Overlapping Environmental Policies and the Impact on Pollution

Transcription

1 Overlapping Environmental Policies and the Impact on Pollution Kevin Novan Contributed presentation at the 60th AARES Annual Conference, Canberra, ACT, 2-5 February 2016 Copyright 2016 by Author(s). All rights reserved. Readers may make verbatim copies of this document for non-commercial purposes by any means, provided that this copyright notice appears on all such copies.

2 Overlapping Environmental Policies and the Impact on Pollution Kevin Novan UC Davis, Agricultural and Resource Economics

3 Overlapping Policies Electricity generators emit many pollutants (CO 2, NO X, etc.) - Prices have been established for a subset of pollutants - Almost exclusively using cap-&-trade programs (not pollution taxes)

4 Overlapping Policies Electricity generators emit many pollutants (CO 2, NO X, etc.) - Prices have been established for a subset of pollutants - Almost exclusively using cap-&-trade programs (not pollution taxes) Emission caps are being combined with additional policies: - e.g., Subsidies for renewable electricity or energy efficiency

5 Overlapping Policies Electricity generators emit many pollutants (CO 2, NO X, etc.) - Prices have been established for a subset of pollutants - Almost exclusively using cap-&-trade programs (not pollution taxes) Emission caps are being combined with additional policies: - e.g., Subsidies for renewable electricity or energy efficiency Combined with a binding cap, increasing renewable output... - Will not affect emissions of capped pollutant - What happens to the unregulated pollutants?

6 Overview of Paper Introduce a simple model of an electricity market: - Two pollutants: one regulated, one unregulated Increasing renewable output can increase pollution

7 Overview of Paper Introduce a simple model of an electricity market: - Two pollutants: one regulated, one unregulated Increasing renewable output can increase pollution Application to U.S. EPA s NO X cap-and-trade program: - How would renewable expansions affect CO 2 and SO 2? SO 2 emissions will increase CO 2 falls but by much less than previously thought

8 Single-Period Model Two Forms of Conventional Generation: X 1 (coal) and X 2 (gas) c i (X i ) = convex private costs (µ 1, µ 2) = emission rates of regulated pollutant (i.e. NO X ) (ρ 1, ρ 2) = emission rates of unregulated pollutant (i.e. CO 2)

9 Single-Period Model Two Forms of Conventional Generation: X 1 (coal) and X 2 (gas) c i (X i ) = convex private costs (µ 1, µ 2) = emission rates of regulated pollutant (i.e. NO X ) (ρ 1, ρ 2) = emission rates of unregulated pollutant (i.e. CO 2) Demand and Renewables: D = demand (perfectly inelastic w.r.t. wholesale price) r = renewable generation

10 Single-Period Model Two Forms of Conventional Generation: X 1 (coal) and X 2 (gas) c i (X i ) = convex private costs (µ 1, µ 2) = emission rates of regulated pollutant (i.e. NO X ) (ρ 1, ρ 2) = emission rates of unregulated pollutant (i.e. CO 2) Demand and Renewables: D = demand (perfectly inelastic w.r.t. wholesale price) r = renewable generation Market Clearing Conditions: (1) D r = X 1 + X 2 (2) µ 1 X 1 + µ 2 X 2 = µ ( µ is binding pollution cap)

11 Binding Emission Cap Residual Demand Level Curve

12 Binding Emission Cap Residual Demand Level Curve Iso-Private Cost Curve

13 Binding Emission Cap Level Curve of Capped Pollutant A Residual Demand Level Curve Iso-Private Cost Curve

14 Binding Emission Cap A

15 Increase in r or Decrease in D (A B) = Scale Effect Holding permit price constant B A

16 Increase in r or Decrease in D (A B) = Scale Effect Holding permit price constant (B C) = Composition Effect Holding residual demand constant B A C

17 Unregulated Pollution Increases Positively Correlated Emission Rates: And: A

18 Unregulated Pollution Increases Positively Correlated Emission Rates: And: B A

19 Unregulated Pollution Increases Positively Correlated Emission Rates: And: B A C

20 EPA Clean Air Interstate Rule (CAIR) Data Sources Strategy Thought experiment: Assuming annual NO X cap is binding... - Add 1,000 MW of solar capacity - How do annual CO 2 and SO 2 emissions change?

21 EPA Clean Air Interstate Rule (CAIR) Data Sources Strategy Thought experiment: Assuming annual NO X cap is binding... - Add 1,000 MW of solar capacity - How do annual CO 2 and SO 2 emissions change? NO X price will fall until net NO X = 0 - Total fossil generation falls, but ratio of coal to gas increases - In the end, CO 2 falls (slightly) and SO 2 increases

22 Conclusions With overlapping cap on a subset of pollutants: - External benefit of renewables/conservation is small - Subsidies are a very poor substitute for missing prices

23 Conclusions With overlapping cap on a subset of pollutants: - External benefit of renewables/conservation is small - Subsidies are a very poor substitute for missing prices With overlapping tax on a subset of pollutants: - Composition Effect doesn t occur - Subsidies serve as imperfect substitute for missing prices

24 Conclusions With overlapping cap on a subset of pollutants: - External benefit of renewables/conservation is small - Subsidies are a very poor substitute for missing prices With overlapping tax on a subset of pollutants: - Composition Effect doesn t occur - Subsidies serve as imperfect substitute for missing prices Policy recommendations: (1) Just price pollution already! (2) If we can t then combine subsidies with taxes (not caps) (3) If we can t use taxes: Set tight price collars / Update caps / Expand set-aside programs

25 Thank You Comments/Questions:

26 Data Return EPA s Continuous Emissions Monitoring System ( ) G t = Hourly fossil generation E t = Hourly NO X, CO 2, and SO 2

27 Data Return EPA s Continuous Emissions Monitoring System ( ) G t = Hourly fossil generation E t = Hourly NO X, CO 2, and SO 2 Emission Rates by Technology Combined Cycle Coal Other N 535 1,911 1,029 Median CO 2 Rate (tons/mwh ) Median SO 2 Rate (lbs/mwh ) Median NO X Rate (lbs/mwh )

28 Data Return EPA s Continuous Emissions Monitoring System ( ) G t = Hourly fossil generation E t = Hourly NO X, CO 2, and SO 2 Emission Rates by Technology Combined Cycle Coal Other N 535 1,911 1,029 Median CO 2 Rate (tons/mwh ) Median SO 2 Rate (lbs/mwh ) Median NO X Rate (lbs/mwh ) To simulate hourly output from 1,000 MW of solar: - Observe hourly solar capacity factors in Texas Capacity factor t = (Generation t)/(installed Capacity) r t = (1,000 MW) (Capacity Factor t)

29 Overview of Empirical Application Return (1) Estimate Scale Effect Model - Predict which fossil generators on the margin each hour - Estimate how much CO 2, SO 2, and NO X would be reduced by solar A MWh of solar reduces: 0.7 tons CO 2, 2.4 lbs SO 2, 1.3 lbs NO X

30 Overview of Empirical Application Return (1) Estimate Scale Effect Model - Predict which fossil generators on the margin each hour - Estimate how much CO 2, SO 2, and NO X would be reduced by solar A MWh of solar reduces: 0.7 tons CO 2, 2.4 lbs SO 2, 1.3 lbs NO X (2) Estimate Composition Effect Strategy - NO X price will fall until net NO X = 0 - Estimate how NO X permit price decline impacts CO 2 and SO 2

31 Overview of Empirical Application Return (1) Estimate Scale Effect Model - Predict which fossil generators on the margin each hour - Estimate how much CO 2, SO 2, and NO X would be reduced by solar A MWh of solar reduces: 0.7 tons CO 2, 2.4 lbs SO 2, 1.3 lbs NO X (2) Estimate Composition Effect Strategy - NO X price will fall until net NO X = 0 - Estimate how NO X permit price decline impacts CO 2 and SO 2 Find evidence that NO X permit decrease will increase CO 2 and SO 2 - Total fossil generation falls, but ratio of coal to gas increases - CO 2 falls by 0.4 tons/mwh, SO 2 increases by 4.8 lbs/mwh

32 Predicting the Scale Effect (for 2009) Return E t = f m(g t) + δ m,y + ε t E t = Aggregate hourly NERC NO X, CO 2, or SO 2 G t = Aggregate hourly NERC fossil fuel generation (MWh)

33 Predicting the Scale Effect (for 2009) Return E t = f m(g t) + δ m,y + ε t E t = Aggregate hourly NERC NO X, CO 2, or SO 2 G t = Aggregate hourly NERC fossil fuel generation (MWh) TRE CO 2 (10,000 tons) Avg. Slope = 0.63 tons/mwh TRE SO 2 (10,000 lbs) Avg. Slope = 1.13 lbs/mwh TRE NO X (10,000 lbs) Avg. Slope = 0.76 lbs/mwh RFC CO 2 (10,000 tons) Avg. Slope = 0.78 tons/mwh RFC SO 2 (100,000 lbs) Avg. Slope = 3.48 lbs/mwh RFC NO X (100,000 lbs) Avg. Slope = 1.47 lbs/mwh Fossil Generation (10,000 MWh) Fossil Generation (10,000 MWh) Fossil Generation (10,000 MWh)

34 Predicting the Scale Effect (for 2009) Return E t = f m(g t) + δ m,y + ε t E t = Aggregate hourly NERC NO X, CO 2, or SO 2 G t = Aggregate hourly NERC fossil fuel generation (MWh) Scale effect of adding renewable capacity: (Annual Emissions) = [ˆf (Gt r t) ˆf ] (G t) 2009

35 Predicting the Scale Effect (for 2009) Return E t = f m(g t) + δ m,y + ε t E t = Aggregate hourly NERC NO X, CO 2, or SO 2 G t = Aggregate hourly NERC fossil fuel generation (MWh) Scale effect of adding renewable capacity: (Annual Emissions) = [ˆf (Gt r t) ˆf ] (G t) 2009 Hourly fossil generation falls from (G t) to (G t r t) Hourly emissions fall from f (G t) to f (G t r t)

36 Predicting the Scale Effect (for 2009) Return E t = f m(g t) + δ m,y + ε t E t = Aggregate hourly NERC NO X, CO 2, or SO 2 G t = Aggregate hourly NERC fossil fuel generation (MWh) Scale effect of adding renewable capacity: (Annual Emissions) = [ˆf (Gt r t) ˆf ] (G t) 2009 Key assumption: f ( ) does not change as r increases The dispatch order is unchanged The input prices are unchanged (i.e. NO X price = 0)

37 NO X Permit Prices return To estimate (Emissions)/ (NO X Price) - Need exogenous variation in NO X Price - My approach: exploit discontinuity caused by Ozone Season NO X Expenditures ($/MWh) NO X Price for Median Coal Unit (NO X rate = 2.85 lbs/mwh) Annual NO X Cost Total NO X Cost 0 May 2009 Oct May 2010 Oct May 2011 Oct Date

38 Ozone Season Switch return Event study: - 20-day window around start of Spring 2009 Ozone Season - Estimate change in hourly Eastern Interconnection emissions (α) E t = α Ozone t + f (G t ) + δ h,w + θ Date t + ε t

39 Ozone Season Switch return Event study: - 20-day window around start of Spring 2009 Ozone Season - Estimate change in hourly Eastern Interconnection emissions (α) E t = α Ozone t + f (G t ) + δ h,w + θ Date t + ε t E t = Aggregate hourly NO X (lbs), CO 2 (tons), or SO 2 (lbs) G t = Aggregate hourly fossil fuel generation (MWh) Ozone t = Indicator variable for Ozone Season

40 Ozone Season Switch return Event study: - 20-day window around start of Spring 2009 Ozone Season - Estimate change in hourly Eastern Interconnection emissions (α) E t = α Ozone t + f (G t ) + δ h,w + θ Date t + ε t NO X (lbs) CO 2 (tons) SO 2 (lbs) α -16,809-2,582-90,629 (3,803) (735) (18,754) N R Newey-West standard errors reported (72-hour lag). = significant at the 1% level.

41 Placebo Effect Texas Interconnection return E t = α Ozone t + f (G t ) + δ h,w + θ Date t + ε t Ozone t = Indicator variable for Ozone Season E t = Aggregate hourly TRE NO X (lbs), CO 2 (tons), or SO 2 (lbs) G t = Aggregate hourly TRE fossil fuel generation (MWh) NO X (lbs) CO 2 (tons) SO 2 (lbs) α ,004 (543) (312) (5,467) N R Newey-West standard errors reported (72-hour lag). = significant at the 1% level.

42 Change in Generation by Technology return G i,t = α Ozone t + f (G t ) + δ h,w + θ Date t + ε t Ozone t = Indicator variable for Ozone Season G i,t = Aggregate hourly generation from fuel source i (coal, cc, gt) G t = Aggregate hourly fossil fuel generation (MWh) Coal (MWh) Combined Cycle (MWh) Other (MWh) α -2,361 2, (870) (696) (1,035) N R Newey-West standard errors reported (72-hour lag). = significant at the 1% level.