Market Power in a Hydro-Dominated Wholesale Electricity Market - - PowerPoint PPT Presentation

Market Power in a Hydro-Dominated Wholesale Electricity Market

Shaun McRae and Frank Wolak April 8, 2017

ITAM and Program on Energy Sustainable Development, Stanford University

We study the market power in the Colombian wholesale elec- tricity market: a bid-based, hydro-dominated system

• Majority of generation capacity in Colombia is

hydroelectric

• Susceptible to periodic shortfalls in water inflows during El

Niño events

• Market prices are determined using price and quantity

bids submitted by generation owners

• Other Latin American electricity markets use cost-based

dispatch

1

Much larger increase in wholesale price during 2015–16 El Niño event despite similar thermal utilization rate to 2009–10

Hydro Thermal

20 40 60 80

• Cap. util. (%)

400 800 1200 Jan 2008 Jan 2010 Jan 2012 Jan 2014 Jan 2016

Bolsa price (COP/kWh)

2

Why do we study this question?

• General interest in the use of wholesale electricity

markets as a data-rich “laboratory” for understanding firm behavior

• Policy tension between “regulators” and “markets” in

electricity

• Most restructured electricity markets have an increasingly

large role for regulation in determining short-term prices and long-term investment

• How should we balance the costs and benefits of

regulation in these markets?

3

Colombian electricity market

Share of thermal generation has been increasing since 2009, even in years with no adverse hydrological conditions

5 10 15 2000 2005 2010 2015

Quarterly generation (TWh)

Hydro Thermal Cogen Wind

4

Most new capacity investment in past decade has been in hy- droelectric generation

5 10 15 2000 2005 2010 2015

Generation capacity (GW)

Hydro Thermal Cogen Wind

5

Comparison of El Niño events

What explains the large difference in prices between the two most recent El Niño events?

• Mean price in 2009–10 El Niño: 185 COP/kWh (US$95/MWh)
• Mean price in 2015–16 El Niño: 675 COP/kWh (US$217/MWh)
• Market structure (institutions and actors) changed little

between the two events

• Was there a difference in the hydrological conditions or

fuel prices?

6

Rise in Colombian natural gas prices after the end of price reg- ulation and the opening of the wholesale gas market

End Guajira price regulation

10000 20000 30000 2008 2010 2012 2014 2016

COP per MMBTU

Guajira Cusiana Henry Hub

7

Diesel prices (in pesos) in 2015-16 similar to 2009-10

Barrancabermeja US Gulf Coast

20000 40000 60000 80000 Jan 2008 Jan 2010 Jan 2012 Jan 2014 Jan 2016

COP per MMBTU

8

Note that natural gas was inframarginal during both El Niño events (so diesel prices are what matter)

20 40 60 80 2006 2008 2010 2012 2014 2016

Million MMBTU

Coal Natural Gas Diesel/Fuel Oil

9

Annual reservoir inflows were similar during 2009-10 and 2015- 16 El Niño events

20 40 60 2000 2005 2010 2015

Annual river flow (TWh)

10

Annual average hydro reservoir levels show similar water levels from 2012 to 2015 as during the 2009-10 El Niño event

0.0 2.5 5.0 7.5 10.0 12.5 2000 2005 2010 2015

Mean reservoir level (TWh)

11

Market power in wholesale electricity markets

Every day and hour, electricity generators submit a step function supply curve to the system operator

S1(p)

250 500 750 1000 1000 2000 3000 4000

MW COP/kWh

12

System operator determines market price where the aggregate supply curve crosses aggregate demand (perfectly inelastic)

S(p) System demand 9319 MW System price 322 COP/kWh

250 500 750 1000 5000 10000 15000

MW COP/kWh

13

Residual demand for a firm is the difference between system demand and the aggregate offers of all other firms

SO1(p) System demand 9319 MW DR(250) = 4147 MW DR(125) = 5682 MW

250 500 750 1000 5000 10000 15000

MW COP/kWh

14

Generation firms can choose the optimal (price, quantity) com- bination along their residual demand curve

DR1 Generation 1797 MW S1(p) System price 322 COP/kWh

250 500 750 1000 2500 5000 7500 10000

MW COP/kWh

15

Residual demand and offer curve at 18:00 hrs on September 18, 2015: EPM

Scarcity price Dispatch price EPMG generation EPMG offer RD

500 1000 1500 2000 2500 2500 5000 7500 10000

MW COP/kWh

Hydro Thermal

16

Residual demand and offer curve at 18:00 hrs on September 25, 2015: EPM

Scarcity price Dispatch price EPMG generation EPMG offer RD

500 1000 1500 2000 2500 2500 5000 7500 10000

MW COP/kWh

Hydro Thermal

17

Residual demand and offer curve at 18:00 hrs on October 2, 2015: EPM

Scarcity price Dispatch price EPMG generation EPMG offer RD

500 1000 1500 2000 2500 2500 5000 7500 10000

MW COP/kWh

Hydro Thermal

18

Two measures of ability to exercise unilateral market power based

• n residual demand curve
• Inverse semi-elasticity, ηhk = −

1 100 DRhk(ph) DR′

hk(ph), quantifies

COP per kWh increase in wholesale price from supplier k reducing its actual output by one percent

• ηhk measures the ability of supplier k to raise the market

price at their actual level of output during hour h

19

Two measures of ability to exercise unilateral market power based

• n residual demand curve
• Pivotal supplier frequency is the fraction of hours in the

week that DRhk(∞) > 0, supplier k’s residual demand is positive for all possible wholesale prices, meaning that some of supplier k’s available capacity is required to serve demand during hour h

• A pivotal supplier can raise the price has high it would like

if it is willing to only sell its pivotal quantity, DRhk(∞) > 0

20

Empirical analysis of market power in Colombian electricity market

Offer prices and inverse semi-elasticities: EPM

5 10 15 20

Inverse semi−elasticity

500 1000 1500 2000 Jan 2008 Jan 2010 Jan 2012 Jan 2014 Jan 2016

Offer price (COP/kWh)

21

Offer prices and proportion of pivotal hours: EPM

0.0 0.2 0.4 0.6

Share pivotal hours

500 1000 1500 2000 Jan 2008 Jan 2010 Jan 2012 Jan 2014 Jan 2016

Offer price (COP/kWh)

22

Offer prices and inverse semi-elasticities: Emgesa

5 10 15 20

Inverse semi−elasticity

500 1000 1500 2000 Jan 2008 Jan 2010 Jan 2012 Jan 2014 Jan 2016

Offer price (COP/kWh)

23

Offer prices and proportion of pivotal hours: Emgesa

0.0 0.2 0.4 0.6

Share pivotal hours

500 1000 1500 2000 Jan 2008 Jan 2010 Jan 2012 Jan 2014 Jan 2016

Offer price (COP/kWh)

24

Offer prices and inverse semi-elasticities: Isagen

5 10 15 20

Inverse semi−elasticity

500 1000 1500 2000 Jan 2008 Jan 2010 Jan 2012 Jan 2014 Jan 2016

Offer price (COP/kWh)

25

Offer prices and proportion of pivotal hours: Isagen

0.0 0.2 0.4 0.6

Share pivotal hours

500 1000 1500 2000 Jan 2008 Jan 2010 Jan 2012 Jan 2014 Jan 2016

Offer price (COP/kWh)

26

Wholesale price is higher in those periods where firms have more market power on average

(1) (2) (3) (4) Mean η 18.15 11.90 (6.16) (3.44) Mean pivotal (0/1) 858.22 200.12 (246.13) (138.19) Mean pivotal q (MW) 1.70 1.26 (0.21) (0.18) Hour × year Y Y Y Y Month-of-sample Y Y Y Y Gen bin × year Y Y Y Y Observations 73,742 73,742 73,742 73,742

27

Offer prices for largest firms are higher for those periods in which firms have greater market power

(1) (2) (3) (4) η 14.18 3.18 6.59 6.55 (5.44) (1.79) (1.92) (2.21) Pivotal (0/1) 337.58 58.51 129.20 −28.29 (99.58) (73.09) (44.95) (115.87) Generator AES Chivor Emgesa EPM Isagen Hour × year Y Y Y Y Month-of-sample Y Y Y Y Gen bin × year Y Y Y Y Observations 46,765 73,407 73,649 73,188

28

Conclusions from market power analysis

• By inverse semi-elasticity measure ηhk and pivotal

supplier frequency all suppliers have little, if any ability to exercise unilateral market power until early 2014, even during 2009-2010 El Niño event

• From October 2015 onwards, all suppliers both measures

showed very large increase in unilateral ability to exercise market power

• This fact explains remarkable increase in wholesale prices

during 2015–16 El Niño event relative to 2009–10

29

Discussion

Declining availability factor led to many more hours in which at least one firm was pivotal in the 2015-16 El Niño event

5 10 15 20 2006 2008 2010 2012 2014 2016

GW

Nameplate Availability Availability (N−1) Max demand Pivotal Q

30

Most new generation investment in Colombia has been hydro- electric

• Steady increase in electricity demand and share of

thermal generation in total

• Most new generation is hydroelectric, with relatively

limited storage capacity

• Increased susceptibility to declines in water inflows during

El Niño events

• Reduced buffer to deal with adverse hydrological

conditions

• This becomes even more important as climate variability

increases

31

Current design of capacity market has been less-than-successful at ensuring availability of thermal backup generation

• Capacity markets pay generators for their availability, even

when they are not producing electricity

• Current capacity payment mechanism was set up in 2006
• Several new thermal plants assigned in the capacity

auction were not built or were built behind schedule

• One thermal plant walked away from its obligation to

produce electricity in spite of having received capacity payments for nine years

• Capacity mechanism places regulatory restrictions on

ability of hydro owners to manage their water resources

32

Conclusion

• Major challenges for all electricity markets relying on

renewables

• How to meet system demand during adverse climate

shocks?

• How to limit ability and incentive of firms to exercise

market power during these events?

• Recent large price spikes in Colombian electricity market

due to high level of market power for generators during dry periods

• Underlying cause has been the shortfall in investment in

thermal generation capacity

33