[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries
.

Climate change the past is

ow good are climate models?
Might they be missing non-linear
behaviour that typifies the real
world? Can the climate of the past
guide us in what to expect in future? These and
other questions are discussed in a piece of
research commissioned by the Environmental
Research Group for the UK actuarial profession.
Making use of money made available by the
Research Steering Committee, Dr Stephan Harrison
has produced a discussion note on these
topics for the profession. Stephan’s paper takes
as read that climate change is happening in a
big way. And so it should. We sincerely hope
that this is no longer an area of debate for a
learned profession such as ours.
Is the past a good guide to the
future?
Well, as you would expect the answer is ‘sometimes’.
The past at least shows us the level of
extremes that are undeniable. When challenged
with ‘it couldn’t possibly get that bad’
it is a powerful argument indeed to be able to
say ‘it did before’. It is a brave (or stupid?) person
that assumes that the past cannot or will
not repeat itself.
So, what do we mean by ‘climate models’? In
Stephan’s paper he refers to general circulation
models (GCMs). The most complex of these
now couple together the oceans, the atmosphere,
the biosphere (trees, etc), and other
factors. They subdivide these into a threedimensional
grid; each grid box contains within
it key data such as pressure, humidity, temperature,
etc. These conditions are used to simultaneously
evolve the model one step at a time
using the laws of physics, eg fluid dynamics
equations. The models are therefore based on
‘reductionism’, ie simple models of nature, created
and tested over many years by scientists.
The grid size is important. The more grid points,
the more features the models can pick up. For
example, early models covered the UK in just
four grid cells (horizontally); the latest Japanese
Earth Simulator has many hundreds.
A key quote from Stephan’s paper is: ‘The
interactions between the non-linear atmosphere,
hydrosphere, biosphere and geosphere,
and society are complex and form one of the
main sources of uncertainty in our predictions
of future climate and environmental change.’
Note his mention of ‘society’ in the model. The
actions of society determine the impact of natural
events, for example the design of a city
determines how well it can cope with a flood;
of course, crucially, the actions of society determine
how much carbon dioxide we emit over
the next 100 years with profound implications.
Stephan’s paper considers some limitations of
models in various categories:
? Theoretical limitations a key point that
Stephan raises is that at certain scales the climate
may display non-linearity. He reminds us
that non-linear behaviour means that initial
condition errors are magnified forward in time,
possibly leading to the onset of chaos and
unpredictability. As a result, it may not be possible
to model certain large-scale features of the
environment. Professor Lenny Smith of the
London School of Economics, who also attends
the Climate Change Working Party, speaks at
greater length on this subject in his recent book,
Chaos: A Very Short Introduction. For this reason,
information about the past is a vital tool for
understanding the future.
? Forcing uncertainty we do not know how
much CO2 is likely to be emitted. A recent paper
argues that emissions are at the high end of what
was allowed for in earlier projections, ie the reality
is worse than the previous ‘best estimate’.
? Model inadequacy the models are just
models after all. They are based on our understanding,
which is not perfect; Stephan gives
the example that ice sheets are breaking up
faster than our models say they should. Climate
happens on all scales from large to small. Some
processes are smaller than the grid size and so
are not picked up properly. One of Stephan’s
examples is cloud formation.
? Model uncertainty this is a topic familiar to
actuaries. For many years guidance on financial
condition reports and now individual capital
assessments (ICAs) have urged us to consider
the possibility that our parameters may be
wrong. Actuaries might be more comfortable
calling this ‘parameter uncertainty’. Examples
closer to home are: mortality predictions have
been wrong year after year, and asbestos
reserves rose steadily over time. Sometimes the
real world just doesn’t behave in line with our
expectations! Climate scientists are also aware
of this and run so-called ‘ensembles’ of models
with slightly different initial conditions then
observe the outputs. You may have heard of an
initiative called www.climateprediction.net where
you can download software and run a climate
model on your home PC. One of the main
architects of that model, Dr Dave Stainforth, sits
on our Climate Change Working Party. His
work (et al.) showed that for some plausible sets
of parameters the sensitivity of the climate to
carbon dioxide is much worse than the mean
outcome, and by symmetry some sets show better
than mean sensitivity too. We should bear
in mind that we are urged by our regulator to
consider 99.5 percentile events to ensure the
solvency of the various institutions we work in
(not quite sure why pension funds don’t have
this requirement?). If governments had to do
this they might be considering the extinction
of humanity according to these models. So, if
you take nothing else from this article and
Stephan’s paper, it is that climate change predictions
that you hear about are usually just
‘best estimates’ and they are bad enough.
So, if the climate models are beset with difficulties,
what can we do? Stephan argues that
past climate information can be extracted from
a number of sources: ice cores, tree rings, isotopes
in shells, gravel size in ocean deposits,
there is an ever-growing list. These can show us
what the climate was like in the past and how
it behaved in those times when carbon dioxide
levels were higher, like today. Unfortunately
there is nothing to cheer us up here either,
summed up dryly by Stephan: ‘Unfortunately,
the pattern that emerges does not give us
grounds for optimism.’ The paper notes that
rapid change has happened regularly in the
past. This is particularly the case at regional
levels. Stephan was also one of the authors in
the recently published Lloyd’s paper on rapid
climate change (see www.lloyds.com/360), which
highlights the possibility of rapid changes in
the climate from various sources including sea
level (due to ocean current reorganisation or ice
sheet collapse, flooding caused by sea-level rise
or heavy rain, or flash events and drought). A
key conclusion is that current climate models
understate this risk.
Stephan quotes Wally Broecker, an internationally
respected climate scientist who wrote
in 1999: ‘No one understands what is required
to cool Greenland by 16 °C and the tropics by
4 ± 1 °C, to lower mountain snowlines by
900m, to create an ice sheet covering much of
North America, to reduce the atmosphere’s CO2
content by 30% or to raise the dust rain in
many parts of Earth by an order of magnitude.
If these changes were not documented in the
climate record, they would never enter the
minds of the climate dynamics community.’
The implication being that the world has surprises
to throw at us that we can barely imagine.
And if we can’t imagine them, it is unlikely that
our models will contain them.
The paper concludes with a consideration of
the use of model results by policymakers.
Stephan says: ‘It would appear that end-users of
the models have largely failed to recognise the
lack of predictive skill at the regional level and
the inability of models to produce rapid climate
change and sea-level rise in their predictions.’
This lack of skill should not be taken by climate
sceptics to yet again conclude that we need take
no action, however, because: ‘It seems likely
that the models will underestimate rather than
overestimate the climate sensitivity over the
long run, because they omit relevant variables.’
What should we do?
There are many things we can do as individuals
and these are of course a matter of personal
choice. Fundamentally, as professionals, we
have to ensure that those we advise are aware
of the issues and are basing decisions on consideration
of a world that will be different under
climate change.
Pension actuaries can urge the trustees they
advise to require that their fund managers pursue
responsible engagement strategies with the
boards of companies in which they invest.
Actuaries in this field might also consider the
impact of climate change on the large property
portfolios such funds hold; are these properties
adapted to climate change? How much will it
cost to change them? Are these factors already
allowed for in the price or could a sudden
price correction take place?
Similar issues arise for life insurance actuaries.
With-profits contracts will be affected by lower
returns from asset portfolios in the future; the
Stern report suggests that growth in GDP will
be affected regardless of the action we take. Are
our boards aware of this? Do our ICAs and
reserves allow for lower growth assumptions in
the future?
Are our products suitably resilient to change?
Some are arguing that longevity rates will further
lengthen in the UK as warmer winters
reduce cases of pneumonia in winter among the
elderly. Can we really continue to guarantee
mortality 40 years into the future? Could we
ever? Studies show that catastrophes lead to
economic issues in local areas, which in turn
have an impact on health; do health insurance
contracts allow for this?
General insurance actuaries can ensure that
the capital models allow for an increasing frequency
of extreme natural catastrophe events
and urge their underwriters to ensure the cost
of capital is taken into account in their pricing
calculations. Do liability underwriters anticipate
claims on professional indemnity (PI) policies
from architects and surveyors who did not
allow for climate change in their designs and
costings? What about actuaries’ PI cover?
The key point is that as professionals we must
ensure that consideration of climate change
and indeed other issues in the 21st century, for
example extreme population growth and peak
oil issues, are taken into account in the forward
planning of those we advise. We must encourage
them to move beyond the typical three-year
planning horizon.
Dr Stephan Harrison’s paper helps us see that
climate change may be occurring faster than climate
models suggest, and that rapid change is
possible particularly on a regional level. We are
grateful for his work and commend the paper
to actuaries. We hope its insights into the
volatility of the past climate will be a call for
action among the profession.
The full article by Dr Harrison can be found at
http://climatechange.pbwiki.com/LessonsFromPast
along with some further discussion and comments
by the author. The climate change wiki is
part of the output of the Climate Change Working
Party announced earlier this year. Visit http://
climatechange.pbwiki.com for more debate on
climate change. You can subscribe to the site for
regular updates of new information.

07_09_6.pdf