Exhibit 99.440
A REPORT TO THE CALIFORNIA POWER
EXCHANGE: THE BENEFITS OF A
SIMULTANEOUS VERSUS SEQUENTIAL PX
MARKET FOR ENERGY AND ANCILLARY
SERVICES
March 2, 1999
Dr. Peter H. Griffes
Analysis Group/Economics
-2-
A REPORT TO THE CALIFORNIA POWER EXCHANGE:
THE BENEFITS OF A SIMULTANEOUS VERSUS
SEQUENTIAL PX MARKET FOR ENERGY AND ANCILLARY
SERVICES
I. INTRODUCTION AND SUMMARY OF THE STUDY.................................................... 4
II. AN OVERVIEW OF THE CURRENT ANCILLARY SERVICES MARKET..................................... 7
A. BIDDING AND BID EVALUATION........................................................... 8
B. SELF-PROVISION OF ANCILLARY SERVICES................................................. 9
C. THE PX AND SELF-PROVISION............................................................ 9
D. REAL-TIME DISPATCH OF ANCILLARY SERVICES AND SUPPLEMENTAL ENERGY..................... 9
E. SUPPLY REMUNERATION FOR ANCILLARY SERVICES........................................... 10
F. DEMAND CHARGES....................................................................... 11
III. PERFORMANCE OF THE ENERGY AND ANCILLARY SERVICES MARKETS................................. 11
A. ANCILLARY SERVICES MARKETS........................................................... 11
B. THE PX ENERGY MARKETS................................................................ 12
IV. PROPOSED IMPROVEMENTS TO THE ISO ANCILLARY SERVICES MARKET............................... 14
A. LINEAR PROGRAMMING OPTIMIZATION APPROACH............................................. 14
B. SMART BUYER APPROACH................................................................. 14
C. COMPARISON OF THE THREE APPROACHES................................................... 15
V. THE RELATION BETWEEN PROVISION OF ENERGY AND ANCILLARY SERVICES.......................... 16
A. POLICY OBJECTIVE IS TO GAIN EFFICIENCY WHILE RETAINING UNBUNDLED SERVICES............ 16
B. RELATION OF THE ENERGY AND ANCILLARY SERVICES PROVISION.............................. 16
1. The temporal aspect of these markets.............................................. 17
2. The substitutability of these services in production.............................. 17
3. Application of these concepts to California institutions.......................... 18
VI. A MODEL FOR PX PROCUREMENT OF ENERGY AND ANCILLARY SERVICES.............................. 20
A. DESCRIPTION OF THE HYPOTHETICAL MARKET............................................... 20
B. BIDDING INTO THE PX MARKET........................................................... 21
C. BID TYPE............................................................................. 21
1. Demand side....................................................................... 21
2. Supply side....................................................................... 22
D. PX EVALUATION OF BIDS................................................................ 23
E. NOTIFICATION OF SELECTED BIDS........................................................ 24
F. PRICING OF ENERGY AND ANCILLARY SERVICES............................................. 24
1. FERC Requirements for Pricing Unbundled Services.................................. 25
2. Alternative Pricing Mechanisms.................................................... 26
a. Pricing based on marginal costs................................................... 26
b. Pricing based on the highest bid providing the service............................ 27
c. Pricing based on indifference of markets.......................................... 29
3. Pricing in the ISO ancillary services market...................................... 30
G. SUPPLY REMUNERATION AND DEMAND PAYMENTS.............................................. 31
VII. CHANGES REQUIRED TO IMPLEMENT THE PX MARKET.......................................... 32
-3-
A. ISO OPERATIONS....................................................................... 32
B. PX OPERATIONS........................................................................ 32
VIII. THEORETICAL IMPACTS FROM A JOINT ENERGY ANCILLARY SERVICES MARKETS................... 33
A. POTENTIAL FOR LOWER COSTS............................................................ 33
B. POTENTIAL FOR HIGHER PRICES.......................................................... 34
1. Impact on Prices under Marginal Cost Pricing...................................... 34
2. Impacts on Prices Under Highest Cost Resource Pricing............................. 35
3. Impacts on Prices Under Indifference of Markets Pricing........................... 36
IX. ESTIMATING THE EFFICIENCY GAINS FROM A SIMULTANEOUS MARKET............................... 38
A. SIMULATION ANALYSIS WILL BE REVEAL MORE THAN AN ANALYSIS OF HISTORIC BIDS............ 38
1. Practical reasons why an analysis of historic data is not helpful................. 38
2. Simulation analysis can be enlightening........................................... 38
B. DESIGN OF THE MARKET SIMULATIONS.................................................... 39
1. Fully sequential evaluation methodology........................................... 39
2. Sequential-simultaneous evaluation methodology.................................... 40
3. Simultaneous evaluation methodology............................................... 40
C. EVALUATION CRITERIA BETWEEN THE MODELS............................................... 41
X. NUMERICAL COMPARISON OF THE EFFECTS OF SIMULTANEOUS AND SEQUENTIAL MARKET AUCTIONS....... 41
A. ASSUMED MARKET SUPPLY AND DEMAND CONDITIONS.......................................... 42
1. Representative Market Supply Curve................................................ 42
2. Representation of demand.......................................................... 43
B. DESCRIPTION OF THE SPREADSHEETS...................................................... 44
1. Model inputs...................................................................... 44
2. Model outputs..................................................................... 45
C. MODEL RESULTS........................................................................ 48
1. Impact on Production Costs........................................................ 48
2. Impact on Consumer Costs.......................................................... 49
a. Evaluation techniques for each pricing methodology................................ 50
b. Pricing methodology for evaluation techniques..................................... 52
XI. ASSESSMENT OF SIMULTANEOUS AND SEQUENTIAL EVLAUATION OF ENERGY AND ANCILLARY SERVICES
MARKETS UNDER DIFFERENT PRICING METHODS ................................................. 54
XII. CONCLUSIONS AND RECOMMENDATIONS...................................................... 55
-4-
THE BENEFITS OF A SIMULTANEOUS VERSUS
SEQUENTIAL PX MARKET FOR ENERGY AND ANCILLARY
SERVICES
I. INTRODUCTION AND SUMMARY OF THE STUDY
In its filing to the Federal Energy Regulatory Commission (FERC) in March 1997,
the PX proposed to implement its own ancillary services auction that would
enable the PX to self provide all or a portion of its ancillary service
obligations to the ISO.(1) In October of 1997, the FERC conditionally authorized
the operation of ISO and the PX. The October 30th order supported the
development of both an ISO and PX ancillary services market.(2)
Although the FERC granted the PX the authority to implement an interim
sequential market of its own, Commissioners questioned whether greater
efficiencies would be achieved if the PX implemented a simultaneous market for
ancillary services. To address this concern, the FERC ordered the PX to file a
study that analyzes the merits of developing a simultaneous auction for both
energy and ancillary services. Specifically, the FERC noted:
Whether sequential or simultaneous auctions are more efficient is an
unresolved empirical question. Accordingly, we will approve the
proposal for sequential auctions on an interim basis in order to gather
experience with which to evaluate the proposal more fully at a later
date. To assist us in that evaluation and to address the concerns
articulated above, we will require the PX to conduct further studies,
and to file a report on their results by January 1, 1999, at which time
we may revisit the issue. The studies should analyze and compare
sequential and simultaneous auctions in terms of their abilities to
develop an efficient, least-cost dispatch(3)
This report responds to the FERC request for a study of the merits of sequential
and simultaneous auction models.
While FERC clearly states the analysis should compare simultaneous and
sequential auctions, it is unclear what delineates a sequential auction from a
simultaneous one. There are two ways to
- ----------
(1) PX March 31, 1997 FERC filing, Section 3.3.4.
(2) Docket Number EC96-19-001 et al.
(3) October 30th order, p. 194-195.
-5-
interpret this. First, there could be simultaneous bidding where the bids for
the energy market are also used for ancillary services; however, the evaluation
of the bids would take place in a sequential fashion. Second, there could be
simultaneous bidding and evaluation of bids for energy and ancillary services
where the same bids are used for both markets and are considered together at the
same time to meet the needs in all markets.
Since the opening of the market in April 1998, significant problems have arisen
in the ancillary services markets. Price caps had been imposed, lifted and
re-imposed. As a result of the summer price spikes in these markets, FERC
required a report(4) from the Market Surveillance Committee of the California
ISO (MSC). On top of the MSC's list of recommendations, was for the ISO to adopt
practices that allowed it to "substitute cheaper superior services for more
expensive inferior services in its procurement of ancillary services.(5)" The
discussion in the MSC report makes it clear that consistency between the bids
for the various services would contribute significantly to attenuating the
instability in these markets.(6)
The hypothetical framework in this report imposes this same restriction of
consistency between bids for different PX markets. With a simultaneous bidding
framework, regardless of whether there is sequential or simultaneous evaluation,
the day-ahead energy bids could be used for either energy or ancillary services.
Tying the markets together could result in the greater competitiveness in the
energy market carrying over to the ancillary services markets. The framework
requires that the willingness to sell capacity for energy is the same as the
willingness to sell capacity for ancillary services. Thus, it requires that the
bids for the same capacity are identical regardless of the market, energy,
regulation, spin, non-spin or replacement. This is more restrictive than the bid
consistency rules suggested by the MSC where bids for inferior services could be
no higher than bids for superior services.
There are numerous ways to price energy and ancillary services in such a
framework. Three particular methodologies are outline in this report. They are
marginal cost pricing, pricing based on highest bid accepted and pricing based
on indifference of markets. Marginal cost pricing is obvious. Pricing based on
the highest bid accepted is similar to marginal cost pricing but does not take
into account the joint production of these services. Indifference of markets
pricing sets prices in such a way that sellers are indifferent between selling
energy or ancillary services. Each of these pricing methodologies will produce
different prices depending on whether bid evaluation has been done sequentially
or simultaneously.
Because of the desirability of bid consistency, this report posits a
simultaneous bidding framework. It then examines the question of whether there
are efficiency gains from the
- ----------
(4) Preliminary Report On the Operation of the Ancillary Services Markets of
the California Independent System Operator (ISO), Prepared by the Market
Surveillance Committee of the California ISO, August 19, 1998. (MSC Report)
hereafter.
(5) MSC Report, p. 37.
(6) MSC Report, p. 38-41.
-6-
simultaneous evaluation of energy and ancillary services markets in relation to
the sequential evaluation of the markets.
The analysis addresses three different ways to evaluate these markets under each
of the three pricing methodologies. The first is `fully simultaneous', where all
markets, energy and ancillary services markets, are considered simultaneously.
The second is `sequential, simultaneous' where the energy market is evaluated
before the ancillary services markets which are evaluated simultaneously. The
third is `fully sequential' where the markets are evaluated in the following
sequence: energy, regulation, spin, non-spin, and replacement. The prior
expectation is that `fully simultaneous' will produce the lowest costs, followed
by `sequential simultaneous,' and `fully sequential' will have the highest
costs.
In order to evaluate these options, nine separate spreadsheet models were
constructed. Each spreadsheet uses a different evaluation technique and pricing
methodology for the energy and reserve markets. Identical inputs are assumed for
market supply and demand conditions. This allows for the isolation of the
evaluation technique and pricing methodology as the determinants of the results.
The key inputs of the model are different levels of demand, the market supply
curve, and the ramp rates assumed for each ancillary service for each portfolio
bid.
In the analysis performed, there was an effort to replicate a general PX supply
curve over the entire time period. Similarly, the demands used represent samples
at regular intervals of the demand distribution. However, less care was taken
with ramp rates, which were simply assumed without a firm check in reality. The
magnitude (but not direction) of the numerical results depend significantly on
the particular assumptions made.
There is a distinction between dispatch costs to producers and the out-of-pocket
costs to consumers. Depending on the pricing methodology used, it is possible to
have an efficient dispatch but relatively high costs to consumers. While FERC's
request specifically cites dispatch costs, the analysis also focuses on costs to
consumers of energy and ancillary services.
The results are a bit surprising. The level of costs depends significantly on
perspective. As with most markets, the costs incurred by producers are not the
same as the costs incurred by consumers. The major difference between them is
commonly known as the producer's surplus and depends considerably on how prices
are set in the market. In this analysis, the results vary by whether the
producers or consumers perspective is adopted.
Because the energy market is competitive most of the time, it is not
unreasonable to assume suppliers' bids are indications of their incremental
costs. The efficiency of the dispatch (or cost to producers) follows
expectations partially. Namely, `fully simultaneous' produces the most efficient
dispatch. The dispatch under `sequential simultaneous' is identical to that
under `fully sequential.' This is a result of the particular ramp rates assumed.
In general, this need not be the case and whether it is true in reality depends
on the ramp rates that would result.
-7-
The FERC's concern lies with the efficient dispatch of the generators in the
market. The market simulations show that, on an annual basis, there are
approximately $24 million savings in dispatch costs from using fully
simultaneous evaluation rather than a sequential approach.
However, costs to consumers (or revenues to producers) are a different matter.
Pricing methodologies affect these costs significantly. The dispatch cost saving
comes because relatively inexpensive capacity that would be allocated to energy
under a sequential approach is used to substitute for more expensive capacity
used for ancillary services. The various pricing methodologies influence whether
sequential or simultaneous evaluation is more desirable. Under two methodologies
examined, sequential evaluation produced lower costs to consumers than
simultaneous. However, the opposite was true for the third methodology.
The conclusion of this report is that the efficiency of dispatch does not align
with the costs to consumers. The pricing methodologies that produce the lowest
dispatch costs under simultaneous evaluation also produce the highest costs to
consumers. There are also other features of pricing mechanism that are
attractive such as the ability to induce bidding of marginal cost, incentives to
favor one market or another among others. None of the pricing methodologies and
evaluation techniques meets all of the objectives. The one that should be chosen
depends on the relative benefits and drawbacks of each pricing/bid evaluation
methodology.
The remainder of this report is laid out as follows. Section II summarizes the
existing ancillary services markets at the ISO. Section III discusses the
performance of the energy and ancillary services markets since they opened.
Because of instability in the markets, the ISO is currently examining ways to
improve the operation of these markets. Section IV outlines the improvements
under consideration. Section V discusses the general relation between energy and
ancillary services markets with a particular focus on California. Section VI
outlines a proposal for introducing combined energy and ancillary services
market in the PX, including various pricing methodologies. Section VII examines
some of the changes needed to introducing such a framework in the PX.
The latter sections of the report address the question posed by FREC about the
simultaneity of the markets. Section VIII addresses the theoretical benefits
achievable from a simultaneous evaluation of energy and ancillary services.
Section IX introduces the simulation framework for estimating the benefits.
Section X describes the models and methodology followed in the analysis. It also
reports the results from the models. Section XI evaluates the desirability of
pricing methodologies and evaluation techniques in light of the empirical
results. Section XII concludes the report.
II. AN OVERVIEW OF THE CURRENT ANCILLARY SERVICES MARKET
The ISO is responsible for ensuring that adequate ancillary services exist to
support the dispatch and consumption of power on the grid. Currently, the ISO
operates a day-ahead and hour-ahead competitive market to supply four ancillary
services -- regulation, spin, non-spin
-8-
and replacement power.(7) The ISO determines the quantities of each ancillary
service that will be required based on WSCC and NERC requirements and ISO
estimates of day-ahead forecast load.
A. BIDDING AND BID EVALUATION
Via their Scheduling Coordinators (SCs), generating units, curtailable demand
and external import/export resources may bid to supply ancillary services into
the ISO auction. Suppliers submit two-part (capacity and energy) bids to the ISO
for each of the four auctions in which they seek to bid. For each ancillary
service offered, SCs must include a bid price for energy in the form of a
staircase function composed of up to eleven ordered, quantity-price pairs of
information.(8) Dispatchable load may also bid to provide non-spin and
replacement reserves. Bids must contain information that lets the ISO validate
that the resource offered meets the technical requirements for the particular
service.
The ISO ancillary service market is run as a sequential auction. The ISO
receives bids for all four auctions in the day-ahead and hour-ahead markets and
evaluates and selects "winning" bids in the following order: regulation;
spinning reserve; non-spinning reserve; and replacement reserve.
Each SC may specify the markets into which it wants to bid ancillary service
capacity. With the exception of down regulation, capacity selected by the ISO in
one of the markets is subtracted from the total capacity offered into the market
by a bidder. If designated by the bidder, any capacity that is not selected in
the preceding market may be passed on into the next auction for consideration.
Different capacity prices may be specified for the same capacity in each of the
markets.(9)
The ISO evaluates bids based on the capacity price in selecting the entities
that are designated to provide ancillary services. For the day-ahead auction,
the capacity prices paid for each ancillary service are posted on the ISO
website by 3 p.m. on the day before the operating day.
- ----------
(7) Black-start capability and voltage control are procured by the ISO on an
annual basis under contract. Currently, these requirements are met by
Reliability Must Run (RMR) units. Regulation has been split into up and
down regulation which are currently being evaluated separately. In what
follows, the term `regulation' will generally refer to up-regulation unless
otherwise noted.
(8) ISO Tariff, p. 285.
(9) ISO tariff, p. 79
-9-
B. SELF-PROVISION OF ANCILLARY SERVICES
The ISO's ancillary service requirements may be self provided by SCs.(10) Load
choosing to self-provide may either contract directly with a generator or a
broker to ensure adequate services are supplied. SCs then submit a
self-provision schedule for each ancillary service, designating the unit, load
or system resource that will be called upon to provide the ancillary services in
real time.
Partial self-provision of ancillary services is permitted with any
non-self-provided portion being procured in the ISO's market. For the
self-provided portion of load, the SC must submit a proxy energy bid,
representing price at which the designated resource may be dispatched in real
time. As with bid-in resources, self-provided resources must be certified by the
ISO to comport with the ISO's technical requirements for providing the ancillary
service.
C. THE PX AND SELF-PROVISION
As a SC, the PX has a right to self-provide ancillary services.(11) Self
provision would require the PX to operate its own separate auction for ancillary
services and forward to the ISO information about winning resources selected by
the PX to cover ISO-enforced ancillary service obligations. In addition, the
FERC has acknowledged that PX participants should be allowed to contract
bilaterally with generation resources to meet any of their own ancillary
services obligations without utilizing either a PX or ISO auction.
Currently, the PX neither operates ancillary service markets nor self-provides
ancillary services, but procures them from the ISO market. Any PX resource
meeting the technical requirements for supplying ancillary services may bid into
the ISO auction. After clearing its day-ahead energy market, the PX acts as an
intermediary by accepting the bids from its participants and passes them on to
the ISO but otherwise has no role in the ISO auction. Similarly, in the
hour-ahead market, ancillary bids are submitted to the PX no later than two
hours prior to the dispatch hour.(12)
D. REAL-TIME DISPATCH OF ANCILLARY SERVICES AND SUPPLEMENTAL ENERGY
The ISO dispatches capacity providing ancillary services and supplemental energy
in real-time to ensure that reliability standards mandated by the WSCC and NERC
are met. The ISO calls on resources to supply incremental energy to keep the
system in balance. It also requires generators to decrement generation resources
to correct oversupply.
- ----------
(10) Voltage Support and Black Start may not be self provided under the ISO
Tariff and will be procured by the ISO for all SCs.
(11) See December 1997 FERC decision, p. 23 and October 30 decision, p. 16.
(12) Report on Market Issues in the California Power Exchange Energy Markets,
The Market Monitoring Committee of the California Power Exchange, August
17, 1998, p. 6. (MMC Report).
-10-
Additional energy in real-time is supplied from five possible sources: the
winning bids submitted in the four ancillary service markets and from
supplemental energy bids.(13) The ISO assembles the energy bids from ancillary
service providers and supplemental incremental energy bids into a single
merit-order stack of system-side resources.(14) Any incremental energy needed in
real-time is drawn from this stack when there is under-supply. Similarly, the
ISO assembles a decremental merit-order stack, consisting of scheduled
generation that willing to be decremented in real time. The decremental stack is
used when there is oversupply.
Thus, while ancillary service resources are selected to stand ready to provide
real-time energy on the basis of their capacity bids, these resources are only
dispatched if they are the least-cost alternative available to the ISO in
real-time. If a supplemental energy bid is less costly, the ISO will draw on
this resource first to provide real-time energy.
E. SUPPLY REMUNERATION FOR ANCILLARY SERVICES
Resources providing ancillary services are paid a capacity payment and an energy
payment if called in real-time. For each ancillary service, the unit with the
highest capacity bid selected to provide the service sets the capacity payment.
That is, the last bidder whose capacity is accepted in the day-ahead or
hour-ahead market by the ISO to stand ready sets the market-clearing price for
each ancillary service.(15) Whether or not the resource is actually called to
provide energy in real-time, resources are remunerated for capacity if selected
to stand ready. Suppliers of regulation also receive a Regulation Energy Payment
Adjustment (REPA).(16)
Until early November 1998, cost-based caps limited the capacity remuneration of
some ancillary service suppliers. Since market inception, all utility-owned
generation has been under a cost-based cap for all ancillary services. They
range from between $4.47 to $9.55, depending on the service. Southern California
Edison, for example, is capped at $4.47/MW for replacement reserve bids.
Energy remuneration for ancillary services is based on real-time dispatch.
Instructed deviations are paid the 10-minute price for energy relevant for the
time in which the ISO instructs the
- ----------
(13) Supplemental incremental energy is bid into the ISO by generators that have
uncommitted capacity. They can be submitted anytime after the day-ahead
market but cannot be withdrawn within 45 minutes prior to start of
real-time hour.
(14) The energy bids for spin, non-spin, and replacement reserves are added to
the stack.
(15) If congestion exists, the ISO establishes zonal market-clearing prices in
each ancillary service market. The details of these payments can be found
in the ISO tariff, section 2.5.
(16) Under Amendment 8, filed with the FERC on May 19, 1998 the REPA is equal to
the energy potentially available in the regulation bid (R(up) + R(down))
multiplied by the greater of $20/MWh or the hourly ex-post price. Effective
November 23, 1998, REPA payments were set to zero.
-11-
generator or load to supply ancillary services.(17) Because the ISO may need to
constrain on or constrain off a resource, there are two prices for each
interval, the incremental 10-minute price (for resources called on) and the
10-minute decremental price (for resources constrained off). Each is set at the
price of the last or marginal unit for generation (or load) that is called to
adjust its schedule over a 10-minute period.
F. DEMAND CHARGES
The ISO charges all SCs for their share of the total costs of providing the four
ancillary services that it buys from competitive markets. In its daily
settlement process, the ISO determines the hourly user rates charged for each
service for each settlement period for both the day-ahead and hour-ahead
markets. For each ancillary service procured in the ISO auction, the ISO
calculates charges based on the ratio between the SC's forecast hourly demand
(less any ancillary services self-provided) and the total demand scheduled by
all SCs in that hour for each zone.(18)
III. PERFORMANCE OF THE ENERGY AND ANCILLARY SERVICES MARKETS
A. ANCILLARY SERVICES MARKETS
A number of problems have been observed in the ISO ancillary services market
since its inception, the most publicized of which was large price spikes in the
replacement power market in mid-July. Replacement reserves reached $5,000/MWh on
the trading day of July 8 for power delivered on the 9th. On July 12, prices for
replacement power delivered on July 13 spiked to $9,999/MWh for the hours
between 2 and 6 p.m.(19)
The existence of cost-based caps together with market-based rates for a limited
number of owners has been attributed to causing significant price spikes in the
ancillary services market in July. Because of price volatility, the FERC
authorized the imposition of a market-based cap of $250/MW for ancillary
services. Thus, utilities and IPP continue to be regulated under a cost-based
cap and are paid a maximum of their capped rate for providing service. Suppliers
authorized for market rates can receive no more than $250/MW.
- ----------
(17) The ISO's automated software system, Balancing Energy and Ex-Post Pricing
(BEEP), prices real-time energy. In the original market design, BEEP was
designed to determine the dispatch instruction required to keep the system
in balance on the basis of a five-minute interval. Due to system
limitations, a five-minute interval is not feasible, and thus BEEP
calculates prices on a 10-minute interval.
(18) Prior to mid-August, the ISO procured ancillary services based on scheduled
load, as per the ISO's April tariff, Section 2.5.20.1. Because some SCs
were deliberately under-scheduling load in the day-ahead market to avoid
ancillary service charges, the ISO received approval from the ISO Board of
Governors to adopt forecast SC loads to calculated ancillary service
demand.
(19) In contrast, average prices for replacement power in the months of April,
May and June north of Path 15 were $8.02, $7.93, and $4.28 per MW,
respectively.
-12-
In July, the FERC directed the ISO Market Surveillance Committee and the
California Power Exchange (PX) Market Monitoring Committee to conduct
independent studies regarding the performance of the competitive energy and
ancillary service markets in California. The reports highlighted several areas
in which the current market structure has lead to inefficiencies. These issues
are reviewed below.
The ISO's Market Surveillance Report (MSR) concluded that because investor-owned
utility (IOU) generation has been under price caps for ancillary services since
the start of the market, incentives for utilities to bid into the ancillary
service market have been dampened.(20) Even after divestiture of fossil assets,
IOUs are currently the largest source of ancillary services, and their low
participation rate has provided opportunities for new owners of plant to
withhold capacity to drive up the market price for ancillary services.
The MSR found that the ISO is procuring about twice as much regulation, spin and
non-spin as was procured prior to competition. Most of the over-procurement
relative to pre-market practices takes place in regulation and is a function of
problems with market design. Because the system is self-dispatched, operators
have less control over the actions of generators and the impact on reliability.
Further, there are built in incentives to deviate. Because the energy market
clears without reference to generator ramp rates, awarded schedules may be
difficult to meet given generator ramping constraints. Under the portfolio
bidding structure, bidders are responsible for this, minimizing these effects.
Also, paying instructed and uninstructed deviations different amounts produces
incentives for uninstructed deviations in the ramping hours.
Because the ISO procures services in a cascading, sequential basis, it is often
the case that the least economically valuable ancillary service (e.g.,
replacement) is more highly priced than regulation, which is the most valuable
ancillary resource.(21) The report indicated less rigid rules of procurement
might relieve some of the pressure on the auctions..
B. THE PX ENERGY MARKETS
In comparison to the ancillary services markets, the day-ahead energy market has
been mostly competitive since it has opened. This can be seen readily in the
graph below which plots the price-quantity combination for every hour from the
opening of the market through the middle of December 1998. Each point represents
a price-quantity combination for a single hour. At quantities below about 31,000
MW, the dispersion of points is small and the concentration
- ----------
(20) This is particularly true because utility generation designated as RMR
could earn more under these contracts than through the ancillary service
market.
(21) Quality assertions are made on the basis of how quickly the ancillary
service must be made available. Regulation must be instantaneously
available to the system, whereas replacement reserves must be on line
within two hours of being called.
-13-
tight. This indicates bidding behavior that does not attempt to affect prices
and is consequently competitive.
Above about 31,000 MW the dispersion is much greater and likely represents a
lessening of available capacity. Consequently, it is easier for bidders to
influence market prices. However, it should be noted that only 5 percent of the
hours in the entire eight-month time period have quantities greater than 31,000
MW. This means the day-ahead energy market has been competitive over 95 percent
of the time. Linking the energy market with the ancillary services markets may
provide the opportunity for this competitiveness to spill over into the
ancillary services markets.
The competitiveness of the hour-ahead market is much less pronounced. Its
history is shorter because it only started toward the end of the summer. This
could be a result of the available capacity to participate in the market.
Because the ancillary services market follows the energy market in the day-ahead
timeframe, none of the capacity associated with energy and ancillary services is
available to participate in the day-ahead energy market. Consequently, the
hour-ahead energy market has been very thin, particularly with the ISO buying so
much capacity for reserves.
[PX DAY-AHEAD UNCONSTRAINED PRICE AND QUANTITIES 4/1/98 - 12/15/98 CHART]
-14-
IV. PROPOSED IMPROVEMENTS TO THE ISO ANCILLARY SERVICES MARKET(22)
The ISO has begun efforts to evaluate and improve the operation of its ancillary
services market. As described above, the sequential nature of its auctions and
evaluations has lead to inconsistent prices across the markets. While the
discussions on revising these markets are ongoing at this point, there are two
different concepts under consideration. Despite the fact that the particular
details of implementing each concept has yet to be worked out, it is helpful to
examine the types of improvements under consideration. They can be labeled the
linear programming optimization approach and the smart buyer approach. Each will
be discussed briefly in turn.
A. LINEAR PROGRAMMING OPTIMIZATION APPROACH
Under this approach, the ISO would evaluate the bids for ancillary services in a
simultaneous fashion. Specifically, it would collect the bids for each of the
reserves markets and find the combination of bids that produced the lowest cost
for the provision of the full quantities needed. Because the same capacity can
be bid in each of the reserve markets, the approach allows for a better
allocation of capacity to its highest value across reserve markets. For example,
a low bid for regulation may not be taken in favor of a higher regulation bid
because taking the corresponding capacity for spin displaces a much higher cost
spin bid. The high differential in the spin bids justifies taking the higher
cost regulation bid because the differential is not as great as in the spin
market. In order to implement this type of approach, it would be necessary to
set up a linear programming optimization problem. An example of this approach
will be given below.
B. SMART BUYER APPROACH
Under this approach, the ISO would still evaluate the markets in a sequential
fashion. However, it would not discard any unsuccessful bids from prior markets
in the evaluation of subsequent markets. Specifically, the nearer-term markets,
e.g. regulation, are more valuable than the further-out markets, e.g.
replacement reserves. Because the operating requirements on the further-out
markets are less stringent, any unawarded capacity in the nearer-term markets
could easily fulfill the operating requirements in the further-out markets.
Thus, the ISO, as a smart buyer, would roll over any unsuccessful bids in
nearer-term markets to further-out markets. It would then essentially replace
any higher cost bids in subsequent markets with the unsuccessful lower cost bids
for the same capacity in earlier markets.(23)
- ----------
(22) This section is based on a conversation with Ziad Allywan, Manager of
Market Operations for the ISO on November 16, 1998.
(23) This approach is essentially the same as the `cascading' ancillary service
markets FERC recently ordered for the NEPOOL market.
-15-
For example, suppose a generator's $20/MW bid was not successful in the
regulation market and it bid $30/MW for the same capacity in the spin market. As
a smart buyer in the spin market, the ISO would replace the $30/MW bid in the
spin market with the $20/MW bid for the same capacity that was revealed in the
regulation market. From another viewpoint, the ISO is buying more regulation
capacity (because its cheap) and less spin capacity (because its expensive) and
substituting regulation for spinning reserve in meeting its reserve
requirements. Again a simple example of this will be given below.
C. COMPARISON OF THE THREE APPROACHES
A simple comparison will highlight the differences between these approaches. For
simplicity, assume there are only two reserve markets, spin and replacement
reserves. Further, assume that there are only two scheduling coordinators
bidding into these markets. Their bids and ISO requirements are spelled out in
the following table. In the bid columns, the first number is the quantity and
the number in parenthesis is the capacity bid price.
- --------------------------------------------------------------------------------
Bids Into Each Market
Total Capacity ----------------------------------------
Available Spin Replacement
- --------------------------------------------------------------------------------
SC1 100MW 100MW($1/MW) 100MW($6/MW)
SC2 100MW 100MW($5/MW) 100MW($100/MW)
ISO Requirement 100MW 100MW
- --------------------------------------------------------------------------------
Under the current sequential approach, the ISO first clears the spin market
before evaluating the replacement market. All unsuccessful bids in the previous
markets are discarded. Under this approach, the ISO awards 100MW to SC1 for spin
and 100MW to SC2 for replacement. The cost of this is $10,100 to the ISO ($1/MW
x 100MW + $100/MW x 100MW). SC1 takes home $100 while SC2 receives $10,000.
Under the optimization approach, the ISO considers the markets together to
minimize the total cost of spin and replacement, subject to available capacity.
By jointly evaluating the markets, the ISO realizes that by accepting an
additional cost of $4/MW in the spin market it can achieve savings of $94/MW in
the replacement market. Consequently, it awards 100MW of spin to SC2 and 100MW
of replacement to SC1. This costs the ISO $1,100 ($5/MW x 100MW + $6/MW x
100MW). SC1 gets $600 while SC2 is paid $500.
Under the smart buyer approach, the ISO considers the markets in sequence, first
clearing the spin market. In the spin market the ISO awards SC1 100MW. The ISO
then proceeds to the replacement market; however, the losing bids from the spin
market are not discarded. They are carried over into the spin market and
compared on a SC by SC basis. Because SC1 was awarded 100 MW in the spin market,
it has no capacity left for the replacement market. The ISO, however, compares
SC2's bids between the markets. Because SC2 was willing to sell 100MW of spin at
$5/MW and that capacity is still available, the ISO replaces SC2's $100/MW bid
for replacement with SC2's $5/MW bid for spin. It then awards SC2 with 100MW in
the replacement market at a price of $5/MW. The cost to the ISO is $600 ($1/MW x
100MW + $5/MW x 100MW). SC1 gets $100 while SC2 is paid $500.
-16-
An alternative way to view the smart buyer approach is that the ISO decides to
buy 200 MW of spin and 0 MW of replacement. It still is meeting its reserve
requirements because the operating requirements for spinning capacity are more
stringent than for replacement reserves. In this case, the cost to the ISO is
$1000 ($5/MW x 200MW + 0/MW x 0MW). Both SCs get $500.
It is clear from this example that the lack of a requirement for consistency
between bid values across the reserve markets can significantly increase costs.
V. THE RELATION BETWEEN PROVISION OF ENERGY AND ANCILLARY SERVICES
A. POLICY OBJECTIVE IS TO GAIN EFFICIENCY WHILE RETAINING UNBUNDLED
SERVICES
There are two policy goals of procuring energy and ancillary services. The first
is to have the least expensive resources available provide the energy and
ancillary service requirements for the system. This least cost provision saves
society's resources.
The second is to unbundle the consumption of energy and ancillary services into
separate products. This unbundling of services means that consumers only pay for
the particular service they consume. To achieve effectively this separation, it
is necessary to price separately energy and each of the ancillary services. It
may also require supply and demand to respond to the incentives to produce or
consume each of the markets separately. While consumers cannot appreciably
affect the ancillary services they consume, they can choose different providers.
Similarly, producers respond to prices when deciding on their allocation of
capacity to each market. The relation between these markets will be discussed in
greater detail below.
Depending on the pricing rules employed, these objectives may be at odds with
each other. Specifically, in order to produce separate prices for each service,
it may be necessary to depart from the least cost provision. Similarly, it may
well be the case that the least cost provision requires that energy and reserves
be paid the same price.
B. RELATION OF THE ENERGY AND ANCILLARY SERVICES PROVISION
In order to unbundle effectively, both suppliers and demanders of these services
must respond to market forces. While this should not be of great concern on the
demand side of the market, suppliers will watch each market closely. This
section will briefly describe two aspects of these markets that hinder their
complete separation.
-17-
1. THE TEMPORAL ASPECT OF THESE MARKETS
There is a timing difference between the provision of energy and the provision
of ancillary services. This timing difference may be highlighted by the
structure and timing of the markets for energy and ancillary services.
As noted above, reserve services are capacity markets. Specifically, the service
is unloaded generating capacity in the real time dispatch of the electrical
system. This unloaded generating capacity is available to be used to generate in
the event of an unanticipated change in supply and/or demand conditions. Having
this service significantly reduces the likelihood of a real time failure of the
system.
Consequently, reserve services markets keep generating capacity available but
unloaded, for use in real time. Because the services are procured in advance of
needed generation, they are forward markets for generating capacity. In
contrast, an energy market has no such requirement for being a forward market.
Energy markets could be forward or spot markets.
This difference highlights how these two services could be compensated. Because
of the forward nature of the ancillary service markets, it is necessary to pay
the owner for holding capacity idle. It may also be necessary to pay for any
energy that the capacity may produce in real time. A two-part payment scheme has
arisen for compensating providers of ancillary services. First, there is a
payment for holding the capacity idle; second there is payment for energy if and
when produced. It is not necessary to pay for ancillary services in this way. If
the up-front payment is high enough, then it could additionally cover the cost
of providing the energy in real time. Energy payments only require one-part
payment for a forward or spot sale.
In California, the institutions providing these services have been well
established. The PX operates forward energy markets. The ISO operates a real
time spot energy market as well. In procuring ancillary services, the ISO uses a
two-part payment. It makes both a forward payment for holding capacity idle and
a real time payment for any energy produced.
As will be discussed in greater detail below, the nature of the supply for these
services and how they are compensated dictates how the markets will be related
to each other.
2. THE SUBSTITUTABILITY OF THESE SERVICES IN PRODUCTION
Often, reserve services and energy can be provided from the same sources. A
generator producing energy may retain some capacity to be able to provide
reserve services. In most cases, generators providing reserves must also provide
energy simultaneously since they have minimum levels of output. Further,
generators can switch from providing one to the other quite
-18-
simply and almost costlessly.(24) In many cases, there is perfect substitution
between the production of spinning and non-spinning reserves and energy. Of
course, a unit's operating constraints limit the amount of substitution. There
is a less than perfect relationship between energy and replacement reserves.
This is because a generator does not have to be on line to provide non-spin or
replacement reserves. The only exception to this is demand-side resources that
can only provide non-spin and replacement reserves. However, their role in these
markets has been (and is likely to continue to be) limited.
This ability to substitute easily allows generators to get the most for their
generating capacity. In particular, they can sell capacity in either the energy
markets or in the reserve services markets. Unbundling the two markets may or
may not lead to energy prices that are determined using a different methodology
than that used for determining prices in the ancillary services.
However, the profit maximizing motives of generators leads to arbitrage between
the markets. In particular, generators will provide capacity to the market they
believe will provide them with the greatest return. Thus, arbitrage on the
supply side of these markets will produce compensation for generators that, in
equilibrium, will be the same whether the generator sells into the energy or
ancillary services markets. This arbitrage will be between the returns from each
market and not necessarily the price. As will be discussed in greater detail
below, this has particular consequences for the introduction of an ancillary
services market run by the PX.
Consumers do not have the same opportunities for arbitrage in these markets.
This is because consumption of energy does not provide the same benefits to
consumers as the consumption of ancillary services. Ancillary services enhance
the consumption of energy by reducing the likelihood of an interruption. Thus,
they increase the quality of the energy. If consumers also had the ability to
engage in arbitrage as suppliers can, then arbitrage on both sides of the market
would lead to different market results. In this case, there would be
countervailing forces working on both markets. With supply-side arbitrage only,
only generators move from one market to others seeking higher compensation. With
arbitrage on both sides of the market, consumers will move between the markets
to reduce their costs.
3. APPLICATION OF THESE CONCEPTS TO CALIFORNIA INSTITUTIONS
The existing structure of the California energy and ancillary services markets
has a significant impact on how much more efficiency and arbitrage can be
squeezed from these markets. There are currently five separate markets here, the
ISO's day-ahead ancillary services market, the PX's day-ahead energy markets,
the ISO's hour-ahead ancillary services market, the PX's hour-ahead energy
market and the ISO's real-time energy market. If the PX were to self-provide its
ancillary services, it would also introduce a day-ahead ancillary services
market.
- ----------
(24) Of course, the exception to this is regulation because it requires AGC
equipment. Further, the cost of providing regulation comes at the cost of a
less efficient heat rate. Compensation for regulation may have to include
an adder to maintain the margin between it and other services.
-19-
Currently there are arbitrage opportunities between the existing energy markets.
Of course, bidders can arbitrage between the PX's (and ISO's) temporal energy
markets. This is a two-sided arbitrage since both suppliers and consumers can do
it.
More importantly, producers can engage in arbitrage between the PX's energy
market and the ISO's ancillary services markets. However, it is difficult to
call the current arrangement arbitrage because suppliers know the PX's day-ahead
energy prices and allocations before and bidding capacity in the ancillary
services markets. Whatever is not scheduled in the energy market can be bid in
the ancillary services market. Further, energy prices are known when the bids to
the ancillary services market are made. Thus, it is easier to gauge the level of
demand for ancillary services and energy revenues. Suppliers only have
competition from other suppliers in limiting their bids from high levels.
Currently, uncertain demand levels and energy revenues do not limit the bids for
ancillary service capacity. Greater uncertainty should reduce the bids for
ancillary services because bidders are less able to assess whether their
capacity will be chosen at the higher bid prices.
In whatever ancillary services market the PX develops, it must take advantage of
being operated in conjunction with the energy market. There is the further
challenge of addressing the potential arbitrage between the PX and ISO's
ancillary services markets. In the structure adopted, it will be difficult to
keep suppliers from participating in both markets. Thus, the structure and
compensation in the ISO's market influence participation in the PX's market.
Unless the return is at least as great as in the ISO's ancillary services
market, the PX ancillary services market may not attract any sellers.
The existence of arbitrage makes it difficult to design a PX market that
exploits the efficiencies of joint energy/ancillary services market. Because the
ISO does not have an energy market, it fundamentally has different structure.
The two-part nature of the ISO's market may have to be altered in order for the
efficiencies of a joint market to be realized. From a public policy view, it is
not clear whether the costs of designing a PX market and altering the ISO's
market to reduce arbitrage opportunities will be offset by the efficiency gain
from a joint energy/ancillary services market at the PX.
As described above, the ISO's ancillary services market makes two payments, one
for capacity and another for energy (when produced). Because the PX market will
have to compete with the ISO for suppliers, its payment structure must be
similar. Thus the PX ancillary service market must make a capacity payment for
standing ready as well as an energy payment when dispatched.
Simply imitating the structure of the ISO's ancillary services market provides
no efficiency gains. Achieving cost savings requires reallocating the mix of
generation to meet energy and ancillary services requirements at significantly
lower costs (as revealed through bids). The task at hand is to design a joint
energy and ancillary services market for the PX market that
-20-
introduces efficiencies relative to the existing structure. The challenge is
further complicated by the requirement that it must compete with the ISO
ancillary service market.
VI. A MODEL FOR PX PROCUREMENT OF ENERGY AND ANCILLARY SERVICES
This section develops a framework for the PX introducing an ancillary service
market that complements its day-ahead energy market. The model is based on one
very significant change from the way the energy and ancillary services markets
currently function. Under the current structure, bidders do not have to reveal
their willingness to provide ancillary services until after they know the
results of the energy auction. This framework assumes that, by bidding in a
combined energy/ancillary services market, producers are revealing their
willingness to provide generating capacity the following day, whether it
produces energy or ancillary services by standing ready to produce energy.
Consequently, bidders make only one bid for both energy and ancillary services.
It should be highlighted that this hypothetical structure is adopted for the
sole purpose of investigating the questions posed by FERC. Specifically the
model addresses the relative efficiency of jointly procuring energy and
ancillary services. If the PX were to introduce ancillary services markets, it
is likely that the structure would be much different than the one discussed in
this report.
The description of the market below ignores the complicating factor of how the
hypothetical market structure would fit together with congestion management
procedures. This complication is ignored for clarity of exposition in setting
out the framework. Of course, these details would have to be addressed in any
market structure the PX may adopt.
A. DESCRIPTION OF THE HYPOTHETICAL MARKET
Under this mechanism, the bidding by suppliers will be the same regardless of
evaluation treatment. There will be differences in the PX's determination of
capacity awards under the three mechanisms. There may also be differences in the
way the PX determines the market clearing prices for each of the reserve and
energy markets.
To begin, suppliers will still bid for energy in the way they currently do.
However, bidders will also indicate ramp rates for each section of each of the
portfolios. The PX will examine the demand bids and project ancillary service
requirements. This forecast of ancillary service requirements will be discussed
more fully below.
Under each of the different evaluation techniques, fully sequential,
simultaneous-sequential and fully simultaneous, bids can be evaluated and awards
made that allocates capacity to energy, regulation, spin, non-spin and
replacement reserves. The particulars of the methodology used under each of
these techniques will be discussed in greater detail below.
-21-
Regardless of the evaluation technique, the PX will determine the market prices
for energy and each reserve service once the capacity has been assigned to the
energy and reserve markets. At that point, the PX will inform the supply bidders
of their awarded energy and ancillary services quantities and prices. The
bidders would then be responsible for submitting generator specific schedules to
meet the awards. Suppliers would be allowed to submit any unallocated capacity
into the ISO's ancillary service auctions as they do today. Various details of
the market will be discussed below in the rest of this section.
B. BIDDING INTO THE PX MARKET
This framework requires a consistency among bids that does not currently hold in
the existing markets. Specifically, in existing markets, supplier's ancillary
services bids need not have any relation to their willingness to sell energy.
They can bid whatever capacity prices and quantities they desire. This leads to
inconsistency in the bids and willingness to provide. For example, more capacity
can be offered for up-regulation, than is offered for spin. Similarly, the
prices offered for regulation could be higher or lower than those offered for
spin from same generator. The mechanism described below requires consistency in
the willingness to sell in these markets.
As described above, the objective of the hypothetical structure is to promote
consistency between bids for energy and ancillary services while facilitating
lower cost provision of these services through joint procurement. The structure
allows for portfolio bids to be offered for both energy and reserve capacity. It
also allows suppliers the option to participate in the energy market only or to
whatever extent they wish in the PX's ancillary services market. This is
accomplished by indicating a ramp rate for each reserve market for each bid
segment. An indication of no ramping capability indicates the supplier does not
want to offer capacity in that particular reserve market.
C. BID TYPE
1. DEMAND SIDE
Demand side bids will operate in the same way they do currently. However, there
will be changes in the options open to the demand side for procurement of
ancillary services. Currently, consumers do not have the option to arrange for
its own self-provision or partake in the PX's self-provision. It can only have
the PX purchase its requirements from the ISO's market. In the future, it will
have more options. These are: arranging for their own self-provided ancillary
services, having the PX purchase services in its auction, and purchasing its
requirement from the ISO's market.
The demander can indicate what portion of its requirement it has previously
arranged and purchase the rest from the PX's market or the ISO's market. Notice
the PX will not allow demanders to split their purchases between the PX and ISO
markets. For example, a demander
-22-
needs 100 MW of ancillary service and has arranged for 40 MW from a source that
will self- provide it. The demander has the further option of getting the
remaining 60MW from the PX's market or the ISO's market. It cannot split the
residual requirement between the two markets. In this way, the demand side can
also influence the degree of arbitrage between the PX's and ISO's reserve
markets. This adds another level of data information and cost to the PX since it
must keep track of transactions in separate markets for each of its demand
bidders.
Once the PX gets the demand bids and indications of how much ancillary services
demanders wish to be self-provided, there is still uncertainty about its
quantity responsibility at the ISO. This is because reserve requirements are not
simply a set percentage of load. Although they are often indicated as such,
reserve requirements will vary depending on the sources of power. For example,
WSCC requires different amounts of reserves for hydro generation than fossil
generation. Similarly, imports into the system are treated differently in the
calculation of reserve requirements.
In this PX market, the PX will not know the sources of the energy being
provided. This is because the PX still will be accepting portfolio bids that are
not required to specify the generation source. Thus, it is impossible for the PX
to know exactly what its reserve requirements will be at the ISO and how much to
buy in its market to self-provide. Nonetheless, this simply gives the PX a
degree of discretion in deciding how many reserves it wants to schedule as
self-provided. Anytime its estimate is too high and it agrees to purchase more
reserves from its suppliers than it needs in for self-provision, the PX can
offer these resources in the ISO's ancillary services market. Similarly, anytime
the PX itself fails to purchase enough reserves to meet its self-provision
requirement, it can purchase these from the ISO's market. The PX's discretion in
this also acts as a mechanism to keep the prices for reserves in line between
the two markets.
2. SUPPLY SIDE
The form of the PX bids on the supply side will retain its current structure.
Specifically, each bidder puts in a piecewise linear (upto 15 segments), bid for
each portfolio it wishes to have the PX consider.
Under this new market, the bidder also submits ramp rates for each reserve
market for each segment in its bid. For example, a bidder offers a portfolio
with three segments. For each of the three segments, the bidder must specify how
quickly capacity in that segment can be ramped up to meet each of the reserve
requirements: up-regulation, spin, non-spin, and replacement reserves. If
capacity represented in that segment has AGC capability, then the ramp rate bid
indicates how much capacity is available in that segment for up-regulation.
Similarly, ramp rates for spin indicates how much capacity in that segment is
available for providing spinning reserve.
As is clear from the description, it is possible for the bidder to opt out of a
particular reserve market by setting a ramp rate for a particular segment to
zero. When bidders choose this
-23-
option, the PX would not consider segments with zero ramp rates available for
the particular market. However, it would consider the segment for the energy
market and other reserve markets as indicated by the ramp rates for the
particular reserve service.
A segment's bid price signals the bidder's willingness to assign capacity from
that segment for whatever service that might be assigned to it. For example, an
incremental price of $2/MW associated with a particular segment means the bidder
is willing to sell capacity for any of the markets at that price. The price bid
for capacity to produce energy, up-regulation, spin, non-spin and replacement
reserves are all the same, in this case, $2/MW. Of course, the ramp rates and
time horizons for each service limit the amount of capacity that can be offered
from that segment to each of the markets.
By requiring the bids for energy and reserves to take this form, there is a
built in consistency between the bids for energy and those for reserves. While
the bidders can control the quantity offered for each service from an ex ante
perspective, they cannot alter the prices, selling essentially the same capacity
at different prices depending on the particular market. Under the current
structure of the ISO's ancillary services market, bidders do have flexibility on
both price and quantity.
Notice that portfolio bidding would still be available to bidders. Thus, there
could be no check on whether the ramp rates were reasonable for each section of
the bid curves. Bidders would take this risk on themselves. Bidders would be
required to provide the capacity from generators that would meet the
requirements that they were assigned in the auctions. This requirement and risk
puts a greater burden on bidders to bid in ways that would facilitate compliance
with the awards from the market. In portfolios with multiple generators
represented, this is a formidable problem. However, the problem becomes
significantly easier if fewer generators are represented in a single portfolio.
D. PX EVALUATION OF BIDS
As described briefly above, evaluation will depend on which methodology the PX
decides to adopt. The methodologies are similar in that each minimizes the cost
of a service or set of services. However, the evaluation methodology varies
across the three approaches under consideration.
The main difference between them has to do with how the energy and reserve
markets are evaluated. In the fully sequential treatment, each market is
evaluated separately in sequence. The ISO evaluates each of the reserve markets
sequentially after the PX has evaluated the energy market. This methodology
differs significantly from today's practice because bids are required to be
consistent with each other. In the other two approaches, the ancillary services
markets are evaluated simultaneously. In the sequential-simultaneous approach,
the energy market is cleared before ancillary services markets are evaluated. In
the fully simultaneous approach, energy is evaluated at the same time as the
ancillary services. The particulars of
-24-
evaluation under each methodology will be discussed in much greater detail below
in the analysis discussion.
E. NOTIFICATION OF SELECTED BIDS
The PX will determine which portfolios have been selected through its processes
to provide energy and each of the ancillary services. Notification will take
place in much the same way it does now. The PX will post energy and ancillary
services prices and tell each participant its allocation in each of the markets.
The suppliers will then be required to turn these awarded capacities into
generating unit schedules. The PX will have to check that the schedules
submitted for these services meet the technical requirements for such services.
The PX must also identify these resources as self provision resources to the
ISO. For the generators that make up its self-provision schedule, the PX must
also designate proxy-energy bids for the ISO's real-time dispatch. The level of
these bids will be the energy bids for the winning capacity in the day-ahead
joint market.
After it receives the generator schedules it should be able to calculate its
required provision for each ancillary service. Any excess capacity can be bid in
the ISO's ancillary services market. Any shortfalls will simply show that the PX
is only partially self-providing.
There are two options for how the PX could interact with the ISO's market when
it has over-purchased ancillary services. It could act as a price taking bidder
or a normal bidder.
If the PX followed price taking behavior in the ISO's ancillary services market,
it would bid its excess capacity at a low price. This would ensure that the ISO
scheduled the capacity and that the PX would receive something for it. As
discussed below, the PX would still be responsible for paying its winning
capacity the PX's market clearing prices for ancillary services. Any shortfalls
or excesses due to price differences between the ISO and PX would flow through
to the consumers in the PX's ancillary services market.
Under the second option, the PX would bid any excess ancillary service that it
may have purchased at the PX's market clearing price. If accepted, then it will
be assured that the price it will be paid by the ISO for the capacity will cover
what it has agreed to pay for it. If rejected in the ISO auction, the PX would
have to pay all of its suppliers and charge the excess costs to demanders. Or it
may be able to adjust its awarded quantities (and price) downward so there is no
excess self-provision. The particulars of this type of mechanism have not been
explored.
F. PRICING OF ENERGY AND ANCILLARY SERVICES
In the discussion so far, little has been said about how the PX will set its
prices for ancillary services. There are a number of ways to set prices for
these markets. Economic theory suggests a correct way to price these services.
However, there are also considerations of
-25-
existing institutions and fairness that influence the pricing rule to be
adopted. All of these factors will be taken into account in discussing pricing
rules. Further, FERC has set forth objectives of any market mechanism. Because
the pricing mechanism dictates how the market participants will behave in a
market, alternative pricing mechanisms will be discussed in relation to the
evaluation techniques.
First, the FERC's objectives for market mechanisms will be discussed. Then,
three different pricing mechanisms will be described and considered in
combination with evaluation techniques. The pricing mechanisms are marginal
cost, highest bid providing service, and unbiased
1. FERC REQUIREMENTS FOR PRICING UNBUNDLED SERVICES
FERC in its December 18 order wrote, "We note that Order No. 888 requires that
all ancillary services be unbundled."(25) Thus, the FERC is intent on having the
energy market unbundled from the ancillary services markets. FERC further points
out that unbundling simply means that services can be purchased separately and
that the amount paid separately for each unbundled service is lower than the
amount paid for the bundled service.
As described in its December 18 order, FERC's objective for the market mechanism
is to permit the PX to develop efficient preferred schedules for both energy and
ancillary services, and the ISO to develop efficient final schedules for both
energy and ancillary services. This objective implies that the mechanism should
provide the services at the lowest cost possible to society. This refers to
dispatch costs that producers incur. However, there is difference between the
costs to producers and the costs to consumers. Consumer costs are producer's
revenues. Costs to consumers need to be evaluated differently from costs to
producers. This distinction will be described in greater detail below.
Further, FERC laid out four principles to evaluate mechanisms for selecting and
compensating energy and ancillary service providers. First, "energy and
ancillary service providers should have incentives to bid a reasonable
approximation of their marginal costs."(26) Namely, the mechanism should provide
incentives for bidders to reveal their marginal costs of providing the service.
Second, "suppliers should not be biased in their choices among supplying energy
and various ancillary services."(27) This means that bidders should be
indifferent between selling energy and ancillary services.
- ----------
(25) FERC December 18 Decision, p. 22.
(26) Ibid, p. 21.
(27) Ibid.
-26-
Third, "suppliers and buyers should not be biased in their choices between the
PX and bilateral deals."(28) This has implications for the prices charged to
consumers and paid to producers. For indifference to hold, they should be the
same. If they are not the same in the centralized market, there is an incentive
to bypass the market and consummate a bilateral arrangement. For example, if the
market price to producers is say $2.5 and consumers pay $3.0, there is an
incentive for them to negotiate a bilateral price that would split the $.5
difference. Consequently, prices should reflect market clearing.
Fourth, "investment incentives should be consistent with efficient dispatch for
both energy and ancillary services over the long run."(29) This requires the
short-run operating incentives to be consistent with the long-run incentives to
invest in generation and transmission.
These FERC objectives and principles can be used to evaluate each of the pricing
mechanisms and evaluation techniques. Given the number and diversity of these
evaluation criteria, it may be difficult to propose a mechanism that satisfies
them all. Three different pricing mechanisms will be considered. They are based
on: marginal cost, highest bid providing service, and indifference of returns.
2. ALTERNATIVE PRICING MECHANISMS
a. PRICING BASED ON MARGINAL COSTS
The theoretically ideal pricing methodology would set prices for energy and
ancillary services at their marginal costs. Marginal costs are defined as the
change in total costs if another unit of service were to be provided. This
signal provides the correct information to consumers and producers of
electricity. Specifically, price equal to marginal cost signals consumers the
cost to society of consuming an additional unit and pays producers only the cost
of providing that unit.
Despite its simplicity in concept, applying marginal cost pricing to jointly
produced products that are consumed at different times is a difficult
proposition. The joint production aspect of the services complicates determining
marginal costs. Similarly, the temporal consumption of the different services
also complicates applying marginal cost pricing.
As described above, both energy and ancillary services can be provided from the
same generating units. Further, generators providing some of the reserve
services, regulation and spin in particular, are required to be on line and
producing energy in order to provide ancillary services.(30) Thus, generators
must be producing both energy and ancillary services. This joint production can
complicate the marginal cost of a particular service, such as spin, because the
- ----------
(28) Ibid.
(29) Ibid.
(30) While this is not true for all types of generators, it is general enough
for this discussion of joint production.
-27-
marginal cost of spin is tied to the marginal cost of energy. Although
separately quoted, these marginal costs are jointly determined.
Similarly, the temporal consumption implied in these markets also complicates
determining marginal costs as described above. On the one hand, reserve services
can be viewed as containing two parts, holding capacity idle to be ready to
produce and the energy they do produce when called. On the other hand, the
energy market combines the capacity to produce with the actual production. Thus,
reserve services are often paid on two different bases, capacity and energy
while energy is paid on one basis, the energy produced.
To accommodate joint production in a marginal cost pricing framework, it is
necessary to change the view of payment for ancillary services. Instead of
employing a two-part structure to compensating reserve services, a one-part
compensation is needed. Specifically, a payment is made for reserve services
that covers both the costs of holding capacity idle and the energy produced when
called. Under this mechanism, capacity providing ancillary services is paid for
energy even if it is not called in real time to produce energy. This may be an
undesirable feature of pricing ancillary services in this manner.
b. PRICING BASED ON THE HIGHEST BID PROVIDING THE SERVICE
An alternative pricing mechanism is to set market prices based on the highest
bid of capacity accepted to provide the service. After allocating capacity to
each of the markets, the price for each market is based on the highest bid for
capacity chosen to provide the service. For example, if the highest bid to sell
energy were $1.9/MWH then the price would be set at this level.
This is close to marginal cost pricing, however, it is not the same in the case
of simultaneous evaluation. This is because evaluating markets simultaneously
leads to joint allocation decisions. Determining the marginal cost of one
reserve service depends on reallocating capacity between all services. Pricing
based on the highest bid providing the service ignores this reallocation effect
between the markets and assumes each market is autonomous and separate from the
other markets.
Unlike marginal cost pricing, this mechanism allows the separation between the
single payment for energy and a two - part payment for reserve services. Namely,
separate payments can be made for reserving capacity and for the energy
generated, if any. Pricing in this manner requires the capacity payment to be
the difference between the highest cost bid accepted for a reserve service and
the energy price. The argument for setting price at this level is that the
opportunity cost of selling energy is the profits that the capacity would have
received if it had been accepted in the energy auction.
Also, paying this quantity to all providers of the reserve service would result
in a market clearing result. Specifically, none of the producers has an
incentive to move to a bilateral arrangement because all consumers are paying
the same amount that all producers receive.
-28-
This capacity pricing arrangement can be seen in Figure 1. The energy price is
set at the level of highest cost of capacity allocated to provide energy. In
this case, PE is the price for energy. The capacity payment to the reserve
service is set at the difference between the highest bid accepted to provide the
service (P(S)) and the energy price. In particular, the capacity payment is
(P(S) - P(E)) for each MW of reserves purchased. If called to provide energy in
real time, this capacity would also receive a payment for the energy produced.
[FIGURE 1 CHART]
Despite the favorable features of this pricing mechanism, it does have some
drawbacks. Under this mechanism, suppliers would rather provide ancillary
services than energy. They are likely to bid in such a way to increase the
likelihood of selling the services with the highest returns. Bids will not
reflect marginal costs. The ancillary services provision is favored because
there is a differential in returns to producers for supplying energy and
ancillary services. The margin of a supplier selling energy is the difference
between the energy price and its bid. The margin for reserve services is the
difference between the highest bid providing the service and the energy price.
This will be larger than the return for energy, particularly for generators near
the margin of the energy market.
-29-
[FIGURE 2 CHART]
This can be seen in Figure 2. Again, the market-clearing price for energy is
P(E). P(S) represents the highest-cost winning bid for the reserve service.
Sellers of energy only receive the amount depicted by A while sellers of the
reserve services receive the amount labeled B. Clearly, generators will have
greater profits providing reserves than energy.
c. PRICING BASED ON INDIFFERENCE OF MARKETS
The third pricing mechanism considered addresses the desired property of having
the returns to selling energy and ancillary the same. This mechanism will also
induce bidders to reflect their marginal costs in their bids and allows for
separate pricing of energy and ancillary services. However, as will be seen,
this mechanism also has its drawbacks, particularly, when it comes to providing
a market-clearing price for reserve services. This mechanism can be described as
indifference of market returns.
This mechanism sets the price of energy at the highest bid of the capacity
chosen to supply any of the reserve services. Namely, if the highest-bid
capacity provides spin or non-spin or any of the other reserve services, the bid
sets the price for energy. In turn, the energy price is the basis for
calculating the capacity payments for the reserve services. The mechanism allows
separate energy and capacity payments to be made for reserve services.
The objective of this mechanism is to keep producers indifferent between selling
in the energy market and in the reserve markets. In order to achieve this
indifference, the returns from each market must be the same regardless of the
market. In the energy market, the return to suppliers is the difference between
the energy price and the bid to provide that service. Consequently, in the
ancillary services markets, the return to suppliers must be the difference
between the energy
-30-
price and the bid. Thus, this is the price paid to each supplier of reserve
service, namely, the difference between the energy price and its bid.
This can be seen in Figure 3. The price of energy is set at P(E), the highest
cost bid of all bidders providing energy and reserve services. The payment made
to supplier who is providing reserve services is the difference between P(E) and
the suppliers bid. In the figure, reserve supplier B receives the difference
between the energy price and its bid. This mechanism keeps suppliers indifferent
between providing energy and reserve services.
[FIGURE 3 CHART]
However, the price to consumers is a different amount. Because different
suppliers will have different bids and be paid different amounts, there is no
single price paid for reserves. In order to maintain revenue neutrality, the
price paid by consumers is the average cost of each service. Specifically, the
average of the prices paid to the suppliers for that service.
This reveals a major drawback to this pricing mechanism. Because individual
consumers pay something different from what individual producers are paid, the
mechanism does not provide a market clearing equilibrium for the reserve
services. In particular, producers who are paid less than average will have an
incentive to create bilateral arrangements with consumers who are paying the
average. Nonetheless, it does induce bids close to marginal costs and does not
bias the choice of markets for producers.
3. PRICING IN THE ISO ANCILLARY SERVICES MARKET
In setting up the pricing mechanism in the PX, an additional consideration is
how the ISO remunerates capacity providing ancillary services. As described
above, the ISO takes a two-part bid for providing reserves: capacity and energy
components. However, the ISO only uses
-31-
the capacity bid in choosing which resource will provide capacity. The prices in
these markets have been as high as, if not higher than, the energy prices.
If the PX were to pay something significantly less than what the ISO pays,
arbitrage would prevail and suppliers would simply move to the ISO's market and
not participate. Also, the objective of introducing PX self-provision is to
lower the cost of ancillary services in the market.
However, the ISO's ancillary services markets are undergoing significant
changes. While it does not appear the ISO will change the structure of its
market mechanisms for ancillary services, it is uncertain exactly what reforms
the ISO will make to its ancillary services markets.
An alternative way to look at the reserves markets is as options markets. The
buyer is purchasing an option to call on the supplier to provide electricity at
a particular price. Thus, there is a reservation payment to stand ready as well
as a strike price for delivery. Under this type of framework, there is two-part
bidding for ancillary services, a capacity component and an energy component.
While this mechanism may seem closer to the existing ISO structure, the correct
way to evaluate such a bid is to combine these components in choosing the
suppliers in the day-ahead framework.
While it would be possible to set up a two-part framework for the PX market, it
involves a more complicated mechanism in allowing for the joint evaluation of
energy and reserve markets. Because of the added complexity, this framework was
not pursued here.
G. SUPPLY REMUNERATION AND DEMAND PAYMENTS
Under this framework, payments to suppliers and charges to consumers are
relatively straightforward and work as in a standard market. The main difference
is that the PX will have to track and settle with buyers and sellers who opt to
participate in the ISO's market rather than its own. There is the further
complication of pricing and paying for any reserve requirements that are sold or
purchased through the ISO's market. Each of these aspects will be discussed
briefly.
It is clear that the PX should pass on the ISO's prices to suppliers and
demanders who have opted out of the PX's self-provision market for ancillary
services. The PX prices may be higher or lower than the ISO's prices. By
requesting participation in the ISO market, suppliers and demanders will receive
those prices.
It is less clear how the PX will deal with sellers and purchasers in its own
markets, particularly since there is the risk that the PX may purchase too much
or too little in its own market. The PX itself must remain revenue neutral with
respect to transactions in the ISO's markets. Any costs or benefits that accrue
due to differences in ancillary service market prices and over- or under-selling
should be borne by the demanders who have opted for the PX market. The PX
-32-
will pay winning generators the PX price it has determined for each service,
regardless of whether it needs to sell that output into the ISO market or not.
However, demanders will pay a blended rate for their services. The blend may
include the cost of ISO-provided services if the PX has purchased too little in
its own auction. It may also include a credit if the PX has over purchased and
sold its excess into the ISO's market at a price higher than its own. Similarly,
the blended rate could include costs if the PX has over purchased and sold its
excess into the ISO's market at a price lower than its own. This latter case
provides an incentive for demanders to opt out of the PX's market for the ISO's
market because the PX's price is higher plus there is an adder for over
self-provision.
VII. CHANGES REQUIRED TO IMPLEMENT THE PX MARKET
A. ISO OPERATIONS
The hypothetical framework assumes the ISO is not planning to make major changes
to its ancillary services markets. This is probably not the case. The ISO is in
the process of evaluating its ancillary service market mechanisms. Although the
ISO may change its evaluation procedures, it is not clear what changes will be
made to the bidding procedure, the form of the bids, the timing of the market,
or pricing of these services. Further, the framework does not require that the
ISO make any changes in any of these areas. This report has assumed that
suggesting such changes is out of its purview and has not addressed how the ISO
might change its market operation and pricing rules for ancillary services.
B. PX OPERATIONS
Under any arrangement the PX might adopt for self-providing ancillary services,
the required changes are significant, both in the PX's tariffs and settlement
system.
Changes will have to be made to the PX's tariff to describe how the PX would
operate its reserves market and how it would self-provide these services.
Because of the inertia built into the current system, these changes may be
difficult to get approved. Any change will produce winners and losers from the
producers and consumers currently in the market. Consumers may gain benefits
from having a source for ancillary services other than the ISO's market.
Producers may face increased costs of deciding how to bid in each of the markets
open to them.
Further, the PX must build the infrastructure to handle such a market. There are
three parts to building this infrastructure. First, the software must be written
to evaluate the energy bids for ancillary services. While the algorithms are
similar to the energy market, ramping rate adjustments must be made. Further an
algorithm to optimize the markets simultaneously may also have to be built.
-33-
Second, algorithms and software will have to be developed because the bidding
structure of the PX differs from the ISO's. The PX energy bids are a
15-piecewise linear curves while the ISO's ancillary services bids are an 11
piece step-wise bid. In order to bid, provide bids into the ISO's market, the PX
must convert its bids into ISO bids.
Third, a new settlement system must be built to track the choices individual
market participants make. The settlements system must track the quantities each
buyer self-provides, purchases from the PX or purchases from the ISO. Similarly,
the PX will have to track the output from generators as well, in terms of being
dedicated to self-provision, selling into the PX market and selling into the
ISO's market. Further, on the demand side, the PX will have to calculate the
appropriate price for its consumers. Because the PX itself may be a net-supplier
or net-demander with regard to the ISO's market, it must calculate a blended
rate of PX and ISO prices for its demanders. These software modifications could
impose a significant cost burden in implementing this framework.
VIII. THEORETICAL IMPACTS FROM A JOINT ENERGY ANCILLARY SERVICES MARKETS
In the discussion so far, there has been some general recognition that joint
provision of energy and ancillary services can lower the total costs of
producing each. This section explicitly discusses the impacts of a joint market.
There is the possibility of lower costs to society; however, these cost savings
may result in higher prices for consumers depending on the pricing rule adopted.
The impact of joint procurement on cost savings will be discussed. Similarly the
effect of different pricing rules on sequential and simultaneous markets is
considered.
A. POTENTIAL FOR LOWER COSTS
There are efficiency gains that can be achieved from a joint energy and
ancillary services market in relation to a similarly designed sequential market
as described above. These efficiencies result from having a wider scope of
choices between uses of capacity.
In a sequential energy and ancillary services market, it is possible to miss
efficiencies from joint procurement. This is because joint procurement allows
capacity substitution across all markets, reserve and energy. In fully
sequential markets, the energy market is cleared before consideration of any of
the reserve markets, and the regulation market before consideration of the other
reserve markets, and so on. This means capacity allocated to energy (or an
earlier reserves market) cannot be substituted for capacity in the later reserve
markets. This missed opportunity for substitution is a missed opportunity to
lower costs. A fully simultaneous market does not miss this opportunity.
For example, in a sequential market, the procurement of energy ignores the ramp
rates for suppliers. However, they are an important component of the procurement
of ancillary services.
-34-
Because the ancillary service market follows the energy market, it is possible
to award energy production to capacity that has fast ramping capability. This
means only slower ramping capacity is available in the reserves market. The
joint cost of producing energy and reserves may be higher because the
lower-cost, fast ramping capacity cannot substitute for the higher-cost, slower
ramping capacity. In a simultaneous market, it is possible for this substitution
to take place, reducing the overall cost of energy and ancillary services.
B. POTENTIAL FOR HIGHER PRICES
The simultaneous provision will also have an impact on the pricing of energy and
each of the ancillary services. As discussed above, one of the purposes of
unbundling is providing separate prices for each of the services being provided.
Thus, the objective includes pricing services as well as minimizing costs.
Depending on which of the three pricing rules described above is adopted, there
is also a difference in energy and reserve prices between procuring them under
simultaneous and sequential evaluation. The relative impact on prices to
consumers will also differ between evaluation techniques.
1. IMPACT ON PRICES UNDER MARGINAL COST PRICING
Under sequential evaluation, the selection of capacity for each market is done
without regard to the impact on other markets. The markets evaluated earlier in
the process will be allocated the less expensive capacity. In the cases
analyzed, the energy markets were evaluated first. Consequently, energy prices
should be lower relative to other services.
Under simultaneous evaluation, explicit consideration is given to how the
capacity allocated in one market affects the available capacity in the other
reserve markets. Similarly, the marginal cost calculations reflect this
consideration. Needing more spin capacity could well reallocate available
capacity to energy, non-spin and the other markets. Thus, the marginal costs of
each of these services will tend to be equated and the prices for the services
will be equated.
In comparing the resulting prices from the two evaluation techniques, the
following should be true under marginal cost pricing. Energy prices should be
lower under sequential evaluation than under simultaneous evaluation. Also,
prices for reserve services should tend to be lower under simultaneous than
sequential since lower cost capacity has been used.
This can be seen in Figure 4, which compares the two techniques. Under
sequential evaluation, the energy price is set first and reserve prices are
generally higher. Under simultaneous evaluation, all prices are equated. In
order to aid in illustration of the results, this graph (and the following ones)
incorporates two simplifying assumptions that do not change the results in
general. First, demand is assumed to be inelastic. Second, it is assumed that
the ramp rate constraints for capacity do not bind in constructing the market
supply curve.
-35-
[FIGURE 4 MARGINAL COST PRICING CHART]
2. IMPACTS ON PRICES UNDER HIGHEST COST RESOURCE PRICING
The results for pricing under Highest Cost Resource Pricing are similar. Recall
under this technique, explicit prices for capacity are calculated as the
difference between the highest cost resource providing the service and the
energy price.
Again, in a sequential market, the energy market is cleared before the reserves
markets are considered. Thus, the lowest cost capacity is allocated to energy
leading to a relatively low energy price. In pricing reserves, the capacity
payments depend on the cost of the highest resource providing the service and
the energy price. Because the energy price is low, reserve prices under
sequential evaluation are relatively high. Of course, the results depend on the
shape of the bid curves.
Under simultaneous evaluation, again the substitution of capacity between the
energy and reserve markets allocates relatively higher cost capacity to the
energy market. Also, the substitution also means the highest cost capacity
needed is likely to be less costly than the highest capacity needed under
sequential evaluation.
-36-
This combination of factors leads to lower prices for energy under sequential
evaluation than under simultaneous. Similarly, because the energy prices are
lower and higher cost resources are needed, reserve prices should be higher
under sequential than simultaneous evaluation. This is the same result seen
under Marginal Cost Pricing.
Figure 5 illustrates the comparison between evaluation techniques graphically.
PE is the energy price in both cases. MCAS is the highest bid accepted for
ancillary services. The difference between MCAS and PE establish the capacity
price for the ancillary service PAS. The sequential approach has lower energy
prices, but higher ancillary services prices.
[FIGURE 5: HIGHEST COST BID PRICING CHART]
3. IMPACTS ON PRICES UNDER INDIFFERENCE OF MARKETS PRICING
Indifference of Markets pricing is qualitatively different than the other two
pricing mechanisms considered. This is because the energy price is not set at
the highest bid accepted for energy, but at the highest bid accepted for any of
the reserve markets. The simultaneous approach allocates capacity more
efficiently than the sequential approach. Thus, less costly capacity is
-37-
needed to meet the combined demand. This yields a lower price for energy under
simultaneous evaluation than under sequential.
Recall under this approach there is no market-clearing price for reserve
services. Suppliers are paid the difference between the energy price and their
bids. Prices to consumers are average prices of those paid to producers. There
is no a priori reason why these differences should be greater or lesser under
simultaneous evaluation in comparison to sequential. The exact nature of the
relative difference will depend on the shape of the supply curve. Consequently,
reserve prices may be higher or lower under simultaneous in comparison to
sequential under this type of pricing.
Figure 6 illustrates these results. Under sequential, P(E) is higher than under
simultaneous, reflecting the less efficient dispatch. Again the capacity payment
for reserve services is the difference between the bids and the energy price. It
is difficult to say that this difference will in all cases be less under
simultaneous than sequential. This result is different from the other two
pricing mechanisms discussed.
[FIGURE 6: MARKET INDIFFERENCE PRICING CHART]
-38-
IX. ESTIMATING THE EFFICIENCY GAINS FROM A SIMULTANEOUS MARKET
The purpose of this report is to address the added benefit from a simultaneous
ancillary services and energy market. This section discusses the methodology
used to estimate such benefits. There are several alternative approaches
including, among others, an empirical analysis of actual bids into these markets
and simulation of the different market designs.
A. SIMULATION ANALYSIS WILL BE REVEAL MORE THAN AN ANALYSIS OF HISTORIC
BIDS
Although the energy and ancillary services markets have been in operation for a
number of months, it is not wise to rely on an analysis of actual market data.
Instead it will be more enlightening to use a market simulation analysis to
quantify the benefits from simultaneity.
1. PRACTICAL REASONS WHY AN ANALYSIS OF HISTORIC DATA IS NOT HELPFUL.
First, the markets are new. Participants have been learning how to operate in
these markets and take advantage of their opportunities. Their bidding behaviors
have changed over time as they have learned more about the market rules.
Second, there have been significant changes in the markets over this period,
particularly with the ancillary services markets. Initially, all ancillary
service bids were capped. Then some bidders had their caps removed. An overall
price cap was introduced as well as a supplemental payment for regulation.
Recently, caps were lifted for all participants. Currently, the ISO is examining
how it might better operate its ancillary services markets. On the energy side,
an hour-ahead market was introduced.
Third, the PX does not have access to the ancillary service bids for all bidders
in the ISO market. Even though the PX passes through its own participant's bids,
these data are confidential and it may be difficult to examine the nature of the
relationship between energy and ancillary services.
Finally, the objective of the analysis is to examine the benefits of
simultaneous procurement relative to sequential purchasing. The current
methodology is based on sequential markets. Historic and current bids are
formulated under the rules that exist for sequential procurement. A change in
the procurement rules should elicit a change in bidding behavior. Similarly, the
rules and formulation of bids in a joint energy and ancillary services market
would elicit different bidding strategies. Thus, using current bids is not
helpful in determining the efficiency gains.
2. SIMULATION ANALYSIS CAN BE ENLIGHTENING
It may be helpful to set up hypothetical markets with the bidding structure to
examine how large a difference the simultaneous structure makes in procuring
ancillary services. Simulations can illustrate the potential savings from
different markets incorporating degrees of
-39-
simultaneity. Simulations could also be used to study the behavior of market
participants if hypothetical markets are actually set up and the participants
are allowed to bid in them to establish equilibria. These experimental
simulations could prove fairly accurate estimates of the differences in the
markets. However, the depth needed to establish highly accurate experimental
results is beyond the scope of this analysis. The following sections undertake
the more modest goal of providing a rough estimate of how much more efficient
simultaneous evaluation is than sequential.
B. DESIGN OF THE MARKET SIMULATIONS
In order to isolate the impact of the degree of simultaneity, the level of
energy and ancillary service demand remains constant between treatments.
Similarly, the bid prices and quantities also stay unchanged. The resulting
differences in costs of provision will indicate the magnitude of the benefits
from a particular level of simultaneity. The differences in efficiency will be
calculated for representative levels of demand in order to illustrate how
efficiency varies over different seasons. Each of the problems and treatments
are set up as cost minimization problems.(31)
It should be highlighted that this research approach does not capture
differences in bidding behavior that may result from the different market rules
and opportunities in the ISO's markets. As discussed above, such behavior
changes may appear at higher levels of demand. The competitiveness of the energy
market should dominate any behavioral difference at lower levels of demand.
The above-described framework for procuring energy and ancillary services is
applied to three different degrees of simultaneity in evaluation. They are:
fully sequential evaluation (fully sequential); a partially sequential,
partially simultaneous evaluation (sequential-simultaneous); and a fully
simultaneous evaluation (fully simultaneous). The fully sequential approach
evaluates in the following order: energy, regulation, spin, non-spin, and
replacement reserves. The sequential-simultaneous approach evaluates the energy
market before evaluating the reserve markets simultaneously. The fully
simultaneous approach evaluates energy and all reserve markets together. Each of
the three treatments are described in greater detail below.
1. FULLY SEQUENTIAL EVALUATION METHODOLOGY
The fully sequential approach means the bids are used to meet the requirements
of each service in order. First, the PX will create a market supply curve from
the bids and clear the market for energy in the same way it does today. The PX
will then take the remaining bid curves and adjust them for the regulation ramp
rates indicated by the bidders. Using this adjusted supply
- ----------
(31) In what follows, the discussion focuses on cost minimization given a level
of demand. The results and applications described will also be applicable
to an analogous surplus optimization where demand bidding is explicitly
taken into account. Such an abstraction is warranted given the limited
bidding activity on the demand side of the market and the clarity of
exposition of the cost minimization framework.
-40-
curve and its estimated regulation requirements, the PX will then clear the
regulation market. Next, the PX will adjust the remaining supply curve for
energy and regulation awards and the ramp rates indicated for spinning reserve.
The spinning reserve market will then be cleared using this adjusted supply
curve and estimated needs for spinning reserve. Similar adjustments and clearing
take place for each of the remaining reserve markets in order.
This approach most closely matches the manner in which energy and ancillary
service awards are evaluated today, by the combination of the PX and ISO.
However, there is one very important distinction. The suggested methodology
requires bid consistency between the ancillary services. Because the PX works
off the bids used in the energy market, it can impose consistency by requiring
any capacity unused in the energy market to bid exactly the same price for all
the reserve markets for which it might be eligible. In this regard, the approach
is also close to the ISO's proposed smart-buyer methodology that attempts to
impose this consistency on bids through administrative rules.
2. SEQUENTIAL-SIMULTANEOUS EVALUATION METHODOLOGY
Sequential-simultaneous procurement means the bid curves are first used to
minimize the costs of meeting the required level of demand. Once an allocation
of energy to portfolio bids has taken place, then the quantities in the
portfolios are adjusted for awarded capacity and ramp rates for each segment.
The entire set of reserve markets is then evaluated, minimizing the costs of
meeting the ancillary service requirements subject to the ramping constraints.
Under this methodology, the PX is able to substitute higher cost suppliers for
lower cost suppliers in one market if such a substitution allows the PX to save
costs in another market. This substitutability does not apply to the energy
market. Rather it only extends to the other ancillary services markets.
This approach is very similar to the combination of the PX's current energy
market and the ISO's proposed LP optimization with one very important exception.
Again, the ISO does not require consistency between the bids made for the
various services offered, as the PX market requires. Although it has been
discussed, this requirement has not been imposed on the current market.
3. SIMULTANEOUS EVALUATION METHODOLOGY
Fully simultaneous procurement means the bid curves and accompanying ramp rates
are used to meet energy and ancillary service requirements using a single
optimization problem. The objective function of the minimization problem is the
sum of the energy and ancillary service costs. Because the energy market is
considered along with the ancillary services markets, it is possible to reduce
total overall costs by substituting relatively low cost capacity that would be
used for energy for much higher cost capacity that would be used for reserves.
As will be discussed in greater detail below, this is the primary theoretical
advantage of a simultaneous approach.
-41-
Because of the trade-off between the energy and reserves markets, this type of
approach cannot be replicated by the ISO. The ISO cannot set up the conditions
that would allow such a substitution because energy schedules are fixed by the
time the ISO receives the bids for the ancillary services markets.
C. EVALUATION CRITERIA BETWEEN THE MODELS
The measure of efficiency is the level of costs incurred for the given level of
provision of services. In this case, who is incurring the cost matters. The
pricing rules adopted affect the out-of-pocket costs for consumers and the
revenue streams to producers. As will be seen, different rules produce different
impacts on consumers and producers. These may be simple wealth transfers between
parties and have no impact on the overall resource cost to society.
Efficiency measures taking the societal perspective are most appropriate. In the
framework described below, a market supply curve is assumed. The measure of
societal costs is the minimized costs, as revealed by bidders. This is
essentially the area under the supply curve for the capacity scheduled for use
in the energy and reserves markets. Society is better off if the resources with
the lowest costs are committed. However, it should be highlighted that capacity
selected for reserves may not have to incur their full bid costs in standing
ready to generate. As described above, the compensation structure pays them as
if they were producing energy. Restricting the analysis to lower levels of the
supply is appropriate because bidders are more likely to be acting competitively
and bidding their true costs.
The analysis also focuses on the costs to consumers, which is the same as the
revenue to producers. Pricing rules may affect the costs to consumers and
revenues to producers without having a material impact on the overall costs to
society. For each of the pricing rules described above, the costs to consumers
and revenues to producers will be calculated for each evaluation technique. This
distribution of benefits may also affect the desirability of instituting a
particular evaluation methodology and pricing structure.
The next section describes a common model. It is an improvement over the
existing structure in that it requires consistency between the bids offered for
energy and each of the ancillary services.
X. NUMERICAL COMPARISON OF THE EFFECTS OF SIMULTANEOUS AND SEQUENTIAL
MARKET AUCTIONS
This section describes nine simple spreadsheet models(32) that were built to
examine the difference between the three evaluation approaches and three pricing
methodologies. The models have been used to compare the cost and pricing results
under exactly the same inputs
- ----------
(32) Copies of the spreadsheet models used are available from the author upon
request.
-42-
with the only variation being in how the markets are cleared and how prices are
calculated. In this empirical analysis, the complication of congestion
management and pricing has been ignored. The analysis focuses on the cost
differences between different evaluation techniques and pricing rules. The
magnitude of the impact of congestion is likely to be a second order effect to
the main result.
A. ASSUMED MARKET SUPPLY AND DEMAND CONDITIONS
1. REPRESENTATIVE MARKET SUPPLY CURVE
The market supply curve was estimated by the PX using data from the opening of
the market through September 1998(33). As described above it is assumed that the
market supply curve reflects the costs of the generators providing the service.
While there are issues of fixed and avoided cost recovery, this is taken as a
proxy to the long-run marginal costs in the market. As was described above, this
is not an unreasonable assumption particularly at levels of demand below 31,000
MW. At loads above this level, the curve may well reflect costs; however, it may
also contain the effects of diminished competition in supply. It should be noted
that load was only above this level 5 percent of the time between the opening of
the market and the middle of December 1998. This supply curve is assumed to
reflect long-run marginal costs.
In order to fit the curve into the analysis at hand, it was necessary to
estimate a stepwise function that approximates the supply curve. The reason for
this change in functional form is two fold. First, the spreadsheet models
developed solve more easily as a stepwise function. Second, the ISO's ancillary
services market uses a similar stepwise functional form. The market curve was
segmented into 16 different steps each of equal size and fitted to the
functional from of the supply curve discussed above. While it is not imperative
to have equal sized steps, such an assumption simplifies the analysis. The
numerical results will depend on the approximation to the supply curve. Both the
supply curve assumed for the market and its stepwise approximation are
illustrated in the following diagram.
- ----------
(33) The exact specification of the curve is Price =
3.92135232170356E-11*Load(3)-2.58757653414328E-
06*Load(2)+0.0581499351022064*Load-423.576761520161. It was taken from a PX
Document entitled "Buying Strategies in the PX: Reflections on Using the
Day-Ahead and Day-Of Markets Together."
-43-
[ESTIMATED PX SUPPLY CURVE AND STEP FUNCTION APPROXIMATION CHART]
2. REPRESENTATION OF DEMAND
The spreadsheet models are set up to calculate the costs and revenues using the
market supply curve and different levels of demand. The purpose of the analysis
is to illustrate the potential savings from having simultaneous evaluation of
the markets. Consequently, using various levels of demand over the entire range
of PX quantities will illustrate the likely savings from simultaneous markets.
The levels chosen for this analysis reflect the range of PX demands from April
1998 through the middle of December 1998. The unconstrained market clearing
quantities ranged from a low of 14,542 MW to a maximum of 36,376 MW. Six
different levels of demand were used, representing the quintiles, median, and
90th percentile of demand in the PX. The minimum was not used because of the
likely occurrence of over-generation at that time. The maximum was not used
since the markets are not likely to be sufficiently competitive at that level of
demand. The following table summarizes the demands used.
Demand Level (MW)
20th Percentile 18475.76
40th Percentile 20685.92
50th Percentile 21799.95
60th Percentile 22726.94
80th Percentile 24803.46
90th Percentile 27724.76
-44-
B. DESCRIPTION OF THE SPREADSHEETS
As discussed above, the spreadsheets take inputs from bidders whose bids
collectively constitute the market supply curve and the PX and uses optimization
methodology to allocate capacity awards for energy and reserves. The bid
evaluation methodologies are cost minimization problems where costs are
represented by the supplier's bids. Given the competitiveness of the energy
market, this is not an unreasonable assumption.
There are three spreadsheet models (fully simultaneous, simultaneous-sequential
and fully sequential) for each of the three pricing mechanisms (marginal cost,
highest bid accepted and indifference between markets). The fully simultaneous
models minimize the costs of energy and reserves in a single optimization. The
sequential-simultaneous models first minimize the cost of energy before solving
a subsequent cost minimization problem for the rest of the reserves markets. The
fully sequential models solve them in order. In all cases, the demand side of
the market is significantly simplified as an inelastic level of demand to be
met. Because all of the effects to be investigated are on the supply-side of the
market, this representation does not reduce generality. The rest of this section
will describe each of the inputs and outputs from the models.
1. MODEL INPUTS
PX inputs are primarily on the demand side of the model. They include the level
of energy demand to be met and the level of reserves the PX wants to purchase.
The desired reserve requirements are specified as percentages of energy demand
and vary by ancillary service. For example, in the analysis presented below, the
PX was assumed to be purchasing 1% of load as up-regulation, 3.5% of load for
each of spinning and non-spinning reserves, and 5% of load for replacement
reserves. The analysis was undertaken by varying the level of demand to be met
and comparing the results across the two models.
SUPPLY BIDS
Quantity Maximum Ramp Rates (%)
Bid Price Maximum Minimum Regulation Spin Non-Spin Replacement
Portfolio 1 5.74 16500 0 0 0 0 0
14.37 727.5 0 0.1 0.5 0.5 5
18.90 727.5 0 0.3 0.75 0.75 5
21.74 727.5 0 0.4 1 1 5
25.37 727.5 0 0.5 2 2 5
32.21 727.5 0 0.5 2 2 5
44.72 727.5 0 0.5 2 2 5
65.34 727.5 0 0.5 2 2 5
96.52 727.5 0 0.5 2 2 5
140.70 727.5 0 0.5 2 2 5
200.33 727.5 0 0.5 2 2 5
-45-
277.86 727.5 0 0.5 2 2 5
Portfolio 2 9.22 727.5 0 0.1 0.5 0.5 5
16.22 727.5 0 0.3 0.75 0.75 5
19.91 727.5 0 0.4 1 1 5
22.75 727.5 0 0.5 2 2 5
27.17 727.5 0 0.5 2 2 5
35.63 727.5 0 0.5 2 2 5
50.57 727.5 0 0.5 2 2 5
74.44 727.5 0 0.5 2 2 5
109.68 727.5 0 0.5 2 2 5
158.74 727.5 0 0.5 2 2 5
224.07 727.5 0 0.5 2 2 5
308.10 727.5 0 0.5 2 2 5
Portfolio 3 12.07 727.5 0 0.1 0.5 0.5 5
17.69 727.5 0 0.3 0.75 0.75 5
20.83 727.5 0 0.4 1 1 5
23.92 727.5 0 0.5 2 2 5
29.42 727.5 0 0.5 2 2 5
39.77 727.5 0 0.5 2 2 5
57.41 727.5 0 0.5 2 2 5
84.80 727.5 0 0.5 2 2 5
124.38 727.5 0 0.5 2 2 5
178.59 727.5 0 0.5 2 2 5
249.88 727.5 0 0.5 2 2 5
340.70 727.5 0 0.5 2 2 5
Bidder inputs are somewhat more complex. The model assumes there are three
suppliers, each with its own portfolio bid. Each portfolio has 12 parts. In
order to facilitate computation, each portfolio uses a step function. This is in
contrast to the piece-wise linear specification described above. Besides bidding
the price level and quantity of each step on the function, bidders also provided
ramp rates for each step for each reserve market. The ramp rates specify the
percentage increase each step could contribute to providing the reserve. As
described above, these ramp rates allow suppliers to indicate how much they
would like to bid in the PX's reserve markets. The ramp rates used in the
analysis were simply assumed with no significant check on the reasonableness of
their magnitudes. As will be discussed below, many of the empirical results rely
on the ramp rates assumed. The table shows supply side bids used in all the
models constructed.
2. MODEL OUTPUTS
Each of the models solves its respective cost minimization problems to allocate
capacity. This is a standard cost-minimization allocation problem where the
suppliers bids represent their willingness to provide capacity in each market.
For the models using simultaneous approaches, the ramping percentages are used
to establish constraints in the cost minimization problem. The table below
represents the model's dispatch quantities for the first quintile level of
-46-
demand, 18,475.76 MWH and the fully sequential evaluation technique under
highest cost bid pricing.
SUPPLY
Awarded Quantities
Energy Regulation Spin Non-Spin Replacement Total
16500 0 0 0 0 16500
521 7 36 36 127 728
0 22 55 55 70 201
0 29 73 73 0 175
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
728 0 0 0 0 728
0 22 55 55 364 495
0 29 73 73 0 175
0 25 146 146 0 316
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
728 0 0 0 0 728
0 22 55 55 364 495
0 29 73 73 0 175
0 0 83 83 0 166
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
Each model calculates prices based on the respective pricing methodology as
described above. While the dispatch for a particular evaluation technique may
not differ between pricing methodologies, the resulting prices will. These
prices as reported in the spreadsheets are illustrated in the table below. Again
the table reports results for a demand of 18,475 MWH and the fully sequential
evaluation technique under highest bid accepted pricing methodology.
-47-
MARKET PRICES ($/MWH)
Energy 14.37
Regulation 8.37
Spin 9.55
Non-Spin 9.55
Replacement 4.52
The following table reports the portfolio revenues implied by the model. These
revenues to producers are also the out-of-pocket costs to consumers. The table
contains the revenues that correspond to the prices and quantities for the 18475
MW level of demand and the fully sequential evaluation technique and highest bid
accepted pricing methodology.
PORTFOLIO REVENUE
Revenue from Each Market
Energy Regulation Spin Non-Spin Replacement Total Total
237173 0 0 0 0 237173
7485 61 347 347 573 8814
0 183 521 521 315 1539
0 244 695 695 0 1633
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
244658 487 1563 1563 887 249159
10457 0 0 0 0 10457
0 183 521 521 1645 2869
0 244 695 695 0 1633
0 207 1389 1389 0 2985
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
10457 633 2605 2605 1645 17945
10457 0 0 0 0 10457
0 183 521 521 1645 2869
0 244 695 695 0 1633
0 0 791 791 0 1582
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
-48-
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
0 0 0 0 0 0
10457 426 2007 2007 1645 16541
C. MODEL RESULTS
1. IMPACT ON PRODUCTION COSTS
The dispatch of capacity for each evaluation technique did not vary between the
pricing methodologies. This was to be expected because the optimization problem
did not vary between pricing methods, only over evaluation techniques. Further,
there is assumed to be no behavioral affect on bidding due to evalution
technique. Specifically, dispatch for the fully simultaneous evaluation
technique under the highest bid pricing methodology was exactly the same under
marginal cost and indifference to markets pricing methodologies. Similarly, the
dispatch under the fully sequential evaluation does not vary with the pricing
method. Consequently, only one set of production costs is reported here in the
table.
- --------------------------------------------------------------------------------------------------------------------
PRODUCTION COSTS
- --------------------------------------------------------------------------------------------------------------------
80th Percentile 90th Percentile
---------------------------------------------- --------------------------------------------
full simul seq-sim full sequen full simul seq-sim full sequen
- --------------------------------------------------------------------------------------------------------------------
Demand 24803.46 24803.46 24803.46 27724.76 27724.76 27724.76
Energy $267,946 $246,040 $246,040 $379,083 $332,588 $332,588
Ancillary Services $75,443 $99,593 $99,593 $126,552 $179,338 $179,338
Total $343,389 $345,632 $345,632 $505,635 $511,925 $511,925
- --------------------------------------------------------------------------------------------------------------------
- --------------------------------------------------------------------------------------------------------------------
50th Percentile 60th Percentile
---------------------------------------------- --------------------------------------------
full simul seq-sim full sequen full simul seq-sim full sequen
- --------------------------------------------------------------------------------------------------------------------
Demand 21799.95 21799.95 21799.95 22726.94 22726.94 22726.94
Energy $185,527 $177,804 $177,804 $211,267 $197,484 $197,484
Ancillary Services $56,212 $66,206 $66,206 $57,858 $72,965 $72,965
Total $241,740 $244,010 $244,010 $269,125 $270,449 $270,449
- --------------------------------------------------------------------------------------------------------------------
- --------------------------------------------------------------------------------------------------------------------
20th Percentile 40th Percentile
---------------------------------------------- --------------------------------------------
full simul seq-sim full sequen full simul seq-sim full sequen
- --------------------------------------------------------------------------------------------------------------------
Demand 18475.76 18475.76 18475.76 20685.92 20685.92 20685.92
Energy $120,669 $117,609 $117,609 $162,347 $155,614 $155,614
Ancillary Services $41,124 $45,591 $45,591 $49,935 $58,828 $58,828
Total $161,793 $163,200 $163,200 $212,282 $214,443 $214,443
- --------------------------------------------------------------------------------------------------------------------
As expected, the fully simultaneous evaluation process produces the lowest
overall dispatch of the evaluation techniques. This is because it allocates
higher cost supply to energy while lower cost supply is allocated to reserves.
This leads to an overall reduction in dispatch costs. The other two approaches
allocate lower cost supply to energy before examining the need for reserves.
-49-
This can be seen in the relative costs of energy and reserves across demand
levels. The cost of energy is higher under the simultaneous evaluation than
under the other two techniques. Further, the cost of ancillary services is lower
under simultaneous evaluation than under the other techniques. The combination
of the two yields an overall lower cost for the combined products.
The relative magnitude of this impact is apparent in the table. For example,
there is a difference of approximately $2,100 in production costs for the 40th
Percentile demand of 20,685 MW, or about 1 percent. Assuming these percentiles
hold for an entire year, there are annual savings of approximately $24 million
in dispatch costs from using a fully simultaneous evaluation procedure rather
than a fully sequential one.
Upon inspection of the table, the costs for both energy and ancillary services
are identical for the sequential-simultaneous and fully sequential evaluation
techniques. While this is not surprising for the case of energy, it is
surprising for the other ancillary services. Because energy is optimized first
in both cases, the two approaches will always produce the same energy costs. The
cost-reducing substitution can only take place among ancillary service markets
under the sequential-simultaneous approach. Thus, it is possible for the
sequential-simultaneous approach to produce lower costs for ancillary services
than the fully sequential approach.
However, at all levels of demand, the fully sequential and
sequential-simultaneous approaches produce the same results. The cost savings
can only be realized if lower cost capacity, allocated in the early markets, can
be freed up to substitute for higher cost capacity in the later markets. Given
the ramp rates assumed in the analysis, the early markets are tighter than the
latter markets. This means the early markets must access higher cost capacity to
meet their own level of demand. Thus, there is no opportunity for the cost
saving substitution to take place. Under different assumptions about the ramp
rates, it is possible that dispatch cost savings would result.
This simply highlights the fact that the ramp rates assumed in the analysis have
been arbitrarily chosen. These results show that the ramp rates used in the
analysis have a significant effect on the magnitude of these results.
Alternative ramp rates will alter the supply and demand mix in each of the
reserve markets and allow for more substitution between services. Alternative
ramp rates could change the size (but not direction) of the results
significantly.
2. IMPACT ON CONSUMER COSTS
There are two different ways to look at the results. Given a particular pricing
methodology, which evaluation technique produces the lowest costs to consumers?
Alternatively, given an evaluation technique, which pricing methodology produces
the lowest cost to consumers? The pricing and consumer costs will be presented
to answer both questions.
-50-
Nine different models were constructed in order to capture each combination of
pricing methodologies and evaluation techniques. Each model was run with six
different demand levels corresponding to particular percentiles of the demand
distribution. Because many of the results are qualitatively the same, prices and
costs will only be reported for two levels of demand.
a. EVALUATION TECHNIQUES FOR EACH PRICING METHODOLOGY
- --------------------------------------------------------------------------------------------------------------------
PRICING METHODOLOGY: MARGINAL COSTS
- --------------------------------------------------------------------------------------------------------------------
40th Percentile 90th Percentile
------------------------------------------- -----------------------------------------------
full simul Seq-sim full sequen full simul Seq-sim full sequen
- --------------------------------------------------------------------------------------------------------------------
Demand 20685.92 20685.92 20685.92 27724.76 27724.76 27724.76
Prices
Energy $21.74 $18.90 $18.90 $65.34 $35.63 $35.63
Regulation $22.75 $25.37 $25.37 $65.34 $84.80 $84.80
Spin $23.92 $27.17 $27.17 $65.34 $74.44 $74.44
Non-spin $23.92 $27.17 $27.17 $65.34 $74.44 $74.44
Replacment $21.74 $21.74 $21.74 $65.34 $57.41 $57.41
Consumer Costs
Energy $449,797 $390,862 $390,862 $1,811,536 $987,820 $987,820
Ancillary Services $61,835 $67,079 $67,079 $235,500 $247,567 $247,567
Total $511,632 $457,942 $457,942 $2,047,035 $1,235,387 $1,235,387
- --------------------------------------------------------------------------------------------------------------------
Under marginal cost pricing, energy prices are higher with the fully
simultaneous evaluation than under sequential approaches. This can be seen both
demand levels. In most cases, ancillary service prices are lower under fully
simultaneous than under sequential approaches. The exception to this is
replacement at the higher level of demand where plenty of low cost replacement
capacity is available. The price differentials are large. Given that much more
energy is purchased than reserve services, higher energy prices lead to larger
overall consumer costs even though the costs of ancillary services is lower
under the fully simultaneous evaluation.
These results are consistent with those discussed above. Specifically, low cost
capacity substitutes for higher cost capacity in providing ancillary services.
This increases the marginal costs for energy. When ramping constraints do not
bind, this leads to price equality across energy and ancillary services as seen
in the higher demand level. These examples show identical pricing for the
sequential-simultaneous and fully sequential evaluation techniques due to
assumptions about ramp rates.
The results are the qualitatively the same for the highest bid pricing
methodology as for marginal cost pricing, but consumer prices do not get as
high. Specifically, energy prices are higher (and ancillary services generally
lower) with fully simultaneous evaluation than with sequential evaluation.
Recall that, under this pricing mechanism, reserve prices are set at the
difference between highest cost resource and the energy price. Overall costs are
lower under the fully sequential evaluation than under fully simultaneous
evaluation. At the higher level of
-51-
demand the ancillary services cost less under fully simultaneous than fully
sequential, but energy costs are significantly higher.
- ----------------------------------------------------------------------------------------------------------------------
PRICING METHODOLOGY: HIGHEST BID ACCEPTED
- ----------------------------------------------------------------------------------------------------------------------
40th Percentile 90th Percentile
-------------------------------------------- -----------------------------------------------
full simul Seq-sim Full sequen Full simul Seq-sim full sequen
- ----------------------------------------------------------------------------------------------------------------------
Demand 20685.92 20685.92 20685.92 27724.76 27724.76 27724.76
Prices
Energy $21.74 $18.90 $18.90 $50.57 $35.63 $35.63
Regulation $12.52 $6.47 $6.47 $41.35 $49.17 $49.17
Spin $12.52 $8.27 $8.27 $29.74 $38.81 $38.81
Non-spin $12.52 $8.27 $8.27 $29.74 $38.81 $38.81
Replacment $7.37 $2.85 $2.85 $26.65 $21.78 $21.78
Consumer Costs
Energy $617,482 $494,178 $494,178 $1,402,059 $987,820 $987,820
Ancillary Services $47,169 $18,708 $18,708 $117,386 $119,150 $119,150
Total $664,652 $512,885 $512,885 $1,519,445 $1,106,970 $1,106,970
- ----------------------------------------------------------------------------------------------------------------------
The results are very different for pricing under market indifference. Under this
pricing methodology, energy prices are set at the highest marginal cost of
capacity providing any of the ancillary services. Because the fully simultaneous
evaluation allows for substitution of low cost energy capacity for high-cost
ancillary service capacity, this marginal cost is lower than under the other
evaluation techniques. This was the same result discussed in the pricing section
discussed above.
Consequently, under a market indifference pricing methodology, energy prices are
lower under fully simultaneous than under sequential techniques. The prices
listed for ancillary services are the average price for each service based on
the capacity providing the service. As can be seen from the two levels of
demand, the total ancillary service costs may or may not be lower under fully
simultaneous evaluation in comparison to sequential techniques.
- ----------------------------------------------------------------------------------------------------------------------
PRICING METHODOLOGY: MARKET INDIFFERENCE
- ----------------------------------------------------------------------------------------------------------------------
40th Percentile 90th Percentile
--------------------------------------- --------------------------------------------
full simul Seq-sim full sequen full simul Seq-sim full sequen
- ----------------------------------------------------------------------------------------------------------------------
Demand 20685.92 20685.92 20685.92 27724.76 27724.76 27724.76
Prices
Energy $23.92 $27.17 $27.17 $65.34 $84.80 $84.80
Regulation $4.76 $5.12 $5.12 $34.10 $29.62 $29.62
Spin $4.41 $4.44 $4.44 $30.45 $33.33 $33.33
Non-spin $4.41 $4.44 $4.44 $30.45 $33.33 $33.33
Replacment $7.56 $6.52 $6.52 $22.66 $38.53 $38.53
Consumer Costs
Energy $494,859 $562,029 $562,029 $1,811,536 $2,351,107 $2,351,107
Ancillary Services $15,191 $14,235 $14,235 $97,001 $126,306 $126,306
Total $510,051 $576,264 $576,264 $1,908,537 $2,477,414 $2,477,414
- ----------------------------------------------------------------------------------------------------------------------
-52-
These results indicate that the pricing methodology plays as large a role in
determining the overall costs to consumers as the evaluation technique. Under
different pricing rules, the costs resulting from each evaluation technique can
reverse themselves. This means that the combination of pricing rules and
evaluation techniques needs to be considered when deciding how the markets
should be operated.
It is important to refrain from making comparisons of ancillary services prices
between these tables. While the energy prices refer to the day-ahead energy
forward market for each technique, the ancillary services are different products
under each pricing methodology. As described in the pricing section, the
ancillary services prices for the marginal cost methodology is both the capacity
payment for standing ready and the energy payment for producing if needed. This
approach does not allow for the separation of the two. The other two
methodologies do allow for such a separation. In the next section, adjustments
are made to allow a comparison across pricing methodologies.
b. PRICING METHODOLOGY FOR EVALUATION TECHNIQUES
This section addresses the question of which pricing methodology yields the
lowest costs to consumers given the evaluation technique. To compare ancillary
service prices between methodologies, assumptions must be made about real-time
energy payments. In the tables in this section, real time energy prices are
assumed to be the same as day-ahead prices and each service is assumed called
for its full quantity in real time with certainty. These per unit energy
payments have been added to the capacity costs of ancillary services reported
above.
The results are contained in a single table that summarizes the prices and costs
for each pricing methodology for two levels of demand. The
sequential-simultaneous technique is eliminated because its results are
identical to the fully sequential technique.
The table shows that marginal cost pricing and highest bid pricing produce
identical results under fully sequential evaluation. This is because there is no
substitution between services and the highest cost service ends up being the
same as those providing marginal energy. Thus, there is no difference between
marginal cost pricing and highest-bid pricing under sequential evaluation.
Not surprisingly, the indifference to markets pricing and fully sequential
evaluation produce the highest costs to consumers. In order to keep suppliers
indifferent between the two markets, it is necessary to pay more for energy and
ancillary services. It should be highlighted that this analysis ignores the
behavior component of bid formation. Because the marginal cost and highest cost
bid pricing treat ancillary service providers more favorably, the bids under
these approaches will be altered to attempt to equate the returns between
selling energy and ancillary services in equilibrium. Thus the results for these
two pricing methodologies are NOT equilibria and will not hold when adjustments
in behavior are taken into account.
-53-
- ----------------------------------------------------------------------------------------------------------------------
EVALUATION TECHNIQUE
- ----------------------------------------------------------------------------------------------------------------------
FULLY SIMULTANEOUS
-----------------------------------------------------------------------------------------
40th Percentile (20685 MW) 90th Percentile (27725 MW)
---------------------------------------- ---------------------------------------------
Marg cost High bid Market in Marg cost High bid Market in
- ----------------------------------------------------------------------------------------------------------------------
Prices
Energy $21.74 $21.74 $23.92 $65.34 $50.57 $65.34
Regulation $22.75 $34.26 $28.68 $65.34 $91.92 $99.44
Spin $23.92 $34.26 $28.33 $65.34 $80.31 $95.79
Non-spin $23.92 $34.26 $28.33 $65.34 $91.92 $92.74
Replacment $21.74 $29.11 $31.48 $65.34 $77.22 $88.00
Consumer Costs
Energy $449,797 $449,797 $494,859 $1,811,536 $1,402,059 $1,811,536
Ancillary Services $61,835 $86,817 $79,523 $235,500 $299,654 $332,501
Total $511,632 $536,614 $574,382 $2,047,035 $1,701,713 $2,144,037
-----------------------------------------------------------------------------------------
FULLY SEQUENTIAL
-----------------------------------------------------------------------------------------
Prices
Energy $18.90 $18.90 $27.17 $35.63 $35.63 $84.80
Regulation $25.37 $25.37 $32.29 $84.80 $84.80 $114.42
Spin $27.17 $27.17 $31.61 $74.44 $74.44 $118.13
Non-spin $27.17 $27.17 $31.61 $74.44 $74.44 $118.13
Replacment $21.74 $21.74 $33.69 $57.41 $57.41 $123.33
Consumer Costs
Energy $390,862 $390,862 $562,029 $987,820 $987,820 $2,351,107
Ancillary Services $67,079 $67,079 $87,299 $247,567 $247,567 $431,950
Total $457,942 $457,942 $649,328 $1,235,387 $1,235,387 $2,783,058
- ----------------------------------------------------------------------------------------------------------------------
One of the major points of this exercise is to unbundle energy and ancillary
services. Thus it is appropriate to examine the impacts of each approach to
consumers of energy-only, consumers of ancillary services only, and consumers of
both energy and ancillary services. Of course, the relevant numbers are the
relative costs under each pricing methodology and evaluation technique.
For customers who only consume energy, the lowest cost is achieved under the
highest-bid (or marginal cost) pricing methodology and sequential evaluation.
This combination sets prices sequentially and does not allow for substitution
between energy and reserve services in the allocation of capacity.
For customers who only wish to purchase ancillary services from the PX, the
desired configuration of the market is quite different. They benefit from the
substitution found under simultaneous evaluation. They also benefit from the
lower cost produced by marginal cost pricing. Consequently, these customers
prefer marginal cost pricing with fully simultaneous evaluation.
Customers of both energy and ancillary services have the same preference as
those consuming energy only. They prefer sequential evaluation under marginal
cost pricing. Although they pay
-54-
more for ancillary services, these higher costs are more than made up by lower
prices for energy.
XI. ASSESSMENT OF SIMULTANEOUS AND SEQUENTIAL EVLAUATION OF ENERGY AND
ANCILLARY SERVICES MARKETS UNDER DIFFERENT PRICING METHODS
This section summarizes the various characteristics of each of the pricing
methodologies and evaluation techniques. The following table summarizes the
numerical findings with the qualitative findings in meeting the objectives FERC
set out for judging market mechanisms. The first four rows address whether the
combination pricing methodology and evaluation technique minimizes costs both to
producers and consumers. The last four rows indicate how well the technique does
against FERC's four principles.
- ----------------------------------------------------------------------------------------------------------------------
PRICING METHODOLOGY / EVALUATION TECHNIQUE
------------------------------------------------------------------------------
Attribute Fully Simultaneous Fully Sequential
- -------------------------------------- ------------------------------------- ---------------------------------------
Marg Cost High bid Market In Marg Cost High bid Market In
------------------------------------- ---------------------------------------
Minimizes dispatch costs? Yes Yes Yes No No No
Minimizes costs to consumers
Energy? No No No Yes Yes No
Ancillary Services? Yes No No No No No
Total? No No No Yes Yes No
Incentive to bid marginal costs? Yes No Yes Yes No Yes
Indifferent to selling energy or A/S? Yes No Yes Yes No Yes
Provides No Bilateral Bias? Yes Yes No Yes Yes No
Separate capacity and energy payments? No Yes Yes No Yes Yes
- ----------------------------------------------------------------------------------------------------------------------
The results for the first four rows come directly from the discussion of prices
and costs to consumers. Simultaneous evaluation minimizes dispatch costs, but
sequential evaluation minimizes total costs to consumers.
The incentive to bid marginal costs is apparent from whether there is a gain
from bidding above marginal costs. Under highest bid pricing, there is an
incentive to raise bids for capacity that may provide reserve services since the
capacity payment is the difference between the highest bid and the energy price.
Under market indifference pricing, reserve payments depend on the size of
difference between bids and marginal costs. Payments are larger with lower bids.
The incentive to sell ancillary services is strong under highest-bid pricing
since the return to marginal generators for providing reserves is much greater
than for energy. The market indifference pricing was designed to avoid such a
bias between markets. Marginal cost pricing often comes up with the same price
for energy and reserves.
The market indifference pricing does provide an incentive for bilateral trades.
This is because it does not calculate a market-clearing price. Suppliers are
paid based on their bids and consumers pay an average of these prices. Low cost
capacity has the incentive to split the
-55-
difference with another supplier. The other pricing techniques do not since they
produce market-clearing prices.
Marginal cost pricing does not provide a convenient way to split the payment for
ancillary services between a capacity payment for standing ready to produce and
energy when output is produced. While this feature is not on FERC's list, it is
a desirable feature because of the two-part design of the ISO's ancillary
services markets. Such an ability to split between the two will allow the PX's
market to mesh better with the ISO's ancillary services markets.
None of the combinations of pricing methodologies and evaluation techniques has
all the desirable properties of an ideal market. In adopting a particular market
mechanism, these trade-offs will have to be discussed and weighed in the
policy-making arena.
XII. CONCLUSIONS AND RECOMMENDATIONS
The framework presented here has required that the suppliers of energy into the
PX market are willing to sell AS to the PX. Although the bidders ultimately have
the choice of whether to participate in the PX's energy only or energy and
ancillary services markets, they are required to bid identically capacity to
produce energy capacity to produce ancillary services. This is a significant
improvement on the existing framework regardless of whether the bids are
evaluated sequentially or simultaneously.
However, this approach is not without costs. While bidders still have the option
to make portfolio bids, this approach requires them also to incur more costs in
the formulation of these bids than do their current techniques. Bidders must now
take into account how they plan to commit units to meet both energy and
ancillary service requirements in deciding their bid prices, quantities and ramp
rates. Further, there are potential difficulties in dis-aggregating their
awarded quantities into unit schedules. These problems can be mitigated
significantly by reducing the number of units contained in a portfolio.
Similarly, this framework imposes burdens on producers and consumers by
extending their choices for the provision of ancillary services. While, in
general, more choices are preferred to fewer, they come at the cost of
understanding the differences and making a choice.
Further, the approach imposes more costs on the PX. In particular, there are the
issues of over- and under-self-provision and how they should be handled. To a
large degree, this discretion opens the PX to taking positions in ancillary
services markets. However, to the extent that the PX makes mistakes, its
customers and suppliers can choose not to participate in the PX's markets.
Ultimately, arbitrage between the PX's and ISO's markets will work to make both
markets more efficient. This arbitrage may also increase the volatility of
ancillary service prices in both markets.
Market pricing mechanisms need to be examined in relation to the desirability of
a sequential or simultaneous evaluation technique. The impacts of the pricing
methodology and evaluation
-56-
technique are not independent. The desired results whether they be dispatch
costs or costs to consumers are also important.
It is clear that the PX's energy market is competitive most of the time. By
linking the ancillary services market to the competitive energy market, the
competition in ancillary services may be enhanced. Under the consistent bid
framework described, the only way a supplier could exercise market power in the
reserves markets would be to withhold capacity. It may be relatively easy to
detect such behavior under the described market.
Overall, there may be a benefit from the PX introducing a competing ancillary
services market, as described here, if the implementation costs can be kept at
reasonable levels. However, more research should be done to gauge the magnitude
of the benefits.
The general conclusion of this report is that the impact of adopting a
simultaneous market for energy and ancillary services depends largely on the
pricing methodology also adopted. Under two pricing methodologies examined here,
marginal costs and highest bid, simultaneous markets produced lower dispatch
costs and higher consumer costs than similar sequential market. One pricing
methodology, indifference to markets, produced lower dispatch costs and lower
costs to consumers.
These results are subject to the following caveat. The analysis did not
explicitly take into account how bidding behavior might change under these
mechanisms. One of the three pricing methods provides incentives to alter
bidding behavior from what was assumed here. Another attempted to provide
incentives not to alter behavior while the incentive for the third technique was
unclear.
None of the pricing methodologies meet all of the desirable pricing
characteristics as put forth by FERC. Trade-offs between the desirability of
each market characteristic is left for the policy makers to decide. Nonetheless
more analysis may better inform this policy decision. Besides examining
different levels of ramp rates, it would also be instructive to examine how
bidding behavior might differ between these market structures.