California's Market Problems
California's power problems is slow regulatory processes for approval of new
plants and a dysfunctional deregulation program. However it seems price
increases in California's electricity markets have changed because of - unit
outages, congestion effects, weather, and, in particular, significant increases
in natural gas prices. The average Day-Ahead Unconstrained Market Clearing Price
(UMCP) for January 2000 was 49% higher than the UMCP for January 1999.
Similarly, the average UMCP for February 2000 was 58% higher than the UMCP for
February 1999. The increase was most noticeable for the offpeak period in which
the increase from 1999 to 2000 was 58% for January and 77% for February. Several
external factors contributed to the increase in price including the following:
1. Higher load forecast due to increased population and economic growth.
2. Lower generation output from in-state hydroelectric units requiring greater
reliance on higher priced resources.
3. Higher natural gas prices across the state.
4. RealTime Pricing
5. Generation owners were withholding some of their capacity from the
Day-Ahead market in anticipation of higher prices in the Real Time market.
Higher Load
The load forecast for the ISO system increased by 6.5% from January 1999 to
January2000 and by 5.6% from February 1999 to February 2000. The increase in
load reflects the general economic growth in California over the past year.
Despite this growth in the load forecast, the Unconstrained Market Clearing
Quantity (MCQ) in the CaIPX Day-Ahead market did not increase by the same
magnitude. The average MCQ increased from 1999 to 2000 by 4% in January and by
only 1 % in February.
Supply Resource Comparison
The change in resources with the most influence on the increase in prices
between 1999 and 2000 was the reduction of in-state hydroelectric generation in
2000. Approximately 45% less hydro was available in January 2000 and 32% less in
February 2000 as compared to the same months in 1999. The volume of Qualifying
Facilities (QF) resources also decreased, but this reduction was most likely the
result of a change of existing resources from the QF classification to SC
Transfer classification caused by the divestiture of geothermal resources or the
buyout of QF contracts. The nuclear units and SC Transfers increased volume from
1999 to 2000. In 1999, San Onofre 1 nuclear unit was off-line for most of
January and February, and Diablo 1 nuclear unit was out of service for most of
February 1999. All nuclear units were back on-line in the first two months of
2000. The increase in nuclear units was somewhat offset by the decrease of QF
resources, both of which are must-run, zero bid units. SC Transfers, including
energy from gas generating units of New Generation Owners (NGOs), also increased
resource volume. For consistency, the energy from these gas units bid by the OUs
prior to divestiture is also included in this classification. The increase in SC
Transfer energy likely contributed to the increase in supply offer price and the
resulting increase in MCP. This was caused by the increase in the cost of the
natural gas used to fuel most of the SC Transfer energy. The increase in gas
price not only affected the bid price for SC Transfers, but also the opportunity
cost of imports, which increased volume from January 1999 to January 2000.
Natural Gas Prices
The natural Gas prices at two western locations Southern California Borderwith
supplies from the Southwest region, and Surnas with supplies from the Pacific
Northwest- showed significant increases in 2000 vs. 1999 for the first two
months of the year. Both prices include interstate transmission costs to
California, but do not include in-state local distribution charges to deliver
the gas to the bumertip of the generating resources. Gas prices show an increase
of nearly 30% from January 1999 to 2000 or an increase of about $0.30/MMBtu.
From February 1999 to February 2000, gas prices increased an average of nearly
50% or $0.80/MMBtu.
The price of natural gas would have the greatest impact on the MCP if gas units
were setting the market price as opposed to coal units or hydro based imports It
is not possible to verify exactly that gas units are the marginal resource of
energy through the CaIPX because most of the gas based energy is bid as a SC
Transfer, which is not bid by a specific resource. In general NGO market share
dropped in the first quarter of 2000. This can be explained, in part, by higher
winter natural gas prices. Because most NGOs own gas-fired units, it is likely
that unfavorable spark spreads led to decisions not to generate as much as they
would in the summer months. Despite electricity demand growth and a lack of new
generators, several factors put the state in a better position this year than
last. The improved factors include more contracts for interruptible supply,
greater public awareness of the serious situation and a year of experience for
new owners of generators.
Real Time Pnces.
Real Time volumes were 41% higher than the comparable period for 1999. The
effect of high Real Time volumes is this: when real-time volumes are high, ISO
dispatchers must deviate from their hour ahead schedules to focus on actual load
instead. When this occurs, ISO dispatchers periodically must work well through
the stack to meet the needs of actual load. The more ISO dispatchers use up
bids, the less liquid the market becomes, in turn placing upward pressure on
prices. High Real Time volumes in Q1 2000 appear to be unrelated to ISO
Forecasted and Actual Loads.
The Latest
There will Energy Company representatives and FERC members testify before the
House Energy and Power subcommittee to attribute California's market problems
This week. Other news: San Diego County and the City of San Marcos have both
begun "preliminary research" into the feasibility of creating municipal electric
utilities. Heat and heavy demand in California sent the State into a Stage Two
Emergency Thursday from 1400 until 1900 PDT, following a transmission capacity
curtailment of several hundred megawatts. The line reportedly sagged due to hot
temperatures and heavy loads, igniting a tree, and cutting the amount of
available transmission from the Northwest. Moreover, the California ISO
indicated that 3964 MW of unidentified generation was off-line in the Golden
State due to mechanical failure or unplanned maintenance. "Our weather service
is reporting temperatures in the 90s in San Francisco and above 100 in Los
Angeles," said one trader. "Reserves are tight, and power distributors are
starting to fret over potential rolling blackouts." With the mid-month daily
market strong, trades for the balance-of-September were up from deals done
Wednesday. At the California-Oregon Border, trades for the month ranged between
180 and 1 88$/MWh, while SP-1 5 contracts were seen as low as 164$/MWh and as
high as 168.5$/MWh. NP-15 September OTC contracts sold for between 165 and
167$/MWh. In unit news, Diablo Canyon #2 (1087 MW) was in Mode 2 startup
Thursday morning after spending nearly 10 days in hot standby due to a forced
maintenance outage. The unit was expected to reconnect to the grid by Thursday
afternoon.
Study on Zonal Price Correlations -
A brief background discussion on congestion resolution processes in California
and how market participants must engage in valuation process to deal with
congestion.Figures 11 and 12 show that zonal prices were much more volatile in
Q1 2000. On March 14, 2000, the SP15 zonal price spiked to $126.49/MWh when PV
experienced line de-rates and heavy congestion.
On February 1, 2000 the ISO implemented a new zone referred to as UZP26." This
zone is entirely located between the two existing zones, NP15 and SP15 and has
no interties that allow for imports.
When the implementation of this zone was being considered, the rationale was
that the zonal price for ZP26 would always be the lower of NP1 5 or SP1 5 zonal
prices. This view stemmed from the following:
1. If congestion occurs in the South to North direction in California, Path 15
(the Northem-most of Path 15 and Path 26) should be the constrained line. The
resulting usage charge should then cause the NP15 zonal price to be higher than
the SP1 5 and ZP26 zonal prices.
2. If congestion occurs in the North to South direction, Path 26 (the
Southern-most of Path 15 and Path 26) should be the constrained line. The
resulting usage charge should then cause the SP1 5 zonal price to be higher than
the NP15 and ZP26 zonal prices.
However, as seen in the insert in Figure 14, the zonal price for ZP26 has been
higher than both NP1 5 and SPI 5 for 20 percent of the total number of hours in
February and March. The average ZP26 zonal price for the entire period was
higher than that of either NP15 or SP15.
When a zone has a higher price than surrounding areas, the argument is that a
higher level of demand existed in that zone. The fact that ZP26 has higher zonal
prices is odd, given that ZP26 has roughly only 300 MWh of demand and 3,093 MWh
of generation capacity.
One important caveat is that, this report has only two months of data available
for analysis. The variance from expectations noted in this report may be a
function of limited data and the newness of ZP26. However, the variance signals
a need for careful tracking of ZP26 market effects in the next quarter.
IOUs experienced decreases in supply market share because of the unit
divestitures outlined previously. In general NGO market share dropped in the
first quarter of 2000. This can be explained, in part, by higher winter natural
gas prices. Because most NGOs own gas-fired units, it is likely that unfavorable
spark spreads led to decisions not to generate as much as they would in the
summer months.
Real Time prices. In Q1, 2000, Real Time volumes were 41% higher than the
comparable period for 1999. The effect of high Real Time volumes is this: when
Real-Time volumes are high, ISOdispatchers must deviate from their hour ahead
schedules to focus on actual load instead. When this occurs, ISO dispatchers
periodically must work well through the BEEP stack to meet the needs of actual
load. The more ISO dispatchers use up bids, the less liquid the market becomes,
in turn placing upward pressure on prices.
High Real Time volumes in Q1 2000 appear to be unrelated to ISO Forecasted and
Actual Loads, as shown in Figure 30. One explanation for this is that generation
ownerswere withholding some of their capacity from the Day-Ahead market in
anticipation of higher prices in the Real Time market.
Figures 26 and 27 show the ISO Actual System Load. The average load for all
hours increased by about 5% from Q1 1999 to Q1 2000. In looking only at the
peaks, off- peak hours actually showed a larger increase (6%) than on-peak hours
(2.4%). This is consistent with Table 3 i.e., PX off-peak prices increasing more
than on-peak between the two quarters.
All Ancillary Service markets had some price spikes this quarter as Figures 32 -
36 show. Spikes were generally due to what the ISO calls "bid insufficiency."
Finally, in March 2000, the average prices for Regulation Up were higher than
Regulation Down for the first time since the redesign of Ancillary Services
(Table 18).
Further analysis will be undertaken if this pattern persists.
In contrast to the previous quarterly report, the NP1 5 zonal prices were not
the highest in Q1, 2000. Prices in Southern California were actually higher than
in Northern California in all listed markets, with the exception of Dow Jones.
On average, all prices tracked more closely than last quarter. The largest
difference in price, when comparing all price interactions, was only $3.43/MWh.
The $3.43/MWh price spread occurred between the Day-Of unconstrained price and
the ISO constrained NP1 5 price.
Concurrently, the CaIPX Compliance Unit has begun to focus on understanding how
traders in the California market system value transmission and related
congestion. Ultimately, regardless of the design, competitors seeking to
maximize their profits and enhance their market positions must factor congestion
into their bidding strategies. Two important questions must be answered:
* Are participants in CaIPX markets abusing congestion-related market
dynamics, in turn undermining the efficiency of CaIPX markets? N Will
redesign proposals have any material effect on the underlying market
calculus used by CaIPX market participants to maximize their profits and
enhance their market positions?
This preliminary study analyzes correlations between the Day Ahead UMCP, zonal
prices and the prices at the California, Oregon Border (COB) and Palo Verde (PV)
3
COB and PV are considered to be the equivalent of hubs for the trading of
electric power in the western United States, specifically California. COB
references import and export activity between California, the Pacific Northwest,
and, to some degree, the Rocky Mountains. PV references similar trade activity
between California and the Southwest.
By examining the relationships within California and between California and
these hub-equivalent points, a comprehensive view of congestion related market
values can be developed.
The core analysis rests on the creation of a correlation matrix for each of the
two years of PX market history to compare the impact of increased transmission
congestion.
4 In addition, a scatter diagram plotting the price at one location against the
price during the same period at a different location is created to provide a
visual depiction of key price relationships.
Prices between the NW1 import zone, NP1 5 and COB during three operating states
of the transmission system were compared as well. The three operating states are
as follows
o No congestion between NP15 and SP15.
* South to north congestion on Path 15 where NP15 prices were greater than
SP15.
The ISO holds an energy market and sorts bids in price merit order and calls
upon the bids when necessary to adjust the balance between generation and load.
Each hour is divided into six-ten minute BEEP (Balancing Energy and Ex-post
Price) intervals. A separate incremental (ISO pays SC to increase generation)
and decremental (SC pays ISO to decrease generation) price may be set during
each trading interval. Instructed deviations receive the ten minute interval
price and un-instructed deviations receive or are charged the hourly ex-post
price.