EXHIBIT 99.502

RESPONSES TO JULY 18, 1997
REQUESTS FOR ADDITIONAL INFORMATION
("DATA RESPONSES")

MARKET DESIGN

1. CONSISTENT WITH YOUR RESPONSE TO QUESTION 14 OF THE APRIL 29, 1997
REQUEST BY FERC STAFF FOR ADDITIONAL INFORMATION, PLEASE PROVIDE A COPY
OF THE EXPERIMENTAL DESIGN, THE QUANTITATIVE DATA AND STATISTICS, AND
THE SUMMARY STATISTICS FROM PROFESSOR PLOTT'S AUCTION EXPERIMENTS.

A copy of the experimental design is found in Appendix B of Robert
Wilson's report "Activity Rules for the Power Exchange, Phase 3:
Experimental Testing," dated March 3, 1997 attached hereto as
Attachment 1. The quantitative data and statistics requested can be
found in Attachment 2. Attachment 3 is a summary table of results and
statistics for those experiments with fixed cost components prepared by
Professor Plott.

2. CONSISTENT WITH YOUR RESPONSE TO QUESTION 14(B) OF THE FERC REQUEST,
PLEASE PROVIDE A COPY OF THE REPORTS ON FOREIGN ELECTRIC MARKETS AND A
BIBLIOGRAPHY OF STUDIES OF THE UK MARKET.

Copies of the reports, which are still in draft form, were put on the
World-Wide web in January. Although the reports have never been
finalized, copies are now provided as Attachment 4. A bibliography of
studies of the UK market is provided as Attachment 5.

3. IN THE RESPONSE TO QUESTION 15 (P.2), LONDON ECONOMICS STATES THAT,
"LONDON ECONOMICS COMPARED THE RESULTS OF ITS ITERATIVE PX SIMULATION
MODEL TO THE UNIT COMMITMENT SCHEDULE PRODUCED BY A CONVENTIONAL
OPTIMIZING MODEL. THE RESULTS WERE BROADLY SIMILAR AND CONFIRMED THAT
THE ITERATIVE PX AUCTION WAS PRODUCING REASONABLY EFFICIENT PRICES
UNDER THE CONDITIONS SIMULATED." PLEASE SUPPLY THE REFERENCED STUDY.

London Economics compared the generation schedule derived of the
iterative auction with the schedule produced by a conventional pool
model in which the commitment order is based on a price that includes
an allocation of start-up costs and no load heat costs. The allocation
principles used for these short-run fixed costs were the same as those
used in the England and Wales Pool (E&W), which allocates start-up
costs over the expected output of a unit within a single peak. The
simulation differed from that used in the E&W Pool in so far as
transfers of no load costs from off-peak to peak prices was not

included. One should also note that the simulation testing in the
original London Economics work took a fixed hydro generation profile,
so the comparison with the conventional pool model did not examine the
way in which hydro generators would bid in the market.

For the day used in the auction simulation, the total cost difference
between the two approaches was 1.6%(1), and the units that were
committed were the same excepting one unit Hunters Point 4. On this
basis the outcomes were described as broadly similar. Table 1 shows the
unit commitment profiles for each simulation technique.

- --------

(1) The estimated costs for the two cases was $8.1m for the Auction
approach and $8.0m for the conventional pool model approach.

TABLE 1

HOURLY PX AUCTION PRICES IN EACH ITERATION

<Table>
<Caption>
1 2 3 4 5 6 7 8 9 10 11 12
----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- -----

Initial $20.7 $20.6 $20.6 $20.6 $20.7 $20.8 $21.4 $21.6 $22.2 $22.6 $23.2 $24.7
Iteration 1 $20.7 $20.6 $20.6 $20.6 $20.7 $20.8 $21.4 $21.6 $22.2 $22.6 $23.2 $24.7
Iteration 2 $20.7 $20.6 $20.6 $20.6 $20.7 $20.8 $21.4 $21.6 $22.2 $22.6 $23.2 $24.7
Iteration 3 $20.7 $20.6 $20.6 $20.6 $20.7 $20.8 $21.4 $21.6 $22.2 $22.6 $23.2 $24.7
Iteration 4 $20.7 $20.6 $20.6 $20.6 $20.7 $20.8 $21.4 $21.6 $22.2 $22.6 $23.2 $24.7
Iteration 5 $20.7 $20.6 $20.6 $20.6 $20.5 $20.7 $20.9 $21.5 $22.1 $22.1 $22.7 $23.9
Iteration 6 $20.7 $20.6 $20.6 $20.3 $20.3 $20.4 $20.8 $21.4 $21.9 $21.8 $22.4 $23.6
Iteration 7 $20.7 $20.6 $20.4 $20.1 $20.2 $20.3 $20.8 $21.2 $21.6 $21.7 $22.3 $23.4
Iteration 8 $20.7 $20.4 $20.3 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.1 $23.4
Iteration 9 $20.7 $20.3 $20.1 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $23.4
Iteration 10 $20.5 $20.1 $20.0 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $23.4
Iteration 11 $20.5 $20.1 $20.0 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $23.4
Iteration 12 $20.5 $20.1 $20.0 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $23.2
Iteration 13 $20.5 $20.1 $20.0 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $23.0
Iteration 14 $20.5 $20.1 $20.0 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $22.9
Iteration 15 $20.5 $20.1 $20.0 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $22.7
Iteration 16 $20.5 $20.1 $20.0 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $22.6
Final iteration $20.5 $20.1 $20.0 $20.0 $20.1 $20.2 $20.8 $21.1 $21.5 $21.6 $22.0 $22.6

13 14 15 16 17 18 19 20 21 22 23 24
----- ----- ----- ----- ----- ----- ----- ----- ----- ----- ----- -----

Initial $25.9 $28.6 $29.7 $38.4 $31.8 $28.0 $25.8 $23.6 $23.0 $22.4 $21.6 $21.3
Iteration 1 $25.9 $28.6 $29.7 $38.2 $31.8 $28.0 $25.8 $23.6 $23.0 $22.4 $21.6 $21.3
Iteration 2 $25.9 $28.6 $29.7 $38.2 $31.8 $28.0 $25.8 $23.6 $23.0 $22.4 $21.6 $21.3
Iteration 3 $25.9 $28.6 $29.7 $38.2 $31.8 $28.0 $25.2 $23.6 $23.0 $22.4 $21.6 $21.3
Iteration 4 $25.9 $28.6 $29.7 $38.2 $31.8 $28.0 $25.2 $23.6 $23.0 $22.4 $21.6 $21.3
Iteration 5 $25.2 $28.6 $29.7 $38.2 $31.8 $28.0 $24.9 $23.4 $22.6 $22.0 $21.4 $20.8
Iteration 6 $25.1 $28.6 $29.7 $38.2 $31.8 $28.0 $24.7 $23.2 $22.3 $21.7 $21.2 $20.7
Iteration 7 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.8 $22.2 $21.6 $21.1 $20.7
Iteration 8 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.6 $22.1 $21.5 $21.1 $20.7
Iteration 9 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.5 $22.0 $21.5 $21.1 $20.7
Iteration 10 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.4 $22.0 $21.5 $21.1 $20.7
Iteration 11 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.4 $22.0 $21.5 $21.1 $20.7
Iteration 12 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.4 $22.0 $21.5 $21.1 $20.7
Iteration 13 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.4 $22.0 $21.5 $21.1 $20.7
Iteration 14 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.4 $22.0 $21.5 $21.1 $20.7
Iteration 15 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.4 $22.0 $21.5 $21.1 $20.7
Iteration 16 $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.4 $22.0 $21.5 $21.1 $20.7
Final iteration $25.0 $28.6 $29.7 $38.2 $31.8 $28.0 $24.5 $22.4 $22.0 $21.5 $21.1 $20.7
</Table>

4. IN THE RESPONSE TO QUESTION 15 (P. 8), LONDON ECONOMICS STATES THAT,
"LONDON ECONOMICS FOUND THAT BIDDERS COULD RELIABLY INTERNALIZE THEIR
DAILY FIXED COSTS IN THEIR BIDS (PARTICULARLY THEIR START-UP COSTS)
USING INFORMATION READILY AVAILABLE FROM THE PX AND THEY COULD RAPIDLY
IDENTIFY THEIR PREFERRED FINAL BID." LONDON ECONOMICS CONCLUDES THAT
"THE PX OUTCOME WOULD BE INDISTINGUISHABLE FROM AN OPTIMIZED OUTCOME."
PLEASE PROVIDE THE RESULTS OF THE HOURLY PRICES IN THE ITERATIVE
AUCTION AND THE OPTIMIZED OUTCOME AND EXPLAIN THE BASIS FOR CONCLUDING
THAT THEY ARE INDISTINGUISHABLE.

The analysis on which London Economics' conclusions were based is
described in "PX Auction Testing. A report for the California PX
Restructuring Trust", London Economics, March 3, 1997. The analysis
included tests:

in which each supplier offered hourly bids that covered all
costs, if the generator bids were successful. That is, for the
peak hour of the day, the generator offers a price which would
cover the costs of one start, NLH and incremental costs in
just that hour. For the second highest hour the bidder would
offer a price that covered those same costs, but over two
rather than one hour. And so on until the lowest demand hour,
in which the bidder would offer a price that covered all daily
costs over a 24 hour period.

In other words, the bidders internalized their fixed costs based on
their expected duration of operation if the bidder was called to
operate in a particular hour. This form of bidding was investigated for
a number of day types including a summer peak working day and on types
of days in which the pattern of demand was irregular. The results of
the tests showed that by adopting this approach a bidder could:

o avoid premature withdrawal from the auction in circumstances
where the final PX price ends up above the level at which the
bidder would make operating profits in the day;

o avoid situations in which the bidder was forced to run at a
loss over the day; and

o manage technical constraints.

In the tests, bidders formulated their initial bids by reference to
their forecast of system demand; thus, if they expected the peak hourly
demand to occur between 1pm and 2pm, they would offer a bid in that
hour which covered their fuel costs and start costs in that hour.

However, if bidders were uncertain of the precise pattern of demand,
they could start the auction bidding at this price or above across a
range of hours, and refine their bids from information that was
released during the auction. The information needed to submit bids of
this form is readily available, namely:

o a demand forecast; and

o in each round of the auction the hourly PX prices from the
previous round.

The auction activity rules are key factors which enable bidders to
internalize their short-run fixed costs. In particular, the constraints
that bidders must participate in the initial round, that bids must
better the previous iteration PX price, that bidders can reduce their
bid price, and that withdrawals are irrevocable reduce the degree of
inter-iteration price volatility that would prevent bidders from
internalizing their short-run fixed costs.

Table 1 shows the PX prices on a summer working day in which thermal
bidders internalize their fixed costs in the manner described above,
and in which hydro generators operate as price takers. This is shown in
graph form in Figure 1.

FIGURE 1

[GRAPH]

The analysis presented in the report dated March 3, 1997 assumed that
hydro generation operated as a price taker, such that it determined its
preferred operating schedule and bid a zero price for its preferred
quantity in each hour. In addition, the analysis did not consider
generator ramp constraints, and did not explore the specific problems
of single station independent thermal generators operating at the
margin.

THE OPTIMIZED OUTCOME

London Economics did not repeat the analysis with an optimizing model,
not least because of the difficulty formulating the optimization. The
simulations are designed to test an auction in which individual bidders
maximize their own profit. This cannot be simply represented in a
conventional optimizing model. However, the results of the simulations
were optimal in the sense that, under the imposed assumptions, no
bidder could increase profits by changing its bid(2), and accordingly,
all bidders have maximized their profits (including those that have no
accepted bid, and therefore have zero profits).

5. IN THE STAGE I DESIGN FOR THE BALANCING MARKET, SETTLEMENT OF AS ENERGY
WILL BE BASED ON AN HOURLY AVERAGE EX POST PRICE INSTEAD OF A
FIVE-MINUTE PRICE.

(a) PLEASE EXPLAIN WHY SETTLEMENT OF AS ENERGY WILL BE BASED ON AN
HOURLY AVERAGE EX POST PRICE INSTEAD OF A FIVE MINUTE PRICE IN THE
STAGE I DESIGN FOR THE BALANCING MARKET.

The use of an hourly balancing market price for Ancillary Services
energy pricing and settlement is a staging issue, until the software
capability for pricing and settlement of five minute energy in the
balancing market is established.

(b) PLEASE PROVIDE ALL ANALYSIS DONE TO SUPPORT ANY MODIFICATIONS TO
THE ALTERNATIVE AS PROTOCOL TO MAKE IT COMPATIBLE WITH THE USE OF
HOURLY AVERAGE EX POST PRICES FOR AS ENERGY.

No quantitative modeling of the Ancillary Services protocol has been
conducted. The theoretical basis for the design of the London Economics
alternative protocol was explained by Professor Robert

- ----------

(2) This is strictly true only if the supply function is continuous.
However, within the limits of cost differences between adjacent bidders
in the merit order and any minimum bid change rule imposed on the
auction, no bidder can increase its profits.

Wilson in his paper Priority Pricing of Ancillary Services which was
attached to the ISO/PX Reply Comments as Attachment 1.

Ideally all AS energy, whatever the design of the protocol, would be
priced at a five minute price, since Ancillary Services are typically
employed on a sub-hourly basis. Consider, for example, a gas-fired
combustion turbine brought on to meet a short-term contingency. While
the turbine can run for a full hour, the ISO will take the turbine
off-line as soon as other thermal units can be ramped up to meet the
new net demand. If additional, even lower cost steam units with a
slower ramp rate are brought on after that, the effective balancing
market price would drop again. This progression across the hour is
illustrated in Figure 2 below. For prices to be completely reflective
of costs the prices must be set at sub-hourly intervals.

The average Balancing Market ("BM") price across the hour is shown as
the dashed line in the figure. Under the current software staging
proposal, the ancillary service bidders would be paid this average
hourly price, not the five minute prices. This raises several issues:

o for some bidders the balancing market price paid would be less
than their operating cost in the hour. In general, it is
important to ensure that the incentives on ancillary services
providers are compatible with their responsibilities to the
ISO to provide Energy as required on short notice. The use of
an hourly BM price will strengthen the need for a penalty
system to ensure that bidders do not renege on ancillary
services commitments;

o as generators will be paid the average hourly price, rather
than the five minute price, this can create a net surplus or
loss for a generator across the hour. In the deliberations
leading to the changes in the Ancillary Services Markets, it
was considered that Ancillary Services bidders would
internalize their expected profits or losses into their
reservation price bids, to the extent that they could be
predicted.

Figure 2 Ancillary services and balancing market prices

[GRAPH]

CONGESTION, ADJUSTMENT BIDS AND PX ZONAL PRICING

6. PLEASE RECONCILE THE APPARENT INCONSISTENCY BETWEEN ZONAL PRICING IN
THE PX AND UNIFORM PRICING FOR END USE CUSTOMERS. SPECIFICALLY, IN THE
COMMISSION'S NOVEMBER 26, 1996 ORDER AUTHORIZING THE ESTABLISHMENT OF
THE ISO AND PX, THE COMMISSION STATES THAT,

WE FIND THE COMPANIES ZONAL PRICING PROPOSAL UNCLEAR WITH
REGARD TO THE PRICES THAT BUYERS WOULD PAY FOR ENERGY FROM THE
PX. FOR EXAMPLE, IT IS NOT CLEAR WHETHER, DURING PERIODS OF
TRANSMISSION CONGESTION, THE COMPANIES PROPOSE THAT PX BUYERS
IN DIFFERENT ZONES WOULD PAY THE SAME PRICE OR DIFFERENT
PRICES FOR ENERGY. ON THE ONE HAND, THE COMPANIES'
MARKET-BASED RATES FILING SUGGEST THAT WHEN TRANSMISSION
CONGESTION EXISTS, PX BUYERS IN DIFFERENT ZONES WOULD PAY
DIFFERENT PRICES FOR ENERGY. ON THE OTHER HAND, THE SAME
FILING STATES THAT THE COMPANIES WILL AVERAGE THE COST OF
ENERGY AMONG THE CUSTOMERS THEY SERVE. (77 FERC P. 61,204)

THE CALIFORNIA PUBLIC UTILITIES COMMISSION ELECTRIC RESTRUCTURING ORDER
ISSUED ON DECEMBER 20, 1995 STATES THAT,

THE MARKET-CLEARING LOCATIONAL PRICES WILL BE OBTAINED FROM
THE ISO (BY A TIME CERTAIN) AS PART OF THE INTEGRATION AND
COORDINATION OF THE ALTERNATIVE NOMINATIONS AND BIDS. EVERY
WINNING GENERATION BIDDER WILL BE PAID THE MARKET-CLEARING
PRICE AT ITS LOCATION, . . . THE POWER EXCHANGE WILL AVERAGE
THE LOCATIONAL CLEARING PRICES: END USE CUSTOMERS SERVED BY
THE EXCHANGE WILL SEE ONE CLEARING PRICE. (CPUC DECEMBER 20,
1995 ORDER, P.50)

TWO ACTIVE ZONES HAVE BEEN DEFINED IN THE STATE OF CALIFORNIA. HOW WILL
A PUBLIC UTILITY THAT PURCHASES ENERGY THROUGH THE PX AND PAYS A PRICE
FOR WHOLESALE ENERGY THAT IS ABOVE THE AVERAGE RECONCILE THIS WITH THE
AVERAGE ENERGY PRICE THAT WILL BE COLLECTED FROM END USE CUSTOMERS?

The PX will not average the price across Zones and Scheduling Points.
In relation to the apparent inconsistency between zonal pricing in the
PX and uniform pricing for End Use Customers, PX customers in different
Zones will pay the applicable PX price for each individual Zone or
Scheduling Point. However, the Local Regulatory Authority determines
the price paid by End Use Customers of a regulated Utility Distribution
Company ("UDC") that purchases its Energy from the PX, and if the Local
Regulatory Authority requires the UDC to average the

price it paid the PX for Energy when charging its retail customers in
different Zones, the UDC's retail customers will see a single average
price. The imposition by the Local Regulatory Authority (CPUC, City
Council, etc.) of average pricing could blunt the locational dimension
of the PX marginal cost signal and may also encourage retail customers
located within the zone where the PX price is below the average price
charged by the UDC to change suppliers. However, the imposition of such
a requirement appears to be within the jurisdiction of the Local
Regulatory Authority (that is, the retail customer market).

With respect to the CPUC December 20, 1995 Order ("CPUC Order"), it
should be noted that AB 1890 was not yet in force. AB 1890 provides for
a rate freeze in California that will continue through to the earlier
of the end of 2001 or the recovery of stranded costs. For the duration
of that rate freeze, customers of UDCs will see a rate equal to the
present rate for each customer class. That is, they will not see the
effect of the different price paid for Energy by their UDC in the zone
in which they are located. However, the rate for each customer class
will include the PX price as an unbundled cost component. Retail
customers will have an opportunity to reduce their cost of Energy by
purchasing it from suppliers that have a price less than the PX price.

Also since the CPUC Order was issued on December 20, 1995, there has
been a small variation with respect to that part of the CPUC Order
which describes the ISO as setting the locational Market Clearing Price
("MCP"). Although the ISO manages Congestion, it will not, under the
present proposal, set the PX MCP for a Zone or for a scheduling point.
The ISO, through its Congestion Management protocols, will establish
the Usage Charge for each inter-zonal transmission path when there is
Congestion. The Usage Charge price information will, at the end of the
Congestion Management process, be conveyed by the ISO to all Scheduling
Coordinators (including the PX) using the congested path. The PX will
use this price information to establish and publish the price for each
Zone and each scheduling point. If the PX is using the congested
transmission path, the PX price for each Zone and Scheduling Point will
differ from the adjacent Zone or Scheduling Point by an amount equal to
the Usage Charge.

The CPUC has participated in the on-going activities that have led to
the current proposal, and indeed has encouraged the joint participation
activities of the Trust Advisory Committee and, therefore, has been a
party to the evolutionary changes described herein and in the Tariffs.

7. DO RETAIL CUSTOMERS WHO BUY ENERGY DIRECTLY THROUGH THE PX PAY PX ZONAL
PRICES OR THE AVERAGED RATE PAID BY END USE CUSTOMERS AS IDENTIFIED BY
THE CPUC IN ITS DECEMBER 20, 1995 ORDER?

Retail customers who buy Energy directly through the PX will pay the PX
zonal price.

8. ON THE JANUARY 1, 1998 EFFECTIVE DATE, WILL THE ISO ANNOUNCE A 5 MINUTE
ZONAL BALANCING MARKET PRICE OR SIMPLY AN HOURLY PRICE?

On January 1, 1998, the ISO will announce only an hourly zonal
balancing market price. The hourly price will be the Energy weighted
average of the twelve 5 minute prices. However, until the required
software is in place, payments will be calculated using only the hourly
price, and announcement of the 5 minute price is therefore unnecessary.

9. IN SECTION 7.2.5.2.2 OF THE ISO TARIFF IT STATES THAT, ". . . THE USAGE
CHARGES WILL BE CALCULATED AS THE MARGINAL VALUES OF THE CONGESTED
INTER-ZONAL INTERFACES. THE MARGINAL VALUE OF A CONGESTED INTER-ZONAL
INTERFACE IS CALCULATED BY THE ISO'S COMPUTER OPTIMIZATION ALGORITHM TO
EQUAL THE TOTAL CHANGE IN REDISPATCH COSTS (BASED ON THE ADJUSTMENT
BIDS) THAT WOULD RESULT IF THE INTERFACE'S SCHEDULING LIMIT WAS
INCREASED BY A SMALL INCREMENT."

Section 7.2.5.2.2 of the ISO Tariff explains the calculation of the
Usage Charges in the simple case of a radially connected zonal network.
The ISO Congestion Management optimization algorithm is designed for
the general case of a zonal network configuration that may include
loops. In this general case, the Usage Charge between any two Zones
will be calculated as the marginal cost of generating Energy in one of
the Zones and consuming the same amount of Energy in the other zone. In
the simple case of a radially connected zonal network, the Usage Charge
between two connected Zones is equal to the marginal cost of their
Inter-Zonal Interface. In the general case, the Usage Charge between
two connected Zones is equal to the marginal cost of their Inter-Zonal
Interface, multiplied by the percentage of the power transfer between
the Zones that flows on their Inter-Zonal Interface (as opposed to
loop-flow).

(a) EXPLAIN HOW REDISPATCH COSTS ARE CALCULATED. WHAT INFORMATION
IS USED? DESCRIBE HOW THE COMPUTER MODEL MAKES ITS
CALCULATIONS. WHAT IS THE COMPUTER MODEL ATTEMPTING TO
OPTIMIZE? WHAT CONSTRAINTS ARE USED IN THE OPTIMIZATION
CALCULATION?

The computer model calculates adjustments in the submitted
schedules to eliminate constraints on the Inter-Zonal
Interface. These adjustments minimize the total cost of
re-dispatching resources, as calculated by reference to the
respective Adjustment Bids. (The computer model is attempting
to optimize customer surplus.) The constraints used in the
optimization calculation are the following:

o the power balance equality constraints in
each node of the power systems network;

o the power balance equality constraints in
all Scheduling Coordinator portfolios,
except one that is arbitrarily taken as
reference;

o the Inter-Zonal Interface power flow limits
(inequality constraints).

(b) PLEASE PROVIDE AN EXAMPLE USING SEVERAL HYPOTHETICAL
SCHEDULING COORDINATORS, EACH SERVING MULTIPLE GENERATING
UNITS AND LOADS IN DIFFERENT ZONES, TO ILLUSTRATE HOW THE
USAGE CHARGE WOULD BE CALCULATED.

Please see the example set out in Attachment 6.

(c) IN THE DAY-AHEAD MARKET, WHAT IS THE RELATIONSHIP BETWEEN THE
USAGE CHARGE AND THE ZONAL DAY-AHEAD ENERGY PRICES IN THE PX.

Please see the example set out in Attachment 6.

10. WILL THE TRANSMISSION USAGE CHARGE FOR TRANSMISSION OVER CONGESTED
INTERFACES BETWEEN ZONES BE CALCULATED BY THE ISO FROM ADJUSTMENT BIDS
IN DIFFERENT ZONES THAT ARE CONTAINED WITHIN ONE PORTFOLIO OR FROM
ADJUSTMENT BIDS IN MULTIPLE PORTFOLIOS?

There appears to be confusion about portfolios and schedules. A
portfolio refers to bidding in the PX auction. A seller may offer a
block of Energy from a portfolio of generation it owns during the
auction without specifying the individual Generators that will produce
the Energy. However, the ISO is concerned with individual Generators
from a control standing point. Successful sellers in the PX auction
must identify the individual generators from their portfolio that will
produce the Energy at the end of the auction for inclusion by the PX in
the PX Balanced Schedule that is submitted to the ISO.

The ISO will calculate the Transmission Usage Charge between Zones
using Adjustment Bids submitted by each Scheduling Coordinator, but
will do so in such a manner as to maintain a balance of Load and
generation in each Scheduling Coordinator's schedule. This constraint
has been put in place so that the ISO does not impose trading between
Scheduling Coordinators.

11. CAN ANCILLARY SERVICE BIDS BE USED FOR CONGESTION MANAGEMENT?

Ancillary Services bids are for Capacity reservation in the forward
markets and Energy Dispatch in real time. Therefore, Ancillary Services
bids cannot be used for Congestion Management. Energy from committed
Ancillary Services capacity, and Supplemental Energy bids can, however,
be used to alleviate Congestion in real time.

12. CAN CONGESTION MANAGEMENT BIDS BE USED FOR ANCILLARY SERVICES?

No. Adjustment Bids submitted in the forward markets are used only for
Congestion Management in the respective forward market.

13. CAN ANCILLARY SERVICES BIDS AND ADJUSTMENT BIDS BE SUBMITTED FOR THE
SAME CAPACITY?

No. A Scheduling Coordinator will not know when it submits Adjustment
Bids whether its units/resources may also be selected to provide
Ancillary Services. Therefore, if it offered the same capacity in both
markets it would risk being unable to fulfill its obligation when
called upon.

14. WHO OWNS THE INTERFACE BETWEEN THE TWO ACTIVE ZONES DEFINED IN THE ISO
AND PX MARKETS?

PG&E owns the interface between the two active Zones. As the interface
is part of the Pacific AC Intertie, other ISO participating and
non-participating parties have contractual rights to use transmission
capacity over that interface.

15. WILL THE REVENUE COLLECTED BY THE PX THROUGH ITS ZONAL PRICING
MECHANISM ALWAYS BE EQUAL TO THE USAGE CHARGE THAT MUST BE PAID TO THE
ISO? PLEASE EXPLAIN WHY. (SCE IN ITS JULY 8, 1997 COMMENTS SUGGESTS
THAT THERE ARE CIRCUMSTANCES UNDER WHICH THEY WILL NOT BE EQUAL.) HOW
WILL THE PX DEAL WITH ANY REVENUE DIFFERENCES?

Please see the example set out in Attachment 6.

16. HOW WILL THE PX CALCULATE ZONAL PRICES WHEN THE ISO DETERMINES THAT
THERE IS INTER-ZONAL CONGESTION AND SETS A POSITIVE USAGE CHARGE?

The PX will calculate zonal prices as set out in the example
calculation of Usage Charge in Attachment 6.

17. WILL THE PX USE ONLY ADJUSTMENT BIDS --- OR THE BIDS ORIGINALLY
SUBMITTED TO THE PX USED TO DEVELOP THE PX'S PREFERRED SCHEDULE --- TO
SET PX ZONAL PRICES? PROVIDE AN EXAMPLE.

When there is no congestion, the PX will use only bids originally
submitted to the PX to set PX zonal prices. When there is congestion,
the price will be based on Adjustment Bids. In the example calculation
of Usage Charge in Attachment 6, the prices shown illustrate Adjustment
Bids.

18. PLEASE PROVIDE A DETAILED EXAMPLE THAT SHOWS THE CALCULATION OF THE
ACCESS CHARGE USING ADJUSTMENT BIDS FROM BOTH THE PX AND OTHER
SCHEDULING COORDINATORS AND THE SUBSEQUENT DETERMINATION OF PX PRICES
USING THE ACCESS FEE AND PX ADJUSTMENT BIDS.

(a) PROVIDE AN EXAMPLE FOR THE CASE IN WHICH NOT ALL PX GENERATORS
SUBMIT ADJUSTMENT BIDS.

(b) PROVIDE AN EXAMPLE FOR THE CASE THAT EVERY PX GENERATOR
SUBMITS AN ADJUSTMENT BID.

There appears to be confusion regarding Access Charge and Usage
Charges. The Access Charge is based on the transmission revenue
requirement of Participating TOs. The calculation of the Access Charge
is not directly related to Adjustment Bids from either the PX or other
Scheduling Coordinators. There is an indirect relationship because
congestion revenues resulting from the Usage Charges, are in the case
of the Participating TOs, credited against the transmission revenue
requirements. An example that illustrates the relationship would be
quite complex. However, the congestion revenue resulting from the
example calculation of Usage Charge in Attachment 6 example 1 would be
150 MW x $10/MW = $1500 during the hour covered by the example.
Congestion revenues from Usage Charges are credited monthly to a
utility's TRBA and will be reflected in access charges in the following
calendar year pursuant to Section 5.5 of the TO Tariff. The Access
Charge will recover that portion of the Participating TO's Transmission
Revenue Requirement not recovered through the Usage Charge.

19. REFERENCE SECTION 7.2.7 OF THE ISO TARIFF

THIS TARIFF SECTION ADDRESSES ZONES WITHIN THE ISO GRID. HAS EITHER THE
ISO OR SDG&E PERFORMED AN ANALYSIS COMPARABLE TO THAT DESCRIBED IN THE
REFERENCED SECTION OF THE ISO TARIFF WITH REGARD TO THE ESTABLISHMENT
OF A SEPARATE CONGESTION ZONE FOR THE SAN DIEGO BASIN? IF SO, PLEASE
PROVIDE A COPY OF ANY SUCH ANALYSIS, INCLUDING ANY WORK PAPERS THAT
SUPPORT THE ANALYSIS. IF NO ANALYSIS HAS BEEN PERFORMED, PLEASE EXPLAIN
WHY SUCH AN ANALYSIS IS NOT NECESSARY BASED UPON TRANSMISSION
CONGESTION INTO THE SAN DIEGO BASIN DURING CERTAIN HOURS.

San Diego Gas & Electric Company is providing the response to this
question.

20. REFERENCE SECTION 7.3.1.3 OF THE ISO TARIFF

PLEASE DEFINE WHAT IS MEANT BY AN "UNUSABLE" ADJUSTMENT BID. IN
ADDITION, PLEASE IDENTIFY AND DESCRIBE THE CIRCUMSTANCES UNDER WHICH A
BID WILL BE DEEMED TO BE UNUSABLE.

Section 7.3.1.3 of the ISO Tariff refers to "inadequate or unusable"
Adjustment Bids. An Adjustment Bid would be deemed unusable when the
Congestion Management algorithm determines that the congestion cannot
be relieved by a change in Energy output of the Load or generation
submitted. For instance, if a transmission path is congested in the
northbound direction, an offer by Load in the southern area to decrease
consumption will not be usable because it will not help reduce
congestion. The purpose of this section is to provide a means to set a
Usage Charge in the event the ISO is not provided with adequate
Adjustment Bids to solve the problem.

21. IS THE PX CONSIDERED ONE LARGE PORTFOLIO OR CAN THERE BE MULTIPLE
PORTFOLIOS WITHIN THE PX?

Please refer to the response to Question 10 above. The PX is not a
"portfolio", but PX Participants bidding in to the PX may submit
portfolio bids. At the end of the auction, the PX will require bidders
to convert their portfolio bids into unit schedules before the
Preferred Schedules are submitted to the ISO. See Section 3.3.2 of the
PX Tariff. Individual suppliers bidding into the PX will be permitted
to submit multiple portfolios.

22. IF ANY SCHEDULING COORDINATOR MANAGES MULTIPLE PORTFOLIOS, MUST THE
INDIVIDUAL PORTFOLIOS REMAIN BALANCED AFTER THE EXECUTION OF ADJUSTMENT
BIDS BY THE ISO TO MANAGE CONGESTION?

The Scheduling Coordinator must submit a Balanced Schedule to the ISO
regardless of the number of "portfolios" it has. Congestion Management
will be done on these Balanced Schedules, not the "portfolios" within
the Balanced Schedules.

23. CAN A PARTICIPANT WITH A SINGLE GENERATING UNIT SUBMIT A SUPPLY
SCHEDULE OF PRICE/QUANTITY INFORMATION AS AN ADJUSTMENT BID THAT IS
DIFFERENT FROM THE SUPPLY SCHEDULE OF PRICE/QUANTITY INFORMATION
SUBMITTED IN THE PX ENERGY AUCTION USED TO DEVELOP THE PX PREFERRED
SCHEDULE? IF NOT, WHY DOES THE PARTICIPANT NEED TO SUBMIT AN ADJUSTMENT
BID?

CAN A PARTICIPANT WITH SEVERAL GENERATING UNITS SUBMIT SUPPLY SCHEDULES
OF PRICE/QUANTITY INFORMATION AS ADJUSTMENT BIDS THAT ARE DIFFERENT
FROM THE PORTFOLIO SUPPLY SCHEDULE OF PRICE/QUANTITY INFORMATION
SUBMITTED IN THE PX ENERGY AUCTION USED TO DEVELOP THE PX PREFERRED
SCHEDULE?

The answer to both questions is yes. The PX is designed to give buyers
and sellers flexibility in this regard when trading in the PX. However,
the Adjustment Bids must be structured relative to the PX Market
Clearing Price.

ANCILLARY SERVICES

24. WHAT CRITERIA (ECONOMIC AND RELIABILITY) WILL BE USED TO CHOOSE THE
LOCATION OF SPINNING, NON-SPINNING AND REPLACEMENT RESERVES IN SEPARATE
ZONES WHEN THERE ARE TRANSMISSION CONSTRAINTS?

So far as concerns economic criteria, the ISO will purchase spinning,
non-spinning and replacement reserve capacity from the cheapest
available sources. See Section 2.5.8 ISO Tariff. Thus, when there are
transmission constraints, the ISO will take into account Usage Charges
when evaluating bids for these services in its Ancillary Services
Markets. See Section 2.5.4 ISO Tariff and the further detail for each
of those Services in Sections 2.5.15, 2.5.16 and 2.5.17. The ISO is
developing protocols governing Congestion Management which will also
deal with Ancillary Services. As will be seen from the proposed
amendments to Section 2.5.4 now set forth in the restated ISO Tariff
filed contemporaneously with these responses, the computer software

will not be in place by January 1, 1998, to allow the ISO to reserve
transmission capacity for Ancillary Services. Pending introduction of
the relevant software, therefore, when congestion is present, the ISO
will purchase the relevant capacity from the cheapest source available
within the relevant Zone.

So far as concerns reliability criteria, the ISO will purchase
spinning, non-spinning and replacement reserve capacity from sources
which will allow the ISO to meet the WSCC requirements and
contingencies on the ISO Controlled Grid in the hours concerned. The
ISO is currently developing the protocols further defining these
technical criteria. As stated, in Section 2.5.4 of the ISO Tariff, the
actual location of these resources will depend, among other things,
upon the available transmission capacity, the locational mix of
generation and historical patterns of transmission and generation
availability as well as the locational spread of demand. When
Congestion is present, therefore, the ISO will take into account in
deciding upon locational quantities of these reserve services, not only
the inter-zonal congestion, but also any forecast Intra-Zonal
Congestion caused, for example, by planned transmission or generation
outages. The ISO will be able to make use of historical data on the
operation of the various parts of the ISO Controlled Grid received from
the three IOUs which will be placing transmission assets under the
control of the ISO. As the ISO's experience of the operation of the ISO
Controlled Grid grows, it will be able to develop historical patterns
and guidelines which will enable it to select the locational quantities
of reserve.

25. WHEN WILL THE TECHNICAL, LEGAL, AND CONTRACTUAL REQUIREMENTS NECESSARY
FOR PURCHASES OF ANCILLARY SERVICES FROM RESOURCES OUTSIDE OF THE ISO
CONTROLLED GRID BE AVAILABLE?

It is anticipated that the appropriate arrangements will be available
by November 1, 1997.

26. PLEASE DESCRIBE THE PRINCIPLES OR PROCEDURES FOR HOW A REACTIVE POWER
CHARGE WILL BE ASSESSED ON A WHEELING TRANSACTION.

There are no current plans to levy a charge for reactive power for a
wheeling transaction. If the ISO Governing Board decides to do so,
after January 1, 1998, the ISO will seek authorization from the

Commission, in which case a Federal Power Act Section 205 filing will
be made.

27. PLEASE CLARIFY THE CIRCUMSTANCES UNDER WHICH THE RESERVATION COSTS OF
REPLACEMENT RESERVES WILL BE RECOVERED THROUGH IMBALANCE ENERGY.

The Replacement Reserve capacity reservation payments in the forward
markets are not going to be recovered through Imbalance Energy charges.
They will be allocated to Scheduling Coordinators according to their
respective non self provided Replacement Reserve requirements, taking
into account Imbalance Energy deviations as follows: The capacity
reservation cost for the portion of Replacement Reserves that are
dispatched in real time will be allocated to Scheduling Coordinators
according to their respective positive Imbalance Energy deviations,
multiplied by their respective non self provided Replacement Reserve
requirements. The cost for the portion of Replacement Reserves that are
not dispatched in real time will be allocated to Scheduling
Coordinators according to their respective non-self provided
Replacement Reserve requirements. The Replacement Reserve requirements
are allocated to Scheduling Coordinators according to their respective
scheduled Demand in each of the forward markets. Positive Imbalance
Energy deviations are due to either less than scheduled generation or
more than scheduled Demand in real time.

28. WHEN WILL THE ISO TARIFF PROVISIONS RESPECTING VOLTAGE SUPPORT BE
CLARIFIED AND WHEN WILL THE PROTOCOLS FOR THE TREATMENT OF VOLTAGE
SUPPORT BE COMPLETED?

The restated version of Section 2.5 of the ISO Tariff filed
contemporaneously with these responses provides the amendments to the
sections regarding Voltage Support (2.5.3.4 and 2.5.18) anticipated in
the ISO's June 23 1997 Reply Comments at pp. 193 to 197. It is
anticipated that protocols for the treatment of Voltage Support will be
completed by November 1, 1997.

29. SECTION 2.5.21 OF THE ISO TARIFF STATES:

SCHEDULING COORDINATORS SHALL NOTIFY THE ISO AT LEAST TWO
HOURS PRIOR TO THE OPERATING HOUR THE SPECIFIC IDENTITY OF
GENERATING UNITS AND OTHER RESOURCES SELECTED TO PROVIDE
REGULATION

THIS STATEMENT APPEARS TO CONTRADICT OTHER PARTS OF THE TARIFF. IF
SECTION 2.5.21 APPLIES TO THE DAY AHEAD MARKET THEN IT IS INCONSISTENT

WITH SECTION 2.5.10.1 WHICH STATES THAT BIDS MUST BE SUBMITTED TO THE
ISO BY 10:00 ON THE DAY PRIOR TO TRADING AND SECTION 2.5.14 WHICH
STATES THAT THE BID INFORMATION MUST CONTAIN THE NAME AND LOCATION OF
THE RESOURCE. IF IT APPLIES TO THE HOUR AHEAD MARKET THEN IT IS
INCONSISTENT WITH SECTION 2.5.10.2, WHICH SAYS THAT BIDS FOR HOUR AHEAD
REGULATION MUST BE RECEIVED ONE HOUR PRIOR TO OPERATION. PLEASE
RECONCILE THESE APPARENT INCONSISTENCIES.

The Restated ISO Tariff filed contemporaneously with these responses
provides amendments to Section 2.5.21 to delete the sentence referred
to and the following sentence. The information required will be
provided to the ISO pursuant to Sections 2.5.20.4 and 2.5.20.5 of the
ISO Tariff.

30. IT IS NOT CLEAR FROM THE TRUSTEE'S PROPOSAL WHETHER REAL TIME
IMBALANCES FOR A PORTFOLIO BIDDER ARE DETERMINED ON A GENERATING
UNIT-SPECIFIC BASIS OR ON AN AGGREGATE PORTFOLIO BASIS. PLEASE INDICATE
WHICH BASIS IS USED TO CALCULATE REAL TIME IMBALANCES.

The information used by the ISO to calculate real time imbalances is
derived from Meter Data obtained pursuant to Section 10, but the
imbalances are allocated by the ISO to Scheduling Coordinators rather
than to specific Generating Units, Loads, or Demand take-out points. It
will be up to Scheduling Coordinators to determine how they wish to
allocate real time imbalances among their participants.

31. IN ORDER TO MINIMIZE ANCILLARY SERVICE COSTS TO USERS OF THE ISO GRID,
SECTION 2.5.12 OF THE ISO TARIFF STATES THAT "THE ISO SHALL SELECT THE
BIDDERS WITH LOWEST BIDS WHICH MEET ITS TECHNICAL REQUIREMENTS,
INCLUDING LOCATION AND OPERATING CAPABILITY." IN ATTACHMENT IV, P.8,
ROBERT WILSON STATES THAT, "THE CONTRAST BETWEEN THE OLD AND NEW BID
EVALUATION RULES SHOW THAT THE CORRECT RULE DEPENDS ON THE DESIGN
OBJECTIVE. IF THE OBJECTIVE IS TO MINIMIZE THE ISO'S COSTS THEN
SOMETHING LIKE THE OLD RULE IS REQUIRED. IF THE OBJECTIVE IS TO PROMOTE
THE EFFICIENCY OF THE MARKETS OVERALL THEN THE NEW RULE IS SUFFICIENT.
THE ADOPTION OF THE NEW RULE ENDORSED THE EFFICIENCY OBJECTIVE RATHER
THAN THE COST-MINIMIZATION OBJECTIVE."

(a) IS THE ISO ABANDONING THE COST-MINIMIZATION OBJECTIVE?

(b) HOW DO THE NEW BID EVALUATION RULES PROMOTE EFFICIENCY OF THE
MARKETS OVERALL?

The ISO is not abandoning the "cost minimization object". The objective
function of the ISO's Ancillary Services markets is to procure the
lowest cost capacity offered to meet the ISO's technical requirements
and this can be achieved by its Day-Ahead and Hour-Ahead auctions.

Robert Wilson's paper explains how the revised approach to the
evaluation of Ancillary Services bids promotes efficiency of the Energy
and Ancillary Services markets overall. He explains that if the ISO
were to evaluate bids based on a combination of the prices for capacity
and Energy (as under the old system), then some additional suppliers
with low marginal costs would be attracted away from the Energy
markets. This would reduce the ISO's costs but increase prices in the
Energy markets. Where the new approach to the evaluation of Ancillary
Services is used, bidders' optimal strategy will be to offer Ancillary
Service bids which reflect the opportunity costs of not offering
capacity, which is not sold to the Energy market, into other markets.
Those suppliers who are marginal in the Energy markets have the lowest
opportunity costs. Thus a supplier who is less likely to succeed in the
Energy market is more likely to be successful in the Ancillary Service
markets. The result should be that the capacity reservations accepted
in the Ancillary Services markets are from precisely those suppliers
whose resources are less valuable in other markets, thus promoting
overall market efficiency.

Market efficiency and cost minimization are not inconsistent. They are
complementary. The ISO believes, however, that what Robert Wilson was
saying in the Attachment IV was that cost minimization of Ancillary
Services to the ISO is not necessarily consistent with cost
minimization of generation, transmission and distribution of
electricity to customers. Although the new approach to the evaluation
of Ancillary Service bids may not necessarily minimize the overall cost
of Ancillary Services (including the Energy component) to the ISO, it
does encourage and promote overall cost minimization across all markets
for Energy and Ancillary Services.

32. CAN A GIVEN RESOURCE BID DIFFERENT RESERVATION PRICES, R, AND DIFFERENT
ENERGY PRICES, P, INTO EACH OF THE ANCILLARY SERVICES MARKETS DESCRIBED
IN THE ISO/PX JUNE 23, 1997, COMMENTS?

The answer to this question is yes. Section 2.5.13 of the ISO Tariff
has been amended to make this clear. The proposed amendment can be
found in the Restated ISO Tariff filed contemporaneously with these
responses.

33. CAN A MARKET PARTICIPANT THAT OPERATES IN THE CALIFORNIA MARKET AS A
METERED SUBSYSTEM (MSS) SELF-PROVIDE ANCILLARY SERVICES EFFECTIVE
JANUARY 1, 1998?

34. IN THE PHASE II FILING SUBMITTED TO THE COMMISSION ON MARCH 31, 1997,
THE MARKET CLEARING PRICE FOR SPINNING RESERVE (SECTION 2.5.15 OF THE
ISO TARIFF) IS CALCULATED AS,

PSP(IJT) = MCP (XT)- ENBID(ITT)

WHERE PSP(IJT) IS THE PRICE PAID TO SCHEDULING COORDINATORS BY THE ISO
FOR SPINNING RESERVE, MCP(XT) IS THE MARKET CLEARING TOTAL BID PRICE, X
IS THE WEIGHTING FACTOR THAT REFLECTS THE PROBABILITY OF USING SPINNING
RESERVE, X AS A SUBSCRIPT INDEXES THE ZONE, AND ENBIDIJT IS THE BID
PRICE OF GENERATION FROM RESERVED CAPACITY. EXPLAIN HOW THE VALUE OF X,
THE WEIGHTING FACTOR THAT REFLECTS THE PROBABILITY OF USING SPINNING
RESERVE, IS DETERMINED.

Under the revised approach to the evaluation and pricing of Ancillary
Services, there is no longer any need for the probability factor x.
Section 2.5.15 of the Restated ISO Tariff filed contemporaneously with
these responses shows how the Market Clearing Price for Spinning
Reserve capacity is determined.

SUPPLEMENTAL ENERGY BIDS

35. IN SECTION 2.5.22.4 OF THE ISO TARIFF THERE IS A DISCUSSION OF
SUPPLEMENTAL ENERGY BIDS.

(a) HOW DO SUPPLEMENTAL ENERGY BIDS INTERACT WITH ADJUSTMENT BIDS?

Adjustment Bids are used to adjust schedules in the Day-Ahead and
Hour-Ahead Markets for Congestion Management purposes. Supplemental
Energy bids are bids for an increase (or decrease) in output in real
time. An Adjustment Bid becomes a Supplemental Energy bid if it is left
standing after the Hour-Ahead Market has closed and it contains the
information required for Supplemental Energy bids, e.g., ramp rates. An
Adjustment Bid and a Supplemental Energy bid do not coexist
simultaneously.

(b) WHO CAN SUBMIT A SUPPLEMENTAL ENERGY BID?

Any Scheduling Coordinator in possession of a current Ancillary
Services certificate for the resource concerned can submit a
Supplemental Energy bid for the increase or decrease in output or
Demand from that resource.

(c) IF AN ADJUSTMENT OR ANCILLARY SERVICES BID WAS SUBMITTED FOR A
RESOURCE, BUT THAT RESOURCE (OR PART OF THAT RESOURCE) WAS NOT
SCHEDULED IN THE HOUR AHEAD MARKET, CAN A SUPPLEMENTAL ENERGY
BID BE SUBMITTED FOR THAT RESOURCE?

Yes. A Supplemental Energy bid may be submitted for a resource (or part
of a resource) for which an Adjustment Bid or Ancillary Service Bid has
not been Scheduled.

36. DURING THE INITIAL STAGE OF OPERATION OF THE ISO THERE WILL BE NO HOUR
AHEAD MARKETS FOR ENERGY, CONGESTION MANAGEMENT OR ANCILLARY SERVICES.
WILL GENERATORS BE ABLE TO SUBMIT SUPPLEMENTAL ENERGY BIDS 30 MINUTES
BEFORE THE HOUR DURING THE INITIAL STAGE?

Unfortunately, this is an incorrect statement. The ISO will run an hour
ahead Ancillary Services market. The ISO does not run an Energy market,
but will be able to carry out Congestion Management in the hour ahead
scheduling process. However, due to staging the PX will not have an
hour ahead market. Scheduling Coordinators will be able to

submit Supplemental Energy bids 30 minutes before the hour during the
initial stage.

TRANSMISSION AND METERING

37. REFERENCE SECTION 7.1.3 OF THE ISO TARIFF

PLEASE EXPLAIN WHETHER PARTICIPATING TRANSMISSION OWNERS WILL RECEIVE
CREDIT UNDER THE PROPOSED SELF-SUFFICIENCY TEST FOR TRANSMISSION
FACILITY INVESTMENTS THAT ARE NOT DIRECTLY CONNECTED TO THE LOAD THAT
ENTITY SERVES. FOR EXAMPLE, UNDER THE PROPOSED SELF-SUFFICIENCY TEST,
WILL TRANSMISSION RIGHTS (E.G., ENTITLEMENTS TO THE CAPACITY OF THE
CALIFORNIA OREGON TRANSMISSION PROJECT) BE INCLUDED IN THE
DETERMINATION OF EACH PARTICIPATING ENTITY'S FIRM IMPORT
INTERCONNECTION TRANSMISSION CAPACITY (FIITC)? IF NOT, PLEASE EXPLAIN
THE RATIONALE FOR EXCLUDING SUCH CAPACITY FROM THE SELF-SUFFICIENCY
TEST DETERMINATION.

Page 220 of the Reply Comments stated "FIITC is the transmission import
capacity directly connected to the TO's system" (2nd paragraph). The
requirement that the import capacity be directly connected to the TO's
system was objected to by various parties. However, the question refers
to capacity which is not directly connected to "load" served by the TO,
rather than capacity not directly connected to the TO's system.

The ISO Tariff does not impose any such restriction in relation to
FIITC (although it does for Dependable Generation). The definition of
FIITC refers to firm transmission capacity associated with transmission
facilities owned by a Participating TO "or contracted to the
Participating TO under an Existing Contract which allows Generating
Units that are not directly interconnected to the ISO Controlled Grid
to deliver Energy to that Participating TO." Perhaps this has been
interpreted as requiring the import capacity to be directly connected
to the TOs system in order to be able to say that the Energy is
delivered to a Participating TO.

The Companies' Joint Answer to Motions to Intervene, Comments, and
Protests Filed on June 6, 1997, dated June 23, 1997, page 27, "The
Self-Sufficiency Test Includes The Appropriate Transmission Resources"
suggests that the Self-Sufficiency Test is designed to determine if a
Participating TO (PTO) has sufficient transmission resources to deliver
its generation resources (including reserves) to its own load. They
argue that if the transmission resources do not connect - either
directly or via an Existing Contract - to the Load that a PTO "serves",
then the generation resources associated with such

transmission cannot serve that Load without additional transmission
facilities. Therefore, such transmission capacity does not count as
FIITC under the Self-Sufficiency Test.

This proposition can be demonstrated by the example of a PTO that has
100 MW of transmission service on the COTP, and has an Existing
Contract to bring 75 MW of that capacity "home" to the Load it serves.
Under this example, the PTO would receive credit for 75 MW of FIITC in
the Self-Sufficiency Test.

According to the Companies, the reason that the PTO in the above
example would get credit for 75 MW (and not 100 MW) is that the PTO can
only serve 75 MW of its Load without depending on transmission
facilities owned by another utility. The TO would have to purchase
transmission service to bring the additional 25 MW home to serve its
Load.

Ultimately, the issue is one of equity and the avoidance of cost
shifting. The objective is that a PTO should be no better and no worse
off in relation to its transmission rights as a result of the creation
of the ISO Controlled Grid. The benefits of the Californian
restructuring should flow from the introduction of markets and not from
the reallocation of costs and benefits arising from transmission
rights. However, there are a large number of different circumstances
(including the varied nature of the Existing Contracts) and it is not
easy to design a single "one-size-fits-all" Self-Sufficiency rule that
satisfies everyone. Doubtless, the parties will express their views on
this subject to the Commission, and the Commission will decide. The ISO
Governing Board has taken no particular position on this subject.

38. APPENDIX B OF THE TRANSMISSION CONTROL AGREEMENT CONTAINS A PRELIMINARY
LIST OF ENCUMBRANCES (EXISTING CONTRACTS) ON THE TRANSMISSION
FACILITIES THAT PG&E, SCE AND SDG&E ARE PLACING UNDER THE ISO'S
OPERATIONAL CONTROL. PLEASE PROVIDE A FINAL COMPREHENSIVE LIST OF ANY
ADDITIONAL ENCUMBRANCES.

Responses to this inquiry are being provided by PG&E, SCE and SDG&E
under separate cover.

39. PLEASE DELINEATE, IN AS MUCH DETAIL AS POSSIBLE, THE AMOUNT OF
TRANSMISSION CAPACITY THAT, (1) WILL BE RESERVED TO ACCOMMODATE
EXISTING CONTRACTS, AND (2) WILL BE AVAILABLE FOR USE BY THE PX AND
OTHER SCHEDULING COORDINATORS.

Responses to this inquiry are being provided by PG&E, SCE and SDG&E
under separate cover.

40. AS OF JANUARY 1, 1999, INDIVIDUAL CUSTOMERS WILL BE PERMITTED TO USE
MORE THAN ONE SCHEDULING COORDINATOR. HOW IS IT POSSIBLE FOR MULTIPLE
SCHEDULING COORDINATORS TO BE RESPONSIBLE FOR TRADES THROUGH A SINGLE
METER?

The Reply Comments stated (at page 258) that:

Originally, it had been thought that the software could not
deal with more than one Scheduling Coordinator being
responsible for trades through the same meter. It may,
however, be possible for Scheduling Coordinators to develop
rules for making such allocations and to notify these to the
ISO.

Initially, the ISO will only deal with one Scheduling Coordinator in
respect of each meter. The relevant Scheduling Coordinators, on the
basis of rules to be developed by the Scheduling Coordinators will
nominate the responsible Scheduling Coordinator for allocating trades
through a single meter. The ISO will not be concerned with the
development or application of those rules, which will be a matter for
the Scheduling Coordinators to decide as the rules will relate to the
trading they wish to undertake between themselves.

At present, there is no fixed date on which the "one Scheduling
Coordinator per customer rule" will be changed. The Reply Comments
stated that the changes will be made as soon as the software permits,
which will probably be after the first year of operations. While it is
now expected that the software will be able to deal with multiple
Scheduling Coordinators and that the Scheduling Coordinators will be
able to develop rules for the allocation of trades through a single
meter, if these goals can not be met then it is unlikely that the `one
Scheduling Coordinator per customer rule' will be changed.

MARKET POWER AND MUST-RUN CONTRACTS

41. IF THE COMMISSION FINDS THAT THE APPLICANTS' MUST-RUN PROPOSAL DOES NOT
ADEQUATELY ADDRESS MARKET POWER CONCERNS IN THE SAN DIEGO BASIN, DO
ALTERNATIVE MARKET POWER MITIGATION PLANS EXIST? PLEASE DISCUSS ANY
SUCH ALTERNATIVES THOROUGHLY.

SDG&E is responding to this question separately.

42. PLEASE EXPLAIN HOW THE FIXED COSTS RECOVERED UNDER EACH OF THE THREE
PROPOSED MUST-RUN AGREEMENTS WILL AFFECT THE RECOVERY OF STRANDED COSTS
(UNDER THE CBC) BY PG&E, SCE, AND SDG&E STRANDED COST RECOVERY UNDER
THE CBC. PROVIDE SAMPLE CALCULATIONS USING EACH OF THE THREE MUST-RUN
OPTIONS FOR EACH OF THE UTILITIES.

Responses to this inquiry are being provided by PG&E, SCE and SDG&E
under separate cover.

43. THE MONITORING PLAN SUBMITTED AS APPENDIX 7 OF THE PHASE II FILING
PROVIDES A LISTING OF INFORMATION THAT WILL BE COLLECTED BY THE
COMPLIANCE DIVISIONS. ONLY VERY GENERAL PRINCIPLES ARE PROVIDED
REGARDING THE CRITERIA THAT WILL BE USED TO IDENTIFY AN EXERCISE OF
MARKET POWER (APPENDIX 7 P. 13-14). PLEASE SPECIFY WHAT CRITERIA WILL
BE USED TO IDENTIFY AN EXERCISE OF MARKET POWER.

Generally, market power is the ability to profitably sustain an
increase in the market price. Several ways a Scheduling Coordinator may
be able to do this in California include strategies, such as
strategically withholding capacity causing higher priced generation to
set the market price or scheduling generation in such a way as to cause
a transmission constraint that forces higher zonal prices and allows
the offending Scheduling Coordinators to profit. In addition, some have
also argued that the California IOUs will have an incentive to employ
predatory pricing strategies causing market prices to fall in the near
term and CTC payments to increase. The ISO and PX are committed to
identifying all such practices and developing specific criteria and
mitigation strategies as appropriate.

As the compliance divisions are formed, they will develop specific
guidelines and criteria for identifying market power in their
respective markets. Two principle areas where specific criteria will be
developed include:

(a) Bidding strategies of generation plant availability. In order
to monitor and identify strategic withholding of generation
capacity from the various markets, the compliance divisions
will monitor both the bidding strategies and dispatch patterns
of generators. Appendix 7 of the Phase II filing contains
several indices (please refer to ISO Tariff Appendix 7, pages
21-22 that was filed March 31, 1997) that the compliance
divisions will develop in order to identify potential
withdrawal strategies including:

o Comparing the amount of capacity a participant bids
into a market to the total capacity registered to the

participant. The compliance divisions' objectives
will be to identify if a particular entity is
withholding capacity from one or more of the markets.

o Identifying patterns where lower priced generation is
withdrawn and higher priced generation is offered.

(b) Bid pricing strategies. While a specific percentage or
threshold has not been set, the compliance divisions will
monitor for large fluctuations in bids for specific resources
in each of the markets. Emphasis will be on detecting large
price swings when congestion is present. The compliance
divisions will be able to use this information as a possible
trigger to the need for further investigation. For instance,
the ISO compliance division would be able to assess if a
particular generation has been able to or has the potential to
exercise market power due to a local transmission constraint
which may warrant the use of a must run contract or similar
call contract.

We wish to assure the Commission that the ISO and PX take this
responsibility most seriously. The ISO and PX are in the process of
forming the compliance divisions. Personnel with the proper skills and
experience are being recruited so that they may address the market
power issues related to their respective markets.

44. THE MITIGATION SECTION OF APPENDIX 7, P.16, STATES THAT "MANY PRACTICES
THAT MIGHT BE VIEWED AS ABERRANT OR AS GAMING THE MARKET STRUCTURE
WOULD NOT NECESSARILY BE ILLEGAL OR VIOLATE EXISTING LAWS, SUCH AS THE
ANTITRUST LAWS. HENCE, IT IS NOT PROPOSED TO PUNISH AGGRESSIVE
COMPETITORS FOR PRACTICES THAT TAKE ADVANTAGE OF WHAT MIGHT IN DUE
COURSE BE REVEALED AS DESIGN FLAWS OR INEFFICIENCIES IN THESE MARKETS
BUT WHICH PRACTICES ARE NOT AT THE OUTSET INDICATED AS ILLEGAL OR
IMPROPER."

(a) WHAT, IF ANY, DISTINCTION IS THERE BETWEEN "AGGRESSIVE COMPETITIVE
PRACTICES" AND THE EXERCISE OF MARKET POWER?

Practices which are "aggressive competitive practices" are not an
exercise of market power. Indeed, they are central to creating a
competitive market. Aggressive competition involves generators seeking
the highest value for its product while providing it at a price which
the consumer believes is competitive. Furthermore, aggressive behavior
should tend to lower prices, whereas, the exercise of market power
would tend to raise prices. It is this distinction upon which the ISO
and PX compliance divisions will focus. The compliance divisions must
seek out such anticompetitive behaviors, however, they must do

so in a way so as not to create rules or restrictions that would thwart
aggressive behavior.

(b) IF AN AGGRESSIVE COMPETITOR OBTAINS MARKET POWER (I.E., THE ABILITY
TO RAISE PRICES OR OTHERWISE INCREASE ITS PROFITS ABOVE WHAT IT WOULD
RECEIVE IN A COMPETITIVE MARKET) DUE TO A DESIGN FLAW OF THE MARKET,
WHAT TYPES OF CORRECTIVE STEPS WILL BE IMPLEMENTED BY THE ISO AND PX?

When the compliance divisions identify that a Market Participant is
exhibiting behavior that is not illegal but is taking advantage of a
design flaw, their immediate tasks are to:

o Recommend to the relevant Governing Board changes to the
appropriate rules and protocols so as to eliminate the design
flaw as quickly as possible.

o Assess whether or not a temporary mitigation strategy, such as
modified bidding activity rules or the use of a must run type
contract by the ISO, is required to prevent further abuses
during the time required to implement the recommended
corrections (e.g. during the time required to make appropriate
filings to amend the Tariff).

45. IN RESPONSE TO STAFF'S APRIL 29, 1997 DATA REQUEST, 76(a), THE ISO
INDICATED THAT THE IMPOSITION OF A MUST-RUN CONTRACT ON A GENERATOR
WOULD NOT OFTEN BE USED AS A SANCTION. IF NOT, WHAT OTHER ACTIONS WILL
BE TAKEN IF IT IS DETERMINED THAT:

(a) A MARKET PARTICIPANT CAN RAISE THE MARKET PRICES IN ITS AREA OF THE
NETWORK BY WITHHOLDING A KEY GENERATOR OR BIDDING THAT GENERATOR AT A
HIGH ENOUGH PRICE THAT IT WILL NOT BE DISPATCHED?

(b) A MARKET PARTICIPANT CAN CAUSE A TRANSMISSION CONSTRAINT IN A PEAK
PERIOD THAT ISOLATES A PORTION OF ITS GENERATION THAT IS NOT DESIGNATED
MUST-RUN NORMALLY BY THE ISO.

The Commission may have misinterpreted the response to Question 76(a)
contained in the May 20, 1997 data responses. The question asked if a
must run contract would be used as a mitigation measure to market
power. The response said that it would be used "in the case where a
specific unit has local market power due to transmission constraints."
The intent of the response was an attempt to clarify the situations in
which a must run contract could be used and that the ISO intends to use
these contracts both to meet its reliability needs and as

a mitigation measure for circumstances where local market power exists
due to transmission constraints. Clearly, this addresses the
hypothetical scenario posed in part (b) of this question.

The scenario posed in part (a) questions what action the ISO would take
in the event a Market Participant strategically withdraws plant in
order to raise prices. As stated in the response to Question 43, the
ISO compliance divisions will monitor for precisely this type of
behavior. Strategic withholding of generation in order to exercise
market power can be dealt with in several ways, including design of
bidding activity rules, the imposition of bid price caps under
specified scenarios, the use of cost-based call contracts, or perhaps
some combination of these measures. Prior to implementing such
strategies, however, the ISO Governing Board would have to approve such
measures and they would then be submitted for the Commission's
approval.

46. THE RESPONSE TO STAFF'S APRIL 29, 1997 DATA REQUEST 87 IMPLIES THAT A
UNIT UNDER A MUST-RUN CONTRACT WILL ONLY BE CALLED IF A SERIOUS
RELIABILITY PROBLEM EXISTS.

(a) PLEASE CLARIFY YOUR EARLIER RESPONSE.

(b) IF CALLING THE GENERATOR WOULD LOWER MARKET PRICES BY
ELIMINATING THE SUSTAINED EXERCISE OF MARKET POWER BY THAT
GENERATOR, WHY SHOULDN'T THE UNIT BE PUT UNDER A MUST-RUN
CONTRACT? DO ALTERNATIVE REMEDIES EXIST?

Part (a) of this question requests the ISO to clarify how the ISO
intends to call on units under must run contracts. The must run
contracts have been designed to provide sufficient flexibility for the
ISO to obtain its required services while attempting to minimize the
costs of the services. The provisions for the ISO to call on a must run
unit vary by the type of contract as follows:

o Contract A - This contract is the most competitive from the
ISO's perspective. The ISO is not required to pay any
reservation fee to the unit. The unit is paid only when it is
called by the ISO. Furthermore, the ISO does not call on the
unit until after the PX market clears. Thus, if the unit is
accepted by the PX, and therefore has offered a competitive
price in the PX, the ISO will not call the unit under the must
run contract and incurs no expense. As stated in the response
to Question 87 part (b) of the staff's April 29, 1997 data
request, the ISO will not use its must run call option for a
unit that has

been accepted in the PX even if the must run contract price is
lower than the PX Market Clearing Price. It will not use the
must run contract in this case because the unit provided a
competitive bid price to the PX and should therefore be able
to receive the PX Market Clearing Price. However, if the unit
had not been accepted by the PX because its bid price was too
high, the ISO would be able to call on the contract and would
only pay the contract price.

o Contract B - Under this contract, the ISO would pay a
reservation fee to the must run unit in return for an agreed
level of availability and being subject to non-performance
penalties. The contract has a cost-based energy dispatch price
and there is a credit back mechanism which allows the ISO to
recover 90% of any revenues earned by the unit in any market.
Similar to Contract A, the ISO would not call the unit under
its contract to the extent it was competitive in the PX
market. Under Contract B, however, the ISO does receive a
rebate back of any revenues in excess of the must run contract
price. Such rebate revenues would then be used to offset the
ISO's reservation fee under the contract.

o Contract C - This is the most expensive contract for the ISO.
Under Contract C, the unit may not participate in any market
and may be called on by the ISO at any time subject to the
contractual limitations. The ISO is obligated to pay the
entire costs of the unit in return for providing an agreed
level of service availability. The unit is paid only the
contract price when it is dispatched and is subject to
penalties for non-performance.

Part (b) of the question seems to suggest the ISO would not use a must
run contract it had with a Generator as a means to eliminate the
exercise of market power. The ISO disagrees with this premise. Clearly,
if the ISO has a must run contract with a Generator, and the Generator
attempts to exercise market power by demanding high prices, for
instance when a transmission constraint exists, the ISO would indeed
call the unit under its must run contract and therefore pay the unit
only the contract price and not the unit's bid price.

47. PLEASE INDICATE WHAT INFORMATION WILL BE COLLECTED BY THE ISO AND THE
PX TO MONITOR THEIR RESPECTIVE MARKETS BEGINNING ON JANUARY 1, 1998.

Please refer to pages 21-22 Appendix 7 of the ISO Tariff where the type
of monitoring activities the compliance divisions will undertake

has been indicated. The type of information that must be collected is
referenced in the descriptions of the indices.

Once the ISO and PX become operational on January 1, 1998, the ISO and
PX will automatically start to develop a wealth of information on the
functioning of the markets they administer emanating from their
day-to-day operations - information as to energy price bids, Adjustment
Bids, Supplemental Energy Bids, the actual performance of plants, etc.
Once the compliance divisions have developed an effective system for
handling and analyzing this information (an exercise that will begin
before January 1, 1998 but will have to continue for a period after
January 1, 1998 as experience with the information received is
developed), they will be able to make use of published or available
historic cost data to establish baselines and to provide other
reference points of use in their analysis.

48. WITH RESPECT TO THE MONITORING PLAN,

(a) EXACTLY WHAT INFORMATION WILL BE MADE PUBLIC?

(b) WHO WILL DECIDE WHAT INFORMATION WILL BE MADE PUBLIC?

(d) WHEN WILL THIS INFORMATION BE MADE PUBLIC?

(e) HOW OFTEN WILL THE INFORMATION BE MADE PUBLIC?

(a) The general approach of the compliance divisions will be to make
all information developed from the markets they administer on a routine
basis publicly available, e.g., through regular reports to the
Commission and to other regulatory or government agencies, or as
specifically needed within the administration of their respective
markets or by participants within them. In two types of circumstances,
however, information may not be made public:

(1) where the information is subject to the confidentiality
provisions of Section 20.3 of the ISO Tariff, which includes a list of
the specific types of information provided by Scheduling Coordinators
that shall remain confidential (at Section 20.3.2); and

(2) possibly for certain limited periods during an
investigation of a potential violation of the market rules where
disclosure of certain information might jeopardize the investigation,
e.g., permit tampering with data.

(b) The compliance divisions, in the exercise of their monitoring
responsibilities or the Governing Boards will decide.

(c) Generally, the information will be made public through regular or
ad hoc reports to agencies such as the Commission. The reporting
schedules and procedures for making information available in response
to specific requests will have to be developed by the compliance
divisions once operational. The divisions are able to assess the most
efficient administrative means of information dissemination.

(d) and (e). Other than the regulatory filings already committed to, it
is not currently possible to anticipate exactly when and how often
information will be made public but it is likely to be more often than
annually. The compliance divisions, once operational and able to assess
the needs of regulatory and other government agencies, of Market
Participants and of their own enforcement programs, will develop
dissemination mechanisms to meet these needs.

49. PG&E, SCE, AND SDG&E HAVE EACH SUBMITTED INDIVIDUAL MARKET POWER
MITIGATION AND MONITORING PROPOSALS. PLEASE INDICATE WHICH ASPECTS,
INCLUDING SPECIFIC RULES AND PROVISIONS, WILL INITIALLY BE INTEGRATED
INTO THE ISO'S AND PX'S MITIGATION AND MONITORING PLANS.

To the extent that the Companies have proposed monitoring programs
associated with their market power mitigation plans, these are
currently general and conceptual in nature and, as they are related to
the Companies' overall efforts to satisfy the Commission's market power
concerns, they are subject to Commission review. Should they be
approved by the Commission, the compliance divisions, which are being
staffed up will undertake to work with the Companies to ascertain how
they should best be integrated into the evolving ISO and PX monitoring
plans.

50. PLEASE CLARIFY EXACTLY WHICH PARTIES WOULD BE RESPONSIBLE FOR THE COSTS
OF MUST-RUN AGREEMENTS -- SCHEDULING COORDINATORS, UTILITIES, OR SOME
COMBINATION THEREOF. PLEASE DELINEATE THE COSTS AND SPECIFY THE
RESPONSIBLE PARTY FOR EACH MUST-RUN GENERATOR WHEN THE MUST-RUN
GENERATOR IS (1) OWNED BY A UTILITY AND SCHEDULED THROUGH THE PX, AND
(2) SCHEDULED THROUGH A NON-PX SCHEDULING COORDINATOR.

Please refer to the amendments to Section 5.2.7 of the ISO Tariff
proposed in the June 13, 1997 ISO/PX Reply Comments which are set forth
in the Restated ISO Tariff filed contemporaneously with these
responses. Payments made under Reliability Must Run Agreements by the
ISO are recovered from the utility in whose service area the

Reliability Must Run Unit is located. The relevant costs are the
amounts payable under the relevant Reliability Must Run Agreement after
having first deducted:

(a) payments received by the ISO from those Scheduling
Coordinators whose Energy Schedules are reduced to allow the
Reliability Must Run Generation to be scheduled; and

(b) any payments made by the ISO for Ancillary Services under the
Reliability Must Run Agreement, these payments being recovered
by the ISO as part of the Ancillary Services user charges
levied upon Scheduling Coordinators by the ISO under Section
2.5.28 of the ISO Tariff.

This applies whether or not the Reliability Must Run Unit is owned by
the utility and scheduled through the PX (in which case there may
inevitably be some "circularity" of payments) or owned by some other
entity, whether scheduled through the PX or some other Scheduling
Coordinator. As indicated elsewhere in these proceedings, the ISO's
intention is to eliminate the need for Reliability Must Run Generation
over the long term and in the near term to replace the initial
Reliability Must Run contracts with a more competitive process (See
Section 5.2.2. of the ISO Tariff).

51. WOULD IT BE POSSIBLE FOR A GENERATOR TO DEFAULT UNDER A MUST-RUN
AGREEMENT BUT SIMULTANEOUSLY SELL ENERGY UNDER A BILATERAL CONTRACT OR
INTO THE PX? IF THE ANSWER IS YES, WILL THERE BE ANY PENALTY IMPOSED?
PLEASE EXPLAIN FULLY ANY POSSIBLE PENALTY.

The answer to the first part of the question is yes, and the penalty
for the owner of the Reliability Must Run Unit in question depends upon
the form of Reliability Must Run Agreement in force for the Unit
concerned at the time of the default. If the 'A' Agreement applies,
there is no penalty. However, the ISO may be prompted to consider
switching the Unit concerned to Agreement `B'. If the 'B' Agreement
applies, the penalty is that the Unit is deemed to be unavailable and
loses its Availability Payment to the extent and for the period of the
default. In addition, 90% of the revenue or deemed revenue from the PX
transaction or bilateral contract will be credited against any
remaining part of the Availability Payment in any event; the question

does not arise if the Unit is under the 'C' Agreement since this
Agreement prohibits the Unit from trading in the PX or bilaterally.

52. WILL MUST-RUN GENERATION BE SCHEDULED AS PART OF THE BALANCED SCHEDULES
OF SCHEDULING COORDINATORS?

The answer to this question is yes. The ISO will adjust the Day-Ahead
Schedule of the Scheduling Coordinator for the Must Run Units concerned
pursuant to the Intra-zonal Congestion Management process referred to
in Section 7.2.6 of the ISO Tariff.

53. PLEASE CLARIFY HOW THE PROXY PRICE, USED TO DETERMINE THE REVENUES THAT
WILL BE CREDITED BACK TO THE AVAILABILITY PRICE WHEN MUST-RUN
GENERATORS UNDER AGREEMENT B TRANSACT UNDER BILATERAL CONTRACTS, WILL
BE DETERMINED.

The proxy price will be applied to the Energy sold by the Reliability
Must Run Unit under the bilateral contract in order to derive the
"deemed" revenue notionally received by the owner of the Unit under the
transaction. This sum will then be used to calculate the amounts to be
credited against the Availability Payments payable by the ISO under the
Reliability Must Run Agreement. The proxy price will either be the PX
Market Clearing Price for Energy in the relevant hour or hours, or such
other index as the ISO considers to be appropriate in the circumstances
of the case. Reference is made to pages 90 and 91 of the ISO/PX June 23
1997 Reply Comments for the reasons for not adopting the ISO's Market
Clearing Price for Imbalance Energy as the basis for determining the
proxy price. The ISO will clearly be looking for an index which
consistently and accurately reflects the actual energy prices found in
commercial transactions of the kind concerned.

OTHER

54. PLEASE DESCRIBE, IN AS MUCH DETAIL AS POSSIBLE, HOW THE ISO WILL
ACCOMPLISH A SEAMLESS INTEGRATION WITH THE WSCC REGION. PLEASE PROVIDE
THE NECESSARY OPERATING RULES AND PRACTICES TO MANAGE THE INTERFACE
BETWEEN CONTROL AREAS, INCLUDING CRITERIA FOR SANCTIONS.

The ISO will accomplish a seamless integration with the WSCC region
through the negotiation of interconnection agreements with the
adjoining control areas and through the coordinated timing of energy
and transmission markets. The interconnection agreements are to be
negotiated in the near future.

With respect to the creation of the necessary protocols, the ISO is
already working within the WSCC to ensure that markets for energy and
transmission are properly coordinated. Current discussions are already
focusing on such details as time-of-day schedule submission issues.

55. PLEASE PROVIDE THE SPECIFIC ISO AND PX PLANS AND PROCEDURES FOR
COORDINATING AND INTERACTING WITH THE FERC, CPUC, CEC, AND OTHER
REGULATORY AGENCIES.

The ISO and PX staffs will coordinate and interact with the FERC, CPUC,
CEC, and other regulatory agencies by designing and implementing a
regulatory compliance plan. The first step will be for regulatory
counsel to identify those activities required of the ISO and the PX by
various regulatory agencies under their regulations and governing
statutes. ISO and PX management will then formulate a plan to execute
the required activities. This plan will require obtaining budgeted
funds to be set aside in the overall corporate budget, assignment of
legal and technical staff to perform specific functions, and the
development of a timetable for required actions. Each corporation will
also designate an individual within its staff to serve as its primary
regulatory affairs liaison officer. This individual will monitor
regulatory developments in order to keep the corporation informed of
changing regulatory requirements, serve as the principal contact within
the corporation for correspondence between the corporation and
regulatory agencies, and meet with regulatory staff when appropriate.

56. PLEASE PROVIDE THE ROLES, RESPONSIBILITIES, REQUIREMENTS, AND PROTOCOLS
FOR A METERED SUBSYSTEM (MSS).

Please refer to the answer to Data Request 33. The ISO anticipates
being in a position to make available its technical requirements and
protocols for Metered Subsystems by November 1, 1997.

57. WILL SCHEDULING COORDINATORS BE ABLE TO SCHEDULE INTERRUPTIBLE ENERGY
OUT OF OR INTO THE ISO CONTROL AREA? IF THIS WILL BE ALLOWED, PLEASE
PROVIDE ALL THE ISO TARIFF MODIFICATIONS NECESSARY TO REFLECT THIS
CHANGE.

It is the intention that Scheduling Coordinators should be able to
schedule interruptible imports into the ISO Control Area. Amendments to
Section 2.5 of the Tariff to provide for the Ancillary Services
necessary to allow this to happen are set forth in the Restated ISO
Tariff filed contemporaneously with these responses. Amendments to
Sections 2.5.3.2, 2.5.20.1, 2.5.22.3.2, 2.5.22.5 and 2.5.23.1 are

included. As stated in the amendments, the software required to
implement the facility to schedule interruptible imports will not be
available during the initial stages of operation of the ISO.

58. THE ISO/PX REPLY COMMENTS AT P. 184-85 INDICATE THAT SECTION 2.5.3 OF
THE ISO TARIFF (DEALING WITH OPERATING RESERVES) SHOULD BE CLARIFIED.
PLEASE PROVIDE THE NECESSARY REVISIONS.

The amendments, proposed in the Restated ISO Tariff filed
contemporaneously with these responses, include those required to
reflect changes to Operating Reserve in Sections 2.5.3.2 and 2.5.20.1.

59. REFERENCE ORIGINAL SHEET NOS. 288-289 OF THE ISO TARIFF

DEPENDABLE GENERATION IS CALCULATED BASED ON THE MAXIMUM RECORDED MW
OUTPUT OF UNITS DURING PEAK PERIODS. UNDER THE DEFINITION OF DEPENDABLE
GENERATION, WILL GENERATORS AVAILABLE FOR SERVICE BUT NOT DISPATCHED
COUNT TOWARDS AN ENTITY'S GENERATION RESOURCES AVAILABLE TO SERVE LOAD?

The Joint Comments filed on June 6, 1997, included a proposed revision
to the definition of Dependable Generation which has been accepted and
incorporated into the Restated ISO Tariff filed contemporaneously with
these responses. The amendment seeks to ensure that a Participating TO
will receive credit for available Generation capacity without a
requirement that the output of that capacity is actually being
delivered to the ISO Controlled Grid at the time of the annual system
peak. The effect of the amendment would be that Generation available
for service, but not dispatched would count towards the resources
available on a Participating TO's system available to serve Load. The
amendment provides as follows:

The sum of the maximum amount of Generation capacity, in MW,
from Generators interconnected with the Participating TO's
transmission or distribution system, that a Participating TO
reasonably believes could be delivered to serve Load,
regardless of ownership of the Generation capacity or whether
a contract exists for the purchase of the output from the
Generator.

60. PLEASE PROVIDE THE AUCTION ACTIVITY RULES THAT WILL BE IN EFFECT ON
JANUARY 1, 1998.

The Auction Activity Rules that will go into effect on January 1, 1998,
will be included in protocols scheduled for completion by November 1,
1997.

61. PLEASE SUBMIT TO THE COMMISSION ANY DOCUMENTS FROM THE WEPEX WEB SITE
ON THE INTERNET, THAT PROVIDE MORE DETAILED INFORMATION AND/OR
ILLUSTRATIVE EXAMPLES OF HOW THE PX AND ISO PROPOSALS ARE INTENDED TO
WORK. FOR EXAMPLE, PLEASE SUBMIT THE FOLLOWING IF THEY ACCURATELY
DEPICT YOUR CURRENT PROPOSAL.

(1) 6/11/97 MEETING PRESENTATIONS: SETTLEMENT CASE STUDY

(2) 5/14/97 PXPG MEETING MINUTES: CASE STUDY PRESENTATION

(3) ISO IMBALANCE SETTLEMENTS, JUNE 24, 1997, BY ALEX D.
PAPALEXOPOULOS

Those documents currently found at the WEPEX website are not definitive
examples or illustrations of how the ISO and PX are intended to work.
The three documents listed, for example, are on-going "works in
progress" as are many of the other documents listed at the website. We
ask the Commission to only consider details supplied through the
filings as the most current information on the subject.

62. THE ISO AND THE PX HAVE REQUESTED (IN THE ISO AND PX REPLY COMMENTS AT
P. 12-13) THAT THE COMMISSION FIND THE ISO, PX, AND TO TARIFFS TO BE
JUST AND REASONABLE FOR FILING AT THE COMMISSION'S SEPTEMBER 10, 1997
MEETING, ACCEPT THEM FOR FILING WITH A NOMINAL SUSPENSION, AND GRANT
THE ISO AND THE PX THE AUTHORITY TO BEGIN OPERATIONS PURSUANT TO THE
TARIFFS ON NOVEMBER 1, 1997. EXACTLY WHAT OPERATIONS DO THE ISO AND THE
PX PROPOSE TO COMMENCE ON NOVEMBER, 1, 1997 AND WHAT ADDITIONAL
OPERATIONS WOULD BEGIN ON JANUARY 1, 1998?

The operations that the ISO and PX propose to implement on November 1,
1997 are the acceptance and processing of applications from Market
Participants who will be using the services of the corporations when
they initiate market operations on January 1, 1998. For example, the
ISO proposes to begin accepting applications for the

certification of Scheduling Coordinators on the first business day
following November 1, 1997. Similarly, the PX will begin accepting
applications for PX Participants on the same date. Both corporations
also contemplate that they will be finalizing the process of
negotiating and executing the various contractual agreements needed for
their market activities. For the ISO, such agreements would include
Scheduling Coordinator Agreements, Reliability Must-Run Agreements,
interconnection agreements with adjoining control areas, and the other
agreements necessary to define its rights and responsibilities with
other entities participating in the market. For the PX, such agreements
would include PX Participation Agreements and Meter Service Agreements.
On January 1, 1998, the ISO and PX will commence providing the services
offered under their FERC Tariffs, namely the provision of transmission
and Ancillary Services by the ISO and the establishment of an energy
market by the PX. In the two months prior to that date, training, trial
running and acceptance testing will be taking place.

63. AS STATED IN ISO/PX MAY 20, 1997 RESPONSE, PLEASE PROVIDE THE METHOD
THAT WILL BE USED TO DETERMINE THE ADMINISTRATIVE PRICE (THE ENERGY
PRICE DURING SYSTEM EMERGENCIES).

A market based approach is proposed. This minimizes the risk of
overpayment or underpayment by the ISO, gaming by Market Participants
and arrives at a result which is as near to reality as would have been
achieved had the emergency not occurred. The method of determining the
Administrative Price and its application will be derived from the
following rules which are reflected in the Restated ISO Tariff which is
filed contemporaneously with these responses.

(1) The ISO will not intervene in the Day-Ahead Market except in the
case where there has been a total system collapse and the system is
being restored.

(2) The ISO will schedule and dispatch all resources offered to it in
the Day-Ahead and Hour-Ahead Markets, regardless of price, before it
intervenes and suspends the Hour-Ahead or Real Time Market as
authorized under Section 2.3.2.3 of the ISO Tariff. In short, the ISO
must:

o Schedule or dispatch all generation resources made available
regardless of price, which includes using all available
Adjustment Bids, Supplemental Energy Bids, and all Ancillary
Services; and

o Schedule or dispatch all price-responsive demand that has been
bid into the markets.

(3) When the ISO has exhausted all available resources, it will turn to
involuntary Load Shedding. However, prior to this, the Hourly Markets
will have continued and both generators and demands will have been able
to submit new prices each hour reflecting their opportunity costs for
their offered resources.

(4) The Hourly Market and Real Time Market will be suspended once the
ISO has reached the conditions referred to above and involuntary Load
Shedding has been implemented. The clearing prices in each market
(Imbalance Energy and Ancillary Services) will be set at the previous
hour's clearing price until such time as the ISO has restored all Load
which was involuntarily shed.

64. PLEASE PROVIDE ALL AGREEMENTS THAT DEFINE THE RELATIONSHIP BETWEEN THE
ISO AND A (1) UTILITY DISTRIBUTION COMPANY, (2) GENERATOR, (3) METERED
SUBSYSTEM, AND (4) SCHEDULING COORDINATOR.

The Agreements that define the relationship between the ISO and (1)
Utility Distribution Companies, (2) Generators, (3) owners of Metered
Sub-Systems and (4) Scheduling Coordinators are as follows:

1. ISO and Utility Distribution Companies - UDC Operating Agreement
2. ISO and Generators - Participating Generator Agreement
3. ISO and Metered Sub-System Owners - SC Agreement(3)
4. ISO and Scheduling Coordinators - SC Agreement

The current status of each of these agreements is as follows:

o UDC OPERATING AGREEMENT:

This document is currently being developed and will be filed
for informational purposes by November 1, 1997.

- ----------

(3) The ISO Tariff has been clarified at Section 2.5.20.2 to provide that a
Metered Sub-System must schedule or bid all its Energy and Ancillary
Services either as a Scheduling Coordinator or through a Scheduling
Coordinator. Accordingly, the SC Agreement will be used to regulate the
relationship between the ISO and participating Metered Sub-System
owners or operators who elect to become Scheduling Coordinators
themselves.

o GENERATOR AGREEMENT:

Again, a pro forma of this agreement is currently being
prepared and will be filed for informational purposes by
November 1, 1997.

o SC AGREEMENT:

A pro forma of this document was filed on March 31, 1997. It
appears at Appendix B to the ISO Tariff. Changes to the pro
forma were proposed in the Joint Comments filed on June 6,
1997 and are reflected in the Restated ISO Tariff filed
contemporaneously with these responses.

65. APPENDIX K OF THE ISO TARIFF CONTAINS A SAMPLE ISO PROTOCOL. PLEASE
PROVIDE THE ISO PROTOCOL THAT WILL BE USED WHEN THE MARKET BEGINS ON
JANUARY 1, 1998.

This question is based on the assumption that there will be a single
protocol implementing Sections 2.4.3 and 2.4.4 of the ISO Tariff. In
fact, given the large number and very disparate nature of the Existing
Contracts it is envisioned that many protocols will have to be
developed over the next few months in order to enable energy to be
scheduled utilizing rights under Existing Contracts as envisaged by the
Tariff. Essentially, the purpose behind attaching Appendix K and the
sample protocol was to demonstrate that, from an ISO perspective, it
would be possible to "feather in" the rights to transmission service
under Existing Contracts with the rights to new ISO transmission
service.

The effect of the ISO Tariff Section 2.4.3.1 is to require each
Participating TO, the holder of transmission rights under an Existing
Contract and the ISO to work together to develop operational protocols,
based on existing protocols and procedures to the extent possible, but
at the same time which are minimally burdensome to the ISO. The
principles that are stated are very general and are designed primarily
to allow the contract rights to be honored. However, as explained in
the Reply Comments filed on June 23, 1997 (page 40) if there is some
flexibility, either in the contracts or if the contract parties are
prepared to agree, then it would be beneficial if the most workable
outcome for all affected interests could be achieved.

Section 2.4.4.5 of the ISO Tariff gives further guidance as to how the
parties to the Existing Contracts and the ISO will implement the
provisions of Section 2.4.4.4 of the ISO Tariff. It explains that the
sample protocol in Appendix K illustrates how rights under Existing

Contracts can be integrated with the ISO's transmission service. These
rules are designed to achieve not only workability but also, wherever
possible, consistency of treatment, in an effort to move over time to
the open, non-discriminatory access regime envisaged by Order No. 888.
Essentially, the purpose of Sections 2.4.3 and 2.4.4 is to enable
parties to Existing Contracts to develop workable operational
protocols. Appendix K, as we have said, shows how it can be done from
the point of view of the ISO transmission service. The nature of the
ISO transmission service is clear from the ISO Tariff. By contrast, the
nature of transmission service under the many and varied Existing
Contracts can not be captured in a "one-size fits all" protocol.

The protocols being developed by the parties to the Existing Contracts
fall into several categories. First are path specific protocols which
will address the rights to the specific path of several parties under
one or more Existing Contracts. These will provide steps that ISO staff
should implement when curtailment of transmission across the given path
is required. These protocols will identify the order and manner in
which Existing Contract holder rights should be curtailed to preserve
each Existing Contract party's rights.

Another category of protocol will be related to information pertaining
to individual Existing Contracts. Specifically, lists of the firm,
conditional firm, and non-firm scheduling rights embodied in each
contract along with operating instructions and/or decision rules that
are currently used to allocate, schedule and curtail the various
categories of transmission services.

Finally, some protocols will be established to clarify issues such as
the scheduling flexibilities embodied in the Existing Contract and how
billings and settlements will be handled for Existing Contracts. These
protocols are intended to provide clarity of roles and responsibilities
but may require action on the part of the ISO.

Development of these protocols is under way but will obviously take
time to complete, given the large numbers involved. However, the
contract parties and the ISO are meeting to develop the protocols in
accordance with Sections 2.4.3 and 2.4.4 of the ISO Tariff and with the
guidance provided by Appendix K in order to achieve as much consistency
as possible and to minimize the operational burden to the parties to
the Existing Contracts and to the ISO.

These protocols, when developed, are likely to reflect day-to-day
operational needs and the ISO wonders whether the Commission will find
it useful to review this level of operational detail. However, if the
Commission wishes, the ISO could provide staff with copies of selection
of them for information purposes as and when they are available.

66. PLEASE PROVIDE THE OPERATING AGREEMENTS AND PROTOCOLS GOVERNING THE
CIRCUMSTANCES UNDER WHICH THE ISO CAN DIRECT THE OPERATION OF A UTILITY
DISTRIBUTION COMPANY'S SYSTEM.

As indicated in the reply to question 64, it is intended to provide the
Commission with a copy of a pro forma UDC Operating Agreement by
October 31. The Protocols, which are currently under preparation, will
contain provisions relating to the issue of instructions to UDCs
regarding the operation of their systems.

ATTACHMENT 1

3 March 1997

Report to the California Trust for Power Industry Restructuring

ACTIVITY RULES FOR THE POWER EXCHANGE

PHASE 3: EXPERIMENTAL TESTING*

MARKET DESIGN INC.
PREPARED BY ROBERT WILSON, VICE PRESIDENT

Executive Summary

The scope of the work stipulated for Phase 3 during February includes
experimental testing of the activity rules developed in Phase 2, as well as
identification of problems and proposed remedies. The experimental program
includes construction of a laboratory prototype by February 14, followed by a
series of tests and demonstrations by February 28. This program was undertaken
by Professor Charles Plott, who is a prominent expert on experimental studies of
markets. He directed the construction of the prototype by H.Y. Lee, and he
designed and conducted the tests at the Caltech Laboratory for Experimental
Economics and Political Science. The prototype was completed by February 14, and
experiments were conducted over the following two weeks, often with several
trading sessions per day. The design followed current practice in studies of
market mechanisms. The subjects were Caltech students whose entire remuneration
consisted of their trading profits. Professor Plott conducted demonstrations on
February 21 for members of the PX Team, and on February 28 for members of the
TAC. He is also submitting a companion report with additional detail.

The first task was to establish whether the PX Protocol can be implemented in a
working prototype. The second task was to establish whether the auction's
iterative process converges, the rate of convergence, and the character of its
dynamics. The third task was to measure the efficiency of the auction outcomes.
Each task was divided into studies of single and multiple markets, and cases
without and with fixed costs. The multiple markets correspond to the PX's 24
hourly markets for next-day delivery; the fixed costs correspond to the start-up
and no-load costs incurred by thermal generators. Additional topics included the
role of withdrawals, substitution among markets (as in the case of hydro
supplies), and sensitivity to parameter specifications (such as the minimum bid
decrement). Some tests were conducted using demand patterns and supply
portfolios representative of the California mix prepared by London Economics
using data from the CEC and FERC.

The main conclusions from these studies are the following:

o IMPLEMENTABILITY. We had no difficulty implementing the PX Protocol.
The software requirements are straightforward. Subjects in the
experiments had no difficulty understanding and following the
procedural rules of the auctions.

o CONVERGENCE. In all tests the auction converged. All subjects tried to
game the system but these strategies proved ineffective. After some
experience, several subjects concluded that simply bidding their costs
is optimal, which accelerates convergence. We conclude that the
activity rules succeed in suppressing gaming behavior or rendering it
ineffective.

o EFFICIENCY. In most tests the auction ended with an outcome that was
within a few percent of perfect efficiency. The final clearing prices
and quantities were close to the theoretical equilibrium prices, even
with few bidders. The exceptions were that inefficiencies occurred in
tests that included either a supplier with significant market power or
one with supplies that could be allocated costlessly among the markets.
We conclude that activity rules cannot supplant measures to mitigate
market power.

o RATE OF CONVERGENCE. Progress is substantial in the first five or six
iterations, residual inefficiency is small after eight to twelve, and
full convergence often occurs in ten to twenty. Nevertheless, in
extreme cases, as when the bid decrement is small, convergence can
require forty iterations. Because the PX might restrict the number of
iterations to as few as twelve over two hours, measures are required to
accelerate convergence or to terminate the auction after progress has
slowed sufficiently or when time expires; or, the allowed time might be
increased or the software altered to enable more iterations or
continuous bidding. We conclude that there are sufficient measures
available to close the auction without significant inefficiencies.

o FIXED COSTS. The tests with fixed costs that must be recovered from
multiple markets showed comparable efficiency. Subjects learn quickly
to stay active in those markets where prices are sufficient to recover
their fixed costs. The dynamics follow the scenario predicted by London
Economics: subjects initially load their fixed costs into their bids in
each market, but then later prorate them among the markets in which
they remain active. With this strategy, withdrawals are minimized and
inefficiencies due to premature withdrawals are rare.

London Economics' companion report addresses these and additional topics. In
particular, they conclude that inclusion of additional constraints on
operational feasibility increase the number of iterations required for full
convergence.

Our summary conclusion is that the PX Protocol is a viable design for an energy
market, and the efficiency of its outcomes is impressive. The numbers of bidders
and markets in the tests were small, and we did not replicate the daily
repetition of the market, but we found no fundamental impediment to full-scale
implementation. Further work in Phase 4 should refine the design to accelerate
convergence and assure a timely close.

1. Review of the Activity Rules

In the absence of activity rules the auction outcome could be inefficient.
Bidders could wait until the final iteration to offer serious bids, which
prevents early price discovery and thereby prevents bidders from identifying
their optimal hours of operation - which is essential due to the start-up and
no-load costs of thermal generators. The purpose of

activity rules, therefore, is to encourage early serious offers so that price
discovery proceeds steadily throughout the iterative process. Their design is
subject to the restriction that they cannot impair efficiency; in particular,
they cannot constrain suppliers who choose to offer their actual costs.

The "standard" activity rules used for the experimental tests are summarized in
Appendix A. They are based on the principle of revealed preference. The
Exclusion Rule is key: a bidder cannot offer later a price that improves a
previous clearing price that was not improved at the first opportunity; i.e., in
the next iteration. Thus, if a supplier declines to improve the previous
iteration's clearing price then we infer that this price is below the supplier's
cost for that increment of supply, so the supplier is precluded from offering a
lower price later. This rule is complemented by four additional routine
procedural rules, stated here in the form applicable to suppliers (the rules for
demanders are analogs). The Opening Rule requires that all available capacity is
offered in the first iteration. The Revision Rule restricts revised prices to
those less than the previous clearing price by at least a specified decrement.
The Withdrawal and Closing Rules require that withdrawals are irrevocable, and
they preclude withdrawals after the final iteration.

The effect of these rules on a supplier is to require an irreversible decision.
If its offered price in the previous iteration exceeded the clearing price, then
in the current iteration it must offer a price less than that previous clearing
price or forego all later opportunities to do so. If its cost is sufficiently
low then the supplier's best strategy is to revise its offered price; otherwise,
it's better to decline, in which case it cannot later revise its price unless
the clearing price rises higher. If all suppliers offer their actual costs then
the auction ends after the second iteration, since no offers are revised.

When suppliers bid strategically by offering prices above their actual costs,
several iterations are required to drive their revised offers down to their
costs. The resulting competitive process involves only those suppliers near the
margin. The extra-marginal suppliers must revise their offers (or be frozen
out); when they do so they become infra-marginal, thereby making some previously
infra-marginal suppliers extra-marginal, and now these too must revise their
offers.

The rate of convergence is driven by the difference between the current clearing
price and the equilibrium price. When this difference is large (relative to the
elasticity), one side of the market is "long" by a large amount. If it is the
supply side then extra-marginal suppliers rejected (or rationed by the Rationing
Rule) and they must revise their offers or be frozen. When these suppliers
revise their offers, they eject a large number of previously infra-marginal
suppliers from the merit order. This ensures a decrease in the clearing price by
at least the amount of the specified price decrement. The dynamics of this
process are elaborated in Professor Plott's companion report.

When the difference is small, however, the imbalance may be small too and the
price need not change for one or more iterations (the PX Team that visited the
Lab called this "stuttering"). The difference can be small either because the
clearing price is close to the equilibrium price, or because the supply
elasticity is small. In either case the increase in the total gains from trade
from further iterations is small, and zero if demand is inelastic too. In
California, the supply elasticity is expected to be large (except perhaps in
peak hours in the summer) and the demand elasticity is small. Consequently, one
expects rapid progress in the first few iterations, after which the trading
gains diminish rapidly, indicating that it may suffice to terminate the auction
by invoking a convergence criterion.

EXAMPLE: Suppose that demand is inelastic at 1000 MWh and at the current
clearing price of 20 $/MWh aggregate supply is flat over the range 900 to 1200
MWh. Then 200 MWh is rationed and the suppliers offering this amount must in the
next iteration offer 19 $/MWh or less, due to the specified decrement of 1
$/MWh. If they all elect to do so then in the merit order their revised offers
displace those who previously offered prices in the interval between 19 and 20
$/MWh, including the unrationed 100 MW at 20 $/MWh. If the supply offered at
prices less than 19 $/MWh exceeds 800 MWh then the new clearing price is 19
$/MWh, and otherwise the new clearing price is between 19 and 20 $/MWh - and
surely less than 20 $/MWh if the supply previously offered at prices below 20
$/MWh exceeds 800 MWh.

This sort of example occurs when the initial 20 $/MWh clearing price is
substantially more than the equilibrium price. At the equilibrium price the
imbalance is nil and no rejection or rationing occurs, but if the equilibrium
price is only slightly less than the current clearing price then the amount
rejected or rationed is small and the amount displaced in the merit order by
revised offers is also small, so there is a greater prospect that the clearing
price changes by less than the price decrement. This feature indicates that for
practical purposes it may suffice to adopt a criterion for terminating the
auction when the percentage of rejected or rationed offers is sufficiently
small.

Convergence can be accelerated by specifying a large price decrement for revised
offers. This has the immediate effect that clearing prices move in large jumps
between iterations. It also has a strategic effect that further accelerates
convergence. A supplier realizes that if its offer is above its cost by less
than the magnitude of the decrement then it will be unable to revise its offer;
consequently, there is a stronger incentive to offer revised prices close to its
cost.

The Exclusion Rule is stated above for the case that the auction proceeds in
discrete iterations. If the auction is accelerated by allowing continuous
bidding then this rule can be stated in terms of the time interval allowed for
submission of a revised offer. The February 21 report provides further
elaboration on this and other variants of the basic activity rules. One should
be cautious about these variants, however, since in the two weeks allowed for
the experiments it was possible to test only the basic set of rules.

2. The Experimental Program

The experimental program was designed to address several issues. A preliminary
step was to verify that the PX Protocol is implementable. This was accomplished
by constructing a prototype sufficient for several bidders (e.g., 12) and
several parallel markets (e.g., 2, 3, or 4), although in principle the software
is capable of much larger numbers (one test was run with 20 markets). Further,
the initial tests verified that subjects easily understood the rules and
regularly submitted offers that conformed to these rules. No fundamental
impediments to a full-scale implementation were identified, and indeed the
software requirements are straightforward. Bids were submitted manually at
keyboards so iterations were slow, but since the eventual implementation will
allow computerized submissions this delay can be avoided. The one lacunae found
in the PX Protocol was an inadequate specification of how to handle cases
without a unique clearing price, as can happen when a supplier specifies a
minimum load. This deficiency was patched by using the lowest price for which
supply is not less than demand.

The next step was to establish how the Protocol works in a single market, with
and without fixed costs, and with and without demand-side bidding. The first
week of testing verified that fixed costs did not impair the efficiency of an
isolated market. Demand-side bidding produced no complications: clearing prices
did not converge monotonically in this case, but non-monotonicity did not
disrupt convergence. A significant conclusion from this series of tests was that
the competitive process and the factors affecting the convergence rate match the
theoretical predictions. In particular, the subjects' strategic bidding slowed
convergence but did not prevent near-efficiency at the close. Some tests were
run with data representative of the California mix, prepared by London Economics
using data from the CEC and FERC.

The second week addressed three central issues posed by the peculiar features of
the PX. These were run with multiple markets representing demand configurations
corresponding to peak, offpeak, and shoulder periods; also, to focus on the
issues, demand-side bidding was mostly excluded. The first issue was the effect
of market power. As expected, the activity rules did not mitigate market power:
a large supplier can sustain a clearing price above the equilibrium price by
withholding supply, and at the margin it can capture the difference between its
cost and the next higher cost in the merit order. The second issue was the
effect of the Opening Rule on supplies that can be freely allocated among
multiple markets: the tests showed that efficiency can be impaired by the
restriction that capacity cannot be reallocated among the markets in later
iterations. This test was imperfect because subjects were apparently unaware
that they might initially offer amounts in the several markets that exceeded
their total capacity and then later withdraw from (or freeze their offers in)
those markets with low prices, which is a strategy that overcomes the
restrictions imposed by the Opening Rule. We concluded that subsequent work in
Phase 4 might consider an activity rule for total-energy portfolios; one
candidate is described in the February 21 report, but based on the successful
record of the predominately hydro NordPool system this need not be the first
priority.

The most important tests in the second week studied the role of start-up costs
when there are multiple markets. The issue was whether convergence would be "top
down" and therefore efficient with few or no withdrawals, or "bottom up" and
therefore potentially inefficient if some suppliers withdraw prematurely when
faced with low initial prices when in fact they could operate profitably at the
final prices. This issue is essentially a behavioral question. One hypothesis is
that a supplier's offer in each market will include its entire fixed cost in the
first iteration, and thereafter this fixed cost will be apportioned among the
markets in which the supplier remains active - this strategy implies that
withdrawals are largely unnecessary because a supplier can exit a market by
freezing its offers above the clearing price. The second hypothesis is that a
supplier will offer prices on an incremental-cost basis in each market, hoping
that the eventual clearing prices will be sufficient to recover its fixed cost,
and withdrawing otherwise - this strategy implies that withdrawals are
important, and that efficiency depends on the order in which suppliers elect to
withdraw. The simulation studies by London Economics showed that, due to the
flatness of California's aggregate supply function based on incremental costs,
the second hypothesis implies a potentially significant inefficiency if
withdrawals are premature due to myopic expectations about the final clearing
prices.

The results from the tests conducted to examine this issue support the first
hypothesis. With no prompting, the subjects invariably followed the top-down
strategy. Consequently, the selection of suppliers, and the markets in which
they remained active, were accomplished by freezing offers, and there were no
inefficient withdrawals in the

test runs. Thus, the tests produced no evidence that there might be an inherent
tendency for incremental-cost offers that could cause inefficiencies due to
premature irrevocable withdrawals. This conclusion is reinforced by other
observations that conform to the predictions from the simulations conducted by
London Economics; e.g., offpeak prices converge first and quickly and shoulder
prices last - in some cases requiring many iterations to settle down.

3. Conclusion

In Phase 2 we were asked to "fill in the blanks" in the PX Protocol by
suggesting activity rules that would suppress gaming and promote price discovery
during the iterative process of the auction. In Phase 3 we were asked to assess
the reliability of the design principles in predicting actual outcomes in
experimental tests. The set of activity rules in Appendix A was the candidate
studied in the experimental program during the February.

The test results from over forty sessions indicate that with this amendment the
PX Protocol can be implemented in a small-scale prototype, and that typically it
converges to an outcome that is within a few percent of perfect efficiency.
Although the number of iterations required for complete convergence could exceed
the limits imposed in the 1/1/98 implementation, there are ample tools to
accelerate convergence (e.g., a large price decrement and/or continuous bidding)
or to conclude the auction after progress has slowed (e.g., a convergence
criterion based on a measure of residual inefficiency). In any case the auction
outcome is potentially feasible after every iteration, and the welfare losses
from premature termination are likely to be small. A crucial test was whether
subjects would adopt the top-down strategy to handle fixed start-up costs,
thereby preventing inefficiencies from premature withdrawals, which indeed they
did.

The efficiency obtained in practice will include aspects not considered in the
experimental design. As London Economics has emphasized, the correct measure of
efficiency is based on suppliers' opportunity costs, not running costs. Because
the PX is a forward market, intertemporal efficiency requires taking account of
expectations about subsequent prices in the inc/dec, ancillary services,
hour-ahead and real-time balancing markets. This is a major reason why the
one-part schedules used in the PX Protocol are sufficient: little is added, and
perhaps some subtracted, by multi-part bidding.

The potential inefficiencies seem to be these four: market power (if it is not
mitigated), premature withdrawals (if some suppliers adopt the incremental-cost
strategy), premature termination (if necessary), and inefficient allocation of
total-energy portfolios among the hourly markets (due to the restrictions of the
Opening Rule). Each of these can be handled by appropriate measures developed in
Phase 4. In particular, we recommend that market-power mitigation not interfere
with the remarkable efficiency of the auction. The monitoring of market power
should ensure that the irrevocability of withdrawals is not used by large
suppliers to stalk smaller suppliers at the margin, as described in the February
21 report. The termination rule should be based on an estimate of the welfare
effect to ensure that losses are insignificant, and experimental tests should
measure the effect of end-game strategies.

With these provisos, we conclude that the PX Protocol is a viable market design.
As far as we can tell from the prototype, the auction is capable of high
efficiency and little prospect of successful gaming. No firm conclusion can be
drawn about the full-scale implementation in 1998, but the present evidence is
that the theory and the design principles accurately predicted the outcomes of
the tests undertaken.

Appendix A
Standard Activity Rules Used for the Tests

The following "standard" version of the activity rules is the one used for the
experimental tests. No attempt was made to test the several variants described
in the February 21 report. The Rationing Rule was "first come, first served"
based on the time stamp of each new or revised tender. This version is stated
for supply tenders; symmetric rules apply to demand tenders. The tenders are
assumed to be schedules that are step functions; modifications are required for
piecewise-linear schedules.

TENDERS: Each step of each tender is a binding offer to trade at any price not
less than the offered price. Each tender remains in force until it is withdrawn
or validly revised by the trader, or rejected by the PX. A revised tender
replaces the previous tender for the same portfolio. At the close of the
auction, those steps with prices above the final clearing price are rejected;
ties at the clearing price are resolved via the Rationing Rule. The remaining
steps are accepted, and each becomes automatically a binding contract, with the
PX as the counter-party, for the tendered or rationed quantity at the final
clearing price - except a step at the margin, for which only a portion of the
offered quantity might be accepted.

OPENING RULE: A new tender can be submitted only in the first iteration. After
the first iteration, the only valid tenders are those submitted in the first
iteration or validly revised subsequently that have not been withdrawn.

EXCLUSION RULE: An active step on a supply tender becomes frozen after the
current iteration if its offered price is not validly revised to improve the
previous clearing price, and in the previous iteration its offered price was
above this clearing price - called its Activation Price. A frozen step cannot be
revised. A frozen step becomes active again after an iteration in which the
clearing price is higher than its Activation Price.

REVISION RULE: An active step can be divided into two active steps with the same
offered price. An active step can be revised only by offering a lower price that
improves the previous clearing price. That is, the revised step must offer a new
price for the same quantity interval that is less than the previously offered
price, and also less than the previous clearing price by at least the specified
price decrement.

WITHDRAWAL RULE: After each iteration except the last, each supplier has the
option to withdraw a tender entirely and irrevocably from any hourly market. The
clearing prices are re-calculated after the withdrawal round. For the purposes
of the Exclusion and Revision Rules and setting Activation Prices, these become
the clearing prices for this iteration.

CLOSING RULE: All hourly markets close simultaneously. They close automatically
after an iteration in which no tender is revised. Otherwise, before time
expires, the final iteration is announced, and the results of the final
iteration become binding transactions at the final clearing price. After the
final iteration, an accepted tender cannot be withdrawn and the supplier remains
financially liable for delivery.

Appendix B
The Experimental Protocol

The experimental program was conducted entirely at Caltech's Laboratory for
Experimental Economics and Political Science (EEPS) by Charles Plott. The lab is
equipped with twenty-four 200 MHz HP Vectra PCs, of which one is reserved for
administration of the tests. The software was constructed by H.Y. Lee from
proprietary modules in C++ previously developed and owned by the lab. The
software mimicked the essential features of the PX Protocol and the "standard"
activity rules in Appendix A. The subjects were Caltech students solicited via
an email broadcast announcement; their remuneration consisted solely of their
trading profits - or $20 if the software crashed. New subjects received only
verbal briefings on the procedural rules; most had participated in other
experiments at the lab and therefore they were generally familiar with how such
experiments are conducted. There were no formal de-briefings after the
experiments. Testing sessions were typically conducted two or three times per
day for one or two hours over two weeks.

In a typical session each of four to twelve subjects was assigned the role of a
supplier or a demander. If the role was supplier, say, then the subject received
a two-column list of the supply cost for each of a discrete set of quantities
sold from its portfolio. The multi-market experiments assumed that this supply
function is the same in each market, except for a separate fixed start-up cost
in those sessions with this feature, whereas demand was different in each of two
or three (or four) markets corresponding to offpeak, peak, and shoulder periods.
Some sessions used the inelastic summer demand schedule and the eight
representative portfolios prepared by London Economics to simulate the
California mix. Testing in the early phase used simple linear schedules, varying
the slopes to establish the sensitivity of the outcomes to the demand and supply
elasticities. The total-energy tests were conducted by allowing such a supplier
to allocate its total supply among the three markets.

The display on a subject's PC screen showed a list of the markets, each with the
previous iteration's clearing price and tentatively accepted quantity, and the
time remaining until the close of the current iteration. By clicking on a
market, the subject obtained a smaller screen that displayed rows in which
revised offers could be entered. Each step of a current tender could be divided
into two steps at the same price by clicking on a button. Each revision was
automatically time stamped and then checked to ensure conformity with the
Exclusion and Revision Rules, and to ensure that the revised schedule was
non-decreasing. After the close of an iteration, the markets' clearing prices
were computed and displayed to the subjects, along with the quantities
tentatively accepted from that subject. In the case of multiple offers at the
same clearing price the Rationing Rule accorded priority to offers with the
earliest time stamps. The clearing price was computed as the lowest price for
which supply exceeded demand, or if demand and supply could be equalized, then
it was the greater of the highest accepted supply offer and the highest rejected
demand bid.

The auction administrator's screen showed a complete summary of the status of
the auction. After the close of the auction, summary results were stored in a
tabular form.

- ----------

* This and other reports from this project can be downloaded from
http://www.energyonline.com/wepex/reports/reports2.html

ATTACHMENT 2

TESTS OF THE POWER EXCHANGE MECHANISM

Charles R. Plott
03/10/97

The report summarizes research undertaken as initial tests of the power
exchange auction mechanism. Section 1 is an outline, orientation and overview.
Section 2 is a description of parameters. This section is written primarily for
technical economics content and can be skipped by those not interested in
experimental detail. Section 3 is a listing of the results. This section
contains observations about the efficiency and speed with which the system might
work. The section includes a subsection on problems. Section 4 is a discussion
of issues not studied but may, nevertheless, be important.

A brief comment about experimental methods in economics may be helpful.
The data discussed below were generated by the application of laboratory
experimental methods in which people participated for real money, which was
theirs to keep. The reader can be confident that the people who participated in
these markets were well trained and motivated. The subjects were upper level
students at the California Institute of Technology (primarily engineers and
scientists) who were participating for the money they might be able to earn.
They were well trained in the rules of the auction and in the accounting. If
they lost money they worked off the debts.

1. OUTLINE AND OVERVIEW

The research followed four phases. The first phase was an exploration
of the general feasibility of the mechanism. The issues posed were whether or
not the abstract market mechanism could be distilled for operational content,
implemented and produce a result. The rules must be precise, complete,
internally consistent and consistent with the realities of hardware, software
and human limitations. Practical problems could surface in many forms. For
example, the rules may contain subtle ambiguities or interact in such ways to be
inconsistent under various circumstances. Similarly, the properties of software,
timing, and coordination, might interact with strategic behavior to lead to
bottlenecks, mire the process, or cause it to loop. The initial phase of
experimentation was designed to test those aspects by attempting to produce a
working prototype.

The second phase was to ask about the performance of the system. Does
it discover price in a smooth and efficient manner? Are the allocations
efficient? The issue concerns the quality of the mechanism performance as

measured in economic terms. Is the performance of the mechanism consistent with
the advertisements? The phase in the progression of the research is a type of
proof of principle. Is it possible (in principle) that this type of mechanism
can produce the type of results that are desired?

The third phase was an effort to ascertain the extent to which the
performance of the mechanism is consistent with the theory implicit in the
mechanism design. It is a test of design consistency in the sense that the
results produced by the mechanism be understandable in terms of the classical
theory used in its design. Is the behavior of the process understandable in
terms of basic economic principles on which the theory of the mechanism rests or
were the results due to some sort of fortuitous events? The mechanism must do
the right things for the right reasons. The issue is important because the
principles on which the mechanism is designed provide the foundation for the
study of more complex settings. The theory should help identify strengths and
weaknesses which can be studied, and the accuracy of the theory itself must be
ascertained since it will be the primary tool that is used in making judgments
about how the mechanism might operate in the scaled up and complex settings that
are only found in the field application. When the mechanism goes to the field it
will necessarily be in an untested environment (by definition) so there will
only be theory on which to rest expectations.

The fourth phase centered around a set of "stress tests". These tests
were motivated by the types of problems that one could imagine encountered in
the environments in which the mechanism might find itself operating. How might
the mechanism perform when encountering those situations? These experiments also
involved some slight changes in the rules (such as the size of the delta stated
in the price improvement rules) to obtain some impression about what might
happen if some particular feature of the mechanism changed.

While the phase four tests extended the study to non classical
environments, such as cases in which producers have U-shaped cost, or cases in
which there are multiple markets and firms operate with a common cost, the
environments all retain classical features of the existence of the competitive
equilibrium. Experimental elements that are known to cause markets problems in
general (holes and strong complements that create synergisms) were not included
in the experimental settings even though they may present in the environment in
which the PX will operate.

2. PARAMETERS AND EXPERIMENTAL DETAILS

The overall pattern of experiments are contained in Figure 1. The left
column of the figure is a listing of experimental sessions, indexed by date. The
first row of the left column in the figure, labeled "1st weeks", indicates

the existence of many small experiments conducted for software testing purposes
and with little concern for the economics. These are all associated with
prototype development and while these experiments did provide data about the
operation of the auction they produced no data that can be usefully reported.
The remaining rows of the left column of the table list experiments that were
conducted and produced data that can be used in analysis.

The second and third columns of Figure 1 contain listings of detail
relevant for each of the experimental sessions. The parameters used in any
instance were combinations of specially developed sets of parameters. Thus, the
incentives for buyers were drawn from one set and the incentives for sellers
might be from a different set. The set from which they were drawn in any
particular instant is listed along a buyer's or seller's row in the Figure. In
this row, at the third column, the numbers of subjects who participated as
buyers and sellers are also listed along with the basic set from which the
parameters were drawn. These basic sets of parameters were combined in various
ways to produce different types of economic conditions and for each of the
conditions the competitive equilibrium model was applied to produce a
prediction. That prediction is contained in the row labeled Eq(P,Q) where the P
stands for the price prediction and the Q represents the quantity prediction of
the model. The actual price and quantity observed in the market are listed
(below the solid line) along with the number of rounds that took place.

Each experimental session consisted of a number of periods. The periods
can be interpreted as different market "days" in which the system faced a
particular demand and supply situation for which a price must be discovered and
an allocation determined. Thus, each period can be viewed as a different "test"
of the system. In some cases, the demand and supply situation was identical from
one period to the next and in other cases the demand and supply situation was
substantially different. The possible periods, from one to nine, are at the top
of the remaining columns in the figure. Thus, the entries in the figure give a
complete description of the parameters and results for each period of each
experimental session.

The data reported begin with experiments 021497 and 021597 on rows two
and three of Figure 1. Parameters for these experiments are in Figure 2, called
the (R_ ,C_) set. The basic parameters are a linear demand and supply with
variations of supply and demand to conditions that are steeper and have some
slight nonlinear features.

The demand and supply in Figure 2 represent the case of only two buyers
and two sellers but it is a reasonable representation of the actual experiment,
since the parameters were given to subjects in pairs. For example, if there were
six buyers and six sellers, as was the case with experimental session 021497,
the curves in the figure can be "multiplied"

by three to get the actual situation. This convention applies to all of the
figures and to all reporting of parameters.

Experimental sessions 021697, 021897, and 022097 were based on a
different set of parameters designed to test the system under different patterns
of demand and supply. The parameters used in this stage are indexed as R1,_ and
C1,_ and are found in Figure 3. The parameters R1,1+2 and C1,1+2 are linear. The
set R1, 1'+2' and C1, 1'+2' are also linear but shifted upward by a constant.
(In the notation to be used here, the first number in the series represents a
SET of parameters that were studied together and other numbers are related to
the parameters at the level of the individual agent. The prime usually means
that a constant has been added to some underlying parameter values) These
parameters were used to study the ability of the system to recover from one
level of prices to another as the demands and costs changed across periods. The
sets R1, 3+4 and C1,3+4 were used to drastically alter the elasticities of
demand and supply. The demand and supply schedules were used in various
combinations to study cases in which one side was linear and the other had a
dramatically different elasticity.

The data from the first few experiments indicated that the behavior
between buyers and sellers is symmetric. So, in an effort to use the subject
pool more efficiently, the demand side of the market was simulated in the sense
that the experimenter simply entered the demand market demand function as a
series of bids. That means that there were no strategic behaviors from the buyer
side of the market. The use of this experimental condition is reflected in the
experiments in which the number of buyers is reported to be 1 as listed in the
third column of Figure 1. In all such cases, the market demand function
implemented by the simulated buyer, assumed that the number of buyers was to the
number of sellers.

Aside from different shaped curves, three dimensions of the parameter
space were represented in the fourth phase of experiments. First, the study was
expanded to multiple markets. Within the multiple market environment (usually
two, three or four markets) three different types of cost structures were
studied.

The first test occurred in period six of experiment 021697 and the
again in periods eight and nine of session 022097. The case of interdependent
marginal cost across markets was investigated by opening three markets.
Technically, suppliers had exactly the same costs as they had when operating in
only one market but now the output had to be delivered in three different
markets, with different demands and separate bids and asks. In many respects
these were the most difficult conditions because marginal costs across markets
were not independent. Different demands existed in the three markets but sellers
had marginal cost functions of the form C((SIGMA) xi ) as

opposed to (SIGMA) C(xi). From the point of view of the competitive model this
rather drastic institutional change should have had no economic consequences.
The price should be the same in all three markets and the sum of the quantities
should equal the quantity when only one market was in operation.

A third set of parameters, indexed as R2,_ and C2,_, was used to study
the case of separable marginal costs in a two market environment. The imposed
cost functions were of the form (SIGMA) C(xi). This new dimension also involved
the addition of a fixed/start-up cost. In a single market environment, the cost
function was of the form F+C(x) and in the two market environment it was of the
form F + C(x1) + C(x2). Figure 4 illustrates the parameters for the two markets
with setup costs studied in experiment 022497. In equilibrium one fourth of the
firms should withdraw.

A fourth set of parameters, indexed as R3,_ and C3,_ added two more
markets to the previous design bringing the total number of markets to four. So,
the cost functions were of the form F+ (SIGMA) C(xi). The addition of the
fixed/start-up cost was coupled with circumstances in which some agents should
be excluded from all markets and must therefore withdraw from the bidding. These
parameters will be discussed in greater detail along with the results in the
next section.

3. RESULTS AND ALLOCATION PROPERTIES

RESULT 1. THE AUCTION IS FEASIBLE.

The mechanism can be electronically implemented from the rules. For the
most part, the rules of the auction are unambiguous. They were used to produce a
working prototype system. The prototype system has operated successfully in over
forty sessions, many of which had multiple markets (from one to twenty) in
operation. In all sessions of all experiments the auction produced a result.

However, the rules are not complete. Conditions and circumstances exits
in which the auction will produce no result because of an incompleteness of the
rules. In particular, the nature of the withdrawals and the events that must
take place after withdrawals are not completely clear. Consider, for example, a
case in which all suppliers withdraw during the same round.

RESULT 2. THE AUCTION PRODUCES PRICE DISCOVERY CLOSE TO THE PREDICTIONS
OF THE COMPETITIVE (GENERAL) EQUILIBRIUM MODEL (FOR A MULTIPLE MARKET SYSTEM).
IT OPERATES EFFICIENTLY IN THE ENVIRONMENTS STUDIED.

For the cases studied, prices tend to be near the competitive
equilibrium for a multiple market system. The only possible exceptions occurred
when

the markets were thin and one side was characterized by almost complete lack of
elasticity. The application of the auction was limited to cases in which the
competitive equilibrium exists.

A sense of the price discovery powers can be obtained from Figure 5 and
Figure 6. Figure 5 shows the relative frequency with which the market quantities
deviated by various amounts from the competitive equilibrium quantity. As can be
seen in the Figure, the mode of the results is for the competitive equilibrium
quantity to be the exact outcome. Figure 6 shows the frequency with which the
price in a period deviated by various amounts from the competitive equilibrium
price. The scale of prices is in terms of the minimum sized bidding unit. If,
for example, deviation of price from the competitive equilibrium price is 10 of
these units, it would be only a small amount in terms of a percentage deviation
form the competitive price. As can be seen the data are distributed closely
around the competitive equilibrium.

The fact that the quantities were near the competitive equilibrium
means that efficiencies are near 100%. The study of the efficiency of the
mechanism included cases in which withdrawals were necessary. In almost all
cases the high cost agents withdraw and prices are properly coordinated to
incremental cost, given the suppliers that remain.

A word of caution is needed here. The smooth nature of the process when
withdrawals were observed was accompanied by downward moving prices. The
behavior of the auction when prices are moving up has not been explored. One
could imagine how speculative behavior by sellers, under conditions of upward
movement of prices (caused by demand side adjustments) might keep firms in the
auction when they should be out. Such strategic behavior could conceivably
complicate the efficiency of withdrawals.

An impression of the data can be gained from two cases, single market
adjustment and equilibration when withdrawals are necessary. Figure 7 contains a
time series of typical adjustment paths for the single market case. Shown there
are six periods. The prices are the results of each of the rounds of price
submissions leading to a final price. The white lines are the competitive
equilibria for the different periods. As can be seen, the markets almost always
converge to near the competitive equilibrium prices and quantity. The space
between the competitive equilibrium and the final price can be attributed to a
small transactions cost almost always observed in market experiments. The
exceptions to the general convergence to the competitive equilibrium can be
found in the final two periods, which were characterized by the inelastic demand
(R3+4) in the next to last period and the inelastic supply (C3+4) in the last
period. Figure 8 contains the time paths of one market experiments with
different sets of parameters (the

C1,_+_ and R1,_+_ parameters). As can be seen convergence is always near the
competitive equilibrium.

The "stress tests" of the system began with the introduction of
multiple markets and interdependent markets. The experiments of session 022097
seen in Figure 8 were changed by simply opening three markets. The aggregate
demand that had existed in one market was " split" into three markets so,
collectively, the three were exactly like the one. The cost of each seller was
the same as before only now the C(x) that existed for the one market case became
C(x1+x2+x3). So, the cost of " total production" was exactly the same but the
output must be marketed in three different markets with different demand (but
still summed to the previous one market demand). The results are in Figure 9. As
can be seen, the prices in the markets did not converge to the single price
equilibrium that exists theoretically. Thus, markets with the interdependent
marginal costs do not adjust as rapidly or as completely as the application of
the competitive model might lead one to believe. What might be the explanation
of this or how it might be corrected was not pursued.

Figure 4 contains the parameters of two-market experiments in which
withdrawals must occur. The environment has similarities with the California
situation. One market is a peak demand and the other is not. Set-up costs exist.
The peak demand can be high or low. The results are in Figure 10. The panels are
typical of all markets studied. As can be seen, the markets always converged to
near the competitive equilibrium and in some cases slightly below. The
quantities are at the competitive equilibrium and withdrawals are efficient (in
the high 90% levels in almost all cases). The convergence to below the
competitive equilibrium is interesting and it is not clear at this time if it is
due to the rent seeking dynamics (which marginal costs are upward sloping) or to
the rules of the auction itself.

The issue of withdrawal was pursued more aggressively in the 022597
session. The parameters for the four market experiments are in Figure 11 and the
results of a typical period are in Figure 12. In these markets relatively
substantial fixed/start-up cost existed. About one fourth of the agents should
withdraw at equilibrium in these environments. All markets converged to near the
competitive equilibrium and below. In this case, the below competitive
equilibrium behavior is due to the dynamics of the mechanism itself and not to
the rents. As markets adjust at different speeds, firms that will ultimately be
excluded exist to compete in some markets before withdrawing. The result is that
prices can go below the equilibrium in some markets (but not below marginal
cost) forced there by the competition that will ultimately leave.

RESULT 3. THE DYNAMICS OF THE PRICE ADJUSTMENT PROCESS ARE
UNDERSTANDABLE.

THE UMBRELLA DYNAMICS OF BID AND ASK ADJUSTMENTS.

The dynamics of bid and ask adjustments after the opening round can be
described as similar to the opening of an umbrella. Figure 13 contains an
illustration taken from experiment 021597. The first bids and asks are tendered
with the bids marked down from demand and asks marked up from supply. These are
represented by the lighter lines in the figure. When the price is announced the
bids move up and the asks move down, opening a "tent-like" area (on its side)
with upper side being the bids and the lower the asks. As the rounds progress,
the size of the tent becomes larger and approximates the actual demand and
supply near the market price. The image is similar to an umbrella held
horizontally while opening. This movement is forced by the eligibility rules
that require movement to be at least within delta of the previous price, but the
actual behavior is accompanied by movement beyond the required delta, thereby,
giving the slope to the sides of the tent.

The nature of the price adjustment dynamic is important for
understanding how the system might perform within a variety of environments.
Prices will begin at near the top of the bids since the sellers will begin by
asking high prices and the price rule thus dictates that the price be determined
by demand. (Technically, this is a little unclear since no units are traded.
Presumably no price at the internal margin means that price is dictated by the
lowest point on the external margin which would be the demand.)

THE STUTTERING NATURE OF PRICE ADJUSTMENTS

The market is said to stutter if it does not close but continues to
produce the same price and quantity. The price adjustment path of the mechanism
is often punctuated by a series of such stutters before price movement takes
place. Thus, the time dynamics of prices and quantities are directly related to
the umbrella dynamics.

Figure 14 contains an illustration of how and why this takes place. The bids and
asks are shown as solid lines, while the underlying demand and supply are
dashed. The price is determined and the bids and asks converge to near the
previous price as shown in the upper left of the sequence. In the example the
price is above the competitive equilibrium so the number of asks exceeds the
number of bids. Thus, there is a small "chunk" of asks that are excluded. These
asks return the next period at a price of delta below the previous price,
thereby, pushing "out" a similar sized "chunk" of the old

higher asks, which have not changed since they were in the accepted set and
faced no immediate eligibility requirements. Price remains the same but the
newly "exposed" asks return to the market at the delta below price level,
thereby, pushing out another "chunk" of the old asks, which become exposed to
the eligibility requirements. These return to the market. At each of these
"stutters" it seems as if the market is doing nothing while, in fact, it is
adjusting who is in the accepted group and who is excluded. The process
continues until price is forced down by the sequentially lowered asks.

THE SPEED OF CONVERGENCE: STUTTERING PLUS JUMPS

The patterns of speed of equilibration are summarized in Figure 15.
Shown there is the relative frequency with which various numbers of rounds were
required for the mechanism to stop. These are the percentages of periods that
lasted a given number of rounds. As can be seen the median number of rounds is
between eight and nine. However, some periods required over twice that number.

Figure 16 illustrates a model of the overall adjustment process. At a
given price the number of stutters is equal to the bid quantity divided by the
excluded supply units (that are above cost) at the price. The latter would be
approximated by the number of unrestricted asks (those asks on which the
eligibility rules have placed no constraints and are, therefore, free to move
downward). After the stuttering is finished a price jump of delta takes place.
The number of price changes to ultimate convergence is the delta distance of the
price from the competitive equilibrium and the total number of rounds that will
occur is the sum of the number of stutters at each price change. Notice that the
number of rounds is related to the configuration of demand and supply. In
particular, the number of rounds is inversely related to the slope of the
demand, the slope of the supply, the excess supply, and the total quantity
demanded. If both are steep and if the supply is close to demand, and if the
quantity of demand is large, it will take more rounds for the convergence. A
simple quantitative model captures the idea. Let s = a + r be the number of asks
where a is the temporarily accepted asks and r is the temporarily rejected. The
number of iterations before a price change would be approximated by a/r . If d
is the distance from equilibrium measured in terms of the number of minimal
increments then the time to convergence is da/r. This model does not adjust for
marginal tinkering by people who are already in the accepted set, but do not
want to be out and face the increment and it does not allow for the fact that a
and r are functions of the price. Of course, both aspects of adjustment could
easily be incorporated.

The stuttering-plus-jumps dynamics of a single market can cause a
special type of interdependence in the convergence among markets. For example, a

high peak demand would cause rapid convergence in the off peak. High demand at
peak puts pressure on the off-peak periods by attracting suppliers. The
additional suppliers create more excess supply at off peak and cause prices to
move more rapidly by exacerbating all of the variables than increase convergence
speed.

4. POTENTIAL PROBLEMS

1. Speed might be a problem. The number of rounds to convergence was frequently
between ten and the high teens. If excess supply is low relative to total
demand, then the speed might even be slower than suggested here. Several options
for increasing speed have been discussed. However, such tools for speeding the
process might interact with strategic behavior and should thus be an issue of
concern, e.g., a big change late will increase time because of the stuttering.

2. Holes caused by non convexities will destroy the existence of the competitive
equilibrium. While the competitive model is very powerful it is not so clear
that it is based on sufficiently powerful principles to suggest what might
happen if the equilibrium does not exist.

3. Interdependent incremental costs across markets could be a source of
complications. The few experiments conducted suggest that the adjustments under
such conditions are not a smooth or efficient as in the separable case.

4. ISSUES STUDIED INCOMPLETELY OR NOT STUDIED

1. Big demand adjustments on the bidder side were not explored. If prices are
moving up I am not sure about dynamics. It might be possible to assure downward
movement by administratively starting the market at an arbitrarily high price.

2 The timing rounds and various options for speeding the process were not
explored. Some observations were obtained with a high delta. The limited study
suggests that a high delta might speed the convergence without hurting
efficiency but the implications for multiple market convergence have not been
pursued. Continuous time versions have not been studied. Similarly fixed times
on the number of rounds was not studied. All of the experiments reported here
had an open ending on number of rounds.

3. Various methods of ending the auction were not explored. The choice of ending
rules can have a major impact on the auction itself. There was no study of
probabilistic ending. The stuttering means it might not be a good idea to base
ending on price changes alone.

4. Conventions for dealing with mistakes, user friendly features, etc., were not
investigated. These must be carefully considered because they can become tools
to employ strategically.

5. Ideas about withdrawals were not pursued. If someone withdraws, does the
mechanism keep their old bids and asks around? Features of cheap talk are
possible, as are the potentials for cheaply hurting a competitor. Sticking
around and driving the price down for others becomes cheap talk if exposure to
risk is not present. (A person could put in a very low ask in some markets and
drive down the price for others before withdrawing.) There may also be questions
about the multiple market dynamics and withdrawal policies. (The problem here is
that the markets adjust in sequence and a person does not know if (s)he wants in
some markets until the other markets adjust.)

5. How things might scale up - we know very little about this. There are many
scaling dimensions that might be a source of problems. These include
interdependencies of various sorts.

6. Transmission and other aspects that impact on allocation were not explored.
If these cause delays or interdependencies there could be an impact.

7. Strategic behavior was not fully explored. Repeated play might introduce
unexpected behaviors. It is clear that with experience the process runs
smoother. But, there are other aspects related to the broad competitive
environment not reflected in the auction alone. A seller might be able to keep
the auction open by changing small amounts, even on accepted asks, hoping to
drive someone else out. Depending on the rules, it could even be done without
lowering prices. We have not explored attempts to sabotage the system.

Figure 1: Parameters, Predictions, Results; All Experiments, All Periods

<Table>
<Caption>
Date Periods
- -----------------------------------------------------------------------------------------------------------------------------------
subjects./ 1 2 3 4 5 6 7 8 9
parameters
- -----------------------------------------------------------------------------------------------------------------------------------

1st wks X X X X X X X X X X

021497 buyer 6 /R_ 1,2 1,2 3,4 3,4 3,4 3,4
seller 6 /C_ 1,2 1,2 3,4 3,4 3,4 1,2
Eq(P,Q) 230,30 230,30 230,24 230,24 230,24 220,27
--------------------------------------------------------------------------------------------------------------------
(P,Q) 230,40 223,18 225,14 222,13 221,21 222,17
rounds 4 8 6 5 5 5
- -----------------------------------------------------------------------------------------------------------------------------------
021597 buyer 4/ R1,_ 1',2' 1,2 1',2' 1,2 3,4 1',2'
seller 4/ C1,_ 1',2' 1,2 1',2' 1,2 1,2 3',4'
Eq(P,Q) 330,24 180,24 330,24 180,24 120,12 390,12
--------------------------------------------------------------------------------------------------------------------
(P,Q) 337,24 193,23 333,23 188,22 160,12 350,12
rounds 6 23 11 23 15 4
- -----------------------------------------------------------------------------------------------------------------------------------
021697 buyer 1/R1,_ 1',2' 1,2 1',2'
seller 8/C1,_ 1',2' 1,2 1',2'
Eq(P,Q) 330,48 180,48 330,48
--------------------------------------------------------------------------------------------------------------------
(P,Q) 340,46 190,48 340,12
370,20
340,12
rounds 3 15 18
- -----------------------------------------------------------------------------------------------------------------------------------
021897 buyer 1/R1,_ 3,4 1',2' 1,2x3 1',2' 1',2'
seller 10/C1,_ 1,2 1',2' 1,2 3',4' 3',4'
Eq(P,Q) 180,60 330,60 250,90 390,30 180,30
rd 3,d=30 rd 3,d=30
--------------------------------------------------------------------------------------------------------------------
(P,Q) 167,36 340,66 250,90 400,36 400,35
rounds 14 11 10 4 5
- -----------------------------------------------------------------------------------------------------------------------------------
022097 buyer 1/R1,_ 3,4 1',2' 1,2 1',2' 1',2' 1',2' 1,2 1',2' 1,2
seller 10/C1,_ 1,2 1',2' 1,2 3',4' 3',4' 1',2' 1,2 3',4' 1,2
Eq(P,Q) 180,30 330,60 180,60 390,30 390,30 330,60 180,60 390,30 180,60
--------------------------------------------------------------------------------------------------------------------
(P,Q) 130,30 340,58 190,60 390,31 400,30 340,60 190,60 350,9 170,18
rd 3+, d=30 rd 3+, rd 3+, rd 3+, 415,15 210,30
d=30 d=30 d=30 395,10 180,10
rounds 8 6 10 14 5 5 7 6 16
- -----------------------------------------------------------------------------------------------------------------------------------
022497 buyer 1/R2,_ normal normal high
seller 12/C2/_ peak peak peak
Eq(P,Q) 30,18 30,18 30,18
55,33 55,33 65,52
--------------------------------------------------------------------------------------------------------------------
(P,Q) 30,18 30,18 30,18
60,33 55,39 65,52
rounds 7 12 16
- -----------------------------------------------------------------------------------------------------------------------------------
022597 buyer 1/R3,_
seller 10/C3/_
Eq(P,Q) 105,9 105,9 105,9 105,9 105,9
120,9 120,9 120,9 120,9 120,9
130,9 130,9 130,9 130,9 130,9
140,9 140,9 140,9 140,9 160,9
--------------------------------------------------------------------------------------------------------------------
(P,Q) 110,6 103,9 103,9 105,9 101,9
120,6 110,9 112,9 111,9 105,9
130,7 125,10 125,9 125,10 125,9
140,7 135,10 135,10 136,9 160,11
rounds 3 12 9 19 11
- -----------------------------------------------------------------------------------------------------------------------------------
</Table>

ATTACHMENT 3

EFFICIENCIES, OPTIMAL VOLUME AND ACTUAL VOLUME
ALL EXPERIMENTS FOR WHICH EFFICIENCIES CAN BE USEFULLY MEASURED

<Table>
<Caption>
- ----------------------------------------------------------------------------------------------------------------------------------
Experiment 0215 0216 0218 0225
- ----------------------------------------------------------------------------------------------------------------------------------
Exits
(profit
of
optimal staying
Opt . Opt . Opt . exit extra
Vol. Vol. Eff. Vol. Vol. Eff. Vol. Vol. Eff. number marg.) eff

Period 1 24 24 .887 48 46 .982 36 36 .927 3 2(-2) .96
Period 2 24 21 .958 48 48 .991 71 68 .996 3 2(0) .96
Period 3 24 23 .996 44 44 .935 104 95 .989 3 1(10) .92
Period 4 24 22 .993 38 36 .993 0 0 1.00
Period 5 12 12 .951 38 35 .988
Period 6 12 12 .925
</Table>

Fixed Costs and Withdrawals: Efficiencies and Their Computation

Experiment 022597 was developed to explore the responsiveness of the system to
certain types of fixed costs. Isolated tests in previous experiments had
demonstrated that the auction would perform well for a variety of fixed cost
situations so this experiment focused on a particularly difficult situation in
which given equilibrium prices an agent with high fixed costs could see a profit
but the act of entry of such agents would drive prices below their profitable
level. Thus, one could imagine a market that experienced some type of
oscillation as firms attempted to stay in the market and then be forced to
withdraw. The movement of prices below the competitive equilibrium might force
the wrong agents to withdraw and thus cause inefficiencies.

Four critical periods existed in this experiment. The efficiencies were
respectively .96, .96, .92 and 1.00. In all cases the inefficiencies were due to
the failure of agents to withdraw that should withdraw according to theory with
a result that other agents, with lower fixed costs, supplied less output than
they were prepared to supply. In no case was the efficiency due to an agent
withdrawing who should not withdraw. Respectively the number of agents that did
withdraw relative to the number that should have withdrawn are respectively 2 of
3, 2 of 3, 1 of 3 and 0 of 0. The latter case was one in which there should have
been no withdrawals at all. Even the high fixed cost agents should have stayed
in and they did. The profitability of these agents is revealing. The number that
should have withdrawn and (profits) are respectively 1 (-2), 1 (0) and 2 (10).

The complexity of the experimental situation is as follows. In the experimental
dollars the fixed cost ranged across agents from a low of 20 to a high of 75.
Given the parameters of the experiment the equilibrium prices in the four
markets were 105, 120,130 and 140 respectively, and at this equilibrium in all
periods but the last, all agents with a fixed cost of 75 should withdraw from
all markets. The calculations for the withdrawal decision are as follows. For
all agents the marginal cost and the supply capacity in all four markets is the
same, one unit capacity at a marginal cost of 100. At the competitive
equilibrium prices the revenue of any agent delivering one unit to each of the
markets would be 495. At those prices agents with the highest fixed cost of 75
would have a total of variable costs equal to 400 and an overall profit of 20.
However, if these high fixed cost agents do not withdraw prices in the four
markets will drop to 100,105,125 and 135, respectively. For the high fixed cost
agents, revenue would be 465 yielding a loss of 10. Thus, the high fixed cost
agents are

tempted to stay in at the equilibrium prices but should ultimately be forced to
withdraw by the market dynamics.

Clearly this is a "stress test" of the classical concept of competitive
equilibrium. The extreme experimental parameters were employed because isolated
examinations in earlier experiments suggested that the mechanism would clearly
pass easier tests. Clearly the efficiency levels are high. Exactly why the
agents that did not withdraw were able to remain in the market is not clear at
this time but regardless of the reasons the resulting inefficiencies observed
are small and do not seem to be a result from any structural feature of the
auction.

ATTACHMENT 4

(i) DRAFT PAPER TO LONDON ECONOMICS ON THE POWER POOL IN ALBERTA

(ii) THE AUSTRALIAN ELECTRICITY MARKETS

(iii) OVERVIEW OF NORDPOOL

DRAFT PAPER TO LONDON ECONOMICS ON THE POWER POOL IN ALBERTA

1. BACKGROUND

In the early 1990s, the Government of the Province of Alberta started seeking
structures that were more market driven to replace the previous regulatory
regime applied to the electricity utility business in the Province of Alberta.
In 1993, the Minister of Energy directed his department to work with
stakeholders to develop a new market structure to apply throughout the Province.
A key objective was to retain low energy prices (the so-called "Alberta
advantage") within the Province. Another key objective was retail price
continuity for customers.

Alberta has a maximum demand in the order of 10,000MW, which is met mostly by
thermal generation within the Province. There are major interconnections with BC
Hydro, and smaller interconnections to Saskatchewan. The Alberta Interconnected
System is a member of the Western Systems Co-ordinating Council covering the
Western part of North America, from British Columbia down to Baja California in
Mexico.

Prior to the restructuring, the electricity sector in Alberta was dominated by
three companies:

o TransAlta Utilities;

o Edmonton Power; and

o Alberta Power.

These companies were vertically integrated companies owning generation,
transmission and having responsibility for distribution. They were responsible
for approximately sixty percent, twenty percent and twenty percent respectively,
of the generating capacity in Alberta. In addition, there were two municipal
distributors in Calgary (City of Calgary Electric System) and City of Medicine
Hat.

2. NEW MARKET STRUCTURE

In order to allow regulation by competition in the wholesale generation market,
and to allow a level playing field for new generation investors,

the market was restructured around a competitive bidding pool and
non-discriminatory transmission access.

2.1 TRANSMISSION

The responsibility for administering and charging for the transmission system is
taken over by the Grid Company of Alberta. This company does not own any of the
transmission assets, which remain with their former owners. However, all
transmission operation and access arrangements are co-ordinated through the Grid
Company, and the Grid Company pays transmission owners a contracted sum to
provide, operate and maintain the transmission facilities. The owners are bound
to put the operation of their equipment under the control of the Grid Company.

Recognising that distributors have a limited degree of freedom in responding to
locational signals, the component of transmission prices charged to distributors
is levied on a postage stamp basis. In other words, all distributors pay the
same transmission charge. Transmission charges to generators are zonally
differentiated taking into account both the cost of losses and potential
congestion costs.

2.2 THE POWER POOL

The power pool is a compulsory competitive bidding pool. Bids are accepted on a
daily basis and are binding for the day ahead, and taken as advisory for the
following six days. Prices are set hourly on an ex-post basis. This means that
Pool prices are related to the actual operation of plant as dispatched in the
event. There is no specific capacity component in the Pool prices, and
distributors bid against a price for taking demand as well as generators bidding
to supply it.

2.3 HEDGING CONTRACTS

In order to provide a smooth transition to a competitive market, and to give
some comfort to the existing generators in regard to potentially stranded
assets, a set of administered hedging contracts were put in place between
generators and distributors roughly equivalent to the status quo shortly before
market restructuring. These administered contracts ensured access to generation
by distributors at the same price as they were used to, thus fulfilling the
objective of tariff continuity. It also provided a guaranteed income for
existing generation in order to avoid the problem of stranded generation.

The initial hedging contracts cover only that demand that existed at the time of
restructuring of the market, and the duration of the contracts is tied to the
expected life of the existing assets. These two mechanisms mean that as demand
grows and as assets become life expired, the competitive portion of the market
will increase. It also means that in its first year of operation practically all
of the volume traded through the market was covered by hedging contracts, making
the Pool price largely irrelevant to the income of the four generators or the
costs seen by distributors.

2.4 COMPETITION IN SUPPLY

The market structure in Alberta does not embrace the concept of competition in
supply to final customers. Distributors have a monopoly of retail sales in their
designated supply areas, and retail prices are still subject to regulatory
control (though now through reference to the Pool price for the generation
component of tariffs).

3. OVERVIEW OF THE POOL

The Pool price is set hourly on an ex-post basis. Generators are paid the Pool
price for electricity produced in each hour. All generators, and electricity
importers and exporters, must buy and sell their electricity through the Pool.

The Pool terminology distinguishes between "offers" which are made by generators
to supply the Pool, and "bids" which are made by distributors to take
interruptible loads. Exporters (or importers) bid to take or supply blocks of
demand on a similar basis to generators and distributors.

In all cases, the basic price offer (or bid) is for blocks of generation (or
load) at increasing incremental costs. Dispatch is in discreet MW blocks. This
means that the first block offered by a generator corresponds to the minimum
stable generation at which he is prepared to operate. Offers and bids also
include dynamic performance limitations, principally:

o minimum on time;

o minimum off time;

o maximum ramp rate (MW/min);

o initial shut down time (notice required to start);

o unit run-up time (time from synchronisation to full load).

The highest offer price of any generating block used in any minute defines the
system marginal cost in that minute. Units where marginal cost flags are set are
excluded from determining marginal price (these are usually units where
operation is effectively inflexible through the application of the dynamic
constraints bid).

The Pool price for each settlement period of one hour is the average of the
individual marginal costs each minute, weighted by system demand in that minute.

If non-interruptible consumer demand is actually curtailed, the Pool price
defaults to $Can 1,000/MWh for the period of the curtailment.

4. OPERATIONAL EXPERIENCE

The Pool started operation on 1 January 1996. The managed hedging contracts
which were put in place to ensure price continuity and to underwrite the income
for potentially stranded units cover most of the power exchanged through the
Pool. This means that the income to generators (and the cost to distributors)
has not been greatly exposed to Pool price. This is a very significant point in
interpreting the behaviour of the players in the Pool during its first year.
Behaviour can be expected to be influenced to a much greater extent as the
proportion of trade covered by the managed contracts diminishes over time.

4.1 CORPORATE STRUCTURE

In response to the new market, the three major companies internally separated
their generation businesses from their distribution businesses, and started
monitoring them and setting incentives for management of them as separate
business activities. This is being extended to trading activities.

4.2 POOL PRICES

The history of Pool prices over the first year can be divided into quarters:

o In quarter one, prices followed the pattern predicted by most
observers, and were generally regarded as the product of cost
reflective bidding.

o In quarter two, prices fell well below the levels forecast,
often in the range $Can 2-4. While many argued that this was
dumping by the major generators, others claimed it was only
due to the effects of competition. In fact, there was a glut
of hydro power in the region, and this probably had the
greatest influence on prices.

o Quarter three was characterised by volatile and sometimes high
prices, most of which are blamed on poor co-ordination of
maintenance outages between generators and late returns of
plant to service after overhauls. In addition, generators were
developing more complex bidding strategies to get market value
for surplus generation.

o Quarter four saw higher than expected Pool prices throughout
the hours of the day, mainly due to greater than expected
plant unreliability, high demands both on the Alberta
Interconnected System and throughout the region (resulting in
high export volumes), and high gas prices.

There is a general consensus that the market has operated reasonably well, and
that the higher than expected levels of volatility reflect the real market
situation and are not the result of perverse behaviour by generators. There is
also a strong link between the market price of gas and the market price of
electricity. As both are used for space heating in winter, there is bound to be
a link between prices in both markets and the weather. During the autumn cold
spell, prices in the spot gas market have been very high.

4.3 COMMERCIAL BEHAVIOUR

Perhaps the biggest surprise to the members of the Pool was the extent to which
the hedging contracts exposed generators to commercial risk. Although the
contracts are purely financial instruments, the managed ones are firm power
contracts, and the generator must maintain his difference payments even when his
unit is unavailable. In this respect, the contracts closely resemble the risk
profile of bilateral contracts for physical supply. This effect particularly
hurt the operators of base load plant, where the cost differentials between the
contract price and Pool price could be high.

Some of the larger users of electricity have complained of the generators
exercising market power to control prices unfairly. No formal proceedings have
been initiated, and hard evidence of the abuse of market power is difficult to
find. It is common in newly formed competitive markets for large electricity
users to develop unrealistically

optimistic views of the prices obtainable in such markets, and this can lead to
frustration.

Generators can re-declare the availability of their plant lower than originally
offered. On occasions, this can lead to the dispatcher having to use very
expensive plant (belonging to the same generator) to meet a short term shortfall
in capacity. With ex-post price setting, this means that Pool prices will be
pushed up, and generators benefit. While this phenomenon has occurred on a
number of occasions, most opinion suggests that the outcome of this type of
gaming is so uncertain as to rule it out as a general commercial strategy. Time
will tell!

4.4 OTHER ISSUES

Distributors have expressed a wish to be able to bid prices for all of their
load, not just the interruptible component. They argue this would help them
inter alia set more interesting and adventurous tariffs for retail customers.

While not related to the market restructuring, growth in demand in Alberta has
been unexpectedly high. This will reduce the relative importance of the managed
contracts earlier than had been envisaged, and advance the need for capacity
additions.

All companies have felt the increased volume of work related to regulation. An
issue that may further increase that volume is the retirement of units currently
covered by the managed contacts but felt to be uneconomic by their owners.

A significant voluntary market in hedges has not yet developed, and risk
management is generally confined to swaps (short term floating or fixed).

The original scheduling optimiser was replaced by Cougar in mid-19966. This was
part of a long planned strategy, and did not reflect dissatisfaction with the
existing tool in respect of pool settlement. The replacement was an "expensive"
exercise, but had no discernible effect on Pool prices.

4.5 PRINCIPAL CONCLUSIONS

o The Power Pool is functioning satisfactorily, according to
most observers

o Issues concerning market power and pool rule manipulation tend
to surface as Pool prices increase. These issues have yet to
be fully understood and distinguished from competitive
advantage.

o The market is still evolving, and the Power Pool Council is
considering refinements to Pool design in 1997.

o Pool prices are exhibiting extreme volatility.

o The market for physical and finical hedges is still illiquid.

o Due to high levels of uncertainty, trading is primarily
focused on short term fixed or floating swaps.

THE LEGISLATED OBLIGATION/ENTITLEMENT OPTION BETWEEN THE GENCOS AND DISCOS
ACCOUNTS FOR THE MAJORITY OF AVAILABLE CAPACITY. DURING EXTENDED PERIODS OF HIGH
POOL PRICE, GENCOS' PRIMARY FOCUS IS ON MEETING THEIR OBLIGATION.

1. THE AUSTRALIAN ELECTRICITY MARKETS

The South East Australian power system currently comprises four separate regions
operating independent market arrangements. They are:

o Victoria, which has led Australian power sector reform with
privatisation and restructuring around a power pool;

o New South Wales (NSW), which introduced an experimental
pooling system as early as 1992, and in 1996 restructured its
generation and distribution sector, and introduced a pool
based market;

o The Snowy Mountain Hydroelectric Scheme (`Snowy'), which is
owned by the Federal Government, which currently trades into
both the NSW and Victorian markets through trading
arrangements specific to the each recipient pool; and

o South Australia (SA), which trades with Victoria and uses
Victorian pool prices as a reference for its own power
transaction valuations, but does not have open market
arrangements.

In addition, there is a process of `national' market reform that aims to develop
a single market across Queensland, NSW, Snowy Victoria and South Australia --
the National Electricity Market (NEM). The most likely outcome from this
national reform will be the amalgamation of the Victorian and NSW markets, at
which stage South Australia and Queensland (when finally interconnected with
NSW) will join.(4)

The Australian markets that are of direct relevance to the PX and ISO formation
in California are the Victorian and NSW markets. The NEM is not yet operating
(although key sections of the NEM are of interest, particularly in respect of
congestion), but both the Victorian and NSW markets are evolving towards the
NEM. Unfortunately, neither NSW or Victorian markets have been operating very
long, so it is difficult to develop firm conclusions. However, the Victorian
market has a three year history, and will therefore form the main focus for
discussion.

- ----------

(4) Indeed, several interim market arrangements known as NEM0, NEM1 and
NEM2 have been developed which define the steps for coalescence of the
NSW and Victorian markets.

TABLE 2 SUMMARY OF VICTORIAN AND NSW ELECTRICITY REFORMS

<Table>
<Caption>
STATE OWNERSHIP INDUSTRY GENERATION TRANSMISSION
- ----- --------- -------- ---------- ------------

VIC Private Horizontal separation Prices for Open & non
distribution/ of generation to wholesale discriminatory
retailing. station level, vertical electricity set in access. Prices
Generation separation of pool and by private regulated by
privatisation generation, contract Office of
underway. High transmission (PowerNet negotiation. No Regulator
voltage Victoria), market significant General (ORG)
transmission and operator (VPX), 5 generation entry
system operation privatised distribution restrictions.
public businesses, competition Limits imposed on
emerging from cross ownership.
interstate retailers Vesting contract
cover declining to
2000

NSW All activities 3 generators. Prices for Open &
public. Entry of Separation of wholesale non-discriminatory
private generation transmission electricity set in access. Ring
anticipated and (TransGrid) from pool and by private fencing of
limited new entry generation. contract transmission and
of retailers. Amalgamation of 25 negotiation. market
distributors into 2 Declining vesting operations.
large metropolitan contract coverage, TransGrid
and 4 rural options for revenue capped
distribution/retailers. extending coverage adjusted on CPI-X
being considered

<Caption>
TRADING
STATE DISTRIBUTION SUPPLY ARRANGEMENTS
- ----- ------------ ------ ------------

VIC Accounting Maximum VicPool III
separation within uniform wholesale spot
distribution & tariffs market. All
supply business. subject to energy traded
Open & non price controls through the pool.
discriminatory and in place Establishing a
access to wires. until 2000 for joint pool with
Information the franchise NSW (NEM 1) which
disclosure of market. may migrate to a
accounts. full NEM.
Price cap
regulation of
wires charges.
Revenue based
on return on
assets

NSW Ring-fencing of Gross margin Competitive pool
retail and for the retail in place from
network operations businesses March 1996;
of distributors. distributors franchise market
Open & regulated by regulated.
non-discriminatory CPI-X. Currently
access to wires. Declining establishing a
Regulation of franchise joint pool with
wires charges Victoria.
</Table>

1.1.1 RELEVANT MARKETS

In this discussion, the relevant market are the pool (or spot market), the
contract market which may include formalised short-term forward markets
associated with the power exchange or pool company, and the ancillary services
market.

1.1.2 SOME COMMON MARKET FEATURES

The Victorian market and NSW market have a number of important common features:

o all are based on the ex post pools in which spot prices are
based on the actual operation of the system rather than the ex
ante anticipated operation of the system;

o all power is traded through the pool;

o there are no proscribed contract markets associated with
either the Victorian or NSW pools. All contract transactions
are essentially private. The NEM makes provision for centrally
co-ordinated markets for short-term energy price hedges, but
these have not been implemented in Victoria or NSW;

o both pools are based on a system of self-commitment for slow
start plant. Thus generators that take more than 30 minutes to
synchronise can determine their own commitment schedule;

o there are no markets for ancillary services as such. Rather,
there are appropriate power to secure appropriate ancillary
services. The NSW market rules make greater provisions for
payments and/or charges for ancillary services than the
Victorian market; and

o the pools have been based on a pool or power exchange company
rather than a mutual contract between pool members (the model
adopted in England and Wales).

1.2 THE VICTORIAN MARKET

All wholesale electricity in Victoria is traded through VicPool, which started
operation in July 1994. VicPool is operated by the VPX (the Victorian Power
eXchange) and was established as part of the reforms taking place in the
Victorian ESI.

1.2.1 PARTICIPANTS

There are four main classes of participants in VicPool which will continue to
operate in the NEM:

o the generators;

o the distributors, who purchase electricity from the pool and
sell it onto customers;

o the large customers who purchase energy from the pool to meet
their own energy requirements; and

o the Traders, who deal with the historic contract obligations
of the Victorian ESI prior to reform, such as Loy Yang B,(5)
Snowy, interstate trading and the Victorian aluminium
smelters.(6)

1.2.2 SIZE OF MARKET AND MARKET PARTICIPANTS

The Victorian market is small in overall size, and individual generators are
significant in terms of overall demand. For example, the larger base-load
generators represent as much as 25% of peak-time capacity requirement. For
example, Table 2 shows the thermal generators listed in Schedule 4 of the
Victorian market rules. Table 4 shows the demand bidders. In total this
comprises no more than 9,000MW of capacity in a market where peak demand is
about 7,500MW.

- ----------

(5) A majority stake in the 1000 MW Loy Yang B power station was sold to
the private sector prior to these Victorian reforms.

(6) Electricity is supplied to Alcoa's two aluminium smelters under
long-term contracts with the Government of Victoria signed in 1984,
which predate these reforms.

TABLE 3. GENERATING UNITS

<Table>
<Caption>
PARTICIPANT GROUP OF UNITS UNIT TYPE UNITS

Energy Brix Morwell Complex Notional Units MOR01
MOR02
GenVic Newport PS Actual Generator NPSD
----------------------------------------------------------------------
Generators 1-4 of Actual Generators JLA01
Jeeralang A PS JLA02
JLA03
JLA04
----------------------------------------------------------------------
Generators 1-3 of Actual Generators JLB01
Jeeralang B PS JLB02
JLB03
Hazelwood Generators 1-8 of Actual Generators HWPS1
Hazelwood PS HWPS2
HWPS3
HWPS4
HWPS5
HWPS6
HWPS7
HWPS8
Loy Yang A Generators 1-4 of Loy Actual Generators LYA01
Yang A PS LYA02
LYA03
LYA04
Loy Yang B Trader Generators 1-2 of Loy Actual Generators LYB01
Yang B PS LYB02
Snowy Trader Victorian Snowy Notional Units SWV1
Entitlement SWV2
SWV3
SWV4
SWV5
SWV6
Victorian Hume Notional Units HUME
Entitlement
Southern Hydro Combined output from Notional Units SHLA
Dartmouth, Eildon, West SHLB
Kiewa, Clover and McKay. SHLC
SHLD
SHLE
SHLF
----------------------------------------------------------------------
Combined output from Notional Unit RCC
Rubicon and Cairn Curran
Yallourn Energy Generators 1-4 of Actual Generators YWPS1
Yallourn W PS YWPS2
YWPS3
YWPS4
</Table>

TABLE 4. DEMAND BIDDERS

<Table>
<Caption>
PARTICIPANT GROUP OF UNITS UNIT TYPE UNITS

Smelter Trader Potlines of Alcoa Actual potlines APD01
Portland Smelter APD02
-------------------------------------------------------------------------
Potlines of Pt Henry Actual potlines PTH01
Smelter PTH02
PTH03
-------------------------------------------------------------------------
Snowy Trader Victorian Entitlement Notional Unit T3PMP
of Tumut pumping
-------------------------------------------------------------------------
Victorian Entitlement Notional Unit JNPMP
of Jindabyne pumping
</Table>

1.2.3 ADMINISTRATION

The pool is administered by the Victorian Power Exchange (VPX). VPX is a
statutory authority, although the Government has not ruled out privatising VPX
at some stage in the future. VPX has two main areas of responsibility:

o MARKET OPERATOR: operating and administering VicPool,
including controlling dispatch to ensure generation meets
demand, and providing information to market participants; and

o HIGH VOLTAGE NETWORK: ensuring that system security is
maintained at an appropriate level; operating the transmission
system, and planning the augmentation of the high voltage
network (which is owned by PowerNet Victoria (PNV)).

1.2.4 BIDDING, COMMITMENT AND DISPATCH

VicPool is an ex-post pool which operates broadly in the same way as the
proposed NEM:

o generators and demand side bidders submit bids to VPX,
specifying a price for each quantity - these bids are then
stacked in merit order;

o demand is estimated, and plant is scheduled to meet demand;

o checks are made to ensure that system security is not being
violated, and generators are instructed accordingly in the
provision of ancillary services such as reserve and reactive
power; and

o generators are dispatched to provide active power to meet
demand.

THE STRUCTURE OF THE BIDS

The information relating to generator bids is shown in Table 5 taken from the
VicPool Rules, Amendment 21. The bid structure is essentially the same as the
proposed NEM bid structure.

TABLE 5. INFORMATION IN RELATION TO GENERATING UNITS(7)

<Table>
<Caption>
Value is
Quantity SI Unit applicable(1) for the Permitted Range of Value
- -------- ------- --------------------- ------------------------
UNIT ID UNIT Designated code for UNIT
- ------- ---- ------------------------
SELF-COMMITMENT FLAG DAY ON or OFF
- -------------------- --- ---------

ELBOW 1 MW DAY 0-10,000
ELBOW 2 MW DAY ELBOW 1 to 10,000(3)
ELBOW 3 MW DAY ELBOW 2 to 10,000(3)
ELBOW 4 MW DAY ELBOW 3 to 10,000(3)
ELBOW 5 MW DAY ELBOW 4 to 10,000(3)
ELBOW 6 MW DAY ELBOW 5 to 10,000(3)
ELBOW 7 MW DAY ELBOW 6 to 10,000(3)
ELBOW 8 MW DAY ELBOW 7 to 10,000(3)
ELBOW 9 MW DAY ELBOW 8 to 10,000(3)
INCREMENTAL PRICE 1 $/MWH DAY 0 to VOLL
INCREMENTAL PRICE 2 $/MWH DAY INCREMENTAL PRICE 1 to VoLL
INCREMENTAL PRICE 3 $/MWH DAY INCREMENTAL PRICE 2 to VoLL
INCREMENTAL PRICE 4 $/MWH DAY INCREMENTAL PRICE 3 to VoLL
INCREMENTAL PRICE 5 $/MWH DAY INCREMENTAL PRICE 4 to VoLL
INCREMENTAL PRICE 6 $/MWH DAY INCREMENTAL PRICE 5 to VoLL
INCREMENTAL PRICE 7 $/MWH DAY INCREMENTAL PRICE 6 to VoLL
INCREMENTAL PRICE 8 $/MWH DAY INCREMENTAL PRICE 7 to VoLL
INCREMENTAL PRICE 9 $/MWH DAY INCREMENTAL PRICE 8 to VoLL
INCREMENTAL PRICE 10 $/MWH DAY INCREMENTAL PRICE 9 to VoLL
OVERLOAD PRICE $/MWH DAY INCREMENTAL PRICE 10 to VoLL
OFFLOADING PRICE 1 $/MWH DAY 0 to 1,000,000
OFFLOADING PRICE 2 $/MWH DAY OFFLOADING PRICE 1 to 1,000,000
DAILY ENERGY MWH DAY(2) 0 to 100,000

OVERLOAD BAND SIZE MW SETTLEMENT PERIOD 0 to 10,000
AVAILABLE CAPACITY MW SETTLEMENT PERIOD 0 to 10,000
MINIMUM GENERATION MW SETTLEMENT PERIOD 0 to AVAILABLE CAPACITY
BACKOFF MINIMUM MW SETTLEMENT PERIOD 0 to MINIMUM GENERATION
COMMITMENT STATUS SETTLEMENT PERIOD 0 or 1
INFLEXIBILITY STATUS SETTLEMENT PERIOD 0 or 1
Notes:
</Table>

(1) Values specified in this column as being applicable for a:

(a) DAY must be the same for every SETTLEMENT PERIOD in a DAY and
can be different for each DAY; and

(b) SETTLEMENT PERIOD can be different for each SETTLEMENT PERIOD.

(2) DAILY ENERGY applies to the entire DAY and can be different for each
DAY and can be updated in any SETTLEMENT PERIOD of the SCHEDULING
PERIOD.

(3) Subject to a MINIMUM BAND SIZE of the UNIT.

- ----------

(7) Source: VicPool Rules Amendment 21

The pool rules allow two bids for revenue in the event that a generator is
required to run below its minimum stable generation or below its backoff
minimum. These bid values provide the mechanism for resolving de-commitment
problems in the event that self-commitment results in excessive capacity.

BALANCING

There is no need for a balancing market given the ex-post nature of the market.
Prices are based on actual generation.

RE-BIDDING

A participant may at any time alter or update the bid/offer information in the
bid/offer database in relation to one or more of its Units in respect of a
settlement period which commences after the time at which the alteration or
updating occurs. In altering or updating bid/offer information in the bid/offer
database, a participant must act in good faith.

A participant must not alter or update:

(a) the self-commitment flag; or
(b) the incremental prices; or
(c) the elbows; or
(d) the offloading price 1; or
(e) the offloading price 2; or
(f) the overload price,

stated in the bid/offer database for a unit for a day after 11.00 am on the day
before that day.(8)

A participant must not alter or update:

(a) the commitment status; or
(b) the available capacity,

stated in the bid/offer database for a unit for the settlement periods falling
on a day later than 37 hours before the start of that day, except:

- ----------

8 Source: VicPool Rules Amendment 21

(a) in order to reflect a change in availability of the unit due
to an event or events beyond the reasonable control of that
participant; or

(b) in order to reflect an increase in availability of the unit
due to an event which the participant could not reasonably
forecast; or

The conditions under which key components of the bid can be changed are
therefore limited, reducing the scope for gaming through re-bidding. As far as
we are aware, the limited ability to change bid/offer information does not cause
problems in VicPool.

1.2.5 POOL RULE CHANGES

There have been four main phases of VicPool. These changes have been gradually
undertaken to merge the Victorian pool rules and institutions with those
proposed under NEM market rules. The latest phase is known as VicPool III
enhanced, and commenced operation on 1 September 1996. Several important changes
were made to VicPool III as part of the movement towards the NEM arrangements:

o DAILY BIDDING: previously generators placed weekly bids. Under
VicPool III enhanced generators place daily bids with VPX;

o INCREMENTS TO BIDS: previously the generators were able to bid
their capacity into the pool in three increments. In VicPool
III enhanced generators are able to bid their capacity into
the pool in 10 increments; and

o SELF COMMITMENT: previously VicPool operated on the basis of
central commitment. In their bids generators were required to
submit start up costs, start up times and minimum on and off
times. VPX analysed the costs and times presented by each
generator and took the start up and close down decisions.
Under VicPool III enhanced generators are required to
self-commit.

There are still several key differences between the NEM rules, and the VicPool
rules currently in operation:

o REGIONS: the NEM is a market with a series of regions, linked
by interconnects. VicPool operates in a single region. Trade
with other States is managed in the context of VicPool by the
IOA Trader;

100

o SHORT TERM FORWARD MARKET (STFM): VicPool has no STFM. To fill
the price discovery role performed by the STFM in the NEM, VPX
publishes seven day ahead indicative prices. The generators
are required to submit to VPX indicative bids on a seven day
ahead rolling basis. The generators are not required to submit
actual bids related to these indicative bids. VPX then
calculates and publishes indicative prices, to provide market
participants with an indication of short term prices;

o MOVEABLE ELBOWS: the bids for VicPool and NEM each contain 10
price capacity bands, however unlike the NEM bids, the VicPool
bands do not have `moveable elbows'. This means that the MW
capacity bid into the pool cannot be sculpted by half hour.
Rather the MW capacity in each band are fixed throughout the
day, and only the price of each band can vary;(9) and

o TREATMENT OF LOSSES: in VicPool all customers pay a pro rata
share of losses. In the NEM losses within a region are
calculated with regard to a reference node. Customers and
generators at the same point pay and receive the same price,
but customers and generators at difference points will pay and
receive different prices, depending on loss factors from the
reference node. Under the NEM loss factors between regions
will be determined dynamically.

OTHER CHANGES IN THE POOL RULES

Since the inception of the market there have been a number of changes to the
VicPool rules, to develop upon the initial rudimentary pooling arrangements, and
to bring it more into line with the proposed NEM rules. Two main changes are of
particular significance:

o a change in the structure of generator bids at the end of
1994, whereby generators were allowed to bid three capacity
bands as opposed to the one band allowed to that date; and

o the foregoing move towards bidding and pool rules akin to the
NEM (i.e. 10 price bands for each unit and self-commitment) in
October 1996.

Hence, pools based on the self-commitment of thermal plant have only been in
operation 8 months (since May 1996 in NSW and October 1996 in

- ----------

(9) The NSW market does include moveable capacity bands. This has caused
some unexpected volatility in pool prices and may provide opportunities
for gaming, although it is difficult to draw conclusions over the
limited extent of operation.

101

Victoria). It is therefore difficult to draw definitive conclusions from the
experience to date.

1.2.6 ANCILLARY SERVICES

The generator licenses require the Victorian generators to provide ancillary
services at the request of VPX. The Tariff Order limits VPX's expenditure on
ancillary services to $20 million per annum. This contrasts with expenditure on
ancillary services in the NSW market of around $80 million per annum. As a
consequence generators receive payments for only a small proportion of ancillary
services. Most ancillary services are undertaken by generators to accordance to
with the requirements of the Code.

The ancillary services provided by each generator vary. For example, the gas
turbines are typically used for black starts, while the coal stations are used
for frequency control. VPX allocates the requirement to provide ancillary
services in such a way that the obligation is shared evenly among the
generators.

At present a major review of the provision of ancillary services in Victoria and
NSW is underway, in preparation for NEM1. It is anticipated that the
recommendations from the review will be implemented in July 1997, when NSW and
Victoria adopt a joint system security policy.

1.3 CONTRACTS IN VICPOOL

There are several mechanisms for managing risks in VicPool. Generators and
retailers can hedge against pool price volatility using:

o vesting contracts; and

o contestable contracts.

Generators can hedge against the risk of an outage leading to large contract
liabilities under a fixed contracts under the generator coinsurance scheme.

1.3.1 VESTING CONTRACTS

Each generator in VicPool holds a vesting contract with each distributor, and
the Smelter Trader. The vesting contracts were put in place on 1 July 1995. The
vesting contracts cover consumption by franchise and some Tariff H
customers.(10) The MW cover under the contract declines with the reduction

- ----------

(10) Historically, large commercial and industrial customers were supplied
under a tariff known as Tariff H. When these customers became
contestable they were given the alternative of entering the contestable
market, or remaining on a Tariff H safety net tariff. Those customers
on the Tariff H safety net tariff were covered in the vesting
contracts.

102

in the franchise market set out in Table 6 until the market becomes fully
contestable in December 2000.

The vesting contracts contain two distinct types of contract cover:

o A TWO WAY DIFFERENCE CONTRACT: under the two way difference
contract generators and distributors compensate one another
for movements in the pool price around the strike price. The
difference contract only applies at pool prices less than
$300/MWh; and

o A ONE-WAY NON-FIRM HIGH SMP HEDGE: under the high SMP hedge
generators are required to compensate distributors for pool
prices above $300/MWh (in March 1994 dollars). The contract is
non-firm because generators are only required to compensate
distributors to the extent they are producing at the time. In
return for this contract cover the distributors pay the
generators a monthly option fee.

The distributors are covered for actual franchise and Tariff H consumption
within +/-7.5% of the forecast load. There is also a provision within the
vesting contracts to vary cover should any Tariff H customers move to the
contestable market. The franchise reduction is shown in Table 6.

103

- --------------------------------------------------------------------------------

TABLE 6 VICTORIAN FRANCHISE REDUCTION STRATEGY

- --------------------------------------------------------------------------------

<Table>
<Caption>
ESTIMATED NUMBER OF
INTRODUCTION OF CUSTOMERS AFFECTED
COMPETITION (CUMULATIVE) CUSTOMER LOAD
- --------------- ------------------- -------------

December 1994 47 > 5MW

July 1995 377 > 1MW

July 1996 1,877 > 750MWh

July 1998 7,000 > 160MWh

December 2000 1,960,000 All customers, assuming
no technical or economic
constraints
- --------------------------------------------------------------------------------
Source: Office of State Owned Enterprises Department of Treasury, Reforming
Victoria's Electricity Industry December 1994 and NSW Electricity Reform
Taskforce, Retail Competition in Electricity Supply, June 1996
</Table>

1.3.2 CONTESTABLE CONTRACTS

The decline in the franchise market, set out in the threshold reduction
strategy, has meant a corresponding increase in the contestable market. Most
retailers have signed contestable contracts with generators in order to hedge
the risks associated with supplying their contestable customers. There is no
publicly available information on contestable contract terms and prices.

1.3.3 GENERATOR COINSURANCE

The Victorian Government set up the generator coinsurance scheme with the intent
of providing a mechanism for managing unavailability risk, enabling the
generators to enter into firm contracts. The scheme commenced on 1 July 1995 and
expired on 30 September 1996. All Victorian generators previously owned by the
SECV were required to participate in the scheme.

The scheme acted as a mechanism to protect participants from exposure to high
pool prices by sharing the revenue from high pool prices among the Victorian
generators. Under the scheme, the generator who is contracting for cover pays to
all other generators a premium, and those other generators compensate the
contracting generator during times of high pool prices. The amount each
generator contributes to the hedged generator's difference payment varies with
the electricity they sent out (and therefore the revenue they earned) during the
relevant period.

104

In addition to the generator coinsurance scheme, the generators often signed
`back-to-back' contracts with other generators to manage their exposure to pool
price at times when they were out on maintenance. A number of schemes have been
implemented since the expiry of the generator coinsurance scheme. These
replacement schemes are bilateral arrangements between contract counterparties,
like the back-to-back contracts, and do not involve any central co-ordination or
compulsory participation like the coinsurance scheme.

1.3.4 INDUSTRY CODES OF PRACTICE

The licenses require industry participants to comply with industry codes and
pool rules. These codes are developed by industry participants. There are a
number of industry codes:

o POOL RULES: the pool rules govern the operation of the
Victorian wholesale market;

o SYSTEM CODE: the system code sets out the requirements for
ensuring the safe and secure operation of the Victorian
electricity system;

o WHOLESALE METERING CODE: the wholesale metering code is
designed to ensure that electricity flows are appropriately
measured in order to facilitate the trade of wholesale
electricity through the pool;

o DISTRIBUTION CODE: The distribution code regulates the
physical supply of electricity from a distributor's network
and the way in which customer's installations affect the
network;

o SUPPLY AND SALE CODE: the supply and sale code regulates the
conditions under which distributors sell electricity to
franchise customers; and

o RETAIL TARIFF METERING CODE: the retail tariff metering code
governs the installation of new equipment.

1.4 HISTORY OF VICTORIAN POOL PRICES

Figure 1 shows the monthly, time-weighted average SMP and system demand for
VicPool since the market commenced in July 1994. The results show no obvious
correlation between pool prices and the seasonal patterns of supply and demand,
suggesting that other factors such as generator

105

behaviour, contract cover and regulation are as important as the supply demand
balance.(11)

- --------------------------------------------------------------------------------

FIGURE 1. MONTHLY AVERAGE PRICES AND SYSTEM DEMAND IN VICPOOL

- --------------------------------------------------------------------------------

[GRAPH]

There are clearly two phases in the life history of the Victorian market:

o the `early' period to 1 January 1996 and the period
thereafter. In the early stages of the market, prices remained
consistently above $30/MWh; indeed, in 1995 they were above
$40/MWh. These are prices above the level one would expect in
a competitive market with over-supply, being close to or above
new entrant prices

o the `late' period thereafter, where prices spot fell below
$21/MWh on average, which is consistent with the observation
that the market is over-supplied, so prices should be below
new entrant prices.

- ----------

(11) Similarly, in the UK pool, the majority of major price movements have
been caused by changes in generator behaviour or factors external to
the pool such as regulatory action of the effect of fuel and hedging
contracts.

106

`EARLY' PRICES

The change in market outcomes reflects significant changes in
market circumstances. The early stage of the market was
characterised by:

o modest over-supply particularly when one considers that
opportunity exports from Victoria to more expensive neighbours
were restricted in this period;

o common interests amongst the generators which were effectively
under common ownership;

o public ownership;

o overall levels of contract cover in the market significantly
below the level of expected demand; and

o high levels of vesting contract cover at prices of between
$35/MWh and $40/MWh.

The net effect of these market conditions was that generators were willing to
sacrifice output (in the spot or short-term market) in order to raise pool
prices. Examination of the bidding patterns of generators at that time shows a
remarkable degree of consistency in the bids of supposedly independent
generators, such that:

o base load generators bid between 50% and 70% of their capacity
at a relatively low price, but consistently bid the remainder
of their capacity at prices between $30/MWh and $40/MWh, even
though this price was well above their short-run operating
cost. The bidding tended to maintain the `natural' merit
order; and

o mid-merit and peaking generators bid in accordance with the
bid of the high price bands of their base-load counterparts,
even though these bid prices were well above operating costs.

This was suggestive of significant degree of tacit collusion -- i.e. the ability
to develop common bidding strategies that sustainably increased profits for all
participants above competitive levels without formal communications. The
repeated nature of the bidding into the pool in combination with the information
release rules provided a good environment for such an outcome. The degree of
common interest was also reflected in contract prices over the period which
almost universally offered identical terms as the vesting contracts.

107

There were other contributory factors to pool price outcomes at the time. For
example:

o throughout the first year of VicPool's operation there was
considerable price volatility as generators tested out the new
market. The price dips in October 1994 and March 1994 are
thought to be due to a degree of experimentation of this sort;
and

o the winter of 1995 was unusually cold in Victoria, leading to
above average demand. This increase in demand was accompanied
by an unusual level of unavailability amongst the Victorian
generators which led to very high prices. It is likely that
some capacity gaming was taking place at this time,(12)

but the most compelling explanation is supra-competitive prices through tacit
collusion.

`LATE' PRICES

Prices fell significantly in 1996. In the lead-up to the Hazelwood sale, the
Electricity Supply Industry Reform Unit (ESIRU) commissioned a review of the
factors influencing the price.(13) This review, which was summarised in the
Hazelwood Information Memorandum, identified a number of factors which
contributed to the period of low prices. The reasons cited by ESIRU were:

o THE COMMENCEMENT OF OPERATION OF LOY YANG B UNIT 2: in January
1996 a new 500 MW unit came into full operation at Loy Yang B.
This unit has a relatively low marginal cost and a high level
of contract cover. It was therefore bidding most of its
capacity into the pool at a low price. The early commissioning
of this unit effectively left the market over-contracted.
Hence, other generators were forced to lower their prices
somewhat to ensure that they still covered their contract
allocations;

o GAS STATION TAKE OR PAY CONTRACTS: the gas fired stations at
Newport and Jeeralang purchase their gas under take or pay
(TOP) contracts, but were well short of their minimum take
quantities.(14) Hence, the effective marginal cost of their
fuel was low. They

- ----------

(12) Capacity gaming happens when a generator deliberately makes some plant
unavailable with a view to raising prices. This is more attractive to
generators in periods of high demand when the rise in prices due to
withdrawal of capacity is likely to be at its steepest.

(13) A review of the basis for recent low prices in the Victorian
Electricity Market by Hugh Bannister of Intelligent Energy Systems Pty.
Ltd.

(14) Due to a variety of reasons including the Loy Yang B unit commissioning
and mild, low demand summer.

108

therefore bid low to ensure their TOP quantities were utilised
prior to the expiry of the contract in December 1996;

o UNUSUALLY LOW LOAD: due to a mild summer, peak demand over the
period was 10% below expected values; and

o ETSA CONTRACT: in December 1995 the ETSA contract was
allocated to Hazelwood. As a result Hazelwood changed its
bidding strategy in order to ensure that sufficient capacity
was dispatched to cover this requirement.

In April 1996 the Victorian government removed the TOP obligation from these
generators which allowed the gas stations to resume their previous bidding
strategies;(15) This measure temporarily increased pool prices, but was not
sufficient to permanently increase pool prices. One is therefore tempted to
suggest that, whilst the foregoing did contribute to the fall in pool prices,
the major factor was the breakdown of the market conditions that fostered tacit
collusion. The major factors in this regard were:

o the addition of the 500MW of fully contracted Loy Yang B
capacity; and of equal importance

o the significant change in market imperatives for the
privatised generators, Yallourn and Hazelwood

Central to this discussion is an understanding of the way in which generators
bid in their contract cover into the pool. The major generators in VicPool have
a high proportion of their expected generation covered by hedging contracts.
This gives them an incentive to ensure they are dispatched to meet their
contract commitments, particularly if they anticipate that pool prices will be
low in comparison to contract prices. They therefore tend to bid in that portion
of their capacity covered by contracts at close to their marginal operating
costs.(16)

YALLOURN AND HAZELWOOD

Yallourn and Hazelwood were purchased on the basis of business plans predicting
very high availability and capacity factors -- that is, they were expected to
operate at full output to supply the base-load market in Victoria and
interstate. Their financing arrangements and business plans were not

- ----------

(15) That is, the Government agreed to bear the cost of failing to meet the
TOP obligations set out under the contract, which should have offset
the increase in base-load capacity from Loy Yang B.

(16) Bidding below marginal operating costs could result in operating
losses, particularly in Victoria which has surplus base-load capacity.

109

conducive to reduced output in order to raise pool prices. Accordingly both
stations have taken aggressive positions in the wholesale contract and spot
markets. This is reflected in pool prices and in contract prices which reputedly
are well below $30/MWh at present. Thus, contract prices are currently below
vesting contract prices and below estimates of new entry costs.

It is not clear that either Hazelwood or Yallourn understood the impact of their
own strategies on market outcomes, or whether their business planning was based
on earlier development of the national electricity market (NEM) than has
occurred.(17)

RULE CHANGES

In September 1996 prices dropped which corresponded with the commencement of
VicPool III enhanced. The key change at that time was the introduction of 10
part bids and self-commitment. Some initial volatility in prices might be
expected after such a change as generators varying their bidding strategies to
learn how the new market arrangements influence pool prices. However prices
since then have remained subdued -- weekly average prices have remained low,
often below $20/MWh. And the average pool price over the whole of 1996 is only
$21/MWh.

Hence, one would tend to suggest that rule changes have been less significant a
factor in causing market prices to reflect the market supply demand balance than
the change in commercial imperatives that eroded the conditions for tacit
collusion.

1.4.1 CONCLUSIONS

The price path in VicPool shows several important lessons:

o it is possible for the generators to influence the level of
prices. For sustained periods of the time prices have been
above `competitive levels'. The repeated nature of bidding
into the pool provides an environment for tacit collusion.
This collusion is further facilitated by the availability of
bid information to pool participants. However, this also
relies upon appropriate financial and contract positions by
the market participants;

o in the relatively small Victorian market the pool price is
relatively sensitive to unusual or external events. For
example, direct

- ----------

(17) Access to the NSW market that contains 3 portfolio generators (under
common ownership) that might be willing to sacrifice output could help
to re-establish market conditions conducive to tacit collusion.

110

intervention on the fuel arrangements for Newport (500MW) had
a significant impact on pool prices, as did the industrial
action at Yallourn immediately after its sale;

o once Victoria is operating in an interstate pool the magnitude
of price shocks will decrease, and market conditions that
sustain tacit collusion may return;

o diversity of financial and contract positions amongst market
participants

111

OVERVIEW OF NORDPOOL

Introduction

In January 1996 a joint Norwegian-Swedish trading exchange was opened based on
open access transmission networks and free competition between generators. This
followed the initiation of electricity market reform in both countries in
1990-91. Open access was introduced to the Norwegian market first in 1991
following the 1990 Act. Both countries separated the grid from generation
creating two new companies - Statnett in Norway was established in 1991 and
Svenska Kraftnatt followed in Sweden in 1994. A new Swedish electricity Act was
passed in 1995 allowing the market to be opened to competition in 1996.
Nordpool provides short-term physical markets and medium-term financial futures
markets for trading electricity. These markets exist in addition to bi-lateral
physical contracts between generators and distributors exchange outside Nordpool
and which account for 85% of physical trade. Traded volumes in Nordpool are
increasing and the experience so far has been generally positive. Further
developments of the existing markets are being considered and it is hoped that
the open market area will eventually also include Finland and Denmark.

This paper summarises some of the key components of Nordpool and provides a
brief commentary on the markets' operations to date.

Market Structure

Norway and Sweden have fully committed their electricity industries to Nordpool
- - there is no other rival exchange though of course there are bi-lateral
contracts exchange outside Nordpool. Additional imports/exports involve the
industries in Finland, Denmark, Germany and Russia. Both.

Generation mix

Norway is 99.5% hydro with annual production of some 120 TWh. There is a small
thermal capability of around 300 MW industrial co-generation plant. There is
substantial hydro reservoir capacity of around 80 TWh that fills from May to
August and allows water to be carried forward from wet years into dry years.

Sweden, in contrast, has a more mixed system comprising 50% hydro, 45% nuclear
and about 5% conventional thermal in terms of share annual energy output of
around 150 TWh. In capacity terms nuclear's share comprises 30%, hydro 50% and
thermal 20%. Most of the thermal capacity is industrial co-generation. Sweden
has less reservoir storage capacity than

112

Norway and runs down its reservoirs during the period of peak winter demand.

The neighbouring system of Denmark has only thermal power plant which is mostly
coal-fired. Finland is more like Sweden with a mix of hydro, nuclear and
conventional thermal.

Market participants

Nordpool currently comprises 43 generators, 43 distributors, 16 brokers/traders,
14 industrial producers/consumers and 3 market makers. About 100 of these
participants are Norwegian(18) where the open market system has been established
for longer but Swedish participation is increasing.

The size of the market participants is far from equal. In Norway, about half of
Norwegian output is accounted for by just four companies: Statkraft which at 30
TWh is around 25% of total output, Oslo Energi (6 TWh), Lysekraft (5 TWh) and
Bergenshalvoens Kommunlaer Kraftselskap (4 TWh). Moreover, on the demand-side,
the largest distributor - Oslo Energi (8 TWh) - is over double the size of the
next largest distributor Nord-Trondelag Elecktsiteitsverk (4 TWh). About half of
the 200 distributors also own power plant.

Sweden has a similar structure with around 300 utilities in total. However
Vattenfall alone accounts for just over 50% of generation capacity (17 GW) and
output. The next largest generators are Sydkraft (5 GW) and Stockholm Energi (2
GW) which are substantially smaller. On the demand-side the largest utilities
are more equal in size: Sydrakft (c.6 TWh) and Stockholm Energi (c.6 TWh) are
the largest and Gottenburg (c.4 TWh) is only slightly smaller. These three
distributors each have between 240,000 and 420,000 customers.

Ownership

Ownership is mixed. The largest generation companies in both Norway and Sweden -
Statkraft and Vattenfall - and the grid companies are in state ownership. There
are a few private companies, such as Gullspangs Kraft in Sweden and Norsk Hydro
in Norway. But most companies are municipally owned reflecting the historical
development of the industry from small scale townships. Some have mixed
municipal and private ownership, others are wholly municipally owned such as
Oslo Energi, while others are co-operatives.

- ----------

(18) The Norwegian industry itself comprises around 250 separate utilities
many of which are very small. the size of each participant is far from
equal.

113

The Markets

Nordpool consists of a two main markets: the physical day-ahead spot market and
the financial market for weekly contracts. The system operators Statnett and
Svenska Kraftnatt are responsible for overall system stability and regulation
and operate slightly different balancing markets. These are now briefly
described.

Day-ahead spot market

BIDDING

Each morning, players submit bids to buy or sell for each hour of the following
day. The day runs midnight to midnight. Participants in Norway with generators
or loads in different geographical locations will submit separate bids for the
different locations. These locations are defined weekly by Statnett. Sweden is
treated as a single region (see the section on Constraints below). At noon, the
market closes and no further bids are accepted. The bids are firm and binding.

BID FORMAT

Participants submit a price/quantity curve for each hour. This shows the
quantities in MW that the participants is prepared to supply (a positive MW) or
purchase (a negative MW) from the spot market at different prices. Prices are
usually specified in Norwegian Kroner (NOK) per MWh (though there is provision
for Swedish Kroner to be used).

IMPORTS/EXPORTS AND BI-LATERAL CONTRACTS

Nordpool co-ordinates the bids and planned power exchanges with Finland, Denmark
and Russia. Statkraft provides Nordpool with details of the power flows over the
interconnector which it has contracted with Denmark. All distributors(19)
provide Statnett and Svenska Kraftnatt with a schedule of bi-lateral contracts
for the day ahead to assist in the management of constraints.

PRICE FORMATION

Nordpool balances supply and demand by stacking up the supply and demand curves
of the market participants. A price in NOK is calculated for each hour of the
day ahead by 1300 hours at the latest of the prior day and the exchange notifies
each player of the prices and quantities of their trades. If there any disputes
to be resolved, these should be notified by 1430 - and prices and quantities
recalculated if necessary.

- ----------

(19) Distributors have the responsibility of informing Nordpool of
contracted movements of electricity into and out of their control area.
This is carried out on a weekly basis.

114

TREATMENT OF CONSTRAINTS

In some circumstances there may be transmission constraints. These are
relatively rare since both the Norwegian and Swedish systems are strong. However
different approaches to dealing with the constraints are currently in use.

When the system spot price is calculated, potential load flows are compared with
available transmission capacities. This allows the presence of constraints to be
identified. In Norway, the market is split when there is a constraint and prices
in different regions are calculated. Thus the net exporting region will benefit
from a low price relative to the constrained, net importing region.

In Sweden, Svenska Kraftnatt takes full responsibility for constraints. In other
words, Svenska Kraftnatt will buy power downstream of a constraint and sell it
upstream. This effectively means that Svenska Kraftnatt `subsidises' generators
required to generate because of constraints so that they are in merit. This
different approach is possible because such constraints do not occur often in
Sweden. Moreover Svenska Kraftnatt was concerned that the Norwegian approach
might result in excessively small regional markets being created that would be
open to excessive manipulation. At the same time Svenska Kraftnatt is keen to
see the market develop towards more continuous trading on the spot market that
would mean that the Norwegian approach could not be applied.

COMMITMENT AND DISPATCH

Each generator is responsible for his own commitment and dispatch to meet his
contractual obligations under bi-lateral contracts and day-ahead spot market
trades.

BALANCE ADJUSTMENT IN SWEDEN

Sweden (but not Norway) operates a Balance Adjustment service. This provides
players with the opportunity to make additional trades up to 2 hours before the
due hour for delivery. The bids in the Balance Adjustment take the same format
as in the day-ahead spot market. Balance Adjustment market is cleared and prices
set 2 hours before the due hour.

This balance adjustment had been a part of the Swedish system prior to the
establishment of Nordpool. In contrast to Norwegian generators, Swedish
generators valued the flexibility of being able to alter their bids up the last
minute to take account of unforeseen changes in the balance of supply and
demand. It is arguable that the run-of-river hydro and system of dams in a
cascade makes this particularly useful in Sweden, especially for small
generators that cannot easily optimise their contractual commitments within

115

a large portfolio of generating plant. In addition the thermal generation in
Sweden demands flexibility to adjust its output - and hence its costs - at the
margin (marginal output in Norway from hydro is arguably less costly). However
volumes traded in this market have been small (and there is some possibility
that it may be abandoned following the experience of Swedish participants with
the Nordpool markets).

BALANCING MARKETS

Following the determination of spot market trades, each generator is able to
finalise the planning of their generation schedules. These plans which include
spot market and bi-lateral contract trades are submitted to the system operators
in Sweden and Norway by around 19.30 hours of the prior day. At the same time,
market participants can also submit bids to provide the system operators with
access to regulating energy. This is dealt with slightly differently in Norway
and Sweden.

BIDDING

Bidding into the balancing markets is possible up to half and hour before the
due hour in Sweden and up to three hours in Norway.

BID FORMAT

In Norway and Sweden, participants offer bids to decrease production/consumption
at given prices in each hour. A `staircase' is then constructed of prices in
NOK/MWh in Norway (and in Swedish Kroner per MWh, SEK/MWh, in Sweden) at which
participants are prepared to regulate production/consumption. These bids are for
increments/decrements at 15 minutes notice.

PRICE FORMATION

In Norway, at the end of each hour, the most expensive bid on the
`up-regulation' side (or the least expensive bid on the `down regulation' side)
is paid to all players called upon to regulate up (and/or down). In Sweden a
distinction is made between active and passive regulation. Thus, Swedish
generators/consumers that are called upon to regulate their
production/consumption will be paid the balancing market marginal price. However
generators/consumers that provide assistance to the system through deviations
from their contracted production/consumption but which have not been called upon
to do so, will be paid at the balance-adjustment price. This is intended to
discourage over reliance on the balancing market. Deviations in prices between
the balancing markets and the spot markets are

116

larger in Sweden than in Norway due to the greater proportion of thermal
generation.(20)

FINANCIAL MARKET FOR WEEKLY CONTRACTS

CONTRACT-TYPES

There are two types of contract that can be traded on the financial market:

o BASE LOAD POWER covering 24 hours of each day for a full week;
and

o PEAK-LOAD POWER covering 0700-2200 hours Mondays to
Fridays(21).

These contracts can be trade as SINGLE WEEKS up to between 4 and 7 weeks in
advance, as BLOCKS OF FOUR WEEKS from between 5 and 8 weeks and up to 52 weeks
in advance, and as SEASONS OF SEVERAL BLOCKS 1-3 years in advance. The contracts
are specified as forward contracts struck against the system spot market price
(and not the regional price in the event of constraints).

FORM OF TRADING

Since November 1996 trading is electronic (terminal-based) for both types of
contract. This market is open each week-day from 1130 to 1500 hours. Nordpool
has a help desk for participants not connected electronically. In practice, only
the baseload contracts are traded in any significant volume and the peak-load
contracts may be discontinued.

SETTLEMENT

Settlement is on a daily basis with the appropriate gains/losses being credited
or debited to the a bank account that each player places at the disposal of the
exchange. Players requiring physical delivery of power that hold a financial
contract may simply submit a spot market bid without any price attached to it.

ANCILLARY SERVICES

Ancillary service markets are not included in Nordpool but are provided by the
system operators. Secondary reserve is essentially provided though the balancing
markets where response is required at 15 minutes notice. However

- ----------

(20) There are around 30 bidders into the Swedish balancing market as
opposed to around 7 in Norway.

(21) The contract for off-load power (covering 2200-0700 hours Mondays to
Fridays and 0000-2400 hours Saturdays and Sundays) has been
discontinued due to lack of demand.

117

reserve capacity is not contracted for since the hydro reservoirs are considered
to be adequate for this function. Spinning reserve and reactive power are not
currently paid for by the grid company in Norway but are simply required to be
provided by each major generator. In Sweden Svenska Kraftnatt pays generators
for spinning reserve.

Statnett and Svenska Kraftnatt are both in discussion with generators on
suitable system of payment for frequency control and reactive power. It is
possible that separate markets may be developed for these functions but there
are no proposal to do so as yet.

INFORMATION RELEASE

The bid information of each market participant is held confidential by Nordpool
and is not released to market participants. Nordpool only makes public aggregate
information.

TRANSMISSION PRICING

The transmission pricing systems in Norway and Sweden are slightly different.
This follows from their different treatment of constraints described above.

NORWEGIAN TRANSMISSION PRICING

There are four elements to the Statnett transmission charge:

o CONNECTION CHARGE in NOK/kW of connected generation capacity
(based on winter capacity when rivers may be low) or maximum
load at time of system peak at grid supply point plus embedded
generation;

o POWER CHARGE in NOK/kW of connected generation capacity net of
load or maximum load net of embedded generation capacity;

o ENERGY CHARGE based on estimated marginal energy losses
estimated annually for 5 geographic areas and for three times
of day calculated from loss factors multiplied by the system
pool price. This recovers approximately twice the actual cost
of average losses.

o CAPACITY CHARGE which comprises the excess revenue collected
through the treatment of congestion described above. Statnett
recovers the value of energy paid for on the low price side of
a constraint and sold into the high price side of a
constraint.

118

Swedish transmission pricing

There are three elements in the Svenska Kraftnatt transmission charge:

o POWER FEE in SEK/kW per annum based on ex ante estimates of
maximum input/output at point of grid connection. This charge
varies by geographic location with inputs in the South being
more expensive than inputs in the north.

o ENERGY FEE based on loss factors determined for each grid
connection (150 nodes) by time period at the price Svenska
Kraftnatt has negotiated with generators for supplying it with
losses.

o INVESTMENT FEE to cover one-off investments in special
circumstances such as new connections.

MARKET COMMENTARY

TRADED VOLUMES

The spot market accounts for around 15-20% of energy consumption. The balancing
markets are then a small fraction of the spot market trades. The majority of
electricity transactions are still under bi-lateral contract.

NORWEGIAN-SWEDISH TRADE

At the beginning of 1996 the exceptionally cold weather caused the inter-country
transfers to exceed transmission capability and the markets had to be split.
Bi-laterally contracted transfers were around 1045 MW and additional net
purchases from Nordpool exceeded limits set at 1,800-2,000 MW. The Swedish peak
prices reached 450 NOK/MWh which caused some concern. Especially among consumers
not covered by financial contracts that were struck against the system price.
However from February to about June the system prices have been the same in
Norway and Sweden except for a few isolated hours and one week-end. During the
summer and autumn, the Norwegian price has frequently been higher than the
Swedish price for several hours a day owing to the dry weather conditions.

119

[GRAPH]

PRICE VOLATILITY

Price volatility has increased significantly for Norwegian participants.
Previously prices in Norway were remarkably stable showing a daily variation of
10-20 NOK/MWh. The introduction of Nordpool introduced price fluctuations of up
to 100 NOK/MWh on a day. This follows from the introduction of Sweden's thermal
capacity to the exchange and some cold, dry weather. Previous dry years in
Norway have also led to price volatility - before the inclusion of Sweden.

NORDIC TRADE

During the first part of 1996, Sweden relied not just on Norway. Additional
imports were made from Denmark, Finland and Germany. In the later months of
1996, Norway began to import significantly from Sweden to compensate for low
reservoir levels.

MARKET MANIPULATION

The rising prices in Nordpool during 1996 have prompted some concern about
market manipulation. Some commentators have suggested that large generators have
withheld water in order to drive up the spot price to ensure a favourable
benchmark price against which bi-lateral contracts could be re-negotiated.
However the rising prices can also be justified with respect to the cold and
dray weather conditions.

120

There has also been some concern about pricing up bids behind transmission
constraints in Norway. However such constraints are in any case relatively rare.

SOURCES:

Knut Fossdal (Nordpool) and Roger Kearsley (Svenska Kraftnatt) A
Norwegian-Swedish Trading Exchange for Electricity, paper presented to UNIPEDE,
November 1996

Jan Moen (Director of Regulation, NVE) A Common Electricity Market
Norway and Sweden: Prerequisites, development and results so far, May 1996.

London Economics confidential papers and briefings

121

ATTACHMENT 5

UK MARKET STUDY BIBLIOGRAPHY
(QUESTION 2)

Richard J. Green and David M. Newbery, "Competition in the British
Electricity Spot Market," Journal of Political Economy, 100(5): 929-953, October
1992.

Mark Armstrong, Simon Cowan, and John Vickers, Regulatory Reform:
Economic Analysis and the British Experience, MIT Press, 1994. Chapter 9.

Michael A. Einhorn (ed.), From Regulation to Competition: New Frontiers
in Electricity Markets, Kluwer Academic Publishers, 1994. [see especially
Chapters 3 by Vickers and Yarrow and 4 by Green]

John Vickers and G.K. Yarrow, "The British Electricity Experiment,"
Economic Policy, 12:188-232, 1991.

Nils-Henrik von der Fehr and David Harbord, "Spot Market Competition in
the U.K. Electricity Industry," Economic Journal, 103: 531-546, May 1993.

Richard Green, "The Electricity Contract Market," Cambridge University,
mimeo, May 1996.

Richard Green, "Increasing Competition in the British Electricity Spot
Market," Journal of Industrial Economics, 1997.

David Newbery, "Power Markets and Market Power," Energy Journal, 16(3):
41-66, 1995.

A. Powell, "Trading Forward in an Imperfect Market: The Case of
Electricity in Britain," Economic Journal, 103: 444-453, March 1993.

Catherine Wolfram, "Strategic Bidding in a Multi-Unit Auction: An
Empirical Analysis of Bids to Supply Electricity in England and Wales," Harvard
University, mimeo, January 1997.

Michael Crew (ed.), Pricing and Regulatory Innovations under Increasing
Returns, Kluwer Academic Press, 1996. [see article by Patrick and Wolak,
"Industry Structure and Regulation in the England and Wales Electricity Market"]

122

Frank Wolak and Robert Patrick, "The Impact of Market Rules and Market
Structure on the Price Determination Process in the England and Wales
Electricity Market," Stanford University, mimeo, 1996.

G. McKerron and P. Pearson (eds.), The British Energy Experience: A
Lesson or a Warning, Imperial College Press, London, 1996.

Alex Henney and Simon Crisp, "Lessons for the U.S.? Transmission
Pricing, Constraints, and Gaming in England and Wales," Electricity Journal,
January-February 1997, 17-23.

Richard Gilbert and Edward Kahn, International Comparisons of
Electricity Regulation, Cambridge University Press, 1996. [see chapter on the
privatization and operations of the UK power pool]

Frank Wolak, "Market Design and Price Behavior in Restructured
Electricity Markets: An International Comparison," Stanford University, mimeo.
[This is a cross-country comparison of the systems in England-Wales,
Norway-Sweden, Victoria-NSW, and New Zealand.]

123

ATTACHMENT 6

EXAMPLE CALCULATION OF USAGE CHARGE
(QUESTION 9)

The following simple example shows the calculation of the Usage Charge where two
Scheduling Coordinators both have load and generation in each of two Zones (A
and B) where the Usage Charge is equal to the difference in PX price in the two
zones and the PX is revenue neutral.

The transmission line connecting Zones A and B has a maximum available capacity
of 150MW. Both Scheduling Coordinators wish to use 100MW of the available
capacity. The diagrams below show each Scheduling Coordinators' Preferred
Schedule.

SCHEDULING COORDINATOR 1

[CHART]

Bid prices and adjustment bids are the same.
Inc/Dec pair adjustment bid: 100MW @ $30

SCHEDULING COORDINATOR 2

[CHART]

Bid prices and adjustment bids are the same
Inc/Dec pair adjustment bid: 100MW @ $10

124

Scheduling Coordinator 1's preferred schedule uses 100 MW, and values
the use of the transmission line at $30/MW; that is, it values a
decrement in output of the zone A generator coincident with an
increment of its zone B generator at $30 for each MW. On the same
basis, Scheduling Coordinator 2 values the use of the transmission line
at $10/MW.

Total demand to use the transmission line is 200 MW. The ISO must
adjust schedules 50 MW to obtain a total schedule equal to the max.
transmission line capability. Adjustment bids (bids to reduce
congestion) are: 100 MW @ $10 (SC2), and 100 MW @ $30 (SC1). The ISO
selects 50 MW of the low, $10 bid. The Usage Charge will be $10 in this
case. SC1 receives 100 MW of access and SC2 receives 50 MW of access.

THE OUTCOME

[CHART]

Scheduling Coordinator 1 does not supply any of its loads in Zone B
from generation in Zone B. On the other hand, Scheduling Coordinator 2
was unable to schedule all the transmission it wished to use across the
transmission line between Zones A and B and, therefore, had to supply
part of its load in Zone B from generation in Zone B.

If SC1 were the PX, the adjustment bid based PX price in zone A would
be $30, the marginal cost in zone A, and in zone B it would be $40, the
marginal cost in zone B.

If SC2 were the PX, the adjustment bid based PX price in zone A would
be $10 in zone A, and in zone B it would be $20.

125

EXAMPLE 2

The following example illustrates a case where the PX would not be
revenue neutral. We will continue with the simple two bus radial system
to illustrate the case. In this example, as shown by the following
preferred schedules of the two SCs, the max. transmission line
capability is 100 MW and the total demand for access is 300 MW.

SCHEDULING COORDINATOR 1

[CHART]

SC1's schedule calls for 150 MW of transmission capacity.

SC1's adjustment bid Inc/Dec pairs are: 75 MW @ $15 and 75 MW @ $30.

SCHEDULING COORDINATOR 2

[CHART]

126

SC2's schedule calls for 150 MW of transmission capacity.

SC2's adjustment bid Inc/Dec pairs are: 80 MW @ $10 and 70 MW @ $25.

Total demand for transmission is 300 MW. The ISO needs to make 200 MW of
adjustments to stay within the 100 MW max. capability of the transmission line.
The 4 adjustment bid offers available to the ISO are:

[CHART]

The ISO will select SC2 for 80 MW + 45 MW = 125 MW of adjustment to its original
150 MW schedule.

The ISO will select SC1 for 75 MW of adjustment to its original 150 MW schedule.

That is, SC2 receives 25 MW (150 MW in preferred schedule - 125 MW adjustment)
of the transmission capacity between zones A and B at a Usage Charge of $25/MW,
and SC1 receives the remaining 75 MW of capacity at the same charge.

127

OUTCOME

SCHEDULING COORDINATOR 1

[CHART]

SCHEDULING COORDINATOR 2

[CHART]

SC1's zone A marginal cost, based on its adjustment bids is $30, and its zone B
marginal cost is $60. If SC1 were the PX, since the Usage Charge is $25, the PX
would have a revenue surplus after paying the Usage Charge if it set the zone
price for energy at these marginal costs.

SC2's zone A marginal cost, based on its adjustment bids is $10, and its zone B
marginal cost is $20. If SC2 were the PX, since the Usage Charge is $25, the PX
would under collect revenues to pay the Usage Charge if it set the zone price
for energy at these marginal costs.

128

It should be noted that the dollar values used in these examples have been
chosen to keep the arithmetic of the example simple, although they have the
effect of exaggerating the actual magnitude of the over or under collection that
might occur.

When the PX is the marginal user of transmission (SC2) it would under-collect
its revenue need by setting the zone price for energy at the marginal adjustment
bids. When it is an infra-marginal user (SC1) it would over-collect. The PX may
be in either the position of marginal user or infra-marginal user in any given
hour. The PX will, therefore, adjust the zone adjustment bid marginal cost based
prices up or down in each hour to preserve Usage Charge payment revenue adequacy
in each hour and eliminate over and under collection of its revenues.

129