Claude Crampes and Thomas-Olivier Léautier
RTE (Réseau de Transport d’Electricité, the French Transmission System Operator) has recently started implementation of an electric power capacity mechanism, which will be operational on 1 January 2017. This mechanism aims at ensuring France’s generation adequacy, i.e., ensuring that installed generation capacity will exceed demand under almost every circumstance. Other European countries are implementing or considering implementation of capacity mechanisms. This post describes the genesis of these mechanisms, then examines their pros and cons.
1 Capacity mechanisms
As explained in previous posts, if electricity markets are left alone, a market structure known as “energy-only”, when demand approaches installed capacity, the price rises to the marginal willingness to pay so that demand adjusts to available supply.
Energy-only markets assume that prices are reliable signals of market fundamentals, not subject to any manipulation. Spot prices must rise to several thousand euros per megawatt hour for a few peak hours per year in order to cover fixed costs of generation. Since producers (and demand response operators) are driven by profit, how can we be certain that prices do not rise more than necessary to cover costs or that there are not too many peak hours? The problem is made worse by the fact that electricity demand is not very elastic, which means that producers can push up prices with little reduction in demand.
To mitigate the risk of market power, many jurisdictions in the United States have implemented price caps, which have the adverse effect of preventing energy prices to rise high enough to cover fixed costs. This situation, known as the “missing money” problem, often results in under-investment, hence has led several US markets to implement capacity mechanisms to ensure that adequate generating capacity is connected to the grid.
These mechanisms share many common features: the electricity system operator, which in most cases is also the Transmission System Operator (TSO), sets a target for capacity which must be available at a future date (for example, 100 gigawatts of installed capacity in 2018). It then organises an auction for this future capacity, open to electricity generators and demand response operators. All the “winners” will receive in 2018 the price fixed by the auction (for example 40 000 € per megawatt per year), corresponding to the value of the capacity in 2018 to compensate them financially for their availability, whether or not they are called on to generate electricity or reduce consumption in 2018. Great Britain has recently set up a capacity market following closely the US model.
The mechanism currently being set up in France differs somewhat from this model, and relies on decentralised decisions by market participants. Electricity suppliers (who are Mandatory Participants on behalf of consumers) must prove to RTE that they have sufficient capacity certificates to cover their demand, taking into account a safety margin. For example, if RTE imposes a 15% margin, a supplier who has a peak demand of 20 gigawatts must prove to RTE that it has generating or demand response capacities certificates of 20 gigawatts, plus a 3 gigawatts margin. RTE certifies these capacities.
In some instances, RTE may consider that the sum of the capacities certificates is lower than the total capacity it believes is required, maybe because all the suppliers are anticipating a loss of market share. If that occurs, a centralized tender for the missing capacity is organized, as in the other countries mentioned above.
2 Rationale for capacity mechanisms
Justifications for capacity mechanisms fall broadly under two categories: energy-only markets (i) run the risk of abuse of producers’ market power, and (ii) may lead to lower security of supply than is desired, in particular when intermittent renewables constitute a significant share of the generation mix. Capacity markets, coupled with caps on the spot markets, alleviate these risks.
Market power in energy-only markets
This is the argument used in the United States, as previously discussed. Concern with market power leads to low price caps, which then lead to missing money. A capacity market is then required to restore investment incentives.
Security of supply
The other justification for capacity mechanisms relies on security of supply. Historically, power companies determined their target generation capacity by combining estimates of future demand with a security of supply criterion. First, engineers produced a distribution of maximum demand 5 or 10 years ahead. In these models, demand is primarily driven by economic growth and climate conditions, not by electricity prices. Thus a cold winter growth results in a high maximum demand, while a warm winter produces a low maximum demand.
Second, policy makers set a security of supply criterion. In France for example, this takes the form of a target for the Loss of Load Expectation: when demand exceeds installed capacity, the TSO must implement rolling blackouts, i.e., selectively cut users for some periods of time. In France, the maximum expected duration of these rolling blackouts is set at 3 hours per year on average.
Then, utilities computed the target capacity that meets the security of supply criterion. For example, they computed the capacity that exceeds maximum demand for all but three hours on average.
Would an energy-only market meet the security of supply criterion? This cannot be guaranteed. As all commodity markets, it would most likely alternate between periods of over- and under-capacity. Furthermore, energy-only prices are highly volatile, hence investors may be concerned about risk, and invest less than is required, in particular for peaking units that run only a few hours per year.
On the other hand, under a capacity mechanism, the TSO computes the target capacity as described above. Thus, a capacity mechanism leads to security of supply level desired by the public authorities.
A variation of this argument involves intermittent renewables. Since the early 2000s, electricity production from wind and photovoltaic sources has expanded enormously in Europe, financed by generous subsidies. As a result, conventional plants are facing serious financial difficulties and many are closing.
The replacement of conventional with renewable generation is not a problem in itself; on the contrary it is precisely the desired effect. However, it is possible that the pendulum is swinging too far and that power plants that should be available to produce during wind-less winter evenings are being closed. This argument is a specific version of the previous one. If energy-only markets were perfect, they would lead to the adequate capacity, i.e., there would be precisely enough thermal plants to produce when the wind is not blowing. However, market imperfections could lead to under-capacity.
3 Rationale for energy-only markets
The arguments for energy-only markets can be structured along the same categories: market power can be controlled with other instruments than a price cap, and security of supply is less relevant today than it was in the past, since demand can respond to price.
Vulnerability to market manipulation
Potential exercise of market power in energy-only electricity markets is a serious issue, which can be mitigated in two ways: demand elasticity, and regulatory oversight (other than price cap). Finally, capacity mechanisms are also subject to manipulations.
First, demand has historically been relatively inelastic precisely because retail prices varied very little compared to wholesale prices. If retail prices were to vary significantly, consumers would make arrangements to adjust their demand, i.e., they would invest in information tools and control mechanisms to react to price signals. When airline companies adopted peak load pricing in the 1980s (known as yield management), consumers changed their behaviour. Faced with sharp price variations, electricity demand would become far more elastic, thus significantly reducing unwarranted price increases by producers. Thus, an energy-only market dynamically leads to a reduction of the potential impact of generators’ market power.
Second, the regulatory authorities (in France the Competition Authority and the Energy Regulatory Commission) can and must ensure there is careful scrutiny in order to detect and punish anti-competitive behaviour. The underlying micro-economics are by now well established, hence these controls should be effective.
Third, the experience of the United States shows that capacity mechanisms are also open to manipulation. Generators and demand response operators receive payments for promised capacity. However, this promised capacity is not always used, thus creating an incentive to promise more than can be delivered. The system must therefore be closely monitored.
For example, one of the aims of capacity mechanisms is to encourage the construction of new generating facilities. Let’s assume that a producer commits to building a power plant that will come on stream in 5 years. In the meantime what can the TSO actually monitor? How often should it visit the construction site to control the project’s progress?
What about those unavoidable forced outages? Is it realistic for a producer to commit to being available 100% of the time? For example, in the United States producers had to deal with an exceptional cold weather event in January 2014, leading to many power stations becoming unavailable. What is the acceptable threshold? Outages are less of a handicap for the larger power producers, who can diversify the risks across all their production facilities: it is easier to guarantee 95% availability with 10 power stations than with just one.
The issue of the credibility of commitments becomes even more complicated with demand response. It is already difficult to assess the reality of demand response in a 24 hour period. So what value can be given to demand response promised in 5 years time? The experience of New England shows that providers promised a high level of demand response, which reduced prices on the capacity markets, and consequently the payments from suppliers to producers. But the demand response promises were not all kept and the system experienced real difficulties. If a TSO accepts all demand response promises (referred to as explicit demand response), it is exposing the system to significant risk. To be on the safe side it is therefore likely that TSOs will de facto prefer production over explicit demand response.
These are more than just technical or measurement problems. California’s experience shows that electricity market operators (producers and demand response operators) are ready to take advantage of all the ambiguities in the rules to increase their profits. It would be naïve to assume that the same would not happen in Europe.
The new world of demand elasticity
A strong argument for energy-only markets is that they rely on demand adjusting to supply, and not the opposite. Capacity mechanisms are inspired by the approach used by electric power utilities throughout the XX century (and described above). While it was highly effective then, it does not take into account the information technology revolution: customers can now adapt their consumption to installed capacity through price in (almost) real time. In a somewhat ironic coincidence, RTE unveiled its capacity markets at about the same time that Tesla launched its Powerwall home battery, which will contribute to residential demand’s responding to supply conditions.
Will there be enough demand elasticity to adjust to supply in all possible circumstances? Maybe not initially. Thus, an administrative criterion governing rolling blackouts may be required. Must we then use a failure criterion? No. Rather than a target for the Loss of Load Expectation, TSOs (or policy makers) should use a Value of Lost Load: when load needs to be shed, the wholesale price is automatically set at this value.
The distinction between a Loss of Load Expectation criterion and a Value of Lost Load may appear semantic, since in both cases policy makers set the figure. But the difference is actually crucial: when they use a Loss of Load Expectation criterion, policy makers set de facto the capacity to be installed; when they set a Value of Lost Load, individual decision-makers have the freedom to choose how adapt to this extreme scarcity value, which encourage innovation in new and better ways to manage scarcity.
What about risk? Capacity markets are not risk-less. Investors simply trade market risk for regulatory risk. Experience in the United States shows that capacity mechanisms require complex and detailed rule setting, which creates significant regulatory risk.
* * *
Capacity mechanisms are the most recent example of a worrying trend among policy makers: administrative supervision of electricity markets. We should expect in Europe the cycle experienced in the United States: first, TSOs develop detailed rules governing capacity mechanisms, then market participants exploit the ambiguities in these rules to make profit, hence the security of the system is compromised. Therefore, TSOs draw up even more prescriptive and detailed new rules, which participants will flout… and so the cycle continues.
Another, far simpler approach, is to set the Value of Lost Load at the appropriate level (for example €20 000/MWh, which corresponds to the 3 hour per year criterion), and … trust consumers and investors. Trust is not blind faith, rather it follows the “trust, but verify” principle, which requires the adoption of a rigorous competition policy to ensure that operators do not exercise their market power, especially during scarcity periods. Trust would lead to technical and contractual innovations to better manage scarcity, hence would truly lead us on the path to energy transition.