Subsidizing a failed business model: a premium for residential electricity demand response

On 26 November the French National Assembly has rejected proposed law 3146 "encouraging a reduction in CO2 production through the development of direct load control".[1] This proposal is the latest instalment in a saga which began several years ago, intending to use public money (a tax on consumed electricity) to subsidise a non-profitable business activity carried out by private operators: residential demand response.[2] Since this issue concerns other countries in Europe and North America, the discussion below is of interest to readers outside of the French borders.

1. The history of (poorly) disguised subsidies paid to aggregators

In the near future, demand for electricity will be adjusted to the wholesale market prices: some consumers will reduce their demand during peak hours, when the value of consumed megawatt hours increases. This major change for the electricity industry will be a significant part of the transition to carbon-free electricity.

There are different approaches to make demand sensitive to price. One of them involves the interruption or reduction of residential customers’ consumption for a few minutes, then the subsequent sale of the energy saved on wholesale markets. Currently, this approach is not profitable in continental Europe for two reasons. First, the number of megawatt hours potentially saved by a residential customer is too small to cover the equipment and commercial costs. On the other hand, industrial customers’ demand response is financially profitable as it generates savings roughly a thousand times larger for slightly higher costs. Second, there is overcapacity on the electricity grid in continental Europe. Electricity market prices, thus the value of any megawatt hours saved, are very low. Residential demand response aggregators in continental Europe, specifically in France, have therefore chosen a poor business model.

For several years, aggregators in France (and in the United States) have tried to get consumers to shoulder the cost of their mistakes, convincing lawmakers to grant them unwarranted subsidies. Three arguments have been put forward in turn. First, that the aggregator would not need to pay electricity suppliers for the saved electricity as it was not consumed. This argument initially convinced law makers, but fortunately was rebutted by the "Brottes" law of April 2013.[3]

The second argument, also introduced by the "Brottes" law of April 2013, is as follows: as any non-consumed megawatt hours help reduce the amount of polluting megawatt hours (peak time production uses thermal power plants), the aggregators should receive a premium. The government initially proposed a draft decree indicating a €30 premium per megawatt hour. All the institutions that examined the decree voiced serious concerns: the Competition Authority during its inspection in January 2014 of the draft decree[4], the 'Conseil Supérieur de l’Electricité' voted to reject the premium decree[5] at the start of December 2014, and the Energy Regulatory Commission issued a negative report and proposed a lower premium in mid-December 2014.[6] Nevertheless, the government issued a decree[7] in January 2015 setting the premium at €16 per megawatt hour during peak times, and €2 per megawatt hour outside of peak times.

The third argument is the one put forward in proposed law 3146: non-consumed megawatt hours contribute to reducing the price of electricity for all users, so aggregators could legitimately receive some of the money saved.

None of these three arguments makes any economic sense.

2. The pizza delivery driver example

To get around the fact that electricity is sacred in industrialised societies, let's use a food-based example.  A pizza delivery driver, let's call him Marcel, has agreed to deliver 11 identical pizzas at 7pm at the price of €10 per pizza. After taking the order, Marcel learns that his supplier can only make 10 pizzas for 7pm. As our readers well know: a pizza must be eaten immediately after it has been cooked, it cannot be made earlier and stored. How can Marcel solve his problem?

He could ask one or more of the customers who has placed an order to delay their pizza order or cancel it. Of course, he will have to compensate them for this inconvenience. As it happens Marie, one of his customers, values eating her pizza at 7pm at €20, whereas the ten other customers, who are hungrier, value it at €30. Eating a pizza at 7pm therefore generates a net surplus of €10 (20-10) for Marie, and a net surplus of €20 (30-10) for the other ten customers.

Marcel does not have this information, but he can organise an auction to screen his customers for a candidate willing to delay eating their pizza in return for compensation.[8] No customer accepts compensation of less than €10. When Marcel offers €10.01, Marie (who is unaware of how much the other customers would accept) decides to delay eating her pizza, as she therefore receives one cent more than the net surplus that she would receive if she ate immediately.

We have therefore shown that:

The compensation that a consumer must receive to delay consumption is their net surplus, i.e. the value that they attach to consumption, minus the purchase price of the item.

This result can also be understood by a slightly different argument. Offering compensation of €10.01 is equivalent to selling the pizza to Marie for €10, then buying it back on the pizza spot market at the spot price of €20.01.

This example illustrates the flaws in the first argument given above: a customer taking part in a demand response scheme receives the spot price for any non-consumed megawatt hours, but in return she must pay the contract price to her supplier. In other words, you can't resell what you haven't bought.

The second argument is that delaying or cancelling the pizza order creates an additional benefit, for example an environmental one, and therefore the customers who delays their order must receive a premium. To understand the argument, let's say that Marcel can get an additional pizza from a supplier in the neighbouring town. As it is further away, delivering one of these pizzas generates CO2 emissions and therefore has a higher social cost, for example €25. If Marcel must deliver a pizza sourced from the neighbouring town, the net surplus for all people involved is reduced by €5; so Marie eats a pizza which she values at €20, which now costs €25 to produce and deliver. Delaying Marie's pizza therefore increases the collective surplus by €5. Why shouldn't Marie receive part of this surplus?

The answer is that this premium would encourage strategic behaviour, i.e. market manipulation. Indeed, let's assume that Marie values eating her pizza at 8pm at €22, as she prefers to eat her pizza at 8pm rather than 7pm. Let's assume at first that there is no premium: Marie is simply compensated with €20.01 if she has to delay her 7pm order to another time. If she orders a pizza for 7pm and agrees to delay it, she will therefore receive a surplus of €10.01 as before, whereas if she orders a pizza for 8pm, she eats at 8pm and receives a surplus of €12. If there is no premium, she orders a pizza for 8pm, and Marcel has no problem at 7pm. [9]

If we add a premium, Marie is compensated with €25 if she has to delay her 7pm order to another time. By ordering a pizza for 7pm and agreeing to delay the order, she receives a net surplus of €15 higher than the surplus if she requests delivery for 8pm. Marie therefore orders a pizza for 7pm, although she prefers to eat at 8pm, as there is the possibility that she could be asked to delay it.

If there is no premium, Marcel has no problem; the premium creates the problem.

We have therefore shown that:

Compensating an economic agent based on values that are not their own encourages opportunistic behaviour.

The third argument is in fact a variant of the second: if Marie requested the pizza to be delivered at 7pm, the price on the pizza spot market would amount to €25, the cost of the pizza delivered from the neighbouring town. By delaying or cancelling her pizza order, Marie therefore contributes to reducing the price from €25 to €20.01. Yet again, the counterargument is that if we give her a fraction of this price reduction, she is encouraged to behave strategically: ordering a pizza for 7pm although she would prefer to eat at 8pm.

We have therefore seen that the three arguments given above regarding the compensation of demand response consumers above the value that they give to the product, reflected in the spot market price, are not economically justified.

The pizza example helps illustrate another problem of the aggregators' business model: they commit to reducing their customers' bills, but do not intend to share the earnings made from reselling the energy on the spot market. In the previous example, let's assume that Auguste enters into the pizza demand response market. Auguste resells Marie's order to Marcel and pockets the €10.01. Marie can see that her pizza bill has reduced from €10 to €0, but she is not compensated for the pizza that she hasn't eaten.

* * *

We can now see that the argument included in the proposed law 3146 has no more basis than the previous arguments. It is simply a disguised subsidy, paid to aggregators who have chosen a poor business model, i.e. making others bail out private companies that have failed. In addition, the law indicates that private aggregators receive the entire premium ex ante, and they do not return any amount that they have been overpaid: an excellent opportunity to receive a cash advance.

One of the purposes of the proposed law is to reduce CO2 emissions. The explanatory information suggests "a saving of at least 2 million tonnes of CO2 per year in our country". This figure seems extremely optimistic: to achieve it, peak time electricity production would have to be reduced by at least 2 terawatt hours, while the current demand response volume, including industrial demand response, is around 0.7 terawatt hours. In addition, as our previous articles have shown, the regulation of greenhouse gas emissions involves setting emissions trading prices and not subsidies for specific technologies, whether for renewable energies or demand response schemes.

Finally, this law, as with the premium decree in January 2015, requires RTE to carry out complex calculations to determine the amount of these premiums, in this case the price that would have prevailed on the market if demand response had not occurred.  This calculation is extremely costly: it requires definition of the methodology, data collection and analysis, sending the information to stakeholders, sending and paying invoices, all of which will be paid for by consumers, in this instance through electricity rates paid to RTE.

We are therefore relieved that the National Assembly did not approve proposed law 3146, and therefore avoided an unjustified increase in our electricity b