Costing the Earth: Uncertainty and Climate Policy

Nafees MeahHead of Science

May 2010

Climate Change As ‘Hard’ Problem In Public Policy

As a public policy issue, climate change is a classic example of a ‘wicked’ problem

Notwithstanding the compelling scientific evidence, it is still contested

It is the case that there is and will always be irreducible scientific uncertainty – we cannot do a controlled experiment on the planet

Even if there is consensus on the science, that does not tell us what we ought to do: what are the trade-offs that the decision-makers need to consider?

Outline

Summary of the science of climate changeThe 2 degree target – AVOID ProgrammeKey questions in the economics of climate

changeEconomic modelling and cost-benefit analysisStern Review and its criticsBottom up technical/economic modelsThe task facing the decision maker

Carbon Dioxide Concentrations In The Atmosphere Since The Beginning Of The Industrial Revolution

MacKay (2009)

Evidence that CO2 is Man Made

Last decade has been the warmest since records began

Climate models show the observed warming is only explained by including human effects through GHG emissions

Excluding human influence

Incl

udin

g hu

man

influ

ence

Year to year range of modelled global temperatures

Observed Global Temperature Changes Not Explained by Natural Factors Alone

By 2100 Global Temperature is likely to be1.8 to 4oC Above 1990 Level

The scale of warming depends on emissions:

Low scenario 1.1 – 2.9oC

Best estimate 1.8 – 4.0oC

High scenario 2.4 – 6.4oC IPCC (2007)

Projected temperatures – land and polar regions warm more than oceans

IPCC (2007)

IPCC Fourth Assessment Report 2007

“Warming of the climate system is unequivocal, as is now evident from observations of increases in global average air and ocean temperatures, widespread melting of snow and ice and rising global average sea level” – p2, IPCC Synthesis Report

Temperature, Sea Level and Snow Cover

• The Earth’s surface has warmed by 0.75C since 1900

• Sea levels have risen by 20cm since 1900

• Now: glaciers, snow cover and sea ice are all declining

• Now: more heat-waves, droughts, extreme rain events and more intense cyclones

IPCC (2007)

Arctic Ocean September Ice Extent

Impacts of climate change

1°C 2°C 5°C4°C3°C

Sea level rise threatens major cities

Falling crop yields in many areas, particularly developing regions

FoodFood

WaterWater

EcosystemsEcosystems

Risk of Abrupt and Risk of Abrupt and Major Irreversible Major Irreversible ChangesChanges

Global temperature change (relative to pre-industrial)0°C

Falling yields in many developed regions

Rising number of species face extinction

Increasing risk of dangerous feedbacks and abrupt, large-scale shifts in the climate system

Significant decreases in water availability in many areas, including Mediterranean and Southern Africa

Small mountain glaciers disappear – water supplies threatened in several areas

Extensive Damage to Coral Reefs

Extreme Extreme Weather Weather EventsEvents

Rising intensity of storms, forest fires, droughts, flooding and heat waves

Possible rising yields in some high latitude regions

Cascade of uncertainty

Emission scenario

Atmospheric concentrations

Climate sensitivity

Climate change

Range of Impacts

Impacts may not increase linearly with warming

Lenton (2007)

Climate Sensitivity: Temperature Response of doubling [CO2]

AR4 concluded that best estimate of climate sensitivity was 30C with range of 2-4.50C (ca. 2SD)

IPCC (2007)

Q = F-λ∆T

Where Q = energy balance,F = forcing and λ = feedback parameter

At eqm Q=0

F = λ∆T

For the special case of doubling CO2

F’ = λS Where S = Climate sensitivity

Climate feedbacks include

Feedback

Water vapour This is the most important. Water vapour is a powerful greenhouse gas.

Cloud radiation Complex impact. Several processes involved. Sensitive to structure of clouds

Ocean-circulation Plays large part in determining earth’s climate. Large heat capacity and moves heat around.

Ice-albedo Ice and snow are a powerful reflector of solar radiation

Climate feedbacks affect the sensitivity of the climate.

Why a ‘fat’ tail?

AVOID Programme and the 2 degree target

AVOID examined variations in:

1. The year of peak emissions (2014 to 2030)

2. The emission rates leading up to the peak (BAU)

3. The emissions reduction rate following peak emissions (1 to 5% per year)

4. The net long-term level of emissions (zero to high levels)

Business as usual

Policy scenario

2 degree target agreed at Copenhagen Accord balances risks against technical and social feasibility in an informal way

AVOID Programme: 2 degree trajectories

AVOID uses a ‘tuned’climate model (MAGICC) Global average temperature determined by cumulative emissionsof GHGs (2.63TtCO2e 2000-2500)Approximates to the area under thecurveTake home message is that to stabilise temperature at 2 degrees is going to be a huge challenge - we need to peak soon and STRONG decline thereafter

GHG emission trajectories consistent with 2˚C increase in global average temperature at 2100 at a 50% probability level

Action on Climate Change

Key questions

1.How much will it cost to ‘stabilise’ the climate and avoid dangerous climate change?

2.Will the cost of avoiding dangerous climate change compete with other priorities such as development?

What action do we take in the light of the scientific evidence for climate change?

So if we applied the appropriate discount rate , then we might say that action would be justified on cost-benefit grounds if:

NPV = Present Value (benefits) – Present Value (costs) > 0

Or for a range of alternative policy actions, choose the one with highest NPV

Uncertainties in economic modelling of climate change

This is a formidable challenge because: We do not and cannot know the precise benefits of policy

action given the underlying uncertainty in the science We do not and cannot know what the future cost of the policy

will be given the long time horizons Costs and benefits functions are likely to be highly non-linear

- and we don’t know what they are If standard economic models are based on marginal

changes, how do we account for irreversibilities? Given the very long time horizons, what is the appropriate

discount rate to use?

Economic models for climate policy

Number of different kinds of economicmodels Much of the debate is about IntegratedAssessment Models (IAMS) which seek to integrate science and the economic theory to optimise climate policyThese are utility maximising models which seek to maximize, W, the social welfare, where

W = ∫ exp(-ρt)U[c(t)]dt

Where ρ is the rate of pure time preference,c(t) is the consumption at time t, andU is the utility function specifying how much utility is derived from a particular level of consumption

Outputs from Integrated Assessment Models

Time

Global economic activity

Reference case withoutimpacts

Reference case withimpacts

Cost of policy

Benefits of policy

Stern Review

Uses PAGE 2002 Integrated Assessment Model Takes account of risk and uncertainty through Monte Carlo simulations

on the climate sensitivity parameter, assumptions on risk aversion and equity

Key finding Cost of trajectory consistent with 550ppm CO2e stabilisation

averages 1% of global GDP per year (range -1% to 3.5%) Avoided damages would be 11% of GDP (range 2-27%) for

Baseline climate and 14% (range 3-32%) for High climate This contrasts with other IAMs which suggest a higher level of cost and

lower level of damage – DICE, MERGE, FUND Other models propose ‘policy ramp’ and modest rates of GHG reduction

The critics

Main criticism in the literature has been over the choice discount rate used by the Stern Review – should instead have used a market rate (i.e. 3 – 7%)

In the Ramsay formula, the social discount rate is given by:

Social discount rate = + ( x consumption/cap growth rate)

Reflects pure rate of time preference (which Stern suggest should be 0) and risk of human extinction (which Stern select as 0.1).

Elasticity of marginal utility of consumption (Stern suggest this is 1, which assumes society is moderately adverse to income inequality).

Growth in per capita consumption varies over time and according to extent of climate change damages. For baseline climate scenario with market impacts only, the 5-95% range of time-averaged growth is 1.08% - 1.14%.

Therefore in Stern, discount rate = 0.1 + (1*(1.08 to 1.14%)) = 1.18 to 1.24%

Discount rate have an important effect on the present value of climate change impacts

Value of £100 over time using different discount rates

90.479

81.865

60.577

36.69636.603

13.53313.262

1.7590.623

0.00000010.0030102030405060708090

100

0 10 20 30 40 50 60 70 80 90 100

110

120

130

140

150

160

170

180

190

200

Years

£

0.1% 0.5%

1.0% 2.0%

5.0% 10.0%

0.004

On Extreme Uncertainty of Extreme Climate Change – Martin Weitzman

Implication of the fat tail of climate sensitivity Translating the pdf of climate sensitivity into confidence levels for temperature change as a

function of GHG concentrations gives:

So at 550 ppm there is a 10% of T >4.8 ˚C. This is disturbing and can’t be ignored in formal economic modelling.

Damage function

Thought experiment on the damage function, which often takes the quadratic form in IAMs of:

C*(T) = 1/ 1+aT2

Where C*(T) is defined as the ‘welfare equivalent’ consumption as a fraction of what the consumption would be at T=0, and a=0.003

However, it is impossible to know a priori what the functional form should be for high temperatures

What if we used quartic or exponential form then the estimated damages would be very different

For a quartic exponential function, C*(T) = exp(-bT4), then at 10˚C C*(T) is 0.08% i.e. a catastrophic loss of ‘welfare equivalent’ consumption

Technical feasibility models - McKinsey Marginal Abatement Cost Curve – Bottom up estimates

Generally optimistic – it can be done and at comparatively small cost!

Choices facing the decision makers

Is formalised cost-benefit analysis appropriate for climate change policy?

If the answer is ‘no’ what other approach should we adopt? Given that a 2 ˚C has been adopted, should economic

analysis focus on seeking the cost effective pathway Is a risk based approach formalising the ‘precautionary

principle’ the appropriate way forward? Do we need more scientific knowledge on threshold

temperatures for major discontinuities or catastrophe’s? What else is there any other approach that we should

consider?

Thank you for your attention

Finally....