Exhibit 99.386
A REPORT TO THE CALIFORNIA POWER EXCHANGE:
ITERATIVE BIDDING IN THE PX MARKET
February 9, 1999
Dr. Peter H. Griffes
ANALYSIS GROUP / Economics
- 2 -
A REPORT TO THE CALIFORNIA POWER EXCHANGE:
ITERATIVE BIDDING IN THE PX MARKET
I. EXECUTIVE SUMMARY............................................................................... 3
II. INTRODUCTION AND PURPOSE OF STUDY............................................................... 4
III. DESCRIPTION OF ITERATIVE BIDDING................................................................ 6
A. PURPOSE OF ITERATIVE BIDDING................................................................. 6
B. DESCRIPTION OF ACTIVITY RULES................................................................ 7
1. Opening rule.............................................................................. 7
2. Revision rule............................................................................. 7
3. Exclusion rule............................................................................ 8
4. Withdrawal rule........................................................................... 9
5. Closing rule.............................................................................. 9
C. EXAMPLE OF ACTIVITY RULES.................................................................... 10
IV. STATUS OF ITERATIVE BIDDING IN THE PX MARKET.................................................... 12
A. PRIOR TO THE OPENING OF THE MARKET........................................................... 12
B. SINCE THE OPENING OF THE MARKET.............................................................. 13
V. SUMMARY OF RESEARCH DONE IN JUNE 1998........................................................... 13
A. BENEFITS FROM ITERATIONS..................................................................... 14
1. Review of London Economics Study.......................................................... 14
2. Theoretical Benefits...................................................................... 15
B. COSTS OF ITERATIONS.......................................................................... 16
1. Feasibility constraints................................................................... 17
2. Strategic behavior........................................................................ 22
3. Transactions costs........................................................................ 24
C. INTERVIEWS WITH PARTICIPANTS................................................................. 25
D. CONCLUSIONS FROM JUNE 1998 STUDY............................................................. 26
VI. SUMMARY OF RESEARCH DONE IN NOVEMBER 1998....................................................... 27
A. EXCERPTS FROM `BETA TEST OF ITERATIVE BIDDING IN THE PX DAY-AHEAD MARKET'.................... 28
B. COMMENTS AND OBSERVATIONS FROM NOVEMBER RESEARCH............................................. 31
1. Comments on results....................................................................... 31
a. The report's discussion................................................................... 31
b. The report's omissions.................................................................... 32
c. Interpreting the report's results......................................................... 33
2. Comments on interviews.................................................................... 37
a. Benefits and costs........................................................................ 37
b. Concerns about gaming..................................................................... 37
C. CONCLUSIONS FROM NOVEMBER RESEARCH........................................................... 38
VII. CONCLUSIONS.................................................................................. 39
VIII. SOURCES...................................................................................... 41
APPENDIX A: SURVEY RESULTS........................................................................... 42
APPENDIX B: ITERATIONS RESULTS FOR BETA TEST......................................................... 45
- 3 -
I. EXECUTIVE SUMMARY
This report addresses some of the issues raised in the evaluation and possible
implementation of iterative bidding in the market. In approving the California
market structure, FERC directed the PX to study how iterations improved the
efficiency of its day-ahead energy market. This report describes the research
effort sponsored by the PX on the topic of iterative bidding in the last year.
The research has largely been motivated by the possible implementation of the
market iterations. The PX has not yet incorporated iterations into its market.
Because it is not clear whether the benefits to iterations outweigh their costs,
the PX has postponed implementing iterative bidding indefinitely.
The PX has sponsored two separate research projects. One was done in June 1998,
and the other took place in November 1998. The June research assessed costs and
benefits of implementing iterations. The June research summarized the role of
iterations in the market and the possibility of increasing efficiency through
iterations. The research explicitly laid out the benefits from iterations and
identified three costs to implementation: 1) feasibility constraints, 2)
strategic behavior, and 3) transactions costs.
Iterations are time consuming. Feasibility constraints address whether there is
enough time for a sufficient number of iterations. Strategic behavior addresses
possible gaming of the market rules. Transaction costs relate to how much more
participants would have to pay to undertake iterations profitably. The June
research showed that schedule would have to be very tight. It also identified
gaming opportunities involving fictitious bids that may have to be addressed
before implementation. Further it showed that transactions costs are significant
and are largely born by participants.
The second research project took place in November 1998 after iterations
software was developed. One objective of research was to test it. The PX held
market simulations of iterations for five separate days in early November. Many
of the same issues identified earlier were raised in this research. This
included the feasibility of having enough time for a sufficient number of
iterations, and transaction costs imposed on participants. Because the benefits
and costs of iterations would accrue to participants, the PX conducted a formal
survey of participants after the market simulations were complete.
The November research results indicated the costs of implementing iterations
outweighed the benefits they would produce. The research highlights the benefits
to the demand side of the market of seeing energy prices before deciding how
much to buy from the various energy markets. Further, the November results
display market price and quantity movements that are consistent with strategic
behavior on the supply side of the market. Further investigation would be needed
to determine whether suppliers were manipulating prices in these tests. The
issue of strategic behavior in the market would have to be fully addressed
before iterations could be introduced.
- 4 -
II. INTRODUCTION AND PURPOSE OF STUDY
In the original design of the PX energy market, iterations were a feature of the
auction design. Iterations allow suppliers a chance to adjust their bids to
better achieve feasible schedules. As a condition of approving the market
structure, FERC required further study on the efficiency of the designs of the
iterations in the market.
In its October 30, 1997 decision conditionally approving the market structure,
FERC states:
We will require that the PX conduct additional studies to further evaluate
the proposed auction, which must be completed and submitted to us by January 1,
1999. The studies should analyze whether the Phase II auction results in an
efficient, least-cost dispatch. The effect of limiting the number of iterations
in the Phase II auction (so that the auction may terminate before reaching an
equilibrium) should be explored, since the ISO/PX proposes only one round of
bidding initially, and only five rounds of bidding later on. Also, the effects
of transmission congestion and the existence of no-load costs on the part of
generators should be explored for each auction.
The studies should also examine the effects of the regular repetition of
auctions involving the same participants and similar costs from day to day. The
studies should pay specific attention to the susceptibility of the auction
format to overt and tacit collusion and the potential to exercise market power.
This report addresses some of the issues raised in the evaluation and possible
implementation of iterative bidding in the market. It describes the research
effort sponsored by the PX on the topic of iterative bidding in the last year.
The research has been motivated by the possible implementation of the market
iterations, rather than the questions raised by FERC in its decision.
The PX has not yet incorporated iterations into its market. In evaluating
possible implementation of iterations, the PX and its participants have been
concerned that the costs imposed on the PX and market participants would
significantly outweigh the benefits that may result. Consequently, the PX Board
has postponed the implementation of iterative bidding indefinitely.
In coming to this stance, the PX has undertaken two separate research projects.
The first was in June 1998. The purpose of the June research was to assess the
need for iterative bidding in the market and explore issues pertaining to its
implementation. In particular, there were concerns whether there would be enough
time for iterations and how much iterations would burden participants. Further,
opportunities for strategic behavior (or gaming) were identified and discussed
in the June research. A number of issues would have to be addressed before
implementing iterations.
- 5 -
The June research summarized the role of iterations in the market and the
possibility of increasing efficiency through iterations. It described the
revision rules and how they would work to move prices. The research explicitly
laid out the benefits from iterations and identified three costs to
implementation: 1) feasibility constraints, 2) strategic behavior, and 3)
transactions costs. Previous research efforts concerning each of these costs
were summarized and considered.
The second research project took place in November 1998. Between June and
November, the PX developed software that would allow iterations. One of the
objectives of the research was to test that the software performed as it should.
Thus, market simulations were run with PX participants for five separate days in
early November.
In revisiting iterations, many of the same issues were raised including benefits
of iterations, feasibility of having enough time for a sufficient number of
iterations, and transactions costs imposed on participants. The PX realized that
many of the benefits and costs of iterations would accrue to participants, so it
conducted a formal survey of participants after the market simulations were
complete. As described by participants, the costs of implementing iterations
outweighed the benefits they would produce. Consequently, the PX postponed
indefinitely the introduction of iterations.
The November research highlights the benefits to the demand side of the market
of seeing energy prices before deciding how much to buy from the day-ahead,
hour-ahead and real-time markets. This could be a significant benefit of
iterations.
Further, the November results display market price and quantity movements that
are consistent with strategic behavior on the supply side of the market.
However, they are also consistent with unprofitable portfolios exiting the
market. Further investigation would be needed to determine whether suppliers
were manipulating prices in these tests. Price and quantity movements did not
suggest strategic behavior on the demand side of the market. However, such
behavior is possible. Because the interests are diametrically opposed, it is not
clear what the value of iterations would be when both suppliers and demanders
are behaving strategically. More research is needed on this topic.
The rest of the report is laid out in the following manner. Section III provides
a description of iterative bidding. It specifies the benefits from iterations
and works through examples of the revision rules applicable to the iterative
process. Section IV provides a brief history of the consideration of iterations
in the PX market. Section V describes, in some depth, the research effort
undertaken in June 1998. The research specifies the costs and benefits
associated with adopting iterations into the market. Section VI summarizes the
Beta Tests that took place in November 1998. Besides reporting the conclusion
made at that time, it discusses, in some detail, some of the results and
implications of the research that was done. Section VII concludes the report.
- 6 -
III. DESCRIPTION OF ITERATIVE BIDDING
A. PURPOSE OF ITERATIVE BIDDING
Iterative bidding in the day-ahead PX market allows market participants to
change their bids in response to prices revealed in earlier rounds of the
auction. This price discovery process has several benefits to suppliers. First,
it assists them in securing feasible production schedules. Second, it also
provides them with information, so they can calculate their revenues and costs
to determine whether they will be able to recover their relatively fixed costs
attributable to start-up and no-load. Finally it allows them to adjust their
bids (or withdrawal) to improve their financial position.
The design of the PX day-ahead energy market is a one-part energy bid. The
bidding instrument is a 15-segment linear bid curve that must be increasing in
the price over the entire quantity offered. Bidders can submit separate bid
curves for each of 24 hours in the day-ahead auction. As such, bidders are
expected to bid in such a way as be able to cover their costs with the awards
and prices. Further, the PX evaluation process evaluates each hourly market
independently of all other hours. Thus, an award in one hour does not ensure
awards in adjacent hours.
Some of a generator's costs are fixed once the generator has started, but are
otherwise avoided. In particular, start-up energy costs become a sunk cost once
the generator starts. No-load costs are avoidable if a generator shuts off, but
are fixed for each hour of operation. The bidding structure does not explicitly
account for recovering these costs. These costs may need to be recovered on a
daily basis. Accordingly, bidders would need to consider these costs as they
prepare their bids. However, there is no simple way to do this and be assured
that resulting schedules will be feasible.
Iterative bidding is a way bidders can get information about prices and awards
without becoming financially obligated to meet generator schedules. Iterative
bidding would allow participants to change their bids in response to quantities
and prices revealed in earlier rounds of the auction. Bidders would also be able
to withdraw from the auction entirely if prices are not high enough to cover
costs. This would allow bidders to adjust their schedules through the iterative
process, resulting in feasible production schedules.
Iterations are a risk management tool that hedge against potential losses due to
infeasible schedules. Suppliers with large portfolios can allocate production
across units to meet their awarded quantities and recover costs. Suppliers with
small portfolios or single units are much more vulnerable. They run a greater
risk of not recovering start-up and no-load costs. Similarly, hydro generation
can better allocate scarce water to periods that will provide the most value.
Activity rules to ensure the market converges to a suitable price and quantity
for all participants limit the ways in which bids can be revised. Essentially,
suppliers must lower their bid prices
- 7 -
through the iterations while demanders must increase their bid prices. The
activity rules determine how quickly the auction will converge on an
equilibrium.
B. DESCRIPTION OF ACTIVITY RULES
This section discusses the activity rules proposed by the PX for its iterative
auction. The iterative auction uses these rules to govern how bidders can change
their tenders. The rules were designed to provide quick and decisive price
discovery. They attempt to limit strategic behavior by bidders who may try to
signal false quantity and price information. The bidding rules also encourage
timely convergence to an equilibrium. Five activity rules have been proposed for
the PX market.
1. OPENING RULE
The Opening Rule describes how supply and demand bidders initially submit
tenders. Under the proposed PX iterative bidding protocols, each market
participant wishing to participate in the auction must make a bid in the first
iteration. No new bids will be accepted after the first iteration. The purpose
of the opening rule is to be sure that all quantities in the market are
represented at the beginning of the auction. Without an opening rule, agents
might hold back their bids to discover information about other participants'
valuations. After seeing others' valuations, the withholders can formulate their
bids to work to their advantage. Anyone bidding at the first of the market is
put at a disadvantage by this behavior.
2. REVISION RULE
The revision rule dictates how bidders can change their bids between iterations.
The rule allows for prices, but not quantities, to be revised. If suppliers
choose to revise their bids, they must lower their selling price. Similarly,
buyers who want to adjust their bids must increase their buying price.
The hourly bids into the market take the form of 16 price-quantity combinations
that are connected by straight lines to form the bid curves. Each price-quantity
combination is called a breakpoint. Supplier's breakpoints must be increasing in
prices and quantities. For the breakpoints that specify a Demander's demand
curve, the prices must decrease as the quantities increase. The first and last
points define the bidder's minimum and maximum quantities. To be in the market,
each participant must submit hourly bids for each of the 24-hours of the day
In particular, the revision rule requires that a supplier (buyer) must lower
(raise) the price of a breakpoint in the next iteration if it wants to keep that
breakpoint active. In each iteration i, the PX calculates a market-clearing
price (MCP[i]). This MCP[i] defines what portions of the bid are active or
passive for iteration i+1. The active portion of the bid curve is defined as the
segment with breakpoints above MCP[i]. These breakpoints are extra-marginal. The
passive portion of
- 8 -
the bid curve is the segment with breakpoints below MCP[i]. These breakpoints
are infra-marginal.
Figure 1 illustrates the definition of active and passive breakpoints on a
supply curve. Points A, B, and C represent breakpoints above the MCP from the
first round. The revision rule requires that for these three breakpoints (A, B
and C) to remain active, the supplier must lower their price to a level below
MCP[1]. The bidder need not adjust its bid at all. However, as will be discussed
below in the exclusion rule, the bidder will not be able to adjust these
breakpoints in subsequent iterations of the market.
FIGURE 1: EXAMPLE OF ACTIVE AND
PASSIVE BREAKPOINTS
[LINE GRAPH]
The revision rule requires suppliers to lower the price for each active
breakpoint below MCP[i] if the breakpoint is to remain active. Passive
breakpoints do not have to be revised, but can be if the supplier decides to
lower them as well. An analogous rule applies to demand bidders. Specifically,
the adjustment to any break points must be in an increasing direction.
3. EXCLUSION RULE
As mentioned above, the exclusion rule freezes active breakpoints that are not
revised at the first opportunity in subsequent iterations. Frozen breakpoints
cannot be changed as long as the
- 9 -
market price is falling. This rule requires bidders to lower their prices in
each iteration if they want to be able to adjust the bid in subsequent
iterations.
More formally, any breakpoint on the supply tender becomes frozen in the current
iteration (i) if: the breakpoint was active in iteration (i-1) because the
breakpoint's price exceeded MCP[(i-1)] and the breakpoint's price is not lowered
in iteration (i) to be less than MCP[(i-1)]. This means a supplier must lower
its price bids below the MCP if it wishes to be able to adjust those parts of
the bid in subsequent rounds.
A breakpoint is always frozen relative to the MCP of the previous iteration.
Specifically, suppliers can always adjust a breakpoint if they have adjusted it
below the MCP from the previous round. If a breakpoint has become frozen, it can
only be adjusted if it is "un-frozen" in subsequent iterations if the price
rises. For a breakpoint frozen in iteration i, MCP[i-1] is the breakpoint's
activation price to un-freeze it. A bid is no longer frozen if the MCP in
subsequent iterations (i+1) rises above the activation price, MCP[i-1].
Demand side bidders have an analogous exclusion rule. They must increase the
price portion of the bid to be able to adjust it in subsequent iterations.
The exclusion rule forces suppliers to commit to progressively lower prices when
prices are falling between iterations or to signal that they will not sell at a
lower price. Similarly it forces demanders to commit to progressively higher
prices when prices are increasing between rounds or to signal that they will not
buy at any higher price.
4. WITHDRAWAL RULE
The withdrawal rule allows a bidder to remove its entire tender from an hourly
market at the end of an iteration. A supplier may pull its portfolio if prices
are iterating to levels that may be insufficient to cover its costs.
Alternatively, a buyer may pull its portfolio if prices are iterating to levels
that impose too great of a cost on consuming.
However, there are some restrictions on this withdrawal. All portions of the
hourly tender must be withdrawn. Perhaps more importantly, the tender cannot be
re-offered in subsequent iterations once it has been withdrawn. Further,
withdrawal can occur at the end of any iteration except for the final iteration.
Once the auction has finished, no accepted bid can be withdrawn.
5. CLOSING RULE
The closing rule governs when the auction terminates and awarded prices and
quantities are finalized. All hourly markets will close simultaneously. The
iterative process will be terminated when one or more of the following has
occurred. First, the auction will end when no notable improvement is made in a
MCP from prior iterations. Second, the market closes when no valid bid revisions
are received from the previous iteration. Specifically, the active bidders
choose not to make improvements to their bids. Third, a maximum number of
- 10 -
iterations is reached. Fourth, the time allotted for iterations elapses. Under
these last two rules, the auction may not have iterated to its equilibrium
values.
C. EXAMPLE OF ACTIVITY RULES
This section describes an example of the bidding rules in effect. The purpose of
this exercise is to show how a tender can be changed under these rules. The
example will focus on a supply bid. An analogous example could be constructed
for the demand side of the market.
FIGURE 2: SUPPLIER'S INITIAL TENDER
[LINE GRAPHIC]
Figure 2 describes a supply curve tendered at the opening of the market and the
resulting MCP and awarded quantity at the end of the first iteration. MCP[1] is
the market-clearing price from the first iteration. Three of the bids break
points, A, B and C, are active at this point. The bidder has the choice of
adjusting its bids or remaining satisfied with its bids. It should be
highlighted that other bidders are likely to adjust their bids. Thus, this
bidder would likely end up providing less than Q[1] if it did not adjust its
bids. In this example, suppose the supplier decides to adjust two of these
breakpoints, A and B, before the second iteration. While the bidder could adjust
the price associated with A and B to a level greater than MCP[1], this would
mean that it could not make any further adjustments in subsequent iterations
unless the price
- 11 -
were to go above MCP[1]. Assume the bidder wants to keep these breakpoints
active, it must then set their prices below MCP[1].
This can be seen in Figure 3. A new bid curve results, containing A(1), B(1) and
C. Again C is not revised, reflecting the supplier's unwillingness to supply
Q(C) at a lower price.
FIGURE 3: SUPPLIER LOWERS BREAKPOINTS
[LINE GRAPH]
After bids are submitted in the second iteration, the PX calculates MCP[2]. This
can be seen in Figure 4. In this example, MCP[2] < MCP[1], but this is not
necessary in all cases. By lowering the bid prices associated with A and B, the
bidder has increased its awarded quantity (Q[2] is greater than Q[1].). The
level of MCP[2] renders A passive. As long as subsequent MCPs do not drop below
A, this breakpoint remains passive. Since it is above MCP[2], B is an active
breakpoint. The supplier must lower the price for B in the next iteration or
risk freezing B at its current bid price. Breakpoint C is frozen because its
price was not lowered below MCP[1] when bids for iteration 2 were made. It
cannot be adjusted unless a subsequent MCP moves above its activation price,
MCP[1]. If MCP[i] rises above MCP[1], breakpoint C is unfrozen and becomes
active or passive, depending on the MCP[(i+1)].
- 12 -
FIGURE 4: RESULT OF ITERATION 2
[LINE GRAPH]
This example provides insight into how the bidding rules are designed to work in
the market iteration process.
IV. STATUS OF ITERATIVE BIDDING IN THE PX MARKET
A. PRIOR TO THE OPENING OF THE MARKET
Before the initial filing, iterative bidding came up in the discussions of
market design. A single-part bidding structure with iterations was proposed as
an alternative to a multi-part bidding mechanism. Iterative bidding was part of
the market design that the PX submitted to the FERC in its in March 1997 Phase
II filing. Much of the market rules and mechanism were still being worked out.
In the design and implementation of the market, time constraints became binding.
Part of the problem was the requirement for coding of software to implement the
market before all design issues had been worked out. The impending opening of
the market and resolving difficult design and implementation problems forced the
PX to delay the introduction of iterative
- 13 -
bidding. In October 1997, FERC approved the delay of iterative bidding until
after the opening of the market. It was to be implemented in June 1998.
B. SINCE THE OPENING OF THE MARKET
In June 1998, there were still many issues to be resolved before iterative
bidding could be introduced. With the delay in the opening of the market, other
design features still needed to be introduced. There was some concern about the
introduction of iterations. While there appeared to be theoretical benefits to
iterations, there were also unforeseen costs. These included finding the time
for the market to iterate. At that time, an assessment of the role of iterative
bidding was done.
While this report will examine the results of this assessment in greater detail
below, the research involved examining how iterations would fit into the
schedule. Concerns about the adequacy of time schedule were significant, both on
the part of the PX and its participants. Interviews examining the costs and
benefits of iterations were done with market participants. At that time,
iterations had been scheduled for implementation in August 1998. However, as a
result of that research effort, the PX board decided to delay further the
implementation of iterations. It also decided to undertake more research and
revisit the issue in the fall.
The issue of iterations came before the board again in November 1998. Prior to
this consideration, a beta test of an iterative mechanism was performed. In the
test, market participants were given the opportunity to bid in an iterative
fashion. After the tests, interviews were conducted with participants. Again,
the research done in the fall will be discussed in greater detail below.
In November 1998, the PX board decided to postpone indefinitely the introduction
of iterative bidding into the market. The general conclusion was that, while
some participants may find it helpful, implementing iterations would impose
significant cost on the PX and participants. The PX would have to refine the
mechanism to avoid gaming. Further, market participants would have to incur
significant costs to take advantage of the iterative process. Also, bid
strategies and cost recovery mechanisms from a single shot auction had been
worked out over time.
V. SUMMARY OF RESEARCH DONE IN JUNE 1998
This section will discuss the research that was conducted in June 1998. The
research effort focused on the issue of whether the benefits of the introduction
would outweigh the costs imposed of the market. The research was qualitative in
nature; specifically, it did not attempt to quantify the costs and benefits of
the introduction of iterative bidding. Benefits were examined. Previous
quantitative estimates of benefits were reviewed. Theoretical benefits were
enumerated. Costs were also specified including, feasibility constraints,
strategic behavior, and transaction costs. Finally, interviews were conducted
with market participants to gain their
- 14 -
insights into the introduction of iterations. The findings of each of these
areas will be summarized below.
A. BENEFITS FROM ITERATIONS
1. REVIEW OF LONDON ECONOMICS STUDY
A number of studies were done on the iterative bidding mechanism. Robert Wilson,
Charles Plott and London Economics completed these studies. Each of the studies
has been filed with FERC at various stages of the proceedings approving the
market structure. The research done by Wilson and Plott focused on the
efficiency of the auction mechanism, particularly, the bidding and activity
rules. The research completed by London Economics addressed the overall
efficiency of the market from the delay of various market features, including
iterations. Because of its focus on quantification, the London Economics study
was of particular interest in the current context.
London Economics (LE) completed research in September 1997. Among other market
features, the study examined the loss in efficiency caused by staging iterative
bidding. The LE study considered the efficiency of the PX market with and
without iterative bidding. The methodology used a market simulation technique
that assumed bidding behavior by market participants. The differences between
the results in the single-shot market and the iterative market provide a basis
to quantify the potential benefits of iterative bidding.
The study compared single-shot versus iterative bidding by measuring three
different effects of iterative bidding. First, the study compared the market
prices with and without iterative bidding to determine how much they would
change. Specifically, it measured how well the resulting prices indicate the
cost of the generators that are chosen in the auction. Second, the differences
in total thermal generation costs under single-shot versus iterative bidding
were measured. This measured the efficiency loss on the supply side of not
having the least cost generators producing. Third, the quantity of capacity that
would change under iterative bidding was examined. This variable measures how
much capacity would be producing with iterations that would not be producing
without iterations. It indicates how much trade between generators changes.
In terms of methodology, LE used a simulation model that accounts for underlying
generator costs and formation of bids. Two different methods were used to
estimate results with and without iterations. First, to simulate one-shot
auctions, the model was used to produce optimal bids, market prices and
generation schedules assuming perfect information about demand and supply
conditions for "typical" spring, summer and winter weekdays. The resulting bids
were then fixed. These fixed bids were submitted for a series of 20 days, each
with different demand and supply conditions. Given the variation in supply and
demand conditions, the fixed bids could only approximate the optimal bids for
those conditions. The results of these runs indicate how bidders would fair
without iterations.
- 15 -
These results were compared to the bids produced using iterative bidding. Under
these conditions, the same series of 20 days were used to allow bidders to
iterate to a market solution. The difference between the results without
iterations and the results with iterations measures the impacts on not having
iterations.
Five different scenarios were run for this comparison. These scenarios vary with
the amount and type of uncertainty introduced into the market. One demand side
scenario reflected the possibility of demand being different than expected.
Three supply-side scenarios reflecting different levels of uncertainty in the
availability of supply with cases for 1%, 3% and 5% difference from expected.
Finally, a scenario was constructed with both the demand uncertainty and supply
uncertainty.
Table 1: London Economics Study: Percent Difference from Efficient Dispatch
PARAMETER WINTER SUMMER
- -------------------------------------------------------------------------------------------------
Allocative Thermal Trade Allocative Thermal Trade
Effc'y Cost Effc'y Cost
- -------------------------------------------------------------------------------------------------
Demand uncertainty 0.1% 1.1% 3.8% 0.1% 3.9% 8.1%
1% supply uncertainty 0.0% - - 0.3% - -
3% supply uncertainty 0.0% 1.8% 10.3% 1.0% 3.0% 12.2%
5% supply uncertainty 0.0% - - 0.9% - -
Combined 0.3% 2.5% 10.3% 0.8% 3.8% 12.2%
Table 1 contains the results of the London Economics study for each measure for
each scenario. The deadweight efficiency loss from the lack of iterations
indicates the societal loss from lack of iterations is 0.3 to 0.8% of the total
market value. This means that market prices under a single-shot market are
within one percent of those resulting under iterations. The research showed that
the costs of thermal production could be as high as 3.9% higher without
iterations than with iterations. The trade number indicates how different
generators' outputs were with and without iterations. This difference was as
much as 12% of average market demand. Of course, this benefits some generators
and costs others.
2. THEORETICAL BENEFITS
The theoretical benefits from iterations are much the same as those described
above in section II.A. As described there, iterations enhance the ability to
submit feasible schedules. They also allow better recovery of start-up and
no-load costs. Iterations are a risk management tool for smaller portfolio
bidders. Each of these benefits is based on the reduction of uncertainty bidders
gain from iterations.
Supply bidders face uncertainty as to the prices that will prevail in the
market. With a better idea of prices, it is easier to formulate bids that will
allow sufficient production to ensure the recovery of costs. The benefits from
iterations come from reducing uncertainty. Iterations will
- 16 -
have a greater value if they reduce uncertainty more. It is possible that
iterations could increase uncertainty if they are sufficiently abused to signal
false valuations.
Over time, greater knowledge of the market and introduction of other risk
management tools have reduced of benefits of iterations in the market.
The novelty of the market has been a source of uncertainty. As time passes, the
history of market operation extends, and participants gain greater experience
with predicting market conditions. Thus, participants can better assess their
ability to recover costs and schedule output, and gain less of a benefit from
the introduction of iterations.
The value of iterations also depends on the availability of other risk
management instruments. For example, the introduction of the hour-ahead market
reduced supplier's exposure to infeasible schedules since the quantities
associated with feasible schedules could be more easily bought or sold.
Consequently, as other risk management instruments have been introduced, the
benefits of iterations are reduced.
To summarize the benefits from iterations, smaller portfolio bidders may benefit
since they would be able to better avoid infeasible and unprofitable schedules.
The London Economics study shows a modest gain in societal efficiency from
iterations altering the price of energy. It also showed a reduction in
production costs and reallocation of generators providing power. Finally, the
benefits from the introduction are reduced over time as market experience is
accumulated and risk management instruments are introduced.
B. COSTS OF ITERATIONS
The research effort in June identified three categories of costs of introducing
iterative bidding into the market: 1) feasibility constraints, 2) strategic
behavior and 3) transaction costs. Feasibility constraints address how easily
the iterative procedures could be incorporated into the PX's market and
scheduling process. The value of iterations comes from the ability of bidders to
adjust their offers. If there is insufficient time to make adjustments, the
benefits of iterations could be significantly reduced. Strategic behavior
involves the ability of participants to signal others a false valuation in order
to take advantage of the responses to the false valuation. While this sort of
behavior is not encouraged, participants who do this may be operating well
within the market rules. This type of behavior may produce inefficient prices.
Transaction costs address the cost to participants of operating in the market.
The introduction of iterative bidding may increase the costs born by
participants to buy and sell in the PX day-ahead market. Each of these cost
categories will be discussed in detail.
- 17 -
1. FEASIBILITY CONSTRAINTS
As mentioned above, feasibility constraints address how easily the iterative
procedures could be incorporated into the PX's market and scheduling process.
This discussion presents an overview of the problem and potential solutions. One
of the outstanding issues is the number of iterations needed to ensure
convergence of the market.
Feasibility constraints focus on whether there is enough time to incorporate
iterations into the market. For iterations to be effective, a sufficient number
need to be incorporated into the auction process. However, there was serious
question as to the necessary number and how long each iteration should last. For
iterations to be effective, it is necessary to adjust the auction and scheduling
timetable to get the number and frequency of iterations correct. It appears to
be a challenge to accommodate the correct duration and number into the market
process. Figure 5 illustrates three steps and the time associated with each.
FIGURE 5: THE CURRENT MARKET SCHEDULE
[FLOW CHART]
It is highly likely the market schedule has to be adjusted for iterations in
order to introduce them effectively into the market. Bids are submitted between
5:00 and 7:00 a.m. to the PX. Some participants submit bids the night before.
Most participants refine their bids throughout this two-hour period and
re-submit tenders up until the deadline. The distribution of submission times in
April and May can be seen in Figure 5. Each diamond represents a bid submission.
- 18 -
Most bids are submitted between 6:15 a.m. and 7:00 a.m. It is not clear whether
participants need the full two hours to submit bids. Closing the period for bid
submissions earlier may be one way to make more time for iterations.
FIGURE 6: TIMING OF BIDS
APRIL & MAY, 1998
[SCATTERGRAPH]
After bidding closes at 7:00 a.m., the PX has an hour to calculate the
market-clearing prices for each hour. This is the time frame that is currently
set aside for iterations if they were to be introduced. On average, the PX is
able to calculate and post market-clearing prices by 7:15 a.m. Currently, the
extra time available between 7:15 and 8:00 is used to increase the time for
bidders to break their awards into schedules and develop adjustment and
ancillary service bids. The hour period from 7:00 to 8:00 a.m. may or may not
provide sufficient time for the market to iterate. This will depend on the
number of iterations needed and the time required for effective bid alteration
to be made.
It may also be possible to reduce the time bidders have to schedule. In this 1.5
hour time period, bidders must create initial preferred schedules, ancillary
services bids and adjustment bids. Participants, particularly those with large
portfolios, have difficulty meeting the 9:30 submission deadline. Figure 7
illustrates the distribution of times at which bidders submitted schedules in
April and May 1998. Notice that some participants take nearly the full period
allotted for submitting schedules. It should be highlighted that this period
includes the leftover
- 19 -
time from the PX's calculations. It appears that this is the least feasible
place to be able to increase the time for iterations.
FIGURE 7: SUBMISSION OF SCHEDULES TO PX
APRIL & MAY, 1998
[SCATTERGRAPH]
There are four possible options for altering the market schedules for iterations
to take place. They are: make no change, start the market earlier, extend the
market long and a combination of starting earlier and extending the market
longer. Each option will be discussed briefly.
The first option is to maintain the current schedule. Under this alternative,
iterations would start at 7:00 a.m. All iterations would need to be completed
within one hour (7:00 - 8:00 a.m.). Under the current schedule, the PX would
need to publish its first price by 7:15 a.m., leaving approximately 45 minutes
for subsequent iterations. Depending on the iteration length, this would leave
only a relatively small number (less than 10) iterations for convergence to take
place. This approach would have the risk that the market does not converge
before the hour has ended.
The second option would be to start the market earlier. Instead of the market
being open from 5:00 a.m. to 7:00 a.m., the PX would stop accepting bids at,
perhaps, 6:00 a.m. Iterations would then start immediately after the PX has
calculated an initial price and would last until 8:00 a.m. when the schedule
continued as previously. This would allow approximately 2 hours for iterations.
Starting the market earlier would require generators to begin the bidding in the
early
- 20 -
morning hours. Generators would also have to commit to quantities farther in
advance of delivery, increasing the uncertainly of their operation.
The third option is to reduce the schedule time. Again iterations would start at
7:00 a.m. but would continue into the time period previously reserved for
scheduling until, perhaps, as late as 9:00 a.m. The length of the extension is
limited by the ISO's scheduling process that requires initial preferred
schedules to be submitted by 10:00 a.m. This option would provide as much as two
hours for iterations and convergence. However, this timetable leaves little time
for bidders to translate their awarded quantities into generation schedules and
ancillary services bids. In the present single-shot market, submitting initial
preferred schedules, incs/decs and ancillary service bids takes the most time.
FIGURE 8: POSSIBLE SCHEDULE OPTIONS
FOR THE ITERATIVE MARKET
[CHART]
The final option would be to start the market earlier and extend the market into
the scheduling period. This combines the second and third options, by starting
iterations by, perhaps, 6:00 a.m. and continuing them until, perhaps, 9:00 a.m.
This would provide the most time available for iterations to converge to an
efficient market price. However, this option also imposes the greatest costs on
suppliers to the market since they would have to start early and have to rush to
get schedules in to the PX.
- 21 -
These options are summarized in Figure 8. There are three periods, submitting,
iterating and scheduling. The desirability of any particular option depends on
the number of iterations needed to converge on an efficient price. If a small
number of iterations is needed or if the iterations occur in quick succession,
it should be possible to introduce iterations into the market. However, it is
not clear how many iterations are needed and how long each iteration should
last. Rough estimates indicate that if the maximum amount of time is give to
iterations - about two hours - no more than 24 iterations at five-minute
intervals could be processed. It is unclear whether the market could converge in
this number of iterations.
Research findings suggest that the actual number of iterations needed may be
higher, particularly if there are technical constraints or demand/supply
volatility. A London Economics study in March 1997 used market simulations to
estimate the number of iterations under different scenarios. In "best case"
scenarios, London Economics found that it took about 10 iterations for the PX
auction to converge. In these scenarios, the London Economics model assumed
smooth demand profiles, no technical constraints, and no gaming behavior. Less
ideal scenarios were run to estimate how many iterations would take to achieve
convergence. Technical constraints increased the number of required iterations
to about 18. The model assumed generators operate for the minimum time and only
start once per day. "Unusual" bidding behavior increased iterations to more than
30. In this scenario, the model used two scenarios: hydro generation was
modified to include two additional small peaks and troughs that introduced more
variability in the residual demand met by non-hydro bidders. The number of
iterations was 32. A burst of hydro generation was introduced during peak hours.
The number of iterations was 42. This is a significant number of iterations that
may not fit into the timeframe allotted, even with relatively quick iterations.
The London Economics study noted there are ways to speed up convergence.
Activity rules, such as minimum step size from previous price, can quicken
convergence, but at a cost. There is a trade-off between efficiency and
convergence. When the model was run using a minimum step size for bid or PX
prices, convergence improved. But in the most complex cases, iterations did not
move below 20. As the size of the minimum step increased, so did the
market-clearing price, implying that there is a trade-off between efficiency and
convergence. However, it is unclear how much of an improvement these rules can
provide.
To summarize the feasibility issues, the time for iterations is very limited. In
order to expand it beyond the current 1 hour, it would be necessary to impose
costs on participants either by starting earlier or shortening the scheduling
time. It is not clear whether iterative bidding can achieve convergence within
the current timeframe allotted. No research has been done to show whether too
few iterations are better than no iterations. Seed prices and other activity
rules can speed convergence, but their impact on efficiency needs to be studied
further.
- 22 -
2. STRATEGIC BEHAVIOR
As described above, the purpose of iterations is to disseminate information
about the market. In order for this to happen, participants must be willing to
reveal what they are willing to pay or be paid. One of the problems with
introducing iterations is that participants may reveal misleading information
about their intentions.
While the revision rules were designed to elicit true valuations from
participants, it still may be possible for participants to follow the rules and
mislead others to their own advantage. While this behavior is often termed
`gaming,' no pejorative association should be given to those who follow the
rules. Rather, it is incumbent on market designers to craft rules in such a way
that their intended outcome is achieved.
Gaming opportunities may or may not lead to material adverse effects on the
market. If gaming is pervasive and the rules are not well constructed, the
impact on the market can be significant, leading to inefficient prices and
misallocation of generation. Two of the difficulties of designing market rules
to discourage gaming is knowing how the rules encourage gaming and how the rules
should be adjusted to avoid it. The games may not be apparent until the market
is established.
Research on the proposed bidding structure did not focus explicitly on gaming
opportunities. The inquiries by Wilson, Plott and London Economics on the gaming
issue were limited. In his experimental work, Plott did not explicitly examine
some of the issues surrounding gaming. His report noted that the behavior of the
auction when prices are moving up has not been explored. Similarly, it stated
that withdrawals were not studied and that sticking around and driving price
down for others becomes cheap talk if exposure to risk is not present. Thus,
strategic behavior was not fully explored.
There is at least one way to game these bidding rules using fictitious bids. The
description below lays out this potential source of problems. It is not clear
how much of an impact the described behavior will have on the market results. As
will be discussed below, it is also difficult to address the elimination of this
gaming opportunity through additional rules. The point of the discussion is to
highlight a potential gaming problem and how it may be addressed. Other gaming
opportunities are probably possible with this set of rules.
Fictitious bids could allow some bidder to provide misinformation about their
true valuation. Portfolio bidding does not require the identity of generating
units to be assigned to particular portfolios. Thus, it is possible to bid more
than one portfolio representing the same units. By introducing multiple bids for
the units, suppliers could depress the price in early iterations only to remove
their low cost bid in subsequent iterations to raise price. By this point in the
iterations, some of the rivals who could have profitably provide power at the
higher prices have exited the market because they saw the lower prices set by
the fictitious bid. This leads to higher prices to consumers.
- 23 -
For example, a bidder initially bids two separate portfolios, one at a low cost
and another at a high cost. In early iterations, the bidder adjusts its low bid
portfolio while freezing its high bid portfolio at its initial level. As the
price falls, the low bid portfolio is awarded significant quantity. The
reduction in price forces other bidders to withdraw from the market. In later
iterations, the bidder withdraws its low bid portfolio and the price increases.
The high bid portfolio is awarded output since other bidders have been forced
out of the market. This behavior does not violate the market rules or the
revision rules described above. This Figure 9 illustrates this behavior.
FIGURE 9: EXAMPLE OF GAMING
WITH FICTITIOUS BID
[LINE GRAPH]
It may be possible to eliminate this type of behavior by changing some of the
rules. However, doing so would significantly alter the nature of the market. One
solution would be to limit multiple portfolio bids for the same units. However,
this requires the units are pre-assigned to portfolio bids and eliminates some
sources of power from the market. Another solution would be to limit
withdrawals, but this would not allow uneconomic generation to get out of
market. A third solution would be to have uncertain stopping rule or time, so
bidders gaming in this way may be caught prior to withdrawal. Over time
expectations will form about stopping rules and when they would apply. None of
these options eliminates the ability of bidder to profit from fictitious bids.
- 24 -
To summarize, gaming opportunities are likely to exist in whatever rules are
adopted. Unless the rules are crafted well, these gaming opportunities will lead
to greater price volatility. Research in the area has not focused in depth on
potential gaming opportunities or how they may be addressed. More research could
be done to identify the extent of potential problems and how they may be
exploited to the detriment of the market.
3. TRANSACTIONS COSTS
The introduction of iterations could significantly increase the costs of
participating in the markets will have to bear to bid in the market. These are
transaction costs for bidding in the PX market. From the participant's
perspective, the benefits of introducing iterations into the market have to be
at least as large as the increase in cost of dealing with iterations. It is
likely these costs will increase significantly with the introduction of
iterations.
The bidding process will become much more complex under iterations. In order to
take advantage of iterations, participants must calculate the profitability of
the awarded quantity. This involves estimating revenue and costs associated with
the quantity awarded in each iteration. To do this, bidders must first separate
the portfolio quantity into unit loading levels. Then, bidders must calculate
costs of each unit loading level allocated. From these costs and revenues,
profits (or losses) can be calculated. Bidders must also revise their bids to
reduce losses or increase profits. They must also check to see that bid
adjustments conform to iteration rules. Finally, they must resubmit their bid to
PX. These steps are numerous and complicated. Further, each of these
calculations will have to be done for each iteration. It is likely to be a
challenge for bidders to accomplish this in a short time period. While the
process could be automated, it will still take an effort to automate.
It may be possible for bidders to ignore the opportunity to revise their bids
under the proposed system. This means they would simply maintain the original
bid curve introduced. This may be adequate. However, given the possibility of
strategic behavior, bidders who do not monitor and react to the market during
the iterations run the risk of becoming a victim of the market.
To accomplish all of this analysis in the period of the iteration, it is likely
that participants will be required to increase staff to perform analysis on bids
and changing bids. Iterations will require an increase in communication with the
PX. Undertaking these calculations in a short time period will require more
computation power.
Further, these costs do not fall evenly on all bidders. In particular, these
calculations will be more easily done on smaller portfolios and single unit
bids, so bidders may gravitate to smaller portfolios. Bidders with larger
portfolios are likely to have more staff and computing power. To the extent
there are scale economies in bid revision, larger participants will be able to
take advantage of these lower costs.
Although no study has been done to quantify the costs of iterations on market
participants, they are likely to be significant. It may well be the case that
the increase in market efficiency
- 25 -
arising from iterations will not be large enough to offset the transaction costs
that must be incurred by participants.
C. INTERVIEWS WITH PARTICIPANTS
As a part of the June research, a number of informal interviews were undertaken
with various PX participants. From the period of May 22, 1998 through June 15,
1998, five telephone interviews were completed. Representatives of the following
companies were interviewed: Pacific Gas & Electric, Southern California Edison,
San Diego Gas and Electric, NorAm Energy and Electric Clearinghouse. The purpose
of these interviews was to gain insight into the issues surrounding the
introduction of iterative bidding into the PX day-ahead market. A number of
topics were discussed in these interviews, including many described above.
Specifically, each interviewee was asked about four topics. They are: the
benefits and use of iterations, the time allotted for iterations, the time
allotted for scheduling, and possible gaming of iterations. The results of these
interviews will be summarized below by these four topics.
The participants were unanimous in their belief that iterative bidding could
bring benefits to the market; however, they were divided as to whether they
would use the iterations if introduced. All respondents said that smaller
portfolio generators would benefit since they could avoid infeasibilities. One
respondent stated that the information provided by the iterative process would
help it make market arbitrage decisions. However, only two of the five said that
they would make use of iterations if they were introduced. The reasons were
varied for declining the use of bids. One stated it did not need iterations
because of divestitures of generation. A second said the market was too new to
make significant changes such as adding iterations. It also stated it would need
to add additional resources to accommodate iterative bidding. The third stated
the size of its portfolio was significant and it could not incur the transaction
costs needed to alter bids during iterations.
Although the interviewees gave various opinions with regard to the time needed
for iterations, they were unanimous with regard to one aspect of the timing.
None was in favor of moving the close of bidding any earlier than 7:00 a.m.
Their opinions about how much time should be allotted for iterations varied. One
believed the one hour originally specified was the appropriate time. Another
stated that only one-half hour should be allotted since more time was needed for
scheduling. Two others did not specify a particular time period, but noted the
allotted one hour would not be sufficient. They noted that a significant amount
of time would be taken in uploading and downloading information from the PX.
Further, they would need a sufficient amount of time to complete analysis and
compare the PX market price with other opportunities. The fifth stated that
perhaps as much as two hours were needed, but scheduling time should not be
sacrificed for iteration time.
The respondents also varied considerably in their responses to the need for
scheduling time. While all were unanimous in rejecting an earlier start time,
some thought scheduling time could be shortened to allow more time for
iterations. Others thought more time was needed for scheduling. The two
interviewees who represented entities that were bidding a large number of
- 26 -
plants stated that more time was needed for converting PX awarded quantities
into generation schedules. One stated that it takes about 45 minutes to
apportion awarded quantities to schedules and then another hour to hour and a
half to calculate adjustment bids and ancillary services bids. The other stated
that it on occasion had difficulty submitting its initial preferred schedules
even with the extra time from the 7:00 a.m. to 8:00 a.m. hour. The three other
interviewees thought the time was either adequate or could be shortened. One
stated that no change was needed to the schedule. One observed that the time
needed to schedule depended on whether it bid single generators or portfolios,
with portfolios taking longer to schedule. The third expressed a desire to
shorten the time for scheduling and also to have the ISO push back its
requirement for schedules to 12:00 noon from 10:00 a.m. Given that these
interviews were done approximately two and one half months into the market, it
is to be expected that the assessments of time needed varied widely.
The interviewee's views about gaming were more consistent. Almost all of them
had not given much thought to the specific opportunities for gaming; however,
they were concerned that it may become a problem. One observed that his company
had not thought of any particular ways to game the system, but, given the
incentives, there are probably many ways that no one has yet discovered. This
interviewee suggested that more research be done involving market simulations to
estimate the extent of the problem. Two others stated that they were concerned
that opportunities may arise, but had not thought of any specific opportunities.
One interviewee stated that its firm had little or no concern about gaming.
Finally, the fifth interviewee stated that it did not consider behavior within
the rules to be `gaming;' however, there was concern about being affected by
behavior outside the rules. A strong set of rules would significantly reduce the
possibility of strategic behavior.
D. CONCLUSIONS FROM JUNE 1998 STUDY
The conclusion drawn from the June 1998 research was that the market was not yet
ready for the introduction of iterations. The policy decision was made that they
should be reconsidered in the fall of 1998. This was to allow more time for
bidders to become familiar with the market and its operations. In particular,
bidders were expected to have become better at transforming their awards into
generator schedules. Similarly, they were expected to have learned about how
their bids affected their profitability from auctions repeated on a daily basis.
These are countervailing effects. The transaction costs would be lower, but so
would the value of the information produced by the iterations.
The investigation concluded that it was not clear whether bidders would have
enough time to adjust their bids in the each of the iterations. Participants
believed they needed sufficient time to undertake the proper analysis to respond
to the information they faced. Over time, the required time would be reduced, as
they became more familiar with the process and data flows between themselves and
the PX. Nonetheless, transactions costs for participants were going to be
significant since much data processing would be needed to respond appropriately
to information in iterations.
- 27 -
From this research, it was clear there are potential gaming opportunities that
need to be addressed before implementation could take place. In particular, it
was possible for fictitious bids to signal false valuations. Activity rules need
to be developed to adequately address problems such as those associated with
fictitious bids. Further, more research may be needed to address other potential
gaming problems.
Overall, in June, the market institutions were still too new to have settled
down significantly for iterations to be introduced. At that time, the hour-ahead
market had not yet been introduced. It could have contributed significantly to
reducing risk from infeasibilities. Further, the amount of learning from daily
repeated bidding was limited. Significantly more learning would take place in
the intervening period between June and November. Thus, by putting off
consideration to a later date, the market would be better able to assess the
value of iterations in a more mature market.
VI. SUMMARY OF RESEARCH DONE IN NOVEMBER 1998
Between June and November, the PX developed software to implement iterations.
This involved incorporating the revision rules that had been specified. These
are essentially the rules described above adapted to apply to demand as well as
supply bids. Beta Tests of the software were undertaken to answer a number of
questions. First, did the software work as expected? What modifications would
have to be made before iterations could be introduced? Second, how many
iterations would fit into the time allotted? Could changing the timing of the
market increase the number of iterations? Third, how would its participants
react to the introduction of iterations and necessary modifications to the
market? Fourth, given the answers to all these questions, was the time right to
introduce iterations? What problems might still have to be worked out before
they could be introduced?
Because most of the costs and benefits of iterations would accrue to its
participants, they would be in a much better position to assess whether there
would be a net benefit from introducing iterations. Would their benefits be
large enough to outweigh their costs? The best way to assess this would be to
survey participants after they had been involved in the Beta Test. A survey was
constructed and participants responded. The results of the Beta Tests and the
survey were presented to the PX Board at its November meeting. A brief report
was written to convey the research findings. The PX Board decided to postpone
the introduction of iterations indefinitely.
The author of this report was not involved with the research undertaken in
November 1998. Therefore, this report is not based on any immediate knowledge of
the research effort. However, a copy of the report to the PX Board was provided
to the author of this report. Large portions of that report will be excerpted
below as an explanation of the November efforts. The portions not directly
spelled out are contained in Appendices A and B. These appendices contain,
respectively, the specific answers to the questionnaire and the price and
quantity results for each iteration for each day tested.
- 28 -
The large excerpt will provide a basis to discuss the report and the issues that
the Beta Test was not specifically designed to address. Particularly, are there
any benefits or costs participants identify that have gone unnoticed? Also, is
there any evidence of gaming in the Beta Tests?
The excerpt of the report will be followed by discussion and comment on the
results. After the excerpt from the November report, comments and observations
will be made concerning some of the specific answers given to the questionnaire
and iteration results that were included.
A. EXCERPTS FROM `BETA TEST OF ITERATIVE BIDDING IN THE PX DAY-AHEAD MARKET'
EXECUTIVE SUMMARY
The PX operated a Beta Test of iterative bidding with PX participant involvement
from November 2, 1998 through November 6, 1998. The purpose of the Beta Test was
to provide PX Participants the opportunity to preview iterative bidding and
submit comments based on this preview. The Beta Test also gave the PX an
opportunity to evaluate in a market simulation environment the task of operating
an iterative Day-Ahead Market.
Many of the PX Participants that took part in the Beta Test shared some common
concerns about iterative bidding such as potential for gaming, reduced time for
[Initial Prefered Schedules] IPS's, Adjustment bids, and Ancillary Services
bids, and additional costs for staffing and system modifications. Many felt the
costs to implement iterations outweighed the potential benefits. Over half of
the participants that submitted comments recommended not implementing iterative
bidding at this time. None recommended implementation.
The PX observed that iterative bidding would initially require 30 minutes per
iteration, which would allow for at most 3 iteration if the final iteration
closed at 8:00AM or 1.5 iterations if PX maintained current schedule of
publishing final auction results at 7:15 AM. Over time, this could be reduced to
15 to 20 minutes per iteration, which would allow up to 4 or 5 iteration if the
final iteration closed at 8:00AM or 2 iterations if PX maintained current
schedule. Prior studies indicated that at least 10 iterations, sometimes many
more, would be required to reach convergence. The benefit of 2 to 5 iterations
would likely be small relative to the costs incurred by all stakeholders. The
Majority of PX participants are against modifying the current PX timeline to
allow for additional iterations. Before iterative bidding could be implemented
the PX and PX Participants would incur additional costs associated with software
fixes and modifications, additional testing, and training.
PX PARTICIPANT INVOLVEMENT
On October 26, 1998, the PX published a PX participant Notice (Notice 98-58)
announcing a Participant Beta Test of Iterative Day-Ahead Bidding to be held
from November 2, 1998 through November 6, 1998. The initial conference call took
place on October 28, 1998 to
- 29 -
explain iterative bidding, answer any questions, and discuss the Beta Test. A
total of 14 PX Participants opted to take part in the Beta Test. On average, the
14 PX Participants collectively represent 90% to 95% of the PX Day-Ahead market
volume. A conference call was held after each session to discuss in an open
forum any observations from the Beta Test.
PX PARTICIPANT COMMENTS
At the conclusion of the Beta Test, the 14 participants were asked to respond to
the following seven questions:
1. Do you think you would modify your bids during iterations? 11 responded.
5 might, but not often.
4 probably would not.
1 yes, but stated against implementing iterations.
1 no, and would consider no longer participating in PX market if
iterations became mandatory.
2. How much time should be allowed per iteration? 12 responded.
8 were in the range of 20-30 minutes.
2 felt 15 minutes would be too short, but wanted to re-evaluate with
software bugs fixed.
1 could not answer until software fixes are in place and re-tested.
1 stated that 5 minutes would be sufficient.
3. Are you willing to move the deadline for bids from 7:00 AM to 6:00 AM if
necessary to provide for iterations? 12 responded.
9 said no.
1 said yes.
1 said yes if other participants that would normally participate in
iterations agreed.
1 suggested 6:40 AM.
4. Are you willing to reduce the amount of time available for submitting
IPS's, adjustment bids, and A/S bids, if necessary to provide for
iterations? 12 responded.
8 said no.
2 said yes.
1 said yes, but would be pressed for time.
1 very reluctantly would agree to a reduction in time.
5. Are you concerned about gaming during iterations? 11 responded.
8 said yes.
1 uncertain.
2 said no.
6. What benefits do you see for your company if iterations are implemented?
12 responded.
7 identified potential benefits such as price discovery, potential for
higher profits, opportunity to reduce costs, or more feasible schedules.
5 mentioned minimal benefit, if any.
7. What negative impacts do you see for your company if iterations are
implemented? 11 responded.
- 30 -
11 identified potential negative impacts such as reduced time for IPS's
and adjustment and A/S bids, additional personnel requirements, costs for
developing analytical and data-management software, costs associated with
human errors and computer system or software problems resulting from more
complex iterative process, conflicting timeline with bilateral markets,
higher prices due to gaming, withdrawal of bids or programmed response to
iterative results, and making self-provision more difficult to implement.
Some participant provided unsolicited comments. Six participants recommended not
implementing iterations at this time. Reasons included negative impacts outweigh
benefits and PX should pursue more beneficial changes such as additional forward
markets or improvements to the hour-ahead markets. Various software
modifications or fixes were also identified. If iterations were implemented,
these modifications and fixes would have to be addressed. None of the 14
participants recommended implementing iterations. The following tables include
the participant responses. Some responses were edited at the request of the
participant to exclude bidding strategy. Other comments were summarized or
paraphrased.
[Table of responses reproduced in Appendix A.]
BETA TEST RESULTS AND CONCLUSIONS
The Beta Test was intended as a preview for participants of the iterative
bidding process. This preview was intended to provide additional basis for
participants' comments regarding the implementation of iterative bidding. These
comments were presented in the previous section. The PX also made some
observations concerning operation an iterative Day-Ahead market which are
summarized as follows:
- During iterative bidding, the PX would not have enough time to
manually enter bids for participants having connectivity issues.
This could cause some participants' bids to be come frozen
prematurely and unintentionally.
- Participants will initially require at least 30 minutes per
iteration to successfully submit modified bids.
- Gradually, the time per iteration could be reduced to 15 to 20
minutes.
- PX will likely have to begin iterative bidding earlier or reduce the
amount of time for IPS's, Adjustment Bids, and Ancillary Service
bids to provide enough time for iterations.
- Iterative bidding should eliminate the need for the Anomalous Bid
Protocol.
- If iterative bidding were implemented, additional costs would be
incurred to modify or fix core trading software based on comments
during Beta Test.
- Substantial training and market simulation testing would be
required.
The Beta Test was not designed to prove or disprove theoretical results.
However, the following observations were made regarding the results of the Beta
Test:
- 31 -
- Session 11-03-98 exhibited slight movements in price and volume
until a bid was withdrawn that caused prices to increase an average
of 80 percent while volume decreased by 20 percent.
- Session 11-04-98 exhibited generally increasing prices with
increasing volume.
- Session 11-05-98 exhibited prices and volume that fluctuated
modestly up and down.
- Sessions 11-06-98 and 11-07-98 exhibited a pattern of generally
increasing prices while volume initially increased, but later
decreased.
The following tables and graphs show the movement of princes and volume during
each of the iterative bidding sessions during the Beta Test.
[Tables reproduced as Appendix B.]
B. COMMENTS AND OBSERVATIONS FROM NOVEMBER RESEARCH
The PX decided to set up the software for iterations and let participants get a
feel for how it would work. Because most of the costs and benefits of
implementing iterations would accrue to market participants, it made sense to
survey them to determine whether iterations would bring a net benefit to the
market. The Beta Testing provided participants with a hands-on feel for how
iterations would work and helped them assess the costs and the benefits. It also
helped the PX identify the costs of implementing the bidding system.
It should be highlighted that the report was written as a document to inform the
PX board about the Beta Testing for discussion about the topic of iterative
bidding. As such, it was not conceived as a complete record of the analysis, but
as background for discussion. A presentation was also made to the board on the
Beta Tests. Consequently, the researchers could have answered any questions that
may have arisen concerning the research effort.
1. COMMENTS ON RESULTS
a. THE REPORT'S DISCUSSION
The PX report gives a brief interpretation of Beta Test results. Nonetheless,
the report does make some important observations that are worthy of highlight.
First, the electronic communications links between the PX and participants are
crucial to the successful operation of iterations. Further, implementation
requires invocation of operating procedures and rules when communications fails
during the process.
Second, that it is questionable whether enough iterations could take place to
achieve convergence. Only two iterations are possible if each takes 30 minutes
and there is only an hour for them. With a quickening of the pace to 15 minutes
only yields 4 to 5 iterations. While this
- 32 -
may be enough, it could also prove problematic, as will be discussed in greater
detail below. Altering the timeline may also prove problematic.
Third, further costs would have to be incurred by the PX to implement system.
Specifically, bugs need to be eliminated from software and it needs to be made
friendlier to users. Also, implementation would require conducting more testing
and training of market participants.
Although the report states that the test was not designed to support theoretical
hypotheses, the report's discussion of the results is quite cursory and will be
expanded below.
b. THE REPORT'S OMISSIONS
As discussed above, there is no reason for the report to have contained a
complete description of the research effort. One omission from the report was a
complete description of the participants who took part in the testing. The
report states that there were 14 participants representing, 90 to 95 percent of
the Day-Ahead market volume. However, it does not report how many of the
participants were supply-side, demand-side, or both supply and demand side
bidders. While it is clear that both demand and supply side bidders would have
to be involved to obtain 90 to 95 percent of the market volume. The number and
size of bidders on each side of the market is important in interpreting the
results. The ability to coordinate behavior will be affected by the number of
bidders. Similarly, the incentives differ significantly if bidders are on both
sides of the market.
Another omission in the report was that nothing was said about the closing rule
used to decide which iteration was the last. Bidding strategies would be
different if the number of iterations were announced before the first iteration.
Similarly, bidding strategies would change if the number of iterations were
announced during the auction. In subsequent conversations with researchers, it
was learned that the number and duration of iterations were predetermined before
each auction began. This means the bidders knew which iteration would be the
last and could adjust their bidding strategies to take advantage of knowing when
the last iteration would take place. Strategic behavior is much more likely to
take place under a regime of certainty than if bidders did not know which
iteration would be the last.
Although it is understandable why such information was not reported, it would be
instructive to have examined how each bidder changed its bids between
iterations. This would perhaps reveal too much about individuals' bidding
strategies. A less sensitive number would have been to report how much total
quantity was bid in at each price level. This would have provided information
about the overall prevalence of fictitious bidding. Nonetheless, only
market-level data are reported and not individual bidder data. Thus, it is
difficult to assess what strategies individual bidders are following.
Nonetheless, the market-level data can also be revealing.
- 33 -
c. INTERPRETING THE REPORT'S RESULTS
The report contains market level prices and quantities for each round of each
auction. These data are attached as Appendix A. In conjunction with the bidding
rules for both demand and supply in the market, examination of changes between
iterations provides some insights as to how bidders were acting in the Beta
Test.
In combination with the revision rules described above, the changes in prices
and quantities between iterations indicate what is going on in the market.
Supply side bidders only have two options in making revisions. First, they may
adjust their bids to lower their selling price for each quantity they offer as
described above. Second, they may remove a portfolio entirely from the market.
Ceteris Paribus, supply bid adjustments lead to lower prices and higher
quantities. Similarly, supply bid withdrawals lead to higher prices and lower
quantities. This is illustrated in figure 10. In the initial iteration, P[1] and
Q[1] prevail. In the second iteration, if as a whole, suppliers adjust their
bids, then a lower price, P[2A], and higher quantity, Q[2A], result. However,
if, as a whole, suppliers withdraw portfolios, then a higher price, P[2B], and
lower quantity, Q[2B], ensue.
FIGURE 10: SUPPLY SIDE EFFECTS
[LINE CHART]
- 34 -
Like suppliers, demanders may change their bids in one of two ways: they may
revise them or they may withdraw them. Again, ceteris paribus, demand bid
adjustments lead to higher prices and higher quantities. In like fashion, demand
bid withdrawals lead to lower prices and lower quantities. This can be seen in
Figure 11. In the initial iteration, P[1] and Q[1] are the outcome. In the
second iteration, if as a whole, demanders adjust their bids, then a higher
price, P[2A], and higher quantity, Q[2A], follow. However, if, as a whole,
demanders withdraw load, then a lower price, P[2B], and lower quantity, Q[2B],
arise.
FIGURE 11: DEMAND SIDE EFFECTS
[LINE CHART]
While these effects may take place simultaneously, examining how the prices and
quantities change between iterations can shed some light on how both demand and
supply bidders are reacting to the market. For example, between the first and
second iterations in the 11-04-98 auction, both prices and quantities have
increased. This indicates that, whatever else is going on in the market, the
predominant effect is that demand bidders are adjusting their bids to increase
the amount they are willing to pay for power. This is the only type of change to
the market that can explain both an increasing price and an increasing quantity.
Given this method of interpretation, it is possible to describe the iterations
in greater detail. Each of the daily markets will be described using these
terms.
- 35 -
The results from the Beta Test represent the first chance PX participants had to
experience bidding with iterations. They learned over these five days. The
11-03-98 market was the first day of the test. There were six iterations in this
market. In the second iteration, prices were generally falling and quantities
increasing with the exception of hours 22 and 23. This indicates that supply
adjustments were the predominant activity in this iteration. In the late hours
of the day, prices went up and volumes went down. This indicates that the net
effect was the removal of supply portfolios from the market. Suppliers may have
bid separate portfolios for separate hours and decided not to operate at those
prices, thus, removing the portfolios for those hours. The third iteration shows
generally increasing prices and quantities, indicating demand decided to buy
more at those prices. There is little change with the fourth iteration and
depending on the stopping rule, the auction may have ended. However, in the
fifth iteration, a chunk of 20% of the quantity in the market was removed from
the supply side, leading to much higher prices and lower quantities. After such
a large removal, it took another iteration of supply adjustment before the
market ended.
The results for 11-04-98 show quite a different pattern of change. It too lasted
six iterations. However, between each iteration, prices and quantities
increased. This indicates that the major effect was demand side adjustments.
Specifically, demanders were increasing their willingness to pay at the relevant
levels of demands. However, the magnitude of the price changes diminished as the
auction went on. There was an average 4 percent increase in prices between the
first and second iterations, but only a .25 percent increase between the fourth
and fifth iteration. There must have been little activity in the later rounds.
The results for 11-05-98 show a prominent supply side of the market. Generally,
prices were falling and quantity increasing. This indicates suppliers were
lowering their willingness to bids throughout the auction. The auction lasted
nine rounds, but activity was diminished with less than a 1 percent average
price change by the fifth iteration. Rounds 3 through 6 saw changes only in
hours 1 through 5, with supply adjustments lowering prices and increasing
volumes. Later rounds saw changes in hours 17 through 20. Supply bids were
withdrawn in iteration 6. In the remaining iterations, supply adjusted and
prices fell. Again, it should be highlighted that a couple of additional
iterations may be needed when there is a supply withdrawal.
The 11-06-98 PX auction lasted six iterations and showed mixed results across
iterations and hours. In the second iteration, prices and quantities increased
in most hours. This indicates demand adjustments. However, consistent with
supply competition, morning hour prices fell while quantity increased. In the
third iteration, there was supply competition as prices fell and volumes
increased. The non-morning hours show increases in demand purchases. The fourth
iteration showed a supply withdrawal as prices increased and quantities
decreased in almost all hours. The fifth iteration showed a mix of effects with
some supply withdrawal and demand adjustments. However, the last iteration
revealed another supply withdrawal that increased prices by an average of over 5
percent. Given the limited information in the report, it is impossible to tell
whether unprofitable portfolios exiting or removal of fictitious bids caused
this. However, the latter is wholly consistent with the classic example of the
supply gaming described above.
- 36 -
It should be highlighted that one way for demand to avoid paying the resulting
high prices would be to play the same game. Specifically, for the demand side of
the market to introduce fictitious bids of their own which they would remove in
the last iteration as well. This would protect them from the resulting higher
prices. This would inflate the quantity stated in the early rounds of the
market. Alternatively, this strategy would be harder to follow if there were
uncertainty about when the auction would end. Such uncertainty would not
eliminate the strategy, but only make it more difficult to follow.
The 11-07-98 auction lasted four iterations and show the same pattern as the day
before. Specifically, demand increased its willingness to buy between the first
and second rounds with an average 8.66 percent increase in quantity and a 13.6
percent increase in prices. However, in the last iteration, supply withdrew
portfolios resulting in an average 9.4 percent increase in price and a reduction
in quantity. A stronger case can be made for supply manipulation of prices in
this case. Withdrawal of supply for economic reasons means the portfolio would
be losing money. However, with each iteration, all prices in all hours were
increasing between iterations. The portfolio that was withdrawn in the fourth
iteration had to have been losing money from the start of the market. If this
portfolio was setting the prices in the previous rounds, its initial bid had to
have been below its costs. It is possible that the bidder of this portfolio may
have been waiting to see if prices would increase over iterations to make it
profitable. However, a fictitious bid is also possible. More information about
the specific bids and bidders would be needed to strengthen the argument either
way. However, the result is consistent with supply manipulation of prices.
PRICE AND QUANTITY CHANGES BETWEEN ITERATIONS AND THEIR IMPLIED BEHAVIOR
Iterations 11-03-98 11-04-98 11-05-98 11-06-98 11-07-98
- ---------------------------------------------------------------------------------------------------------
1 - 2 P(down)Q(up) SA P(up)Q(up) DA P(down)Q(up) SA P(up)Q(up) DA P(up)Q(up) DA
2 - 3 P(up)Q(up) DA P(up)Q(up) DA P(down)Q(up) SA P(up)Q(up) DA P(up)Q(up) DA
3 - 4 P(down)Q(up) SA P(up)Q(up) DA P(down)Q(up) SA P(up)Q(down) SW P(up)Q(down) SW
4 - 5 P(up)Q(down) SW P(up)Q(up) DA P(down)Q(up) SA P(up)Q(down) SW
5 - 6 P(down)Q(up) SA P(up)Q(up) DA P(up)Q(down) SW P(up)Q(down) SW
SA = Supply Adjustment DA = Demand Adjustment
SW = Supply Withdrawal DW = Demand Withdrawal
The table summarizes the price and quantity movements for the changes between
iterations for the first six iterations of each round. There were 11 iterations
on 11-05-98, but only first five are shown here because there was little
movement in prices and quantities in the later rounds. As can be seen in the
table, for all days, a supply or demand adjustment started in the early rounds
of the market. As expected, supply withdrawals came near the end of the
iterations. It should be highlighted that demand withdrawals were never a
predominant effect for any iteration on any day. This may indicate demand's
unwillingness to take advantage of the revision rules in the iterative process.
- 37 -
2. COMMENTS ON INTERVIEWS
The participant survey solicited views concerning the introduction of
iterations. The answers are attached as Appendix B. The report did a good job of
summarizing these comments. The survey dealt with two different issues that will
be discussed here. First, what was the participant's perceived benefits and
costs of iterations? Second, how did the participants view gaming in the market?
a. BENEFITS AND COSTS
The bulk of the questions in the survey had to do with the benefits and costs of
implementation. In large part, the answers were conditioned on the conclusion
the participants had come to. For instance, more than one participant stated
that it would not support increasing the time for iterations given it did not
see benefits and probably would not use them.
Three questions addressed the time available for iterations and garnered
generally negative responses. Only one participant was willing to increase the
time for iterations. Thus, the benefits from iterations are great enough to
overcome the transition costs of changing the established process.
Two participants cited price discovery as a benefit of iterations. These were
probably demand side participants who were able to adjust consumption in the
early rounds of iterations. Only two participants cited better adoption of
feasible schedules as a benefit. Most found little benefits from iterations.
Many of the participants cited higher costs of dealing with iterations as a
significant negative impact of iterations.
b. CONCERNS ABOUT GAMING
One of the questions asked explicitly about concerns about gaming in the market.
Three participants cited particular games as a significant problem. Their
comments support some of the observations made above about the possibility and
likelihood.
One explicitly complained that the activity rules are not sufficient to prevent
the type of fictitious bidding described above. Specifically, "Withdrawal Rule
would allow a participant to submit a fake bid to drive price down to force
others to withdraw from the market until the fake bid is withdrawn leaving the
real bid at a higher price...." Further, this participant stated it would not
agree to implementing iterations unless this flaw was fixed.
Another participant stated it believed that there would be gaming, but it would
not diminish the end results relative to a single-round auction. However, they
also stated, "The basic game in iterative bidding involves the submission of
bids that are not physically based for the sole purpose of withdrawing them to
drive prices up." They correctly note that it would be difficult to guard
against this behavior because it is similar to legitimate withdrawals.
- 38 -
It should also be highlighted that there were countervailing views of the
results. One participant cited a higher MCP as a benefit from iterations. As
discussed above, these higher prices may have resulted from gaming, particularly
in the last two auctions. Another participant cited higher prices from gaming as
a significant negative impact of the introduction of iterations.
c. CONCLUSIONS FROM NOVEMBER RESEARCH
Given the responses of participants to the survey, the report comes to the
correct conclusion about implementation of iterations at this time.
Specifically, the current iterative structure and timeline produces more costs
for participants than it does benefits. Nonetheless, there are two new
conclusions that follow from the November research. First, there is a benefit to
demanders from participating in iterations. Second, there is the real potential
for gaming the iterative process. Each of these conclusions will be discussed in
turn.
FIGURE 12: DEMAND PURCHASE DECISION
[LINE CHART]
Iterations provide significant benefits to the demand side of the market as well
as to supply, particularly during the transition period. Because PG&E, SCE and
SDG&E are required to purchase the power from the PX market, they currently have
few options for procurement. Particularly, they can only buy from the Day-Ahead
or On-the-Day markets. CTC recovery
- 39 -
provides an incentive for them to reduce PX prices as much as they can.
Splitting purchases between these markets is one of the only ways they have to
reduce their power costs. The basic problem they are solving is a quantity
allocation problem between the markets. By buying less in one market they will
pay a lower price in that market. However, they must then buy more in the other
market, pushing up the price in that market. To do this effectively, they must
forecast relevant supply elasticities for both markets. The tradeoff can be seen
in Figure 12. Demanders have a total QT they must purchase between both markets.
By moving down the Day-Ahead supply curve, they move up the Hour-Ahead curve.
The optimal allocation is given when the extra they save in one market is
exactly offset by the extra they must pay in the other market.
Iterations in the market allow demanders to obtain better information about the
Day-Ahead price before committing to how they would like to split their load
between these markets. The iterations allow them better control over the
quantity they purchase in the Day-Ahead market. If prices are low, they may
adjust their demand bids to increase their purchases. If prices are high, they
may withdraw some bids to reduce their purchases. This is a significant benefit
from iterations.
It is possible that supply-side bidders manipulated the market in the Beta Test.
The price and quantity patterns in the last two days of the test are consistent
with suppliers removing a `fictitious bid' in the last round of the auction. It
would be impossible to eliminate the possibility of a fictitious bid without
requiring all bidding to be done on a unit basis. The elimination of portfolio
bids would have other consequences on the market and may not solve this problem.
Further, the market results were not consistent with strategic behavior on the
demand side of the market. However, that does not mean that demand could not
behave exactly the same way supply does by bidding fictitious load that it would
pull out of the market on the last iteration. This behavior would tend to reduce
price and quantities. Because the seller's objective is diametrically opposed to
the buyer's, it is not clear how the net result would turn out if both sides of
the market were undertaking strategic behavior simultaneously. More
investigation and data analysis would be needed to determine what the likely
result would be.
VII. CONCLUSIONS
In its evaluation of the gains from implementing iterations in the market, the
PX correctly looks to its participants for assistance in determining whether
iterations should be adopted. Participants bear most of the costs and enjoy most
of the benefits from iterations.
Participants report that the costs imposed by iterations are large in comparison
to the benefits they provide under the present set of revision rules and market
timeline. The costs imposed by iterations include increased staff and computer
resources to process the information. The benefits are minimal except, perhaps,
on the demand side of the market where price discovery allows a better
allocation of purchases between the day-ahead and other markets.
- 40 -
However, more testing of iterations would have provided more information about
the likely impacts of implementing these rules. For instance, there was not
enough study to be able to tell what the impacts would be of strategic behavior.
There may or may not be significant problems with bidders behaving strategically
through the iterative process. While the Beta Test revealed market price and
quantity results that are consistent with strategic behavior on the part of
suppliers, strategic behavior was not apparent on the demand side of the market.
As discussed above, it would be possible for demanders to undertake similar
behavior. Before implementation or revision of rules to address the problem of
fictitious bids, more research should be done to identify the net effect when
both sides of the market are behaving strategically.
It may be difficult to eliminate strategic behavior since it would require the
adoption of unit bidding. Unit bidding would considerably lessen the costs to
suppliers of analyzing their profitability and revising their bids. However, it
may require suppliers to forego the benefits of portfolio bidding.
Given the evidence from both the research efforts in June and November, the
policy decision to forego implementation of iterations at this time is a prudent
one. Changes should be focused on addressing ways to reduce costs to
participants. Any change in the rules or market that reduce the cost of
iterations should make iterations more attractive.
VIII.
- 41 -
SOURCES
London Economics, "PX Auction Testing: A Report for the California PX
Restructuring Trust"
London Economics, Inc., March 3, 1997. (Also included as PX Appendix 3,
Attachment B, PX/ISO filing, March 31, 1997).
London Economics, "Response to Request for Additional Information," Questions 15
and 18, May 20, 1997.
London Economics, Inc., "PX Auction Testing: A Report for the California PX
Restructuring Trust." March 3, 1997. (Also included as PX Appendix 3, Attachment
B, PX/ISO filing, March 31, 1997.)
London Economics, Inc., "The Impact of PX Simplifications on PX Efficiency: A
Report for the California Trust on Power Industry Restructuring," September
1997.
Order Conditionally Authorizing Limited Operation of an Independent System
Operator and Power Exchange, Conditionally Authorizing Transfer of Control the
Facilities on an Interim Basis to an Independent System Operator, Granting
Reconsideration, Addressing Rehearing, Establishing Procedures an Providing
Guidance, FERC Order, October 30, 1997
Plott, Dr. Charles R., "Tests of the Power Exchange Mechanism," March 10, 1997.
(Also included as Attachment 2, August 15, 1997 FERC Filing.)
Power Exchange Bidding and Bid Evaluation Protocol, Revised to Incorporate
Iterative Bidding and Activity Rules, no date.
The FERC's 90 Questions, April 29, 1997
Wilson, Dr. Robert, "Bidding Activity Rules for the Power Exchange: Report to
the California Trust for Power Industry Restructuring" Market Design Inc., March
14, 1997. (Also included as PX Appendix 3, Attachment A, PX/ISO filing, March
31, 1997).
- 42 -
APPENDIX A: SURVEY RESULTS
Participant Do you think you would normally How much time should
modify your bids during iterations? be allowed per iteration?
- ------------------------------------------------------------------------------------------------------
1
2 20 minutes
3 Probably not. 15 minutes appears tight,
but we cannot accurately
assess until technical
difficulties are resolved
4 If iterations become mandatory, Based on the Beta Test, 5
there is a good chance that minutes is sufficient time.
we will no longer participate
in the PX Market because of
the time constraints and
manpower required to do
iterations.
5 Not often, maybe 10% 20-30 minutes is more realistic
of the time. than 15 minutes. PX could open
the market with longer time
period, then move to 15 minutes
after participants gain experience.
6 Currently would not normally Should be re-evaluated
modify bids. Would modify after software bugs are fixed.
bids if economies in market Right now, 15 minutes is too
place made good business short.
sense to make such
modifications.
7 Right now, given the number Cannot answer this question
of resources we need to bid until fixes are made and tested.
and/or schedule, iterative
bidding is not necessarily an
option, since you need to
modify each resource
and modify the appropriate
prices that are active. At this
point, w would incorporate
reviewing the MCP from the
first iteration into our bidding
process. But, as a rule, we
would not automatically modify
our bids.
8 Yes, but only to a limited extent. For the Beta tested process and
However, if we saw evidence of software, absolutely no less than
routine excessive gaming, such 15 minutes. Realistically, 25-30
that only the last iteration was minutes are needed to make any
meaningful, our involvement in but the simplest, essentially
iterations would drop off substantially. pre-designed, changes when
Activity may be limited to the multiple portfolios are involved.
selective withdrawal of bids due to
the primary reason that withdrawal
of bids is simple, quick, and relatively
less subject to human error.
9 Yes. 20 minutes maximum,
preferably 30 minutes.
10
11 Not likely. No less than 20 minutes.
Participants must be allowed
sufficient time to give proper
consideration to bid modifications.
If too little time, participants
will be incented to either not
modify bids, or develop a
programmed bid adjustment
response to each round's results.
Both of these behaviors completely
defeat the purpose of iterations.
12 Not at this time. 30 minutes.
13 Same as Participant 12. Same as Participant 12.
14 Probably, but not very often. Minimum of 20 minutes.
Participant Are you willing to move Are you willing to reduce
the deadline for bids the time available for
from 7:00 am to 6:00 submitting IPS's
am if necessary to Adjustment Bids, and
provide time for iteration? A/S bids, if necessary to
provide for iteration?
- --------------------------------------------------------------------------------------------------------------
1
2 6:40 AM 8:20 AM
3 No. No.
4 No. Rather not.
5 Yes. Yes.
6 Would be acceptable if the Based on other's comments,
participants that would normally would not be acceptable.
participate in iterations are
willing to move the timeline.
7 No. As it is, we are pushing No. This would b an extreme
other entities we work with to hardship, given the additional
provide us appropriate data by work on top of the current workload,
6 am so that we can prepare our bids. to develop IPSs, Adj. Bids, and A/S bids.
8 No. Iterative bidding is not worth Currently, the benefits of iterative
sacrificing time that is currently bidding appear to be far less than
used to make last0minute the benefits of having enough time
improvements to initial bids. to do a good job scheduling and
At most, a few minutes before offering Adj. Bids and A/S bids.
7:00 might be sacrificed (e.g., 6:45). Iteration should not be allowed to
The proposed Say-of market at reduce the time for these activities.
6:00 am would make closing the Iterative bidding is simply not worth
initial Day-Ahead iteration before detracting from the far more complex
7:00 especially impractical. and manually intensive processes of
scheduling and bidding into the
adjustment and A/S markets.
9 Prefer not. Suggest up to four Would very reluctantly agree to
20 minute iterations, first shortening the time to 1.5 hours.
closing at 7 am and the last However, in view of previous comment,
closing at 8 am. Feel that 3 to 4 we don't think it is necessary.
iterations are sufficient for operators
to be able to adjust their schedules
as necessary to run efficient production
and recover costs as required.
A fixed number of iterations provides
greater opportunity for gaming.
10
11 Would oppose, given not Definitely not. DS's, Adj. Bids,
likely to participate in bid modification. nd A/S bids are extremely
important and those with large
number of schedules already have
a hard time finishing on time.
Self-provision of A/S will also consume
additional time.
12 No. Absolutely not. This would
be totally unacceptable.
13 Same as Participant 12. Same as Participant 12.
14 This is not acceptable. May be a possibility,
Wold conflict with bilateral but could be pressed
markets. for time.
- 43 -
Participant Do you think you would How much time should Are you willing to Are you willing to
normally modify your bids be allowed per move the deadline for reduce the time
during iterations? iteration? bids from 7:00 am to available for
6:00 am if necessary submitting IPS's
to provide time for Adjustment Bids, and
iteration? A/S bids, if necessary
to provide for
iteration?
- ------------------------------------------------------------------------------------------------------------------------------------
1
2 Price discovery. Various software fixes
identified. Without
fixes, would not favor
iterations since
negative impacts would
outweigh the benefit
of price discovery.
During Beta Test, the
resulting price change
after 2.5 hours of
iterations was $0.10.
That alone make
iterative bidding look
impractical.
3 Uncertain. Minimal, if any. Inadequate time to Additional testing
process bids. necessary after
technical difficulties
resolved. Various
software fixes
identified.
4 Yes. If production reacts A burden and at times
same as simulation, impossible for the
chance for more profit pre-schedulers and
from higher MCP. dispatchers to do the
process.
Pre-schedulers are
overloaded as it is.
5 No. Better price Additional time and Various software fixes
discovery. Perhaps personnel requirements identified.
some opportunity to with no definite
reduce costs. financial benefit.
6 Difficult to evaluate, None identified at Based on relatively Various software fixes
but would be concerned this time, but would small price movements identified.
about participants not want to eliminate during market
with larger market the ability to have simulation, not much
share starting the this function in the negative impact.
bidding with lower or future when the market However, gaming could
higher prices, knowing becomes more robust cause prices to move
they could re-price and competitive. out of the normal
their energy or market range.
withdraw from the
market altogether.
Some have contracts
based on PX index
pricing, whether the
energy is being bought
through the PX or not.
7 Yes. For now, minimal Timing of the IPS, A/S
benefit. Our main and Adj. Bids. These
concern is congestion bids are the "bread
and this will not and butter" on setting
necessarily address the market for the
it. next day. Concerned
about other issues
related to billing and
settlements,
appropriate treatment
of RMR generation, and
solutions to
congestion.
8 We are not concerned Most of the potential High opportunity cost Iterative bidding
that gaming will benefits result from of reducing the amount should not be
diminish the end better matching our of time for IPSs, Adj. implemented in the
results relative to a financial obligations Bids, and A/S bids. form that was tested
single round auction, with our physical Costs involved in during the Beta Test.
but we do expect supply capabilities developing new The process as tested
gaming and think that and costs. However, analytical and data is far too time
it will substantially the magnitude of these management software as consuming. The
eliminate the speculative benefits well as staffing theoretical
theoretical benefits should not be viewed issues if ours are incremental benefits
of the iterative as that great. Other extended too much of iterative bidding
process.. The net risk/revenue earlier into the should be viewed as
result is very little management tools morning. Finally, doubtful in practice.
payback for the time (including but not costs associated with If the Board decides
and effort consumed. limited to human errors and in favor of
The likely basic game participating in computer system or iterations,
in iterative bidding markets closer to real software problems implementation should
involves the time) exist to better resulting from the occur only after the
submission of bids match physical and more complex iterative successful conclusion
that are not financial positions. process. of another test, any
physically based for The benefits of an software bugs are
the sole purpose of iterative vs. a one fixed, and parties
withdrawing them to time auction process more familiar with
drive prices up. Can are doubtful and senlements problems
only be guarded really can't be have endorsed the
against with demonstrated in any concept. At least one
sophisticated market test we can think of. month lead time should
monitoring (a be provided after
difficult task to final iterative
distinguish between process and software
legitimate withdrawals are established to
and gaming allow participants to
activities). Risk to develop appropriate
gamers of not knowing new analytical and
when the last data management tools.
iteration will occur Iterative bidding
is very minimal since should not be seen as
the gaming strategy a priority at this
could include stage of the maturing
withdrawing enough energy market. It
capacity during each appears unlikely to
iteration to ensure play a significant
the stopping criteria role in improving the
is not met, with the ability of parties to
exception of the hard manage financial
time deadline. risks. Attention would
be better applied to
working with the ISO
to make the hourly
market more
streamlined (closer to
real time) and user
friendly.
- 44 -
Participant Do you think you would How much time should Are you willing to Are you willing to
normally modify your bids be allowed per move the deadline for reduce the time
during iterations? iteration? bids from 7:00 am to available for
6:00 am if necessary submitting IPS's
to provide time for Adjustment Bids, and
iteration? A/S bids, if necessary
to provide for
iteration?
- ------------------------------------------------------------------------------------------------------------------------------------
9 See a potential for Virtually none (keep Significant in terms Considering the
gaming, mostly due to in mind we are a net of time effort and various impacts of the
the fact that while seller and have no related staffing, as iteration process on
sellers can withdraw interest in reduced well as conflicting the market and our own
their bids, buyers prices) compared to timeline with operations, we are
cannot enter a new bid negative impacts. bilateral markets. against implementing
in later iterations. Iteration represents iterations. Answers to
Seems to be somewhat the PX having a "free previous questions
contrary to the effort option" on our energy provided in the event
to improve the prices. for an extra hour. By the PX fixes software
10 11 Yes. Activity the time iterations problems, provides
rules are not are closed, most of sufficient testing,
sufficient to prevent our bilateral and decides to
gaming. Withdrawal transactions are implement iterations.
Rule would allow a completed.
participant to submit
a fake bid to drive
price down to force
others to withdraw
from the market until
the fake bid is
finally withdrawn,
leaving the real bid
at a higher price,
resulting in a MCP
that exceeds the price
at which most
withdrawn capacity
would have been
willing to generate.
This is at least one
flaw that needs to be
fixed. Until then, we
will not agree to
iterations. 12 Yes. 13
Same as Participant
12. 14 Not a
significant concern.
10
11 Yes. Activity rules Few, under the current Higher prices due Not in favor of
are not sufficient to structure of the either to gaming or implementing
prevent gaming. proposed bidding programmed response to iterations at this
Withdrawal Rule would rules iterative results. time. However, do not
allow a participant to Greater possibility of want to foreclose on
submit a fake bid to mistakes, less time to the idea either.
drive price down to create Adj. Ids and Postpone iterations
force others to A/S bids. A process until there is a
withdraw from the that potentially makes demonstrable benefit
market until the fake self-provision more at a minimum risk.
bid is finally difficult to implement Rather the PX pursue
withdrawn, leaving the correctly. development of
real bid at a higher additional forward
price, resulting in a markets prior to
MCP that exceeds the implementing
price at which most iterations as we see a
withdrawn capacity larger potential
would have been benefit in such
willing to generate. development.
This is at least one
flaw that needs to be
fixed. Until then, we
will not agree to
iterations.
12 Yes. We may be able Any reduction of time Recommend shelve
to revise bidding for for IPSs, Adj. Bids, iterations. May be
generation units for and A/S bids would be worth resurrecting at
implementing feasible unacceptable. Price some later date, but
schedules. spikes because of do not want to set an
gaming, withdrawal of implementation date at
bids to raise prices, the time.
etc.
13 Same as Participant 12. Same as Participant Same as Participant Same as Participant
12. 12. 12.
14 Not a significant concern. May be an opportunity Negative impacts far May be a possibility,
to sell more. outweigh the benefits. but could be pressed
Costs include for time.
personnel changes due
to the times involved
and less activity with
the PX due to
bilateral market
timing conflicts.
- 45 -
APPENDIX B: ITERATIONS RESULTS FOR BETA TEST
Simulation Results from 11-03-98 Test
ITERATION #1 ITERATION #2 ITERATION #3
Hour MCP Volume MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng
- -----------------------------------------------------------------------------------------------------------
1 25.5149 13,839.6 25.0684 13,851.8 -1.75 0.09 25.1028 13,875.9 0.14 0.17
2 25.0007 13,786.3 24.3026 13,829.3 -2.79 0.31 24.3700 13,852.3 0.28 0.17
3 21.4308 13,608.3 21.0086 13,620.0 -1.97 0.09 21.0040 13,645.1 -0.02 0.18
4 22.8479 13,660.1 21.3979 13,700.4 -6.35 0.30 21.3824 13,725.8 -0.07 0.19
5 25.4741 13,983.0 25.0107 13,995.3 -1.82 0.09 25.0416 14,019.9 0.12 0.18
6 33.9970 15,359.6 33.9970 15,359.6 0.00 0.00 33.9970 15,359.6 0.00 0.00
7 53.6885 16,792.0 53.6883 16,792.0 0.00 0.00 53.8957 16,793.5 0.39 0.01
8 57.3040 17,830.1 57.3038 17,830.2 0.00 0.00 57.4813 17,836.7 0.31 0.04
9 60.2912 18,139.7 60.2911 18,139.7 0.00 0.00 60.4514 18,147.0 0.27 0.04
10 55.2454 19,002.9 55.2453 19,002.9 0.00 0.00 55.4098 19,008.9 0.30 0.03
11 57.6999 19,086.8 57.6998 19,086.8 0.00 0.00 57.8577 19,093.0 0.27 0.03
12 58.3690 19,068.7 58.3688 19,068.7 0.00 0.00 58.5245 19,075.0 0.27 0.03
13 43.9953 17,938.5 43.1070 18,170.4 -2.02 1.29 43.1998 18,171.2 0.22 0.00
14 44.4718 17,745.7 43.7222 17,941.4 -1.69 1.10 43.7481 17,959.6 0.06 0.10
15 44.3616 17,678.1 43.6423 17,868.3 -1.62 1.08 43.6488 17,891.6 0.01 0.13
16 42.8081 17,753.2 42.2550 17,898.9 -1.29 0.82 42.2095 17,935.6 -0.11 0.21
17 44.7584 18,006.6 43.9977 18,256.2 -1.70 1.39 44.0497 18,266.7 0.12 0.06
18 45.4382 19,188.0 44.9992 19,347.0 -0.97 0.83 44.9998 19,373.6 0.00 0.14
19 45.8005 18,889.7 45.2621 19,081.3 -1.18 1.01 45.3348 19,082.0 0.16 0.00
20 44.9092 18,234.1 44.5556 18,356.2 -0.79 0.67 44.6263 18,357.1 0.16 0.00
21 42.8459 18,055.8 42.4925 18,172.7 -0.82 0.65 42.5547 18,177.3 0.15 0.03
22 39.7595 17,400.5 42.3760 16,619.6 6.58 -4.49 42.3764 16,619.6 0.00 0.00
23 36.6966 16,532.1 39.6236 15,752.1 7.98 -4.72 39.6243 15,752.1 0.00 0.00
24 20.9864 15,086.3 28.0213 14,935.9 33.52 -1.00 27.1650 14,944.5 -3.06 0.06
------------------- ------------------------------------ ----------------------------------
41.4040 406,665.7 41.5599 406,676.7 0.38 0.00 41.5856 406,963.6 0.06 0.07
ITERATION #4 ITERATION #5 ITERATION #6
Hour MCP Volume % MCP % Vol MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng Chng Chng
- ------------------------------------------------------------------------------------------------------------------------
1 25.1028 13,875.9 0.00 0.00 66.5545 11,233.5 165.13 -19.04 66.5545 11,233.5 0.00 0.00
2 24.3700 13,852.3 0.00 0.00 59.7646 11,488.1 145.24 -17.07 59.7646 11,488.1 0.00 0.00
3 21.0040 13,645.1 0.00 0.00 57.5443 11,413.5 173.97 -16.35 57.5443 11,413.5 0.00 0.00
4 21.3660 13,726.2 -0.08 0.00 58.5590 11,433.6 174.08 -16.70 58.5590 11,433.6 0.00 0.00
5 25.0416 14,019.9 0.00 0.00 63.9028 11,649.3 155.19 -16.91 63.9028 11,649.3 0.00 0.00
6 33.9970 15,359.6 0.00 0.00 76.6063 12,274.3 125.33 -20.09 76.6063 12,274.3 0.00 0.00
7 53.8957 16,793.5 0.00 0.00 93.0266 12,954.7 72.60 -22.86 92.8824 12,968.2 -0.16 0.10
8 57.4813 17,836.7 0.00 0.00 89.0012 14,440.4 54.84 -19.04 88.8290 14,459.9 -0.19 0.14
9 60.4514 18,147.0 0.00 0.00 92.0143 14,545.5 52.21 -19.85 91.8674 14,561.5 -0.16 0.11
10 55.4098 19,008.9 0.00 0.00 85.4252 15,458.4 54.17 -18.68 92.6781 14,590.1 8.49 -5.62
11 57.8577 19,093.0 0.00 0.00 87.3385 15,493.3 50.95 -18.85 87.1651 15,515.8 -0.20 0.15
12 58.5245 19,075.0 0.00 0.00 87.7634 15,482.6 49.96 -18.83 87.5943 15,504.8 -0.19 0.14
13 43.1998 18,171.2 0.00 0.00 79.0559 14,156.2 83.00 -22.10 78.7859 14,175.9 -0.34 0.14
14 43.7481 17,959.6 0.00 0.00 80.1324 14,008.7 83.17 -22.00 79.9935 14,018.8 -0.17 0.07
15 43.6488 17,891.6 0.00 0.00 80.0171 13,929.1 83.32 -22.15 79.9645 13,933.0 -0.07 0.03
16 42.1850 17,941.9 -0.06 0.04 77.4086 13,864.7 83.50 -22.72 77.1426 13,883.6 -0.34 0.14
17 44.0497 18,266.7 0.00 0.00 78.8250 14,144.4 78.95 -22.57 78.5635 14,162.9 -0.33 0.13
18 44.9998 19,373.6 0.00 0.00 83.4484 14,819.4 85.44 -23.51 83.1663 14,841.5 -0.34 0.15
19 45.3348 19,082.0 0.00 0.00 76.4652 15,248.0 68.67 -20.09 83.9946 14,654.7 9.85 -3.89
20 44.6263 18,357.1 0.00 0.00 78.0904 14,317.1 74.99 -22.01 78.0202 14,322.3 -0.09 0.04
21 42.5547 18,177.3 0.00 0.00 73.9360 13,926.2 73.74 -23.39 73.8005 13,935.8 -0.18 0.07
22 42.3764 16,619.6 0.00 0.00 67.2876 13,136.1 58.79 -20.96 66.8728 13,163.0 -0.62 0.20
23 39.6243 15,752.1 0.00 0.00 60.2846 12,315.3 52.14 -21.82 60.2846 12,315.3 0.00 0.00
24 27.1650 14,944.5 0.00 0.00 48.0100 11,734.9 76.73 -21.48 48.0100 11,734.9 0.00 0.00
----------------------------------- ------------------------------------- -----------------------------------
41.5839 406,970.3 0.00 0.00 75.0192 323,467.3 80.40 -20.52 75.5228 322,234.3 0.67 -0.38
-46-
Simulation Results from 11-04-98 Test
ITERATION #1 ITERATION #2 ITERATION #3
Hour MCP Volume MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng
- --------------------------------------------------------------------------------------------------------------
1 138.7488 7,948.2 152.1834 7,963.4 9.68 0.19 154.0089 7,965.5 1.20 0.03
2 80.0820 8,257.5 83.5213 8,317.5 4.29 0.73 85.0086 8,333.1 1.78 0.19
3 74.9955 8,013.0 76.0084 8,068.3 1.35 0.69 77.0058 8,111.6 1.31 0.54
4 78.0239 8,149.5 79.0085 8,209.5 1.26 0.74 80.0054 8,263.3 1.26 0.66
5 101.3070 8,441.7 109.2207 8,444.1 7.81 0.03 112.0087 8,445.0 2.55 0.01
6 201.0481 8,538.4 211.1587 8,541.5 5.03 0.04 215.9516 8,543.0 2.27 0.02
7 225.8105 9,192.9 228.8427 9,193.9 1.34 0.01 230.5389 9,194.4 0.74 0.01
8 187.7054 10,587.2 194.7793 10,589.4 3.77 0.02 200.0079 10,591.0 2.68 0.02
9 197.5304 10,685.3 206.9756 10,691.6 4.78 0.06 211.5941 10,694.0 2.23 0.02
10 206.8824 10,730.2 216.4953 10,736.6 4.65 0.06 220.5132 10,739.3 1.86 0.03
11 211.4975 10,748.2 220.1406 10,753.9 4.09 0.05 223.1588 10,756.0 1.37 0.02
12 214.0332 10,641.3 221.7829 10,646.4 3.62 0.05 224.1089 10,648.0 1.05 0.02
13 222.1792 10,463.7 225.6048 10,466.0 1.54 0.02 228.0034 10,467.6 1.06 0.02
14 225.7341 10,190.1 235.1831 10,196.4 4.19 0.06 241.5478 10,200.6 2.71 0.04
15 225.7328 10,113.8 235.1582 10,120.1 4.18 0.06 241.5197 10,124.3 2.71 0.04
16 225.7507 10,081.2 235.1082 10,087.4 4.15 0.06 241.4570 10,091.7 2.70 0.04
17 226.1181 10,328.8 240.1522 10,338.2 6.21 0.09 247.0723 10,342.8 2.88 0.04
18 226.4257 10,944.8 240.5433 10,954.3 6.23 0.09 247.1360 10,958.7 2.74 0.04
19 226.3423 10,863.0 230.1066 10,867.2 1.66 0.04 233.1054 10,870.6 1.30 0.03
20 223.3216 10,833.9 228.5911 10,839.9 2.36 0.06 229.9367 10,841.4 0.59 0.01
21 204.4151 10,720.7 214.0269 10,731.6 4.70 0.10 218.3020 10,736.4 2.00 0.04
22 175.1481 10,428.7 177.0817 10,430.9 1.10 0.02 177.0817 10,430.9 0.00 0.00
23 141.8638 9,598.1 151.6702 9,609.3 6.91 0.12 151.7525 9,609.4 0.05 0.00
24 79.3292 9,282.8 79.9976 9,321.7 0.84 0.42 79.9976 9,321.7 0.00 0.00
-------------------- ------------------------------------ ------------------------------------
180.0011 235,783.0 187.2226 236,119.1 4.01 0.14 190.4510 236,280.3 1.72 0.07
ITERATION #4 ITERATION #5 ITERATION #6
Hour MCP Volume % MCP % Vol MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng Chng Chng
- ------------------------------------------------------------------------------------------------------------------------
1 155.0081 7,966.6 0.65 0.01 156.0076 7,967.8 0.64 0.02 156.0076 7,967.8 0.00 0.00
2 87.0068 8,352.1 2.35 0.23 88.0062 8,355.0 1.15 0.03 88.0062 8,355.0 0.00 0.00
3 79.2806 8,209.4 2.95 1.21 79.2987 8,209.9 0.02 0.01 79.2987 8,209.9 0.00 0.00
4 80.0054 8,263.3 0.00 0.00 80.8139 8,281.6 1.01 0.22 80.8139 8,281.6 0.00 0.00
5 115.0072 8,445.9 2.68 0.01 116.0067 8,446.2 0.87 0.00 116.0067 8,446.2 0.00 0.00
6 220.0005 8,544.2 1.87 0.01 220.2271 8,544.3 0.10 0.00 220.2271 8,544.3 0.00 0.00
7 232.4418 9,195.0 0.83 0.01 234.5451 9,195.6 0.90 0.01 235.0967 9,195.8 0.24 0.00
8 205.0019 10,592.5 2.50 0.01 206.0008 10,592.8 0.49 0.00 206.0008 10,592.8 0.00 0.00
9 215.0037 10,696.9 1.61 0.03 216.0019 10,697.6 0.46 0.01 216.0019 10,697.6 0.00 0.00
10 220.5132 10,739.3 0.00 0.00 220.5132 10,739.3 0.00 0.00 220.5132 10,739.3 0.00 0.00
11 223.1588 10,756.0 0.00 0.00 223.1588 10,756.0 0.00 0.00 223.1588 10,756.0 0.00 0.00
12 224.1089 10,648.0 0.00 0.00 224.1089 10,648.0 0.00 0.00 224.1089 10,648.0 0.00 0.00
13 230.0070 10,468.9 0.88 0.01 232.8327 10,470.8 1.23 0.02 234.0060 10,471.6 0.50 0.01
14 245.5774 10,203.3 1.67 0.03 245.6651 10,203.4 0.04 0.00 245.6651 10,203.4 0.00 0.00
15 245.5527 10,127.0 1.67 0.03 245.6375 10,127.1 0.03 0.00 245.6375 10,127.1 0.00 0.00
16 245.5036 10,094.4 1.68 0.03 245.8242 10,094.6 0.13 0.00 245.8242 10,094.6 0.00 0.00
17 249.4611 10,344.4 0.97 0.02 249.5189 10,344.4 0.02 0.00 249.5189 10,344.4 0.00 0.00
18 249.4218 10,960.2 0.92 0.01 249.4814 10,960.2 0.02 0.00 249.4814 10,960.2 0.00 0.00
19 235.8417 10,873.7 1.17 0.03 235.8417 10,873.7 0.00 0.00 235.8417 10,873.7 0.00 0.00
20 229.9367 10,841.4 0.00 0.00 229.9367 10,841.4 0.00 0.00 229.9367 10,841.4 0.00 0.00
21 218.3020 10,736.4 0.00 0.00 218.3020 10,736.4 0.00 0.00 218.3020 10,736.4 0.00 0.00
22 177.0817 10,430.9 0.00 0.00 177.0817 10,430.9 0.00 0.00 177.0817 10,430.9 0.00 0.00
23 151.7525 9,609.4 0.00 0.00 151.7525 9,609.4 0.00 0.00 151.7525 9,609.4 0.00 0.00
24 79.9976 9,321.7 0.00 0.00 79.9976 9,321.7 0.00 0.00 79.9976 9,321.7 0.00 0.00
------------------------------------ ------------------------------------ ------------------------------------
192.2905 236,420.9 0.97 0.06 192.7734 236,448.1 0.25 0.01 192.8452 236,449.1 0.04 0.00
-47-
Simulation Results from 11-05-98 Test
ITERATION #1 ITERATION #2 ITERATION #3
Hour MCP Volume MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng
- -------------------------------------------------------------------------------------------------------------
1 17.3078 15,390.4 14.4685 15,404.4 -16.40 0.09 12.8492 15,609.4 -11.19 1.33
2 13.9991 14,970.7 12.4184 15,163.9 -11.29 1.29 10.9207 15,319.4 -12.06 1.03
3 12.4703 14,732.1 10.9990 14,891.0 -11.80 1.08 9.8170 14,945.9 -10.75 0.37
4 12.5388 14,851.2 11.0081 14,929.3 -12.21 0.53 9.7430 15,065.3 -11.49 0.91
5 16.7483 15,337.0 14.9915 15,346.3 -10.49 0.06 12.4749 15,554.5 -16.79 1.36
6 24.5189 16,991.3 22.1606 17,004.3 -9.62 0.08 21.9969 17,257.8 -0.74 1.49
7 27.0070 18,938.3 27.0001 19,162.3 -0.03 1.18 27.0001 19,162.3 0.00 0.00
8 27.9966 20,034.3 27.4453 20,050.2 -1.97 0.08 27.3943 20,050.9 -0.19 0.00
9 29.9947 20,760.2 29.4329 20,800.1 -1.87 0.19 29.3229 20,801.1 -0.37 0.00
10 29.3145 21,338.9 28.9956 21,354.5 -1.09 0.07 28.8304 21,356.1 -0.57 0.01
11 30.3527 21,667.2 30.0079 21,693.8 -1.14 0.12 31.9866 21,635.3 6.59 -0.27
12 31.9899 21,723.8 31.9866 21,735.2 -0.01 0.05 33.9961 21,682.6 6.28 -0.24
13 34.9911 21,686.9 34.9917 21,701.3 0.00 0.07 34.9725 21,701.8 -0.05 0.00
14 35.9751 21,714.8 35.7730 21,739.6 -0.56 0.11 35.9715 21,755.0 0.55 0.07
15 36.0015 21,580.8 36.7892 21,556.5 2.19 -0.11 36.1826 21,586.5 -1.65 0.14
16 36.0021 21,324.3 37.5484 21,265.2 4.30 -0.28 36.9764 21,295.2 -1.52 0.14
17 36.0050 21,588.6 39.9924 21,402.5 11.07 -0.86 39.5062 21,441.7 -1.22 0.18
18 42.9643 22,625.4 41.9919 22,694.0 -2.26 0.30 42.4001 22,697.5 0.97 0.02
19 43.5176 22,396.9 42.9027 22,444.4 -1.41 0.21 43.2025 22,453.9 0.70 0.04
20 39.1212 21,954.2 38.5965 21,999.5 -1.34 0.21 38.4921 22,030.2 -0.27 0.14
21 34.9773 21,234.6 34.9783 21,249.0 0.00 0.07 34.9730 21,249.1 -0.02 0.00
22 27.9944 19,823.3 27.4769 19,839.0 -1.85 0.08 27.4811 19,840.2 0.02 0.01
23 24.9585 17,462.8 24.0005 17,472.3 -3.84 0.05 23.4731 17,476.1 -2.20 0.02
24 15.6528 16,055.7 13.1182 16,270.0 -16.19 1.33 11.8672 16,324.0 -9.54 0.33
------------------- ----------------------------------- -------------------------------------
28.4333 466,183.7 27.8781 467,168.6 -1.95 0.21 27.5763 468,291.8 -1.08 0.24
ITERATION #4 ITERATION #5 ITERATION #6
Hour MCP Volume % MCP % Vol MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng Chng Chng
- ----------------------------------------------------------------------------------------------------------------------
1 11.7593 15,812.7 -8.48 1.30 11.3274 15,814.0 -3.67 0.01 11.0067 15,898.0 -2.83 0.53
2 10.0000 15,472.8 -8.43 1.00 9.6949 15,523.8 -3.05 0.33 9.3690 15,538.7 -3.36 0.10
3 8.9705 15,099.4 -8.62 1.03 8.4862 15,101.4 -5.40 0.01 8.0945 15,150.3 -4.62 0.32
4 9.3275 15,217.1 -4.26 1.01 9.0082 15,218.4 -3.42 0.01 9.0083 15,229.5 0.00 0.07
5 12.5672 15,704.2 0.74 0.96 12.0488 15,705.9 -4.13 0.01 12.0688 15,805.8 0.17 0.64
6 21.9969 17,257.8 0.00 0.00 21.9969 17,257.8 0.00 0.00 21.9969 17,257.8 0.00 0.00
7 27.0001 19,162.3 0.00 0.00 27.0001 19,162.3 0.00 0.00 27.0001 19,162.3 0.00 0.00
8 27.3943 20,050.9 0.00 0.00 27.3943 20,050.9 0.00 0.00 27.3943 20,050.9 0.00 0.00
9 29.3229 20,801.1 0.00 0.00 29.3229 20,801.1 0.00 0.00 29.3229 20,801.1 0.00 0.00
10 28.8304 21,356.1 0.00 0.00 28.8304 21,356.1 0.00 0.00 28.8304 21,356.1 0.00 0.00
11 31.2923 21,683.2 -2.17 0.22 31.0315 21,690.7 -0.83 0.03 31.0315 21,690.7 0.00 0.00
12 33.0078 21,719.7 -2.91 0.17 33.0049 21,726.9 -0.01 0.03 33.0049 21,726.9 0.00 0.00
13 34.9385 21,727.8 -0.10 0.12 34.9385 21,727.8 0.00 0.00 34.9385 21,727.8 0.00 0.00
14 35.9715 21,755.0 0.00 0.00 35.9715 21,755.0 0.00 0.00 35.9715 21,755.0 0.00 0.00
15 35.9973 21,595.6 -0.51 0.04 35.9804 21,596.5 -0.05 0.00 36.4949 21,596.5 1.43 0.00
16 36.4947 21,318.7 -1.30 0.11 36.4947 21,318.7 0.00 0.00 36.7719 21,305.2 0.76 -0.06
17 39.5062 21,441.7 0.00 0.00 39.5062 21,441.7 0.00 0.00 39.9936 21,399.5 1.23 -0.20
18 42.4001 22,697.5 0.00 0.00 42.4001 22,697.5 0.00 0.00 43.9960 22,587.7 3.76 -0.48
19 43.2025 22,453.9 0.00 0.00 43.2025 22,453.9 0.00 0.00 44.5863 22,358.1 3.20 -0.43
20 38.4921 22,030.2 0.00 0.00 44.9906 21,639.5 16.88 -1.77 44.9938 21,689.4 0.01 0.23
21 34.9746 21,274.1 0.00 0.12 34.9746 21,274.1 0.00 0.00 34.9746 21,274.1 0.00 0.00
22 27.4811 19,840.2 0.00 0.00 27.4811 19,840.2 0.00 0.00 27.4811 19,840.2 0.00 0.00
23 23.3378 17,476.6 -0.58 0.00 23.1920 17,477.2 -0.62 0.00 23.1752 17,477.3 -0.07 0.00
24 11.9922 16,462.5 1.05 0.85 11.7817 16,474.3 -1.76 0.07 11.7817 16,474.3 0.00 0.00
----------------------------------- ----------------------------------- ----------------------------------
27.3441 469,411.1 -0.84 0.24 27.5025 469,105.7 0.58 -0.07 27.6370 469,153.2 0.49 0.01
-48-
Simulation Results from 11-05-98 Test (continued)
ITERATION #7 ITERATION #8 ITERATION #9
Hour MCP Volume % MCP % Vol MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng Chng Chng
- -----------------------------------------------------------------------------------------------------------------------
1 11.0067 15,898.0 0.00 0.00 11.0067 15,898.0 0.00 0.00 11.0067 15,898.0 0.00 0.00
2 9.3690 15,538.7 0.00 0.00 9.3690 15,538.7 0.00 0.00 9.3690 15,538.7 0.00 0.00
3 8.0948 15,153.0 0.00 0.02 7.7277 15,154.5 -4.54 0.01 7.3987 15,155.9 -4.26 0.01
4 9.0083 15,229.5 0.00 0.00 9.0083 15,229.5 0.00 0.00 9.0083 15,229.5 0.00 0.00
5 12.0688 15,805.8 0.00 0.00 12.0688 15,805.8 0.00 0.00 12.0688 15,805.8 0.00 0.00
6 21.9969 17,257.8 0.00 0.00 21.9969 17,257.8 0.00 0.00 21.9969 17,257.8 0.00 0.00
7 27.0001 19,162.3 0.00 0.00 27.0001 19,162.3 0.00 0.00 27.0001 19,162.3 0.00 0.00
8 27.3943 20,050.9 0.00 0.00 27.3943 20,050.9 0.00 0.00 27.3943 20,050.9 0.00 0.00
9 29.3229 20,801.1 0.00 0.00 29.3229 20,801.1 0.00 0.00 29.3229 20,801.1 0.00 0.00
10 28.8304 21,356.1 0.00 0.00 28.8304 21,356.1 0.00 0.00 28.8304 21,356.1 0.00 0.00
11 31.0315 21,690.7 0.00 0.00 31.0315 21,690.7 0.00 0.00 31.0315 21,690.7 0.00 0.00
12 33.0049 21,726.9 0.00 0.00 33.0049 21,726.9 0.00 0.00 33.0049 21,726.9 0.00 0.00
13 34.9385 21,727.8 0.00 0.00 34.9385 21,727.8 0.00 0.00 34.9385 21,727.8 0.00 0.00
14 35.9715 21,755.0 0.00 0.00 35.9715 21,755.0 0.00 0.00 35.9715 21,755.0 0.00 0.00
15 36.4949 21,596.5 0.00 0.00 36.4949 21,596.5 0.00 0.00 36.4949 21,596.5 0.00 0.00
16 36.7719 21,305.2 0.00 0.00 36.7719 21,305.2 0.00 0.00 36.7719 21,305.2 0.00 0.00
17 39.9936 21,399.5 0.00 0.00 39.9936 21,399.5 0.00 0.00 26.9196 19,770.5 -32.69 -7.61
18 43.7381 22,601.4 -0.59 0.06 43.5624 22,610.7 -0.40 0.04 43.5824 22,610.7 0.05 0.00
19 44.3227 22,371.1 -0.59 0.06 44.0074 22,393.3 -0.71 0.10 44.0022 22,406.5 -0.01 0.06
20 44.9930 21,689.4 0.00 0.00 44.9930 21,689.4 0.00 0.00 44.9930 21,689.4 0.00 0.00
21 34.9746 21,274.1 0.00 0.00 34.9746 21,274.1 0.00 0.00 34.9746 21,274.1 0.00 0.00
22 27.4811 19,840.2 0.00 0.00 27.4811 19,840.2 0.00 0.00 27.4811 19,840.2 0.00 0.00
23 23.2088 17,483.3 0.14 0.03 23.2088 17,483.3 0.00 0.00 23.2088 17,483.3 0.00 0.00
24 11.7817 16,474.3 0.00 0.00 11.7817 16,474.3 0.00 0.00 11.7817 16,474.3 0.00 0.00
----------------------------------- ----------------------------------- -----------------------------------
27.6166 469,188.6 -0.07 0.01 27.5809 469,221.6 -0.13 0.01 27.0230 467,607.2 -2.02 -0.34
ITERATION #10 ITERATION #11
Hour MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng
- --------------------------------------------------------------------------------
1 11.0067 15,898.0 0.00 0.00 11.0067 15,898.0 0.00 0.00
2 9.3690 15,538.7 0.00 0.00 9.3690 15,538.7 0.00 0.00
3 7.3987 15,155.9 0.00 0.00 7.3987 15,155.9 0.00 0.00
4 9.0083 15,229.5 0.00 0.00 9.0083 15,229.5 0.00 0.00
5 12.0688 15,805.8 0.00 0.00 12.0688 15,805.8 0.00 0.00
6 21.9969 17,257.8 0.00 0.00 21.9969 17,257.8 0.00 0.00
7 27.0001 19,162.3 0.00 0.00 27.0001 19,162.3 0.00 0.00
8 27.3943 20,050.9 0.00 0.00 27.3943 20,050.9 0.00 0.00
9 29.3229 20,801.1 0.00 0.00 29.3229 20,801.1 0.00 0.00
10 28.8304 21,356.1 0.00 0.00 28.8304 21,356.1 0.00 0.00
11 31.0315 21,690.7 0.00 0.00 31.0315 21,690.7 0.00 0.00
12 33.0049 21,726.9 0.00 0.00 33.0049 21,726.9 0.00 0.00
13 34.9385 21,727.8 0.00 0.00 34.9385 21,727.8 0.00 0.00
14 35.9715 21,755.0 0.00 0.00 35.9715 21,755.0 0.00 0.00
15 36.4949 21,596.5 0.00 0.00 36.4949 21,596.5 0.00 0.00
16 36.7719 21,305.2 0.00 0.00 36.7719 21,305.2 0.00 0.00
17 26.8840 19,770.7 -0.13 0.00 27.3848 20,018.3 1.86 1.25
18 43.5624 22,610.7 -0.05 0.00 43.5624 22,610.7 0.00 0.00
19 44.0022 22,406.5 0.00 0.00 44.0022 22,406.5 0.00 0.00
20 44.9930 21,689.4 0.00 0.00 44.9930 21,689.4 0.00 0.00
21 34.9746 21,274.1 0.00 0.00 34.9746 21,274.1 0.00 0.00
22 27.4811 19,840.2 0.00 0.00 27.4811 19,840.2 0.00 0.00
23 23.2088 17,483.3 0.00 0.00 23.2088 17,483.3 0.00 0.00
24 11.7817 16,474.3 0.00 0.00 11.7817 16,474.3 0.00 0.00
----------------------------------- -----------------------------------
27.0207 467,607.4 -0.01 0.00 27.0416 467,855.0 0.08 0.05
-49-
Simulation Results from 11-06-98 Test
ITERATION #1 ITERATION #2 ITERATION #3
Hour MCP Volume MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng
- -----------------------------------------------------------------------------------------------------------
1 26.2162 12,311.3 26.5538 12,423.3 1.29 0.91 26.4616 12,450.7 -0.35 0.22
2 25.7248 12,060.2 25.9908 12,193.2 1.03 1.10 25.8920 12,220.6 -0.38 0.22
3 25.0860 11,806.8 25.3588 11,938.2 1.09 1.11 25.2262 11,975.5 -0.52 0.31
4 25.2161 11,910.2 25.5539 12,025.6 1.34 0.97 25.3808 12,073.0 -0.68 0.39
5 25.9579 12,281.9 26.2319 12,408.9 1.06 1.03 26.1490 12,433.1 -0.32 0.20
6 31.0038 12,241.4 31.0108 12,245.8 0.02 0.04 31.0108 12,245.8 0.00 0.00
7 34.4429 12,754.9 34.3287 12,808.8 -0.33 0.42 39.6478 13,094.7 15.49 2.23
8 34.3445 13,648.2 34.2360 13,702.2 -0.32 0.40 39.5576 14,099.7 15.54 2.90
9 35.0502 13,927.8 34.9790 13,967.4 -0.20 0.28 39.6745 14,866.9 13.42 6.44
10 35.8670 14,005.1 35.0125 14,382.2 -2.38 2.69 39.8910 14,943.5 13.93 3.90
11 36.3252 14,136.2 35.0183 14,716.3 -3.60 4.10 40.0238 15,001.3 14.29 1.94
12 36.4531 14,177.9 37.7480 14,949.3 3.55 5.44 38.6936 14,990.3 2.51 0.27
13 35.9794 14,377.1 37.3897 15,227.0 3.92 5.91 37.9921 15,237.7 1.61 0.07
14 36.3902 14,228.4 37.7658 14,955.1 3.78 5.11 38.7239 14,999.4 2.54 0.30
15 36.6297 13,996.2 37.8272 14,765.9 3.27 5.50 38.8606 14,807.0 2.73 0.28
16 36.2164 13,929.9 37.5654 14,703.4 3.72 5.55 38.2704 14,744.2 1.88 0.28
17 35.9998 14,031.4 37.3128 14,887.4 3.65 6.10 37.8089 14,888.0 1.33 0.00
18 36.9472 14,484.0 38.1305 15,190.8 3.20 4.88 39.0371 15,191.8 2.38 0.01
19 36.8297 14,406.3 38.0382 15,114.1 3.28 4.91 38.9951 15,115.2 2.52 0.01
20 36.9401 13,981.2 38.0692 14,688.0 3.06 5.06 39.0097 14,689.0 2.47 0.01
21 36.0095 13,821.7 37.6394 14,451.5 4.53 4.56 38.4370 14,492.4 2.12 0.28
22 34.9741 13,320.4 34.9712 13,330.5 -0.01 0.08 38.1858 13,645.9 9.19 2.37
23 30.0084 13,516.5 30.0084 13,523.4 0.00 0.05 30.0084 13,523.4 0.00 0.00
24 27.2733 13,175.7 27.6873 13,254.5 1.52 0.60 27.6633 13,262.0 -0.09 0.06
------------------- ------------------------------------ -----------------------------------
32.9952 322,530.7 33.5178 331,852.8 1.58 2.89 35.0250 334,991.1 4.50 0.95
ITERATION #4 ITERATION #5 ITERATION #6
Hour MCP Volume % MCP % Vol MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng Chng Chng
- ------------------------------------------------------------------------------------------------------------------------
1 26.4616 12,450.7 0.00 0.00 26.4616 12,450.7 0.00 0.00 29.0627 11,726.1 9.83 -5.82
2 25.8920 12,220.6 0.00 0.00 25.8920 12,220.6 0.00 0.00 27.7294 11,716.5 7.10 -4.13
3 25.2262 11,975.5 0.00 0.00 25.2262 11,975.5 0.00 0.00 26.6901 11,564.2 5.80 -3.43
4 25.3808 12,073.0 0.00 0.00 25.3808 12,073.0 0.00 0.00 26.9927 11,632.6 6.35 -3.65
5 26.1490 12,433.1 0.00 0.00 26.1490 12,433.1 0.00 0.00 27.9153 11,923.9 6.75 -4.10
6 31.0108 12,245.8 0.00 0.00 31.0108 12,245.8 0.00 0.00 33.9861 11,274.1 9.59 -7.93
7 42.5951 13,198.0 7.43 0.79 44.8593 13,355.4 5.32 1.19 46.9592 12,970.3 4.68 -2.88
8 42.5822 13,990.0 7.65 -0.78 44.9342 13,169.3 5.52 -5.87 46.9818 13,801.8 4.56 4.80
9 42.6079 14,680.2 7.39 -1.26 44.9368 13,838.5 5.47 -5.73 46.9856 14,342.9 4.56 3.64
10 42.9135 14,756.8 7.58 -1.25 44.9775 14,915.4 4.81 1.07 46.9966 14,369.3 4.49 -3.66
11 43.1266 14,764.8 7.75 -1.58 45.7784 15,329.1 6.15 3.82 47.9743 14,383.4 4.80 -6.17
12 38.9861 14,750.7 0.76 -1.60 38.9340 14,850.6 -0.13 0.68 39.4554 13,851.2 1.34 -6.73
13 38.5459 15,031.8 1.46 -1.35 38.2888 15,130.7 -0.67 0.66 39.3012 14,135.0 2.64 -6.58
14 38.9916 14,760.6 0.69 -1.59 39.3480 14,862.2 0.91 0.69 41.7694 13,923.1 6.15 -6.32
15 39.0198 14,567.2 0.41 -1.62 39.4967 14,667.7 1.22 0.69 41.9506 13,720.5 6.21 -6.46
16 38.9040 14,504.9 1.66 -1.62 38.8913 14,604.9 -0.03 0.69 41.3500 13,657.6 6.32 -6.49
17 38.3201 14,688.5 1.35 -1.34 38.0648 14,788.2 -0.67 0.68 39.2566 13,789.6 3.13 -6.75
18 39.1529 14,952.0 0.30 -1.58 39.5749 14,052.4 1.08 -6.02 41.3287 14,104.4 4.43 0.37
19 39.1136 14,875.3 0.30 -1.59 39.5498 13,975.8 1.12 -6.05 41.1963 14,027.6 4.16 0.37
20 39.1333 14,449.1 0.32 -1.63 39.0818 14,549.1 -0.13 0.69 39.5830 13,549.6 1.28 -6.87
21 38.9381 14,252.9 1.30 -1.65 39.0622 14,353.1 0.32 0.70 41.3941 13,455.7 5.97 -6.25
22 38.9032 13,413.1 1.88 -1.71 39.1560 13,515.6 0.65 0.76 41.3054 12,637.1 5.49 -6.50
23 30.5898 13,321.1 1.94 -1.50 30.5898 13,321.1 0.00 0.00 33.3942 12,348.7 9.17 -7.30
24 27.6633 13,262.0 0.00 0.00 27.6633 13,262.0 0.00 0.00 30.0092 12,350.8 8.48 -6.87
------------------------------------ ----------------------------------- -----------------------------------
35.8420 331,617.7 2.33 -1.01 36.3878 329,939.8 1.52 -0.51 38.3153 315,256.0 5.30 -4.45
-50-
Simulation Results from 11-07-98 Test
ITERATION #1 ITERATION #2 ITERATION #3
Hour MCP Volume MCP Volume % MCP % Vol MCP Volume % MCP % Vol
Chng Chng Chng Chng
- -------- -------------------- ----------------------------------------------------------------------------------------
1 22.0006 13,737.4 22.0094 15,412.8 0.04 12.20 22.0094 15,412.8 0.00 0.00
2 21.3668 13,360.7 22.0024 15,253.2 2.97 14.16 22.0084 15,253.2 0.03 0.00
3 20.1368 12,919.4 20.8330 14,937.8 3.46 15.62 21.3871 15,042.5 2.66 0.70
4 20.2909 12,997.9 20.9663 15,023.7 3.33 15.59 21.5239 15,109.8 2.66 0.57
5 20.9614 13,472.6 21.9801 15,380.7 4.86 14.16 22.0045 15,384.1 0.11 0.02
6 22.6785 14,271.2 25.0021 15,903.6 10.25 11.44 25.3677 15,976.0 1.46 0.46
7 26.2906 15,634.5 26.7357 17,746.9 1.69 13.51 26.7357 17,746.9 0.00 0.00
8 26.1286 16,356.8 26.6362 18,474.0 1.94 12.94 27.0057 18,549.7 1.39 0.41
9 27.7447 17,028.0 28.3713 18,875.6 2.26 10.85 28.9722 18,893.2 2.12 0.09
10 27.9841 17,555.3 32.2385 18,876.7 15.20 7.53 32.9741 18,937.8 2.28 0.32
11 27.9874 17,893.5 34.5337 19,035.1 23.39 6.38 35.2658 19,085.4 2.12 0.26
12 27.9885 17,969.4 34.8165 18,975.7 24.40 5.60 35.5569 19,026.0 2.13 0.27
13 27.9861 17,868.9 31.5605 19,134.8 12.77 7.08 32.5469 19,138.1 3.13 0.02
14 27.9879 17,798.7 33.2617 18,947.4 18.84 6.45 33.9972 19,018.4 2.21 0.37
15 27.9867 17,592.6 32.9779 18,760.9 17.83 6.64 33.6803 18,891.9 2.13 0.70
16 27.9859 17,443.4 32.9722 18,612.0 17.82 6.70 33.6780 18,729.7 2.14 0.63
17 27.9884 17,892.7 34.6769 18,914.3 23.90 5.71 35.4189 18,964.5 2.14 0.27
18 33.1555 19,161.4 40.0076 19,791.4 20.67 3.29 40.6111 19,900.8 1.51 0.55
19 32.1264 19,051.0 40.0053 19,658.1 24.52 3.19 40.5881 19,824.1 1.46 0.84
20 32.3741 18,498.0 40.0028 19,121.8 23.56 3.37 40.5438 19,347.8 1.35 1.18
21 28.3022 17,904.1 36.7389 18,804.7 29.81 5.03 37.7707 18,809.4 2.81 0.02
22 27.9815 16,709.8 28.8557 18,324.6 3.12 9.66 29.5820 18,330.8 2.52 0.03
23 25.0003 15,147.9 26.8779 16,521.0 7.51 9.06 27.5519 16,599.7 2.51 0.48
24 22.1283 14,245.7 24.7335 16,000.1 11.77 12.32 25.0065 16,041.9 1.10 0.26
-------------------- ----------------------------------------- ------------------------------------------
26.3568 392,510.9 29.9498 426,486.9 13.63 8.66 30.4911 428,014.5 1.81 0.36
ITERATION #4
Hour MCP Volume % MCP % Vol
Chng Chng
- -------------------------------------------------
1 22.0094 15,412.8 0.00 0.00
2 22.0084 15,253.2 0.00 0.00
3 21.3871 15,042.5 0.00 0.00
4 21.5239 15,109.8 0.00 0.00
5 22.0082 15,384.6 0.02 0.00
6 25.3133 15,960.3 -0.21 -0.10
7 30.1917 17,030.7 12.93 -4.04
8 30.0070 17,790.1 11.11 -4.09
9 35.3969 18,131.6 22.18 -4.03
10 39.7867 18,159.1 20.66 -4.11
11 40.0090 18,433.2 13.45 -3.42
12 40.0062 18,405.9 12.51 -3.26
13 38.5634 18,490.7 18.49 -3.38
14 39.9955 18,250.5 17.64 -4.04
15 39.9948 18,050.3 18.75 -4.45
16 39.6743 17,964.0 17.80 -4.09
17 40.0056 18,333.2 12.95 -3.33
18 41.2175 18,915.3 1.49 -4.95
19 41.1927 18,838.6 1.49 -4.97
20 41.1604 18,362.3 1.52 -5.09
21 40.5645 18,152.5 7.40 -3.49
22 36.0073 17,697.8 21.72 -3.45
23 27.5519 16,599.7 0.00 0.00
24 25.0065 16,041.9 0.00 0.00
------------------------------------------
33.3576 415,810.6 9.40 -2.85