CGE models for climate policy simulation: Some critical thoughts - - PDF document

CGE models for climate policy simulation: Some critical thoughts

Satoshi Kojima

Director, Economy and Environment Group Director, Economy and Environment Group Institute for Global Environmental Strategies (IGES)

Climate Policy Simulation Tool (CPST) Brainstorming Meeting 15-16 July 2010, Shangri-La Hotel, Bangkok, Thailand

I nstitute for Global Environmental Strategies

Outline

I t d ti it f iti l i ti f li it ti



Introduction : necessity of critical examination of limitations



Scope of CGE application

 2 types of policy impacts  2 types of policy impacts 

Difficulty in long-run assessment

 Discounting problems 

Policy cost issues (e.g. abatement costs)



Coping with uncertainty

 Policy planning and policy implementation 

Conclusions

2 I nstitute for Global Environmental Strategies

CGE b d li li i l i ? CGE-based climate policy simulation?

CGE b d li t li i l ti l t l l i CPST



CGE-based climate policy simulation may play a central role in CPST.

 Economic impacts of climate change and climate change policy through inter-

sectoral linkages are important. Both major consequences and drivers of environmental and social impacts.

 Multi-regional CGE seems suitable to assess international impacts of domestic

policies as well as impacts of international policy coordination. po c es as e as pacts o te at o a po cy coo d at o



To explore such possibility, need to critically examine following limitations:

 Scope of CGE application

p pp

 Difficulty in long-run assessment  Policy cost issues (e.g. abatement costs)  Coping with uncertainty 3 I nstitute for Global Environmental Strategies

Scope of CGE application: 2 types of climate policy impacts



Direct policy impacts



Direct policy impacts

 Impacts caused by implementation of climate policies itself: e.g. influence of

carbon pricing on market conditions, air pollution mitigation assessment, etc.



Mitigation (or adaptation) impacts

 Impacts of climate change mitigation (or adaptation) induced by climate policies:

e g impacts of more stable food production impacts of less flood impacts of e.g. impacts of more stable food production, impacts of less flood, impacts of less damages from extreme climate conditions, etc. They have different characteristics. y



Time-lag: Direct policy impacts can be generated in much shorter period than mitigation/adaptation impacts.



Uncertainty: Mitigation/adaptation impacts depend on climate change severity and



Uncertainty: Mitigation/adaptation impacts depend on climate change severity and they are associated with much higher uncertainty than direct policy impacts.



Climate change dependence: Direct policy impacts matter regardless of actual li t h t climate change events.

4 I nstitute for Global Environmental Strategies

P li i li i Policy implications



Ends and m eans of policy



Ends and m eans of policy

 Mitigation/adaptation impacts are originally ends of climate policies. However,

assessing these impacts are very difficult due to time scale and high uncertainty. g p y g y

 Direct policy impacts are originally means (or by- products) of climate policies,

but easier to assess. In practice, more justifiable and robust policy ends.



Potential m agnitude of im pacts

 Mitigation/adaptation impacts are potentially huge as potential damage of

t t hi li t h i i ( f th St R i ) B t d t hi h catastrophic climate change is immense (c.f. the Stern Review). But due to high uncertainty not necessarily convincing to justify policy.

5 I nstitute for Global Environmental Strategies

R d d f CGE l i Recommended scope of CGE analysis

Miti ti i t f t t tti



Mitigation impacts for target setting

 Mitigation impacts might not be suitable subject of quantitative assessment for

cost-benefit analysis (CBA). Rather, better to reflect them to target setting y ( ) g g through political process. Inclusion of mitigation impacts in CBA causes controversy. f f

 Impacts of adaptation policies need to be addressed by scientific models, e.g.

hydrological models, ecological models, etc. → These will be done outside CGE analysis → These will be done outside CGE analysis



Direct policy impacts for demonstrating policy benefits

 Direct policy impacts are free from controversy over high uncertainty in climate

ect po cy pacts a e ee

• co t o e sy o e

g u ce ta ty c ate

• science. Once demonstrating net benefits of climate policy based on these

impacts, policy justification is robust. → This is suitable for CGE analysis

6 I nstitute for Global Environmental Strategies

H fl b i CGE? How to reflect abatement costs in CGE?

I t f b t t t



Importance of abatement costs

 Key driver of economic impacts.  Sector specific and country specific abatement costs are crucial information to  Sector-specific and country-specific abatement costs are crucial information to

analyse impacts of climate policies on competitiveness.



Difficulty in estimating abatement costs



Difficulty in estimating abatement costs

 Bottom-up approach estimates expenditures of sector for abatement actions. But

these expenditures are not necessarily “costs” for an economy (could be resource reallocation).

 Top-down approach tries to capture abatement costs for an economy as loss of

d ti it ( GDP i t) lt f b t t ti iti M CGE productivity (e.g. GDP per input) as a result of abatement activities. Many CGE studies employ this approach.

7 I nstitute for Global Environmental Strategies

Ab i CGE di Abatement costs in CGE studies

CGE b ti ti l fl t iti ti h i fl t d b th



CGE-base estimation merely reflect mitigation mechanisms reflected by the models.

 Mainly substitution effects due to price change (e.g. carbon pricing).

y p g ( g p g)

 Many studies do not estimate sector-specific and country-specific abatement

costs, which are key to analyse competitiveness issues.



Challenges to reflect sector specific abatement costs

 Ideally, abatement costs will be modelled as additional inputs (intermediate

inputs such as equipments or factor inputs such as labour and capital) for inputs such as equipments, or factor inputs such as labour and capital) for abatement activities with reduced productivity as these inputs do not directly contribute to production.

 In practice such simulation is difficult as these variables are endogenous  In practice, such simulation is difficult as these variables are endogenous

• variables. Compromise may be just reducing corresponding productivity

parameters.

8 I nstitute for Global Environmental Strategies

CGE d l d l i l bl CGE models and long-run environmental problems

I CGE l i l t f l ?



Is CGE analysis relevant for long-run?

 Model parameters (technology parameters, behavioral parameters such as

preference of commodities) may change for long-run.

 Economic conditions, social structures may drastically change beyond our

expectation.

 Better to restrict short or mid term (say 10

20 years)

 Better to restrict short- or mid-term (say, 10 – 20 years). 

How to reflect long-run environmental impacts like climate change?

 Rely on scientific results (e g climate model) to set policy targets for simulation  Rely on scientific results (e.g. climate model) to set policy targets for simulation.

For example, setting a 20% GHG emission reduction by 2020, considering the potential climate change for long-run.

 CGE model can provide potential impacts of policies to achieve given targets  CGE model can provide potential impacts of policies to achieve given targets. 

Using CGE models for long-run cost-benefit analysis cannot avoid “discounting” problems discounting problems.

9 I nstitute for Global Environmental Strategies

Di i bl Discounting problems



Long run cost benefit analysis faces discounting problems: “Normative vs



Long-run cost-benefit analysis faces discounting problems: Normative vs. Positive”

 For long-run, setting significant discount rate results in ignoring future events.

Example: 100 years consequences: 1%/year -> 37%, 5%/year -> 0.8%!

 Some argue low discounting rate to avoid ignorance of future (Normative

discounting) g)

 Some argue empirical observation indicates quite high discounting (5 – 10%),

and setting unrealistic low discounting rate is irrelevant.



Kojima (2006) argues that normative and positive rates are different and they can coexist.

10 I nstitute for Global Environmental Strategies

Discounting issue: Reconsideration of private utility function

R id th i f i t t l tilit ti i i



Reconsider the meaning of intertemporal utility optimisaion.



Gorman (1957) distinguished current satisfaction, termed as “felicity”, from utility defined as the highest possible average level of felicity.



Strotz (1956) acknowledged that the current utility may depend on the consumption of a later date, and criticized the term “instantaneous utility” as a misnomer.



Let’s define



Felicity: enjoyment (or satisfaction) derived from the current activities or i t environment.



Utility: enjoyment (or satisfaction) derived not only from the current conditions but also from the expectations of future felicity. U: Utility function of a household in time t u(t): felicity function in time t

 dt

t u e U

t



 





11 I nstitute for Global Environmental Strategies

 : pure rate of time preference

C i f 2 di i i l lf f i Coexistence of 2 discount rates in social welfare function



Then, in the social welfare function, normative and positive rates coexist.

dG dt t u e e dG G U e SW

T t G T G

  

     

   

) ( ) (

  

dG dt t u e e dG G U e SW

G

  

      ) ( ) (

G: Gth generation g : a weight for Gth generation’s utility, or in other word “social discounting rate”.



Social discounting rate must be determined from normative perspective. →



Social discounting rate must be determined from normative perspective. Ramsey’s famous aphorism: any nonzero discount rate is “ethically indefensible” (Ramsey 1928).



But is it an appropriate way to treat intergenerational equity? Intergenerational equity must be about conservation of the basis of human survival, and

12 I nstitute for Global Environmental Strategies

intergenerational considerations must be done outside cost-benefit analysis.

Uncertainty issues

U t i t b fl t d i CGE l i b ti li l i



Uncertainty can be reflected in CGE analysis by separating policy planning stage and policy implementation stage.



Policy planning is based on expectation on future conditions (price paths, d i i fl i li i di i ) productivity fluctuation, climatic conditions, etc.).



Impacts of policy implementation can be assessed against “realized”

• conditions. For example, any random shocks can be introduced through Monte

Carlo simulations.



This approach is conceptually simple, and easy to assess various aspects such as robustness against bad events such as robustness against bad events.

 This approach seems particularly relevant for policy simulation for adaptation. 13 I nstitute for Global Environmental Strategies

Illustrative example of this approach

K ji (2007) d l d li d R t th d l t



Kojima (2007) developed an applied Ramsey type growth model to assess sustainable development policies:



Can be regarded as full-dynamic small-scale CGE model (3 production sectors, d h id f bli i h d i i d and the government as provider of public services such as water and irrigated land).



3 sectors are: rain-fed agriculture, irrigated agriculture, and other sectors. Vulnerability of rain-fed agriculture is modelled through a stochastic risk factor.



Moreover, uncertainty is modelled as stochastic foreign exchange rate. Th ti l li i d t i d b d t ti f th d i i



Then, optimal policy is determined based on expectation of thee decision- maker, while policy impacts are derived through Monte-Carlo simulations.

14 I nstitute for Global Environmental Strategies

Illustrative example: Some numerical results

90 Mean Equivalent Variations



CC1: Standard

30 60 90 ita/year] q

Alt4 (B ) Alt4 (CC1)



CC1: Standard deviation of rain-fed agricultural sector’s risk factor increased

• 30

EV [DH/capi

Alt4 (Base) Alt4 (CC1)

risk factor increased 20% from the base scenario, representing

1 3 5 7 9 11 13 15 Year

100 200 Annual EV (Base-Alt4) 100 200 Annual EV (CC1-Alt4)

representing potential negative impacts of climate change

100 100 V [DH/capita/year] 100 100 V [DH/capita/year]

change.



Mean value is not very affected, but fluctuation of

• 200
• 100

1 3 5 7 9 11 13 15 E Year

• 200
• 100

1 3 5 7 9 11 13 15 E Year

fluctuation of possible outcomes is drastically increased

15 I nstitute for Global Environmental Strategies

increased.

Source: Kojima 2007 (adapted)

Conclusion (1)

Thi t ti iti ll i th li it ti / bl f CGE



This presentation critically examines the limitations/problems of CGE models to be applied for climate policy simulations.



To consider the scope of CGE application to climate policy simulations, p pp p y , better to distinguish 2 types of policy impacts: direct policy impacts and mitigation/adaptation impacts.



Direct policy impacts, e.g. influence of carbon pricing on market conditions,



ect po cy pacts, e g ue ce o ca bo p c g o a et co d t o s, can be simulated by CGE models



Assessment of mitigation/adaptation impacts need to be done by scientific models outside CGE models models outside CGE models.



Modelling policy costs (e.g. abatement costs) is a challenge.



Shocking production inputs (endogenous in default) is practically difficult. g p p ( g ) p y



Potential compromise is to reduce productivity of production inputs.

16 I nstitute for Global Environmental Strategies

Conclusion (2)

CGE d l t it bl f l i l ti



CGE models are not suitable for long-run simulations.



Model parameters may change in the long-run.



Economic conditions and social structures may drastically change. y y g



Long-run cost benefit analysis cannot be free from discounting problems.



Intergenerational equity must be about conservation of the basis of human survival not about cost benefit survival, not about cost-benefit.



Better to restrict time frame of CGE analysis to short- or mid-term.



Uncertainty issue can be reflected in CGE models by separating policy



Uncertainty issue can be reflected in CGE models by separating policy planning based on expectation and policy implementation under uncertainty.



This approach is particularly relevant for adaptation policy simulation.

17 I nstitute for Global Environmental Strategies

Th k f i Thank you for your attention.

Contact address: kojima@iges or jp Contact address: kojima@iges.or.jp

18 I nstitute for Global Environmental Strategies

Reference

G Willi M (1957) ‘C i diff d di i i hi i l Gorman, William M. (1957), ‘Convex indifference curves and diminishing marginal utility’, Journal of Political Economy, 65 (1), 40−50. Hanley, Nick. et al. (1997), Environmental Economics in Theory and Practice, Oxford: Oxford University Press. Kojima ,Satoshi (2007) Sustainable Development in Water-stressed Developing Countries: A Quantitative Policy Analysis. Cheltenham: Edward Elgar. Kojima ,Satoshi (2006) ‘Reconsideration of dynamic utility optimisation and intergenerational equity in sustainable development studies’, International Journal

• f Ecological Economics and Statistics, 6:26-36.

g , Ramsey, Frank P. (1928), ‘A mathematical theory of saving’, Economic Journal, 38, 543−59. Strotz Robert H (1956) ‘Myopia and inconsistency in dynamic utility maximisation’ Strotz, Robert H. (1956), Myopia and inconsistency in dynamic utility maximisation , Review of Economic Studies, 23, 165−80.

19 I nstitute for Global Environmental Strategies