May 5, 2004

1
ISO New England State of the Market Report 2003

• May 5, 2004

Robert G. Ethier, Ph.D. Director, Market
Monitoring
2
Fuel Prices and Energy Prices

• Electricity prices were driven to high levels by
  fuel prices, which are the largest component of
  generators marginal costs, and to a lesser
  extent by load levels.
• The next figure shows that monthly energy prices
  for 2002 and 2003 have been driven by fuel price
  trends.
• Natural gas prices were 74 percent higher than
  2002 on average.
• Nearly half of New England capacity is gas-fired
  or gas capable.
• Electricity prices peaked in February and March
  as natural gas prices rose to unprecedented
  levels.
• The July Peak Summer Load was much lower.

3
Fuel Prices and Energy Prices (Continued)

• Electricity prices increased less than gas prices
  because economic dispatch substituted other,
  cheaper fuels.
• Gas-only units were on the margin 52 of the time
  in 2003 versus 55 in 2002, despite approximately
  6,000 MW of new gas capacity added over the two
  years.
• Gas-capable units were on the margin 67 of the
  time in 2003.

4
New England Electricity Natural Gas Prices
2001 - 2003
SMD Implementation
5
Energy Prices in 2003

• The next figure shows real-time price duration
  curves for 2001 to 2003.
• These curves show the percentage of hours when
  the load-weighted price for New England is
  greater than each given price level.
• Price levels were generally higher in 2003 than
  in the previous two years due to higher fuel
  prices.
• In 2003, there were fewer price spikes than the
  two previous years
• In 2003, real-time prices exceeded 500 for 1
  hour, compared to 4 hours in 2002 and 15 hours in
  2001.
• The lower quantity of price spikes was primarily
  due to milder weather in New England combined
  with relatively robust capacity margins.
• Scarcity pricing provisions were implemented, but
  were not triggered in 2003.

6
(No Transcript)
7
Load Profile

• The next figure shows annual load duration curves
  for New England.
• These curves show the percentage of hours in
  which the load is greater than the level
  indicated on the vertical axis.
• In 2003, peak days had far less impact on average
  prices than in 2002. The absence of severe price
  spikes was due to mild summer loads.
• There were only 19 hours in 2003 when actual
  loads exceeded 24,000 MW, compared to 34 hours in
  2002.
• In 2003 there were 200 hours when load exceeded
  21,000 MW compared to 263 hours in 2002.

8
(No Transcript)
9
All-in Energy Prices

• The following figure calculates an all-in price
  that includes the cost of energy, ancillary
  services, capacity, and other costs.
• The all-in energy price is a weighted average of
  various locations within New England, since
  energy prices vary by location.
• Ancillary services includes reserves and
  regulation prior to SMD, and regulation after SMD
  implementation.
• This figure shows that all-in prices rose in
  2003.
• The all-in price rise is primarily caused by
  increased energy prices in 2003, which rose 41
  in 2003 due to higher fuel prices. (from
  41.65/MWh to 55.36/MWh)
• The capacity component fell in 2003 due primarily
  to increases in installed capacity.
• While the energy component increased in 2003, the
  fuel adjusted energy price fell relative to 2002.

10
All-In Price Metric 2001 - 2003 (/MWh)Total per
MWh Energy, Uplift, Capacity, and Ancillary
CostsIncludes Energy and Fuel Adjusted Energy
Energy Costs
Fuel Adjusted Energy Costs
Total Energy
Fuel Adjusted Total Energy
Uplift
Capacity
Ancillary Services
Note Energy Interim Market Period ECP System
Load SMD Period RT Load Obligation RT LMP
11
Economic Incentives for New Investment

• In long-run equilibrium, the market should
  support the entry of new generation by providing
  sufficient net revenues (revenue in excess of
  production costs) to finance new entry.
• We calculated the net revenue the markets would
  have provided to different types of units in
  2003.
• A gas-fired combined-cycle (heat rate 7,000).
• A gas-fired combustion turbine (heat rate10,500)
  .

12
Economic Incentives for New Investment (Continued)

• Even though energy and all-in prices were higher
  in 2003, the net revenue for gas-fired units was
  lower in 2003 than 2002 due to gas price
  increases.
• New capacity added in 2002 and 2003 also reduced
  net revenues.
• These results indicate that the market in 2003
  did not produce sufficient net revenue to support
  investment in a new gas turbine (GT) or a new
  combined-cycle (CC) unit.
• A new GT would only recover 16 - 21 of its
  estimated annual fixed costs for 2003.
• A new CC would only recover 64 -73 of its
  estimated annual fixed costs for 2003.

13
Economic Incentives for New Investment (Continued)

• This was done pool-wide because LMPs existed for
  only a portion of the year
• A unit in Connecticut, for example, would have
  earned additional revenue.

14
Net Revenue Metric 2003
All Values in /MWh Combustion Turbine Unit Combined Cycle Unit
Energy revenues 1 58,773 315,239
Energy marginal costs 2 47,907 241,792
Net Revenue Energy 10,867 73,447
Revenue Capacity 3 1,972 1,972
Revenue Ancillary Services 4 - 1,492
Total Net Revenue 12,839 76,912
Estimated Annual Fixed Costs 60,000-80,000 105,000-120,000

1. Energy revenues are calculated as the revenue per
   MW of a hypothetical unit assumed to be
   dispatched during each hour when the market
   clearing price equals or exceeds the unit's
   marginal cost, adjusted for a 5 forced outage
   rate. Revenues are calculated based on the
   system wide energy clearing price prior to SMD
   (March 1, 2003), and based on the real-time Hub
   LMP from March 1, 2003 onward.
2. Energy marginal costs are calculated as the
   average Massachusetts natural gas daily spot
   price multiplied by the unit's respective heat
   rate the unit's respective variable OM. These
   marginal costs are then adjusted for a 5 forced
   outage rate.
3. Capacity revenues for year ending 12/31/2003 are
   the UCAP revenues derated by the 5 forced outage
   rate.
4. Ancillary service revenues are calculated only
   for Regulation.

15
Forced Outages

• The next figure presents the trend in the forced
  outage rates from the beginning of the operation
  of the New England Markets.
• The forced outage rate is the percentage of time
  capacity is unavailable due to full or partial
  forced outages.
• Total outage rates have declined substantially
  following the implementation of markets in New
  England.
• This is consistent with the incentives the
  deregulated markets provide to maximize
  availability, particularly during high load
  (on-peak) conditions.
• Previous analysis suggests that new
  combined-cycle units initially have high outage
  rates. New England has many new combined-cycle
  units. Improvements in outage rates may be
  expected as these units mature.

16
(No Transcript)
17
Congestion Costs

• The following figure shows the monthly average
  congestion components of LMPs in the day-ahead
  and real-time markets for March through December
  2003.
• Maine had significant negative congestion as
  generation was periodically constrained down due
  to export constraints.
• Connecticut had significant positive congestion
  as it was periodically import constrained.
• Northeastern Massachusetts/Boston experienced
  less congestion than expected based on historical
  data due to significant generation additions and
  transmission upgrades.
• Note that these numbers understate congestion
  costs, as they exclude significant out-of-merit
  local operating reserve costs, which dont affect
  LMPs.

18
On-Peak Average Day-Ahead Congestion March -
December 2003
19
On-Peak Average Real-Time Congestion March -
December 2003
20
Competitive Benchmark Analysis

• Evaluated actual energy clearing price and actual
  cumulative bid-in capacity sorted by ascending
  price (aggregate bid-intercept) versus marginal
  cost-based simulated dispatch.
• Simulated dispatch designed to produce an
  estimate of the perfectly competitive market
  outcome.
• Caution that the estimate is subject to an
  unknown error.
• Metric is increase over perfect market
  outcome (Quantity-weighted Lerner Index).
• Results in 2003 show market continues to function
  well, with modest differences from competitive
  baseline.

21
Competitive Benchmark Results 2003 vs. 2002
Note Energy Clearing Price is the ECP prior to
March 1, 2003 the Real-Time Hub Price as of
March 1, 2003

22
Other Conclusions

• The New England markets continued to perform
  competitively in 2003 with no evidence of
  significant economic or physical withholding.
• Day-ahead and real-time energy prices exhibit
  good convergence.
• Average day-ahead/real-time spread was 1.10 MWh
  during first year of SMD
• Virtual trading volumes were reasonable in 2003,
  contributing to the convergence between the
  day-ahead and real-time prices.

23
Other Conclusions (Continued)

• Real-time prices in adjacent regions continue to
  be inefficiently arbitraged.
• The ISO-NE Demand Response Program provides a
  modest real-time reduction when necessary.
• Mild conditions in 2003 limited the
  implementation of such reductions.
• Regulation was only ancillary service market in
  2003.
• A market flaw was identified in 2003 and
  corrected in early 2004.

24
Other Conclusions (Continued)

• Out-of-merit operation an on-going issue.
• Primarily in import-constrained areas.
• Would be helped by increase in quick-start
  capacity.
• Continuing to investigate unit commitment and
  software resolutions.
• New Forward Reserve Market should help incent
  this capacity.
• Resource Adequacy in constrained areas in an
  on-going issue.
• Clear market deficiency when large numbers of
  units required for reliability do not cover
  going-forward costs.