of season-ahead baseload prices, based on data from the London Energy Brokers’ Association (LEBA) Baseload Index and the Nasdaq Baseload Index. For intermittent generation, the CfD reference price is the volume weighted average of day-ahead electricity prices for the relevant settlement period, based on data from the APX Intermittent Index and the N2Ex Intermittent Index. See FIT Contract for Difference standard terms and conditions for more information.
 For a graphical illustration and explanation, please see EMI Final Report, Appendix 5.3, Annex C, Paragraphs 5-11.
 See EMI Final Report, Appendix 5.3, Annex C and Supplement 1 for full outline of the calculation.