Demand response and contractual distortions in electricity retailing

Active demand response, i.e. the ability for customers to reduce their consumption and resell power into spot markets, is a recent and important development in the power industry, in Europe and in North America. Most observers agree that higher penetration of non-dispatchable renewable energy sources will increase the importance of demand flexibility. Since few customers are expected to enroll in Real Time Pricing contracts, active demand response is considered a promising avenue towards demand flexibility.

This blog’s authors have recently published an academic article on retail contracting when demand response is possible[1], showing that, in certain cases, demand response is detrimental to the efficiency of power markets. This surprising result arises from information asymmetry between retailers and consumers. It illustrates how a superficial analysis of demand response could lead to unintended outcomes. As is often the case, rigorous economic analysis is required to design appropriate policies.

This post first describes briefly this article’s main findings, then discusses other issues where analysis is required to assess the consequences of having more and more consumers participating in market adjustment.

Lack of demand reactivity

Historically, electricity demand was not responsive to variations in wholesale prices. Since liquid wholesale markets for electric power did not exist in most countries until the early 1990s, there was no price signal for users to respond to. Thus, users were accustomed to consume power to respond to their needs, not to the opportunity cost of power at any given time. Over the last 20 years, demand response has become a technical reality. Liquid wholesale markets now exist in most countries, providing the value of power at different times and locations. Most large industrial and commercial users have meters that enable them and their supplier to track their real time consumption. Similar "smart" meters are being deployed for residential and professional users in many jurisdictions.

Demand response has the potential to transform how power systems are designed and operated. For example, if enough users reduce their demand as price increases, there may no longer be a need to implement involuntary curtailment, hence the notion of capacity adequacy, which currently stands at the core of most electric power markets' design, may no longer be relevant. For this reason, demand response has received significant attention from academics and policy makers in Europe and North America.

In most industries, consumers respond to market conditions by purchasing on the spot market and adjusting their consumption to spot price variations. The electricity industry is different, since most consumers have fixed-price contracts with their supplier, which specify a price independent of the spot price. Demand response is thus customers reselling into an adjustment market the megawatt-hours they are entitled to withdraw and do not consume.

It is worthwhile emphasizing that “demand response” is different from “demand management”. In the latter, consumers buy electric power at low price because they accept the risk to be disconnected by their provider. Ex post, they are not the decision maker. By contrast, demand response is a voluntary reduction in electricity consumption by retail customers to react to an increase in the power price or to incentive payments.

Markets’ new design

With consumers becoming active players, the first issue that has attracted attention of academics and policy makers is the design of adjustment markets, in particular the price customers must pay to have access to the market and the price they receive when reselling power. It has led to contentious debates in many jurisdictions. Basic microeconomics (and accounting) show that a consumer who participates in the adjustment market by reducing his expected consumption must be paid the price of the adjustment market provided he acquired the rights to participate, i.e. he paid the retail price for every kWh of his expected consumption or withdrawal profile. This is not clearly understood by some businessmen and regulators (in particular the FERC in the USA) who consider that consumers should not have to pay for the non-consumed kWh. This is obviously wrong since it would mean that consumers are allowed to sell something they do not own.[2]

Even if the adjustment market is designed in line with ownership rights, a second issue of interest is consumers' potential strategic behavior. Customers are paid for the difference between the consumption-that-would-have-happened, called their baseline, and their actual consumption. They thus have incentives to inflate that baseline.

In a particularly illuminating example, the demand response operator at the main Baltimore baseball stadium turned on the stadium's lights to create false demand that it was then paid to reduce. This behavior was detected, and the operator subsequently fined by the Federal Energy Regulatory Commission.[3]

Retail contracts

To address the issue of strategic behavior, a first possibility is to develop algorithms that provide a robust estimation of the baseline. While this approach may work for some customers, it may not apply to all. Another approach is to recognize that some consumers will always have better information than retailers on their baseline, hence to design contracts that explicitly acknowledge this private information. The analysis then requires contract theory to build models of consumers' strategic behavior.

The basic framework developed in the academic article previously mentioned follows the second way. Its timing is as follows. First, suppliers and consumers agree on bilateral supply contracts, that is a quantity to provide and a retail price. Later, a random shock occurs in the system (more consumption and/or less production than expected). Demand and supply must then be rebalanced after the shock is realized. What are the outcomes when compensation for the shock is provided solely by increased production, and when it is provided by an adjustment market in which producers and customers participate?

To answer this question, two features must be taken into account. First, customers have private information on their value for electricity, which gives them the opportunity to behave strategically. Second, the contract price arises endogenously from the bargaining between retailers and consumers. This is true in all markets, but it is essential in the electric power industry where the operators have to fix unexpected imbalances for a non-storable product whereas contracts have already been signed.

Standard economic analysis suggests that, if the adjustment market is well-designed and frictionless, resale is ex post efficient: consumers resell power precisely up to the point where their marginal value for power equals the price in the adjustment market.

Furthermore, if there are no frictions in the retail market, ex post demand response does not distort the retail contract either, which is then efficient. In other words, neither customers nor suppliers have an incentive to distort the retail contract from the efficient one when consumers are allowed to intervene in the adjustment process. Opening the adjustment market to customers thus increases net surplus.

Asymmetric information on profiles

The above result does not necessarily hold under imperfect information. Suppose there are two types of customers with different willingness-to-pay. Contract theory states that suppliers must leave an information rent to customers with the higher willingness-to-pay (high type), and reduce consumption of the low type customers. It results that while the retail contract remains efficient for high type customers, it is no longer efficient for low type. Thus, the possibility of customers' strategic behavior, which gives rise to the information rent, defeats efficiency of adjustment markets. What is the net surplus impact of customers' participation in the adjustment market? In some instances, the information distortion for low type customers is so large that opening the adjustment market to customers reduces net surplus.

Policy implications

In summary, because of information asymmetry on consumers’ preferences joint to the possibility that consumers compete against producers in adjustment markets, suppliers offer doubly distorted contracts with a potential loss of social surplus. The policy implication is not that policy makers should abandon adjustment markets for fear of inefficiency. Rather policy makers and regulators ought to be aware that consumers will exert strategic behavior when possible, and should therefore design markets that accommodate this strategic behavior.

Other problems will arise, in particular market power by aggregators and the control of the information collected on profiles. Thousands, and progressively millions of consumers will have the possibility to intervene in the adjustment market by reducing their demand below their contracted quantity at a price determined by all the bids of reserve producers and ready-to-renounce customers. However, because of their small scale, most customers will not do it individually. They will rather choose to be connected to a remote operator, able to monitor domestic appliances and industrial machines for given duration, at some specific dates where the electric system needs to be rebalanced. Controlling the possibility to disconnect 1 kW at one million locations gives the equivalent capacity of a nuclear generation plant, with the additional advantage of time and spatial flexibility.[4]

In this aggregation business, there are strong economies of size (scale, scope, density) as well as indirect network externalities (compatibility, technological complementarity, potential for two-sided payment). This means that the supply of demand response will most likely be made by a small number of large firms. In all network activities, take-off is a delicate phase, but when successful, the operator can harvest huge rents thanks to its dominant position. Maintaining a minimal dose of competition without impairing the gains of networking is a difficult task that must be prepared beforehand.

As for information on profiles, energy consumers are very heterogeneous: depending of their type (industry, business, household), size, location, technical characteristics of appliances and buildings, they consume large or small, regular or irregular quantities, at peak or off-peak periods. Additionally, depending on the equipment they have installed to produce local energy (photovoltaic panels, wind turbines), they can be net demanders or net suppliers of energy on the grid. The problem is that, electricity being non storable, producing 1 MWh has a much higher cost when it must be supplied in one hour than when the consumer needs 100 kW per hour for 10 hours. Thus, having a good knowledge of consumption profiles allows to install a cheaper and more reliable production mix. Unfortunately, traditional meters only give information on aggregated consumptions, sometimes for the last 6 months. Thanks to smart meters and demand-response programs, large quantities of precise information will be collected from consumption places and processed to be used by energy suppliers and load-shedding aggregators. This raises two questions. One is that consumers will progressively lose the initial informational advantage; it is good news for efficiency, and bad news for equity since energy and service suppliers will be able to extract more rents from consumers. Regulators will probably have to intervene to limit the market power of energy and service suppliers, except if we can enter a world of strong competition in retail. The second question is the trading of data collected on consumers' behavior. Will it be legal to sell data on consumption profiles? Will data on profiles be viewed as an essential facility for new entrants?

Again, it is better to prepare regulations before the problem becomes urgent to solve. Unfortunately, as regards energy policy, most decisions are taken after intense campaigns by lobbyists rather than sound economic analysis.[5]