Written evidence submitted by Professor Richard Green, Imperial College Business School (WIN0063)

What role did the UK grid play in the high domestic prices of winter 2022-23?

  1. President Putin’s decision to reduce Russian gas sales to Europe was the cause of the high energy prices of 2022-23. The fact that the UK bought relatively little gas from Russia before the attack on Ukraine, compared to some countries on the continent, is irrelevant. Western Europe (including the UK) effectively has a single market for gas. Arbitrage between neighbouring countries ensures that the prices in their gas spot markets are very similar, unless the pipelines (interconnectors) between those countries are at full capacity.
  2. The UK trades significant amounts of gas with the Continent; in summer, we tend to export gas (sourced from our own fields, from Norway and from imports of Liquefied Natural Gas (LNG)), some of which is put into storage. In winter, we need to import gas from Continental storage facilities to make up for the lack of storage in the UK. In the summer of 2022, when LNG supplies to Europe rose significantly to make up for the cut in Russian sales, enough of this flowed into the UK that the interconnectors were often at full capacity and the spot price in the UK (“NBP”) could be lower than the price in the Netherlands (“TTF”) that is generally used as the marker price for North-West Europe. Today, the NBP futures price (for next month’s gas) of 86.5 pence per therm is practically identical to the TTF price of 34.4 €/MWh.[1]
  3. The wholesale price of electricity closely follows the price of gas. In any commodity market, prices tend to be set by the marginal cost of the most expensive producer needed to meet demand, and electricity is no different. The use of pay-as-clear auctions in practically all organised electricity day-ahead wholesale markets around the world guarantees this, but it is a feature of markets in general – prices below cost will (eventually) force the exit of some producers, while above-cost prices tend to encourage entry.
  4. The chart below confirms that British gas and electricity prices closely track each other. The axes are aligned so that the price of 1 MWh of electricity is equal to the price of the gas needed to generate it at a thermal efficiency of 48%, the average reported in the Digest of UK Energy Statistics. The difference between the two lines is mainly made up of the cost of carbon permits, and of the Carbon Price Support (a tax on fossil fuels used in electricity generation).


  1. Arbitrage between markets also affects electricity prices. The Single Electricity Market has reached the stage that most electricity auctions in Europe are connected, ensuring that power will flow automatically from cheaper regions to those with higher costs, up to the capacity of the interconnectors. If interconnector capacity is high enough (relative to the cross-area differences in conditions), then prices are equalised. At other times, some markets can be significantly cheaper than their neighbours. Abundant cheap hydroelectricity in the Nordic countries means that power there tends to be structurally cheaper than in France and Germany, while limited import capacity and higher costs in Italy raise prices in that country.
  2. Electricity flows can change from hour to hour, based on movements in demand and in available generation capacity in each region. The UK has traditionally imported electricity, as shown in the chart below, but even in periods with large net imports, there have also been some exports. For three quarters of 2022, the UK was a net exporter, as somewhat higher gas prices on the Continent, and reduced availability of French nuclear stations, drove prices there above those in the UK.

A graph of energy trends

Description automatically generated

  1. The UK’s decision to leave the Single Energy Market means that cross-border arbitrage will be less effective – traders now have to guess the direction of future price differences, rather than having power moved automatically to even them out. If this means that the UK is exporting less electricity than it could, UK-based power stations are asked to generate less and wholesale prices will be lower, bringing savings to buyers (which may be passed on to consumers). However, the loss to producers and traders will exceed these savings. When inefficient arbitrage means that the UK imports too little power, our generation and wholesale prices will be higher, and consumers pay more; again, this loss is only partially offset by profits to generators and traders.
  2. From time to time, there have been suggestions that the price-setting mechanism should be reformed from pay-as-clear, or marginal, pricing to a pay-as-bid system in which every generator receives the price they asked for. Typically, the people making these suggestions notice that some generators (usually those with very low marginal costs, such as nuclear or renewable generators) are submitting price offers far below the market price but receiving that much higher price. Simple arithmetic reveals that the average of some low offers and some high offers would be lower than the current pay-as-clear price. Simple thinking should reveal that if the payment system changed, so would the offers. Instead of signalling their costs, generators would submit their best estimate of what the market price is going to be, and that price would depend on the costs of the generator expected to be just needed to meet demand, as before. To the extent that those estimates are wrong, we risk getting a less efficient dispatch, in which higher-cost stations are prioritised over lower-cost ones.
  3. Could pay-as-bid be made to work by forcing generators to offer some (audited) measure of their costs? For an efficient dispatch, this would have to be their variable cost, but for cost recovery, it would have to be their average cost. Those numbers can be very different, particularly for low-carbon generators, as the variable costs of solar, wind and nuclear plants are very low.
  4. It would be much better to continue with the system we already have, in which short-term wholesale prices are set on the basis of marginal costs and pay-as-clear (giving an efficient dispatch) but are hedged with longer-term contracts. For low-carbon generators, these are financial contracts for differences, so that when the market price is low, the generator is still able to recover its costs, in return for giving money back to consumers if the market price rises above the contracted level. For other generators, the contracts can be based on physical delivery, but if the generator is able to meet their commitment either by generating or buying power in the market, we still have efficient incentives.
  5. Hedging in advance tends to offer good value, although (like most insurance policies) the contracted price might be higher than the expected cost of buying in the spot market, and the actual price will sometimes be below the contracted price (just as some insurance purchases turned out to be “unnecessary”). Hedging during a crisis runs the risk of over-paying (although if future prices turn out even higher than expected at the moment of hedging, that over-payment still represents an ex-post saving). California was able to buy power for the summer of 2001 for (significantly) less than the then-current spot market prices only in the form of multi-year contracts offering more than the then-expected price in their later years.[2]
  6. The UK grid’s limitations do contribute to consumer costs, however. Wind farms are often located far from demand, and when their output is high (which often happens to several nearby wind farms at the same time) there may not be enough capacity to transport all the electricity available. The farms have to be persuaded to generate less power than is physically possible, effectively buying back some of their output (assuming they had sold it already). Generators will find this advantageous if the price they pay to buy back power is less than their marginal cost – it is cheaper to buy themselves out of their sales commitment than to meet it by generating. For renewable generators receiving a subsidy per unit actually generated, however, their marginal cost is negative – their true marginal cost (positive, but low), less the subsidy they receive if (and only if) they generate. These generators would submit a negative buy-back price, which if accepted means that the system operator has to pay them to generate less. This naturally raises the cost of resolving the constraint, as instead of receiving some money from the generator(s) buying back output, which can be used to offset the cost of buying more from a better-located generator, the system operator now has to pay both generators, and recover the money from consumers. This affects generators receiving certificates under the Renewables Obligation and those with Contracts for Differences.[3]
  7. Over the longer term, limited transmission and distribution capacity makes it harder for new generators to enter the market and add supply that would help bring down prices. Nick Winser, the UK’s Electricity Networks Commissioner, has recently made recommendations intended to speed up the process of making connections available. It should be noted that if large amounts of new capacity depress market prices below the level of generators’ costs, some older generators will (eventually) exit to restore the balance between supply and demand at a level allowing the remainder to cover their costs. Nonetheless, the current long delays in getting a connection is impeding the urgent task of decarbonising the power system.

What more could have been done to prevent price shocks being passed to consumer bills?

  1. The government’s Energy Price Guarantee effectively ensured that all consumers faced the same tariffs, whatever their consumption or their income. If a scheme must be introduced quickly, it had best be simple. Income- or consumption-contingent schemes are harder to design and implement, but the past year might have been spent thinking about whether such a scheme was feasible or desirable. The Guarantee’s simplicity ensured that some help went to high-income consumers who did not need it while there were low-income, high-consumption customers who faced hardship from their rising bills, even with the Guarantee in place.
  2. A household’s energy consumption is not tightly correlated with its income; the size of the home, the state of its insulation, the number of people in the household and how long they spend at home all have their effects, and some people have a health-related need for warmth. Energy consumption at second homes is typically low, but these are related to affluence, not to poverty. home size, insulation, time spent at home (particularly at second homes) all important. Energy companies (the natural candidates to implement any scheme adopted by the government) have some information on particular categories of vulnerable customers, but they do not hold income data, and this is right.
  3. Any scheme to subsidise energy consumption must accept that there is a dilemma between giving help to needy households and retaining the incentive effect of high prices. The higher the energy price, the less people will tend to consume (relative to “normal” levels for that time of year). Those reductions are (mostly) not costless; they may represent reductions in comfort or convenience, and low-income households will tend to face particularly acute sacrifices.
  4. Setting the price cap at a low level will be expensive for the government, since the gap between a given market price and a lower cap is higher, the subsidy will have to be paid on more units if consumption is higher, and the higher consumption is also likely to drive up the market price. A higher price cap saves money for the Treasury but imposes hardship on more households.
  5. One compromise between providing help and giving incentives is to lower the price for some of each household’s consumption, but not for all. “Lifeline” or “increasing block” tariffs do this with a lower price for the first units of energy consumed than for higher amounts. All households receive the same amount of help, regardless of need, unless they consume less than the subsidised level of energy. A more sophisticated, but still practical, scheme was implemented in Germany, where support was provided for consumption equal to 80% of the household’s previous year’s demand. Reducing the cost of 100 kWh by, say, 8 p/kWh provides the same support as reducing the cost of 80 kWh by 10 p/kWh. If consumers realise that their last 20% of consumption is costing much more than it used to, they may save more energy than if the cost of every unit was reduced. That will not directly reduce the cost to the government (unless the consumer reduces demand by over 20%!) but may do so if lower demand leads to lower wholesale prices.
  6. Lower wholesale prices as a result of this more targeted support scheme are more likely for electricity than for gas. European gas markets are more integrated, implying that any reduction in demand would have to be larger to have an appreciable impact in the wider market, although the overall impact of several similar national schemes would be greater. Electricity wholesale prices, set on a half-hourly basis in the UK (hourly in most other countries), can reach extreme levels when demand is particularly close to the available supply, and lower demand at those times has a disproportionate effect on prices.[4]
  7. In general, there is evidence that consumers probably respond more to the average level of their bills than to the incentives given by the details of their tariff structure.[5] This would imply that the precise way in which support is given is less important than the amount. However, the salience of energy prices over the last year may be high enough for consumers to appreciate the different incentives given by the German scheme and respond accordingly. In due course, we may have evidence on whether Germany achieved greater demand reductions than its neighbours, but researchers would have to control for many other factors, and I am not aware of any studies published so far.

How effective is the Government's approach towards supporting the sector and delivering a functioning energy market?

  1. Some of my points above (particularly paragraphs 7-11) are relevant if this question is related to the wholesale market. I do not have specific points to offer on the retail market, or on the Committee’s other questions.

August 2023

[1] 85.7p is equal to €1 at the current mid-market exchange rate, and 1 MWh is equal to 34.12 (UK) therms. In other words, €34.4 could be used to purchase 1 MWh in the TTF market or converted to (34.4 x 85.7 = 2,948) pence, which would allow the purchase of (2,948 ÷ 86.5 = 34.08) therms. The gas prices were obtained from EEX.com on 25 August 2023.

[2] California’s electricity prices rose from summer 2000 onwards because of increasing natural gas and air pollution (nitrogen oxides) permit prices and reduced imports of hydroelectricity from the Pacific North-West. Market manipulation (notoriously by Enron) worsened the situation, while a retail price cap that was not accompanied by enough wholesale market hedging drove one major utility into bankruptcy (as happened to minor firms in the UK). Generators fearing non-payment reduced their sales and there were several blackouts early in 2001, accompanied by record prices. As the wholesale market broke down and its operator filed for bankruptcy, the California Department of Water Resources (the least inappropriate agency available) was tasked with procuring electricity, using the kind of contract described above. Soon afterwards, short-term prices had returned to normal levels, helped by the completion of plants that had been under construction during the crisis (easing any capacity shortages) and by falling gas prices.

[3] A related problem is if there is so much renewable generation that the market-wide price becomes negative, signalling that thermal generators would incur costs from having to shut down and start up again later. Renewables Obligation generators still have an incentive to run (and receive subsidies) if the price is not too negative, but CfDs can (in principle) be written so that top-up payments are only made while the market price is positive, ensuring that generators prefer to reduce output while the price is still positive.

[4] Demand response schemes allow for reduced consumption at exactly such times when the market is short of available generating capacity. I am not saying that the German support mechanism was a kind of demand response scheme, as it was not targeted on shortage hours, and most households don’t receive the price or other signals needed to implement demand response. My point is that if the underlying level of consumption is held down by the higher prices faced for each household’s last few kWh of demand, this generally lower demand will lead to lower prices.

[5] For example, Ito (2014) compared the customers of two electricity companies serving adjoining parts of Los Angeles (and hence facing the same weather conditions) to show that differences in consumption between the two groups were far better explained by the differences in the average prices they faced than by those in the prices they faced for the last few units of their consumption. Ito, K. (2014) “Do Consumers Respond to Marginal or Average Price? Evidence from nonlinear electricity pricing”, American Economic Review 104(2) pp 537-563.