Efficiency, Wealth Transfers and Risk Management Under Realtime Electricity Pricing

1
Efficiency, Wealth Transfers and Risk Management
Under Real-time Electricity Pricing

• Severin Borenstein
• Haas School of Business, UC Berkeley
• University of California Energy Institute
• IDEI Economics of Electricity Markets Conference
  June 2-3 2005

2
The Simple Economics of RTP

• Economists favor Real-Time Pricing (RTP)
• RT Metering is not costly for large customers
• RTP sends accurate signals to customers
• Increased elasticity lessens market power
• Political reality retail markets will be, at
  best, a mix of customers on flat-rate service and
  RTP
• Questions
• How Large are the Gains from RTP?
• Who would be Winners and Losers?

3
Simulating A Long- Run Competitive Model of
Electricity Markets

• Demand differs in all hours
• Free entry/exit of generation capacity in very
  small (1MW) increments
• L-shaped production costs of each unit
• 3 technologies differ in FC and MC
• Some customers on RTP, others on flat rate that
  covers its wholesale costs
• all have same time-variation of demand

4
P6
D6
P5
P4
D5
P3
D4
P2
D3
P1
D2
D1
Kb
Km
Kp
5
Long-Run Equilibrium With RTP

• For given capacities, Kb,Km,Kp, solve for SR
  competitive equilibrium
• Then adjust capacities so that owners of each
  type of generation break even
• Then adjust flat rate to retailer break even
• This produces unique competitive equilibrium
• Algorithm to find equilib starts with peaker
  capacity, then mid-merit, then baseload

6
Long-Run Equilibrium Without RTP

• Find the flat rate that covers all costs when
  capacity is efficient for load
• Equivalent to competitive wholesale price spike
  in peak hour equal to fixed costs of peaker
• I assume that the demand distribution used (in
  this case from California ISO) results from
  combination of break-even flat rate and
  break-even time-of-use rate

7
What the model omits

• Reserves
• Plant outages -- increase price volatility
• Market power of sellers
• Risk-aversion of customers
• Cross-elasticity of demand across hours

8
Data Inputs for Simulations

• Demand profile From CAISO for 1999-2003 (five
  years). Very similar results other systems
• Demand elasticities Broad range of estimates,
  most with large standard errors
• use -0.025 to -0.500, constant elasticity demand
• Price used include 40/MWh for TD
• Three production technologies based roughly on
  coal, combined-cycle gas turbine, and combustion
  turbine.

9
Table 1Production Cost Assumptions
10
Basic Results Capacity and Price Effects (table
2)

• Large reduction in peaker capacity. Small changes
  in baseload and mid-merit capacity.
• Very high peak prices with most inelastic demand,
  appx equal to capacity cost of peaker over sample
• With a bit more elasticity (-0.1) peak prices
  below 10,000/MWh.
• Still significant share of annual costs if not
  hedged

11
(No Transcript)
12
Basic Results Welfare Effects (table 3)

• Total surplus increases with RTP, but at a
  decreasing rate as more move to RTP
• 1/3 on RTP gives gt ½ of total benefits
• Both RTP and flat-rate customers benefit, but RTP
  customers benefit more
• Flat-rate customers may not benefit (flat-rate
  may increase) if they have different load shape
• TS gain as percentage of total energy bill is
  modest, but much larger than plausible cost of
  implementing RTP

13
(No Transcript)
14
(No Transcript)
15
Results if elasticity varies with level of demand
(tables 4 and 5)

• Elasticity is linear function of (flat-rate)
  load, but weighted-average elasticity unchanged
• Smallest elasticity is 50 original
• BH show RTP could lower welfare if higher
  elasticity at peak demand time
• But in simulations, benefits are greater with
  larger elasticity at peak
• Larger effects on capacity, lower peak prices
• Reduced effect if elasticity greater at off-peak

16
RTP vs Time-of-Use Pricing

• TOU is just peak/shoulder/off-peak pricing
• TOU captures lt20 of realtime price variation
• No obvious way to set TOU prices
• Quasi-wholesale market with capital cost of
  peakers loaded onto period peak hour
• Average-cost approach, spreads capital cost
• Fixed ratio approach w/ ratio from actual TOUs
• Regardless of TOU method, creates only 10-20 of
  the gains from RTP (ignoring reserves)
• Doesnt address large price mismatch at peak
  times
• BUT Important assumption about demand
  responsiveness to prices with varying notification

17
(No Transcript)
18
Direct Estimation of the Size of Transfers from
RTP Adoption

• Simple analysis assuming no demand elasticity -
  estimating pure transfer effect
• this is a lower bound on losses due to
• ability to respond to price
• market price compression due to RTP
• Data on realtime consumption of 636 large
  customers in California
• randomly chosen among all large customers
• Using a set of realtime prices (actual and
  simulated), calculate customer costs under
  breakeven flat rate and under RTP

19
Changes in Electricity Bills due to RTP(assuming
demand of sample customers has zero price
elasticity)
20
How Hedgeable is RTP Risk?

• Resistance to RTP due to risk
• separate from transfers, leaves bill volatility
• not risk of sustained high prices, which RTP
  reduces
• Why do large customers care about bill
  volatility?
• Why do publicly traded firms buy insurance?
• How much does RTP increase bill volatility?
• How much would hedging reduce it?

21
Empirical Analysis of Hedging

• Same data as for analyzing transfers
• customer load data, actual and simulated prices
• Calculate monthly bills for customers under
  alternative billing regimes
• flat-rate, TOU, RTP, RTP with Hedging
• Study monthly bill volatility
• Focus on most volatile prices from simulation
  with very inelastic market demand

22
Alternative Measures of Volatility

• Coefficient of Variation (std dev / mean) under
  each billing regime
• Maximum/Mean bill faced under each billing regime
• Ratio of measures under alternative billing
  regimes
• same-customer changes in volatility

23
Bill Volatility Measured As Standard Deviation of
Bill
24
Bill Volatility Measured As Maximum Bill
25
RTP and Operating Reserves

• RTP will not eliminate the need for reserves
• so long as price-responsive demand is slower than
  callable supply
• But RTP offers more than peak demand reduction
• demand tilts as well as shifts
• RTP will gradually reduce use of reserves
• as system operators recognize its reliability
• Eventually, RTP will reduce the standard for
  percentage reserves

26
Conclusions

• Conservative estimates of potential welfare gain
  outweigh implementation costs
• Even with very small demand elasticities
• Diminishing return to increased elasticity or
  increased share of population on RTP
• TOU is a poor substitute for RTP so long as there
  is shorter run elasticity of demand
• Recent pilot programs indicate there is
• Most of the transfer RTP causes are already
  taking place under TOU
• RTP does increase bill volatility compared to
  TOU, but most of that increase can be eliminated
  with simple hedging strategies