The Impacts and Costs of Climate Change
homogeneous population with few real outliers. However, the data may not be drawn from a single
population and the mean is sensitive to the tails of the distribution.
This is important for the estimation of climate change impacts as the models show that the distribution
is right skewed, i.e. the mean is higher than the median value46 and there are often outliers. The
median is less sensitive to outliers, and has been regularly quoted in literature studies, but is biased
towards lower values when the probability distribution has a long, high-value tail (as with climate
change impacts- note this may have led to bias in some of the values published.).
Strong and Weak Sustainability
In looking at any social cost of climate change value, it is extremely important to realise what is, and
is not, included in the value. It is also important to understand the trade-offs implicit in the numbers,
i.e. between different regions, or between different positive and negative effects.
The use of a single aggregated value implies an assumption about substitution between categories of
impact. The existing models, consistent with a cost-benefit analysis, assume full substitutability, i.e.
between very different impact categories. This may mean that the aggregated economic cost is the net
of the losses from for example to damages to natural ecosystems, against the positives, for example
from reduced energy for heating.
It is clear that different stakeholders will have different views on whether such substitution will be
acceptable. In order to help examine these issues, we propose that some detailed analysis is
undertaken, showing the balance of positive and negative effects, by region (rather than single global
For policy appraisal (cost-benefit analysis) we are interested in the marginal social costs of climate
change47. The marginal damage cost is the damage from an additional tonne of CO2 emitted.
Specifically, it is the change in the net present value of the monetised impacts, normalised by the
change in emissions. The models used in the analysis have been used to estimate the marginal social
costs, i.e. the models are run with and without additional pulses of emissions to assess the marginal
costs. However, the underlying analysis within the models, such as for loss of land, may not
adequately reflect scarcity, i.e. the models may be underestimating the true marginal costs48. There
have also been concerns that some of the potential changes from climate change are clearly non-
marginal (e.g. the risk of major changes to ocean currents, major sea level rise - note these are also
non-linear)49. Some commentators have responded to this by arguing it still possible to look at
marginal changes around policy decisions in regard to climate change policy, whilst recognising that
non-marginal impacts are not fully represented.
46 Measures based on a cumulative probability function include the quartiles and median. The distribution of the
data is captured in the median and quartiles: The minimum, maximum, and three quartiles (lower 25%, median
or 50% and upper 25%) are derived from the ordered data set. The median is the value for which 50% of the
data are larger.
47 Rather than the total costs of future climate change out-turns, or the average costs associated with for example
a doubling of CO2 concentrations.
48In practice, the SCC estimates from models such as FUND are 'average' marginal damage costs.
49 Threshold effects present particular challenges, both in estimating the physical impacts of climate change and
in determining appropriate WTP/WTAC values for these impacts.
AEA Technology Environment, August 2005
Watkiss, Paul; Downing, Tom; Handley, Claire & Butterfield, Ruth. The Impacts and Costs of Climate Change. Oxford, England. UNT Digital Library. http://digital.library.unt.edu/ark:/67531/metadc29337/. Accessed August 27, 2016.