A Discussion of

1
A Discussion ofAn Empirical Assessment of the
Competitiveness of The New England Electricity
Market(Bushnell, Saravia 2002)

• Bob Ethier, Ph.D.
• Manager, Market Monitoring Mitigation
• May 10, 2002

2
Objectives of the Study

• To determine the extent to which market power is
  exercised by developing a counterfactual
  competitive benchmark against which wholesale
  power prices may be compared.

3
What is the competitive benchmark?

• An estimate of the wholesale market price that
  would exist if no market participant exercised
  market power.
• This estimated market price equals the
  incremental costs associated with the cheapest
  unit that is not needed to serve demand in a
  given hour.
• The benchmark is based on the economic principle
  that a price-taking market participant will
  produce power from a generating unit as long as
  the incremental costs incurred do not exceed the
  revenues from selling that power.

4
What incremental costs comprise Bushnells
competitive benchmark?

• Direct fuel costs reflect spot prices for all
  fuels except for high-sulfur coal.
• Environmental costs are the average traded permit
  prices for SO2 and NOX.
• Variable OM costs are based on publicly
  available sources.

5
How is this benchmark price used in measuring
market power?

• Bushnell compares this benchmark price to 3
  different measures of wholesale market prices
• ISO-NEs reported Energy Clearing Price.
• The price at which market demand intersects the
  aggregate supply curve from all generation units
  price bids.
• The price at which market demand served by
  thermal units intersects the aggregate supply
  curve from thermal units price bids.
• An indeterminate range of error exists around
  benchmark level (e.g., environmentally limited
  units not explicitly considered, hydro assumed to
  be perfectly competitive).

6
How are the generating units modeled in computing
the benchmark price?

• Thermal units
• For a given demand level, marginal costs the
  incremental costs associated with the last
  available unit necessary to meet that demand,
  given the unavailability of other units due to
  random forced outages.
• Monte Carlo technique used to simulate each
  units outage probability.

7
How are the generating units modeled in computing
the benchmark price? (cont.)

• Nuclear, Cogeneration and Small Thermal
• Bid price and actual availability are used to
  proxy the true marginal cost and availability
• Hydro and geothermal
• Actual observed output is assumed to equal the
  optimal (least-cost) output expected in a
  competitive market.
• Consequently, in determining the benchmark price,
  hydro and geothermal production are assumed to be
  optimal. The marginal costs of meeting the
  remaining demand by thermal resources are then
  calculated.

8
Results

• Bushnell calculates price-cost margins for the 3
  different measures of wholesale prices mentioned
  above. These margins are expressed as
  quantity-weighted Lerner indices.
• In this report, these Lerner indices measure the
  percentage of wholesale price over the benchmark
  at a given time, weighted by power demand at that
  time.
• This is averaged over the period of the study
  (May 1999-Sept. 2001)
• On the following page is a comparison of ISO-NE
  with PJM and CAISO for May-December 1999 (the
  only period in which the results of similar
  market power studies are available).

9
Comparison among US electricity markets May-Dec
1999
10
Results

• For the period May 1999 - September 2001, the 3
  measures of wholesale market prices in New
  England produced the following quantity-weighted
  Lerner indices
• Energy Clearing Price (ECP) 12 above benchmark
  
• This includes expected physical dispatch
  inefficiencies and unit operating constraints.
• All-unit aggregate energy bids 4 above
  benchmark
• Cleanest measure of market competitiveness
• Major thermal energy bids and quantities
  supplied 11 above benchmark.
• Ignores bid mitigation due to transmission
  congestion
• These differences reflect various factors such as
  day-ahead and real-time operations, congestion
  management, unit operating constraints, and
  self-scheduled capability.

11
Results (cont.)

• 1. New England wholesale markets compare
  favorably with others in the U.S. However, this
  may change with level of forward contracting and
  reserve margins. New market rules should help.
• 2. Further refinements to the benchmark price
  analysis is desirable. Such refinements include
  removing simplifying assumptions such as optimal
  hydro resource bidding and price-insensitive
  imports.
• 3. Benchmark analysis is a useful tool for
  analyzing market performance, but it is not
  sufficient by itself for determining whether
  intervention in markets is required (long-run
  analysis tool not suited to real-time analysis).