Climate Change Economic Modelling Massimo Tavoni, FEEM, Milano

1
Climate Change Economic ModellingMassimo Tavoni,
FEEM, Milano
Univ. Cattolica, Piacenza, 30 Novembre 2005
2
Summary

• The Climate Change problem
• Introduction to models coupling Economy and
  Climate
• Technological Change
• Modeling Framework WITCH
• Open points/enhancement

3
CO2, energy and income
4
IPCC Third report (2001) Earth temperature

• Most of the earths warming over the last 50
  years (0.6 degrees Celsius) can be attributed to
  human activities
• Global temperature is expected to increase by 1.4
  - 5.8 degrees Celsius over the next century

5
Damages

• Unique and threatened systems (extinction of
  species, loss of unique habitats, bleaching and
  death of coral)
• Extreme climate events (health, property and
  environmental impacts from increased frequency
  and intensity of some climate extremes)
• Distribution of impacts (cereal crop yield
  changes, decreases in water availability, greater
  risks to health, net market sector losses)
• Global aggregate impacts (globally aggregated net
  market sector losses, more people adversely
  affected than beneficially affected)
• Large scale, high impact events (significant
  slowing of thermohaline circulation melting and
  collapse of ice sheets)

Source IPCC-TAR
6
Climate cycle
7
Climate change and policy

• Climate change one of most significant
  environmental problems, though one of the most
  controversial.
• Global, different geographical and sectoral
  effect
• Free-riding

• Long-run phenomenon, intergenerational
• Uncertain, difficult political support

• Costly
• Economic growth, developing countries

8
Costs of Kyoto
9
Economic-Climate Models
10
Modeling variety

• Economic Module
• Static (and dynamic ?) CGE Models (e.g. EPPA,
  MIT)
• Optimal Growth Models (e.g. RICE, Nordhaus
  MERGE, Stanford MIND, Pik)
• Bottom Up Models of the Energy System (e.g.
  MARKAL)
• (Econometric Models (e.g. E3, University of
  Cambridge))
• Coupled with
• Climate Module (e.g. IPCC models at DEA-CCAT,
  Climate Change Advisory Team of the Danish Energy
  Agency)

11
Results variety ..
12
BU vs TD divide
Top-down
Bottom-up

• Climate change technologically complex
• technology detail
• technical change

• Climate change a global long-run phenomenon
• long-term model
• feedback within the model
• welfare maximization
• strategic interaction

13
Endogenous Technical Change
14
Technical Change/1

• Changes in technology believed to bring about the
  de-coupling of economic growth from the
  generation of polluting emissions.
• Economic analysis offers justification for public
  policies to induce ETC. Two market failures
• Climate related externalities not accounted for
  in the market prices of carbon fuels gt direct
  emission policies (taxes, caps, EU)
• Spillover benefits to society gt technology
  incentives (subsidies, US)

15
Technical Change/2

• Investment in energy RD declined by
  approximately 50 worldwide between 1980 and
  1999, Richels et al. (2004).
• BUT - oil price increase
• - energy security issues
• - ratification of the Kyoto Protocol
• - incentives from many governments (e.g. US)
• ?Imply a growing interest for the issue of energy
  technological change and its driving forces

16
Technical Change/3

• Two main driving forces of energy technological
  evolution
• Investment in Research and Development (RD), or
  disembodied technological change typically
  modelled in a TOP DOWN framework.
• Accumulation of experience deriving by change in
  hardware and actual implementation, the so called
  learning by doing (LbD) or embodied technological
  change typically modelled in a BOTTOM UP
  framework.
• The two forces interplay.

17
Main Results in Literature - ETC

• The effect of embodying endogenous technical
  changes (ETC) in an economic climate model is
  ambiguous
• The presence of ETC could favor postponing
  emission reductions, as in Wigley, Richels and
  Edmonds (1996)
• (RD channel)
• The presence of ETC could represent an incentive
  to undertake at least some immediate abatement
  action in order to increase the stock of
  experience and to decrease abatement costs, as
  for example in Grubb (1996)
• (LbD channel)

18
WITCH World Induced Technical Change model
19
General structure
Emissions CO2, other GHGs
Optimal Growth Model
Climate Model
Temperature
Output reduction
Damage
20
Main features

• Top-down neo-classical optimal growth (dynamic,
  perfect foresight)
• Detailed energy input specification (BU)
• Hard-link (stand-alone optimization) hybrid
• Endogenous Technical Change
• World, 12 regions, interacting strategically
  (open-loop Nash)
• Solved numerically

21
The objective function

• Where R(t) the pure time preference discount
  factor
• and c(n,t) is per capita consumption. Each model
  period accounts for 5 years.
• Subject to budget constraint

22
Production

• Kc capital in final good production
• L labour (full employment)
• ES energy services
• Xj,z is the total consumption of fossil fuel j
• Pi is the price of fossil fuel j
  (endogenous)
• Pccs is the cost of CCS

23
Energy representation
24
Electricity
Electricity production via fixed proportions
prod. Function. Parameters represent
technological features of power production.
Capital accumulation through technology-specific
investment cost.
Energy is but capital cumulated over time in
power plants and capital invested for fuels and
OM, which fully depreciates every period.
25
Endogenous Technical Change
1. Learning by Doing (global) in plants
investment cost
2. Energy RD for energy efficiency
Positive externality of knowledge creation but
physical capital crowding out
26
Regional disaggregation
27
Channels of interaction

• CO2 emissions
• Prices of fossil fuels
• Technology spillover (LbD and RD)
• Non-cooperative game solved recursively
• - at each iteration each region optimal choice
  given all other regions (previous iteration)
• - Till convergence, ie best response to all other
  region best responses

28
Algorithm

• Solved with GAMS (NLP, solver CONOPT)
• 13 choice variables, 30 time steps, 12 regions
  (over 4500 variables)
• Solution time 30 mins to 1 hour on Pentium M, 2
  Ghz
• Potential non-convexities due to LbD. Tests on
  objective function, possibly resort to heuristic
  algorithms
• Non-cooperative game robust to regions
  ordering, different starting values

29
A piece of baseline results
30
Exercises

• The structure of the models allows to run
  different scenarios/ evaluate several policies
• CO2 emissions stabilization (450,500,550 ppm)
• emissions caps vs taxes
• permit trading -banking
• RD subsidies vs direct policies
• technology options (CCS, nuclear, renewables)
• region coalitions

31
Pifalls-improvements

• Non-electric energy
• Lbd increasing returns
• Trade
• Uncertainty stochastic version
• Better climate module/feedback

32
massimo.tavoni_at_feem.it