Memorandum submitted by Actuarial Profession's
Resource and Environment Group
Oliver Bettis and Nick Silver recently presented
a paper at the IARU International Scientific Congress on Climate
Change, titled "Risk of Ruin: A framework for reviewing greenhouse
gas stabilisation targets" on behalf of the Actuarial Profession's
Resource and Environment Group.
The paper is directly relevant to the setting
of Carbon Budgets; a copy of the abstract and summary is attached.
The final version of this paper will be peer reviewed and presented
at an official Faculty of Actuaries meeting in January 2009.
The purpose of the paper is to develop a risk
management framework for setting GHG target concentrations.
To summarize our argument; we first define a
"ruin" event, an event or events that might be caused
by climate change that would be potentially catastrophic for the
planet; adaptation would be virtually impossible. Examples include
the melting of permafrost leading to the release of methane hydrates
or the collapse of the Greenland and Antarctic ice sheets.
A risk management perspective demonstrates that
the aim of GHG reductions was to reduce the probability of a "ruin"
event to below a level that could be regarded as acceptable.
To do this, we would have to estimate the threshold
temperature which gives rise to a ruinous event, the atmospheric
concentration of GHG that gives rise to this temperature increase,
and society's appetite for risk.
The existing scientific models do not allow
us to estimate any of these variables with any degree of confidence.
An order of magnitude calculation means that current GHG concentrations
produce at least an order of magnitude more risk than society
is willing to bear.
The Government's current target, 80% reductions
by 2050, is broadly thought to be equivalent to 50% global reductions
by 2050. This is generally agreed to represent a 50% probability
that average global temperatures will ultimately increase by more
than 2C.
This is likely to mean that there is more than
a 50% probability that one or more of the ruinous events that
we have identified could occur. We consider a more appropriate
probability would be a maximum of 5%, possibly less.
Our conclusion is that we must de-carbonize
as quickly as possible, and simultaneously develop so called "geo-engineering"
solutions.
ABSTRACT
The purpose of this paper is to use actuarial
techniques to build a risk framework for use by policy-makers
in formulating an optimum greenhouse gas stabilization target.
Research suggests that society currently underestimates the underlying
risk of climate change and resources required for mitigation.
The paper examines what is the risk of ruin that society might
be prepared to accept, given current available knowledge of the
risk distribution.
The concept of "ruin" is defined in
the context of climate change. Ruin constitutes severe impacts
which have a catastrophic effect on society, such that adaptation
would be extremely difficult or impossible. The time horizon at
interest and the severity of effects are defined. An example of
a situation of ruin would be a steep fall in world food supply
or large scale irreversible ice-sheet melting.
Actuaries have developed tools and techniques
to model and advise on the effect of extreme events on insurance
companies. In recent times insurers have been required to develop
capital models to value risks and set capital requirements such
that the risk of ruin is below a threshold level which is perceived
to be reasonable for the institution to take. This paper adapts
this approach to assess the impact of climate change on society.
The paper sets out how a calculation of stabilization
targets under a risk management framework would be achieved, but
argues that, due to the level of uncertainty of the variables,
this calculation cannot at present be made. The conclusion is
that, due to the inability of estimating the risk of realistic
"ruin" scenarios, only pre-industrial atmospheric greenhouse
gas concentrations should be considered safe.
The implications for climate change policy are
that research should be concentrated on the tail of the climate
sensitivity distribution and the probability of ruinous events
so that the target concentration might be increased; de-carbonisation
of the economy should be undertaken as rapidly as possible, research
is required into methods for removing greenhouse gases from the
atmosphere, and "climate management systems" need to
be investigated as back-up measures if the risk for deployment
can be shown to be less than the risk of ruinous climate change.
CONCLUSIONS
The paper has identified that there is a high
degree of uncertainty in the sensitivity of climate to greenhouse
gas forcing. There is a high degree of uncertainty about the amount
of shielding the Earth currently receives from aerosols, hence
a high degree of uncertainty about the total radiative forcing
that the Earth is receiving and has caused the current amount
of warming. The paleoclimate records show that ice sheets and
hence the sea level is very sensitive to the temperature. There
is a long time delay between increased levels of greenhouse gas
concentrations in the atmosphere into the atmosphere and the full
warming effect.
There is still a large amount of uncertainty
regarding the climate sensitivity, and it might not be possible
to reduce this uncertainty, at least on the timescale needed to
negotiate and implement emissions targets.
IT IS
NOT POSSIBLE
TO RECOMMEND
AS SAFE
ANY GREENHOUSE
GAS LEVEL
ABOVE THE
PRE-INDUSTRIAL
On a risk management basis the only CO2 stabilization
target that we could be certain would have an acceptable risk
of ruin is the pre-industrial level, of around 280ppm CO2.[1]
It may well be the case that a higher target is in fact safe,
but this cannot be ascertained with any degree of confidence at
this time.
We think it very unlikely that a target above
350ppm would carry an acceptable risk of ruin. Therefore we can
be sure that any acceptable CO2 stabilization target will be substantially
below the current atmospheric level; around 385ppm.
We have not attempted to calculate a time-frame
or an emissions reduction pathway for this target, precisely because
these will be subject to the same degrees of uncertainty as any
other calculations.
To the authors' knowledge, a target concentration
of 280ppm is below any that has been published in the literature.
However, this target results because of the risk management framework
that we have applied. The reason for the low target is:
— There are plausible scientific scenarios
that could result in catastrophe for much of humanity.
— There is a high degree of uncertainty
about the temperature trigger points at which these would occur.
— It is not possible to assign a probability
that a given level of atmospheric concentration of carbon dioxide
will not result in a temperature rise beyond a certain threshold.
This means that whatever target probability
that we might assign, current scientific knowledge does not allow
us to ascertain what atmospheric concentration of carbon dioxide
will result in the risk of "ruin" being below this probability
threshold.
1.1 The current atmospheric concentration
of greenhouse gas has an unacceptably high risk of ruin
Assigning the probability level of what risk
society would be willing to bear would also be problematic and
we have not attempted to do this. The UK Financial Services Authority
set this level at 0.5% for the insolvency of a regulated financial
institution. Although this is an annual figure so is not directly
comparable, it seems unlikely that society would be willing to
tolerate much higher levels than this of a climate related catastrophe.
This would lead us to suspect that the probability of ruin for
most emission scenarios currently envisaged is at least an order
of magnitude higher than that which society would be willing to
tolerate.
1.2 Geoengineering and methods of removing
CO2 from the atmosphere should be investigated as a matter of
urgency
This has a number of profound implications for
climate change policy:
1. Research requirements: the target concentration
arises because the calculations required are subject to uncertainty.
If these uncertainties could be removed, then it may be possible
to increase the target GHG concentrations. To undertake this calculation,
the probability distribution of trigger events at different temperatures,
the tail of the distribution of climate sensitivity to GHG concentrations,
and the mechanisms of positive feedbacks on the climate system
need to be understood.
2. GHG emissions reduction targets: a step change
in reductions will be required—to reach a target GHG level
below the current atmospheric level, the economy will have to
be de-carbonised as rapidly as possible.
3. Carbon sequestration: if it proves necessary
to achieve near to pre-industrial GHG concentrations then emissions
reductions alone will be insufficient. Therefore methods of removing
GHG gasses from the atmosphere will need to be developed.
4. Climate management systems: Rapid reductions
in GHG emissions would cause a reduction in the aerosol shield
in the atmosphere leading to a sudden increase in the net warming
effect. Also there is a large time lag between greenhouse gas
increases and the full warming effect. Hence even if pre-industrial
GHG concentrations can be achieved in the long run, it may be
the case that a ruin trigger cannot be avoided. Therefore geo-engineering
options; methods of artificially reducing the temperature may
need to be deployed, if the risk of deploying these options can
be demonstrated to be less than the risk of catastrophic climate
change.
5. Adaptation: localised climate impacts would
need to be understood with sufficient granularity such that adaptation
measures could be put in place to avoid societal collapse in vulnerable
regions. The possibility of adapting to large-scale catastrophic
events that could be caused by climate change, for example multi-meter
sea level rises, sudden increases in temperatures or change in
precipitation levels will need to be considered.
27 April 2009
1 Recognizing that atmospheric CO2 concentration has
varied naturally within a range during the Holocene period with
no catastrophic effect. Back
| 
