When the Long Run Matters:

Transcription

1 When the Long Run Matters: The Joint Effect of Carbon Decay and Discounting Terrence Iverson 1 Sammy Zahran 1 Scott Denning 2 1 Department of Economics, Colorado State University 2 Department of Atmospheric Science, Colorado State University July 2014

2 A long run problem Airborne Fraction of the Total CO2 Release 60% 50% 40% 30% 20% 10% 0% Ocean Invasion Reaction with CaCO 3 Reaction with Igneous Rocks Year A.D. Source: Archer and Brovkin 2008

3 But in typical IAMs, carbon price insensitive to long run Conventional economic discounting

4 But in typical IAMs, carbon price insensitive to long run Conventional economic discounting Present value of $1 billion in 200 years

5 But in typical IAMs, carbon price insensitive to long run Conventional economic discounting Present value of $1 billion in 200 years

6 But in typical IAMs, carbon price insensitive to long run Conventional economic discounting Present value of $1 billion in 200 years Present value of $1 billion in 400 years

7 But in typical IAMs, carbon price insensitive to long run Conventional economic discounting Present value of $1 billion in 200 years Present value of $1 billion in 400 years

8 But in typical IAMs, carbon price insensitive to long run Conventional discounting assumptions 1 Constant discounting parameters

9 But in typical IAMs, carbon price insensitive to long run Conventional discounting assumptions 1 Constant discounting parameters 2 Calibrate to match historical interest rates

10 But in typical IAMs, carbon price insensitive to long run Conventional discounting assumptions 1 Constant discounting parameters 2 Calibrate to match historical interest rates

11 But in typical IAMs, carbon price insensitive to long run Conventional discounting assumptions 1 Constant discounting parameters 2 Calibrate to match historical interest rates

12 Rationale for declining time preference 1 Positive Near-term behavioral evidence (Thaler 1981, Lowenstein 1987) Long-term stated preference (Cropper et al. 1994, Layton and Brown 2000, Layton and Levine 2003) 2 Normative Preference aggregation (Li and Löfgren 2000, Gollier and Zeckhauser 2005, Heal and Millner 2013) Uncertainty (Weitzman 1998, Iverson 2013b)

13 Time Inconsistency [Declining rates] solve one problem by creating another... Suppose an intelligent decision-maker plans a strategy for the long future, beginning today. Five years from now, she reconsiders the strategy, having followed it so far. She will want to change to a different strategy for no other reason than the passage of time.... This sounds like a poor way to run a railroad. Robert Solow in Portney (1999)

14 Avoiding time inconsistency Straightforward in principle (Strotz 1955)

15 Avoiding time inconsistency Straightforward in principle (Strotz 1955) Difficult in practice. Prior climate policy literature assumes: 1 A commitment device (Mastrandrea and Schneider 2001, Guo et al. 2006) 2 Quasi-hyperbolic discounting (Karp 2005, Gerlagh and Liski 2013 ) 3 One state variable model (Karp 2007, Fujii and Karp 2008)

16 Avoiding time inconsistency Straightforward in principle (Strotz 1955) Difficult in practice. Prior climate policy literature assumes: 1 A commitment device (Mastrandrea and Schneider 2001, Guo et al. 2006) 2 Quasi-hyperbolic discounting (Karp 2005, Gerlagh and Liski 2013 ) 3 One state variable model (Karp 2007, Fujii and Karp 2008) Iverson 2013a, which extends Gerlagh and Liski 2013! 4 state IAM in spirit of DICE (Golosov, Hassler, Krusell, and Tsyvinski 2014)

17 Avoiding time inconsistency Straightforward in principle (Strotz 1955) Difficult in practice. Prior climate policy literature assumes: 1 A commitment device (Mastrandrea and Schneider 2001, Guo et al. 2006) 2 Quasi-hyperbolic discounting (Karp 2005, Gerlagh and Liski 2013 ) 3 One state variable model (Karp 2007, Fujii and Karp 2008) Iverson 2013a, which extends Gerlagh and Liski 2013! 4 state IAM in spirit of DICE (Golosov, Hassler, Krusell, and Tsyvinski 2014)! Arbitrary non-constant time preference

18 Avoiding time inconsistency Straightforward in principle (Strotz 1955) Difficult in practice. Prior climate policy literature assumes: 1 A commitment device (Mastrandrea and Schneider 2001, Guo et al. 2006) 2 Quasi-hyperbolic discounting (Karp 2005, Gerlagh and Liski 2013 ) 3 One state variable model (Karp 2007, Fujii and Karp 2008) Iverson 2013a, which extends Gerlagh and Liski 2013! 4 state IAM in spirit of DICE (Golosov, Hassler, Krusell, and Tsyvinski 2014)! Arbitrary non-constant time preference! Solve analytically for time consistent optimal carbon tax

19 Calibration: declining time preference rates Rationale for declining rates Two viewpoints: Stern and Nordhaus Utilitarian social planner aggregates preferences

20 Calibration: declining time preference rates Rationale for declining rates Two viewpoints: Stern and Nordhaus Utilitarian social planner aggregates preferences Calibrate to match mean interest rate of 5.4 percent

21 Calibration: declining time preference rates 4 Time preference rate (in percent) Nordhaus 10% Stern 20% Stern 30% Stern 40% Stern 50% Stern Stern Time horizon (in years)

22 Calibration: carbon cycle Decay function as in Golosov et al d t = γ L + (1 γ L )γ 0 (1 γ) t γ L is fraction of current emissions that remains permanently

23 Quantifying long run importance Cumulative value function: CVF (t) Fraction of the current carbon tax due to damages before date t Depends on both discounting and carbon cycle parameters

24 Cumulative value, zero Stern weight 100 Cumulative value (percent of total) =0 L L =0.1 L =0.2 L = Time horizon (in years)

25 Cumulative value, γ L = Cumulative value (percent of total) Nordhaus Time horizon (in years)

26 Cumulative value, γ L = Cumulative value (percent of total) Nordhaus 10% Stern Time horizon (in years)

27 Cumulative value, γ L = Cumulative value (percent of total) Nordhaus 10% Stern 20% Stern Time horizon (in years)

28 Cumulative value, γ L = Cumulative value (percent of total) Nordhaus 10% Stern 20% Stern 30% Stern Time horizon (in years)

29 Joint effect of carbon decay and discounting Long run emission fraction (in percent) Percent of Total Value Accruing After 200 Years Probability weight on Stern discount rate (in percent)