EC330 AT W3: Climate change damages

Challenges in measuring climate change damages

• Projected changes are historically unprecedented (not known before)

  • require new approaches for assessment, as existing models may not accurately capture the complex interactions and long-term impacts.

• Non-linear responses

• Interactions of sectoral impacts

  • different range of outcomes

• Spatial spillovers

  • effects of climate change in one location may propagate to another

• Measurement issues

Non-linear responses

• Economic activities like labor productivity, crop yields, and labor supply react non-linearly to temperature.

  • This means small temperature increases can lead to disproportionately large impacts once certain thresholds (e.g., 20–30°C) are crossed.

  • To "really capture the effect," it's essential to use temperature distributions rather than just averages. Otherwise, we risk underestimating the real economic impacts of climate change.

    • average temperature data hides extreme heat days, which are the ones that really cause damage.

What are the approaches to capturing non-linear effects?

1. Outcome specific approaches

   1. When to use: If we have good knowledge of how a certain process (e.g. crop growth) reacts to temperature.

   2. Thresholds: Use known temperature thresholds where performance switches from beneficial to harmful (e.g., crops grow well below 29°C but yields drop sharply above that).

2. Non-parametric methods

   1. more flexible as it avoids restrictive functional forms (e.g., assuming the effect of temperature is always linear or quadratic).

   2. How it works: Instead of modeling a single threshold, it groups temperature data into bins (ranges, like 25–27°C, 27–29°C) and looks at outcomes in each.

What are the different range of outcomes of temperature extremes?

1. GDP growth

2. Health and mortality

3. Labour supply and productivity

   etc..

Why might the range of outcomes pose challenges for estimating climate change damages?

• Hard to aggregate across outcomes

  • Different outcomes have different units and impacts

  • crop yield loss vs increased mortality

• Interaction of effects

  • lower crop yields can worsen health through malnutrition

  • may cause double counting

• Not specific

  • damages are hard to define precisely

• Not evenly distributed

  • impacts vary by region, sector and socioeconomic group

  • some groups may be far more vulnerable than others

• Non-linear damages

  • small change in temperature can cause big, unpredictable effects

  • making it hard to predict with simple averages

What may cause spatial spillovers?

• Ecosystem interactions

  • deforestation leads to soil erosion, loss of biodiversity etc

• Human activity

  • our interconnected world means that changes in one location can ripple through economic, social and physical systems

  • ie: migration, trade and production networks can connect human activity and amplify the effects

Types of data

1. Weather station data

   • raw data from ground stations

2. Gridded data

   • statistical interpolation of station data

3. Satellite data

4. Reanalysis data

   • station/satellite data run through a climate model

Advantages of weather station data

• locally accurate

• can generate a long time series, provides detailed observations of climatic conditions and trends over time.

Disadvantage of weather station data

• limited or changing coverage

• non-random station locations

  • tends to be clustered in populated areas or richer countries, causing bias

Advantages of gridded data

• more complete spatial coverage than weather stations

• useful for creating a balanced panel

Challenges of gridded data

• Interpolation techniques vary, leading to uncertainty

• Sparse inputs lead to inaccurate outputs

  • especially in regions with few stations

Advantages of satellite data

High spatial resolution

• detailed coverage across globe, even remote areas

Disadvantages of satellite data

• relatively recent

  • limited historical depth

• cloud cover issues

  • can obscure images, making it hard to detect floods

  • will have data gaps

Advantages of reanalysis data

• more complete

  • combines satellite and model data for global consistency

Disadvantage of reanalysis data

• relies on assumptions and parameters used in model

  • may not fully reflect reality

What is mortality displacement?

occurs when extreme events (like heatwaves) cause deaths to occur earlier than they would have, leading to a temporary spike followed by a dip in mortality.

total deaths over months/years

Integrated assessment models

a tool for assessing potential policy responses to climate change

Steps of building integrated assessment model

1. Model projecting GHG emissions

2. Model mapping GHG emissions into climate change

3. Damage function estimating economic costs of climate change

4. Social welfare function for aggregating damages over time

Cost benefit analysis

aims to assess the desirability of an action by comparing the benefits and costs of the action in a common unit of account

uses expected utility theory

• weighs outcomes by their probability

Why do we need to use a discount factor in conducting the cost benefit analysis?

costs and benefits often occur at various points in time

• carbon abatement has costs today, benefits are far in the future

so, discount factor is used to express a future cost or benefit as an equivalent current cost or benefit

Challenges with implementing cost benefit analysis

• potentially large range of costs and benefits

• difficulties in measuring costs and benefits

• choice of discount rate?

• an outcome of NPV>0 may have undesirable effect on inequality

• benefits may be non-market

• flow of costs and benefits may be uncertain

• how to account for possibility of extreme effects?

• how to account for possibility of irreversible outcomes?

• there is a long timeframe, leading to intergenerational tradeoffs

  • how do we account for new abatement technologies?

Why might a social discount rate < market rate be appropirate?

• accounts for ethical considerations and intergenerational equity.

• should not include risk premium

  • reflecting pure time preference, not risk.

  • not relevant in considering a society-wide long-term impacts like climate change

• evidence behaviour is consistent with hyperbolic discounting

  • tend to discount the near future more steeply than the distant future

  • long-run damages should be discounted less heavily over time.

Normative estimate of discount rate

δ captures preference for today’s generation relative to next generation

ng captures the fact that future generations will be wealthier

• gains to wealthier people should count less than gains to poorer people

  • future < current

the discount rate also reflects the value of incremental future consumption may depend on whether future generations are wealthier than the current generation

δ, pure rate of time preference

measures amount by which utility of a future generation discounted in welfare calculations relative to utility of current generation

Explain the economic and philosophical debate on choice of δ

δ =0

• current and future generation should be weighted equally, only ethically justifiable choice

• (Stern)

δ >0

• preference for present over future utility that individuals exhibit in savings and investment behaviour

n, elasticity of marginal utility of consumption

n=0

• MU of dollar of consumption is constant as income increases

n=1

• 1% rise in income reduces MU of consumption by 1%

it reflects the society’s aversion to consumption inequality

• if higher, that means you really want redistribution to transfer income from rich to poor

Dismal Theorem

under certain conditions regarding uncertainty and preferences, society has an indefinitely large expected loss from high-impact, low probability events.

implies that CBA will be very misleading in the presence of catastrophic outcomes with small, uncertain probabilities

• expected utility theory doesn’t work well — because the rare disasters (low probability) overshadow everything else, making decisions hyper-sensitive or even mathematically undefined..

Weitzman’s argument

Even if traditional models don’t justify strong climate action (e.g. through optimal consumption smoothing), we should still act early because there's a small but real chance of catastrophic outcomes — and these low-probability, high-impact events are not handled well by standard economic tools.

Nordhaus vs Stern

Nordhaus chooses δ and n that reflects a higher preference for present utility

• should be consistent with observed real interest and saving rates rather than normatively acceptable parameters.

Stern chooses δ close to zero

• based on ethical arguments

• >0 only to account for possibility of exogenous extinction of humanity

Stern chooses n = 1

• If someone’s consumption doubles (from c to 2c), then an extra dollar of consumption is half as valuable.

• So, we value extra consumption less the richer someone is.

• shows some preference for equality

Nordhaus wrinkle experiment

If you use a very low (or zero) discount rate, you place almost equal weight on all future generations — even those 10,000 years from now.

• willing to spend a massive amount today to prevent a tiny inconvenience in the distant future

• can be politically and economically unrealistic.

Hence, the experiment highlights how using a low discount rate can lead to inflated present value estimates of future benefits and costs.

Stern’s response to Nordhaus’ argument

Social Rate of Return is not equal to Private Rate of Return, especially for climate change

• choice of market rate is depended on the type of assets invested in

  • long run lending rates on gov bonds is 1.5%

  • private long run rates of return on equities is 6-7%

• social discount rate may fall over time

  • environmental services are income elastic

    • As societies become wealthier, clean air, biodiversity, and climate stability become more highly valued.

    • We’re willing to pay more for environmental goods as incomes rise.

  • price of environmental goods may increase with scarcity over time

    • If something is more valuable in the future, we should discount it less today to reflect the increased importance to future generations