Energy technology diffusion and CO2 emission reduction: An application of the Ramsey model with logi

1
Energy technology diffusion and CO2 emission
reduction Anapplication of the Ramsey model
with logistic process

• Kazushi Hatase
• Graduate School of Economics, Kobe University

2
Effect of economic inertia motivation for this
study
An explanation of economic inertia by Grubb (1997)

• Most capital stock in the energy sector has a
  lifetime of 30-40 years, and once capital stock
  is built, it is not easy to replace
• In addition, there are complex interdependent
  systems in energy infrastructures, and thus the
  energy sector cannot easily respond to the
  requirement of a technology change to help CO2
  emissions reduction
• ? The above raises questions about the optimal
  CO2 reduction pathways calculated by general
  equilibrium models

Objective of this study

• Expressing economic inertia in the energy sector
  by using a logistic curve
• Investigating the change of optimal CO2 emission
  reduction pathways when major parameters are
  varied

3
Ha-Duong, Grubb and Hourcade (1997), Nature, 390
270 273

• This is virtually the sole study which explicitly
  investigates the effect of economic inertia in
  the energy sector
• Using a cost-minimization model, expressing CO2
  abatement cost as
• where
• Ca, Cb constant
• Eref reference CO2 emission (business as usual
  case)
• d decline rate of cost
• e(t) emission reduction rate defined as
• Adjustment costs expressed by
  reflects economic inertia in the energy sector
  

4
Policy implications of Ha-Duong, Grubb and
Hourcade (1997)

• When the stabilization target is 550 ppm,
  deferring CO2 abatement does not affect abatement
  cost so much
• When the stabilization target is 450 ppm, the
  cost of deferral rises sharply as the inertia of
  the system is increased

5
Model of this study
Model

• Global economy is viewed as a two-sector Ramsey
  model
• Energy sector of the model consists of two energy
  technologies
• Fossil energy
• New carbon-free energy
• Diffusion of new energy technology is modeled by
  combining the logistic curve and learning-by-doing

Significance of the model of this study

• The model considers economic inertia and
  endogenous technological change, both of which
  are important for policy decision-making
• Use of logistic curve gives more realistic
  projections for policy studies compared to the
  use of CES production function with fixing
  elasticity between fossil and new energies

6
Model of global economy (the Ramsey model)

• Intertemporal utility maximization
• Production function
• Capital accumulation
• Income accounts identity

7
Logistic curve

• Energy inputs consist of two energy technologies
• Share of the new energy grows following the
  logistic curve
• Modifying the equation above into the inequality
  form
• Finite difference form is used in the computer
  program

8
Logistic curve (continued)

• Coefficient determines the speed of diffusion
  in
• It determines the potential speed of diffusion
  in
• Coefficient a can be interpreted as a parameter
  determining the degree of economic inertia the
  smaller a is, the larger the economic inertia and
  the slower the technological change

9
Energy price and learning-by-doing

• Price of fossil energy increases as a result of
  fossil energy extraction
• Price of new energy declines as experience
  increases

• Data of experience index(source McDonald
  Schrattenholzer, 2001)

10
Learning-by-doing in the computer program

• Using a finite difference form (Anderson Winne,
  2004)
• Substituting Wt by the cumulative installed
  capacity of new energy
• Estimation of W0 (Gerlagh and van der Zwaan,
  2004)

11
Combining the Ramsey model, logistic curve and
learning-by-doing

• Ramsey model

Learning by doing
Logistic curve
12
Climate change model

• Adopt a simple CO2 accumulation model (Grubb et
  al., 1995)
• Anthropogenic CO2 emission
• Natural CO2 emission (adopting DEMETERs
  parameterization)

13
Simulation scenarios

• Simulation is lead to a time path of emissions
  that satisfies the stabilization target of 500ppm
  (cost-effectiveness simulation)
• Investigating how
• Potential speed of technological change
  (coefficient a)
• Leaning rate (experience indexb)
• affect CO2 emission reduction pathways and
  the costs of reduction

• Model runs and parameter settings

STC Slow Technological Change FTC Fast
Technological Change LL Low Learning
HL High Learning
14
Common parameters (mainly adopted from DEMETER
model)
15
Calibration of the production function (based on
MERGE models method)

• Setting up the reference values of Y(t), K(t),
  E(t)
• Differentiating and rearranging the production
  function to obtain a and ß

16
Optimal CO2 emission pathways

• Four emission pathways are not very different
• Learning-by-doing has virtually no effect in STC
  (slow technological change)

17
Optimal CO2 reduction pathways

• FTC HL supports deferring CO2 emission
  reduction
• The other three paths are nearly the same in the
  early 21st century

18
Optimal technology switch timing

• HL (high learning) makes the starting point of
  technology switch earlier
• STC (slow technological change high economic
  inertia), like high learning, favors making a
  technology change sooner rather than later

19
Loss of GWP through CO2 emission reduction

• Loss of GWP primarily depends on the learning
  rate
• Pathways of GWP loss with the same learning rate
  are nearly the same in both the earlier and later
  periods

20
Technology switch and GWP loss under High Learning

• Technology diffusion of STC starts early, but GWP
  loss in the early period is not so different from
  FTC (major difference occurs after 2060)
• Starting technology switch from the early period
  does not make big difference of GWP loss before
  2050

21
Other discussions

• High economic inertia, like high learning, makes
  the starting point of technology switch earlier
• ? This conclusion is quite obvious and the
  qualitative insight that the model provides is
  not really telling us much.
• ? From the policy perspective, we need
  empirical evidence and quantitative analyses on
  the effect of economic inertia, i.e., to what
  extent economic inertia makes technology switch
  earlier.
• Ha-Duong et al. (1997) shows that the cost of
  deferring abatement rises sharply when the
  stabilization target is 450 ppm, while in this
  study economic inertia does not make abatement
  cost so different before 2050
• ? We need to model economic inertia in other
  ways (i.e. not using logistic curve but other
  expressions) to check the difference between the
  studies.
• This study applies learning-by-doing model for
  expressing endogenous technological change.
• ? Combining RD model (Romer, 1990) and
  learning-by-doing model might be an idea to
  improve the model.