Economics and Design of Capacity Markets for the Power Sector

This and the other sections build on—and are partly taken from—earlier work, including Cramton and Stoft (2005, 2006, 2008), Joskow (2007), Stoft (2002), Ockenfels (2007a, 2008b), as well as on practical experiences in designing capacity, reserve and other energy markets. A German summary of this work can be found in Cramton and Ockenfels (2011a).

As we have learned during the crisis, financial markets, too, do not always clear.

But note that there have been substantial advances in pump storage and in the future many will be driving electric cars that offer a large quantity of battery storage when the cars are plugged into the grid.

If the administrator knew that consumers have an average value of lost load of € 100,000/MWh, the optimal capacity level can be calculated by analyzing a well-designed market with administrative price setting. Whenever the supply and demand curves fail to intersect (a market failure so gross it is not analyzed in any economics text), the price should be set by the administrator to € 100,000/MWh to reflect the best interests of consumers. Peakers will earn between € 0 and € 400/MWh in fixed cost recovery for one hour (half an hour on the way up and half an hour on the way down) on hot days. This will result in € 2,000/MW-year of fixed cost recovery out of € 80,000 needed. Consequently € 7,800/MW per hot day must be recovered from VOLL pricing. This requires 0.078 hours of VOLL per hot day. With demand shifting at 2 GW per hour, demand must exceed supply by 78 MW at the peak. Hence the optimal capacity level is 50,000 MW minus 1,000 MW of demand elasticity minus 78 MW of uncovered demand = 48,922 MW of capacity.

That new capacity was endowed with free CO₂ emission certificates reduced net investment costs and thus also incentivized new entry in Europe. Clearly, this too, is not an efficient way to handle reliability issues.

Often, this assertion is based on a misconception of scarcity prices, which do not reflect the incremental cost of the last produced unit, but nevertheless are fully consistent with marginal cost pricing and perfect competition (Sect. 2.3).

That suppliers earn rents on baseload generation is irrelevant for the decision to build peakers. And it is irrelevant for the argument that there is missing money when there is no shortage of resources. In an efficient equilibrium mix of resources, all plants need scarcity prices to cover their total costs. In fact, any type of plant could be used as a benchmark here. See, e.g., Grimm et al. (2008a, 2008b), Stoft (2002), Joskow (2007), Ockenfels (2007a, 2007b) and Ockenfels et al. (2011) for the pricing mechanisms in competitive electricity markets.

More detailed discussions and illustrations of the scope for market power depending on capacity reserves can be found, e.g., in Ockenfels (2007b). In fact, unlike in other industries where capacity constraints are less strict due to cheap(er) storage opportunities, market power measures specific to the electricity sector can be read as decreasing functions in excess capacity (DG Competition 2007a, 2007b, Bundeskartellamt 2011). This holds, for instance, for the Pivotal Supplier Index, which shows for every given hour whether a given supplier is “pivotal,” i. e. necessary to meet demand. A supplier is more likely to be pivotal when the (residual) capacity level is scarce. Accordingly, market power studies typically found more exercise of market power in peak hours (see Müsgens 2006 for Germany, Borenstein et al. 2002 for the US).

See Ockenfels (2008a) for more details and a discussion of this and other flaws in defining and estimating competitive prices in Germany’s electricity market. See Nitsche et al. (2010) for one of the very rare attempts to take long-term considerations into account when judging German electricity prices.

Without long-term contracts or vertical integration opportunism are known to possibly lead to underinvestment, as shown by, e.g., Williamson (1979), Hart (1995), and Joskow (1987).

Price caps that sometimes constrain competitive market prices but overall allow coverage of total costs do not necessarily imply a shortage of capacity, but would induce the wrong mix.

Voltage reductions imply an inefficient use of electric equipment, and so is costly to electricity consumers; these costs, too, are not reflected in market prices.

On the other hand, the costs of a shortage of, say, 2 % of capacity is probably orders of magnitude higher. Suppliers, when left to their own devices, would typically prefer to err on the side of underinvestment.

We note that a bid cap is not necessarily equal to a price cap, because prices could in principle be negotiated outside a market platform that imposes the bid cap. However, this is not critical to our design proposal, because the exact level of the bid cap is not critical, bilateral real-time negotiations are difficult in electricity markets, and because the incentive to withhold capacity in our design are small anyway. What is critical, though, is that—to the extent scarcity is a real-time problem—there is a reliable real-time market price, and that the relevant capacities can be quickly dispatched.

Another possibility would be a vertical demand curve at the target. However, having some elasticity around the target is consistent with what we know about the marginal value of capacity to load and has the additional benefit that it at least slightly mitigates supply side market power.

Lumpy investments are respected: the investor does not have to fear partial acceptance. If multiple bidders drop at the clearing price, the group of bids are accepted to balance supply and demand as closely as possible.

This may not be the case if, for instance, there is only a limited number of suitable sites and these are owned by few suppliers. In this case, other measures to mitigate market power are necessary.

Retirements may be rejected for reliability reasons, but only if the reliability problem cannot be resolved during the planning period with alternative actions, such as transmission upgrades or new capacity.

Why base the hedge on load share? New England has about 30 GW of capacity, but sometimes, due to cold weather, many generators cannot run, and the price has spiked with as little as 20 GW of load. If reliability options covered the full 30 GW, then load would be paid for 30 GW times the $1000 spot price less the $300 strike price. Hence load would profit by $7 million dollars per hour during such an incident. This upsets generators without reason, and causes them to worry that extra capacity will be purchased so load can profit more in this way. Basing the reliability option on load share solves this problem by putting the generators in a nearly balanced position in every hour.

The efficient price would be the value of lost load, which is however not known and if it were know, it would probably not be politically acceptable. In fact, to our knowledge, no such system even attempts to employ value of lost load bidding.

An increasing share of renewables also requires an integration of intraday and balancing markets.

Based on his long experience in regulating the US electricity and other sectors, Joskow (2010) describes what we think is a not an atypical phenomenon in the administration of market regulation and design: “The regulatory process is subject to interest group capture, political influence, and tremendous pressure to engage in (hidden) taxation by regulation […]. The modern field of political economy based on rational actor models of political behavior did not start with studies of regulation by accident. This phenomenon goes well beyond simplistic models of capture by regulated firms and reflects the fact that regulatory agencies have things that they can do to help one interest group and harm others, naturally leading them to become targets of political competition. This phenomenon is exacerbated over time as young ‘expert’ regulatory agencies become dominated by commissioners and senior staff who have come up through the political process and are sensitive to the same political considerations as are their sponsors in the executive and legislative branches and those they regulate. In my view, this has become a more serious problem over time as ‘independent’ regulatory agencies once heavily populated by reasonably independent technocratic experts with clear public interest goals have increasingly come to be populated by commissioners and senior staff with narrower political goals—whether it is on the less regulation or more regulation extremes of the political spectrum depending on which political faction is in power.”

Moreover, the details of the auction and product design of capacity markets (e.g., with respect to the rating of capacity bids) with unclear future transmission constraints and technology mixes are likely to be more complicated than what has been implemented before in other, more settled markets.