Professional Documents
Culture Documents
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
American Economic Association is collaborating with JSTOR to digitize, preserve and extend
access to The American Economic Review
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
American Economic Review: Papers & Proceedings 2008, 98:2,1-37
http://www.aeaweb.Org/articles.php7doi-l0.l257/aer.98.2.l
By Nicholas Stern*
Greenhouse gas (GHG) emissions are exter? quickly as possible, because policy decisions are
nalities and represent the biggest market failure both urgent and moving quickly?particularly
the world has seen. We all produce emissions, following the recent United Nations Framework
people around the world are already suffering Convention on Climate Change (UNFCCC)
from past emissions, and current emissions will meetings in Bali in December 2007. The rel?
have potentially catastrophic impacts in the evant decisions can be greatly improved if we
future. Thus, these emissions are not ordinary, bring the best economic analyses and judge?
localized externalities. Risk on a global scale ments to the table in real time.
is at the core of the issue. These basic features A brief description of the scientific processes
of the problem must shape the economic analy? linking climate change to GHG emissions will
sis we bring to bear; failure to do this will, and help us to understand how they should shape the
has, produced approaches to policy that are pro? economic analysis. First, people, through their
foundly misleading and indeed dangerous. consumption and production decisions, emit
The purpose of this lecture is to set out what GHGs. Carbon dioxide is especially important,
I think is an appropriate way to examine the accounting for around three-quarters of the
economics of climate change, given the unique human-generated global warming effect; other
scientific and economic challenges posed, and relevant GHGs include methane, nitrous oxide,
to suggest implications for emissions targets, and hydrofluorocarbons (HFCs). Second, these
policy instruments, and global action. The sub? flows accumulate into stocks of GHGs in the
ject is complex and very wide-ranging. It is a atmosphere. It is overall stocks of GHGs that
subject of vital importance but one in which the matter, and not their place of origin. The rate
economics is fairly young. A central challenge at which stock accumulation occurs depends on
is to provide the economic tools necessary as the "carbon cycle," including the earth's absorp?
tive capabilities and other feedback effects.
* London School of Economics and Political Science, Third, the stock of GHGs in the atmosphere
Houghton Street, London WC2A 2AE, UK (e-mail: traps heat and results in global warming: how
n.stern@lse.ac.uk). I am very grateful for the advice and
comments of Claire Abeille, Dennis Anderson, Alex much depends on "climate sensitivity." Fourth,
Bowen, Sebastian Catovsky, Peter Diamond, Simon Dietz, the process of global warming results in climate
Ottmar Edenhofer, Sam Fankhauser, Graham Floater, Su change. Fifth, climate change affects people,
Lin Garbett, Ross Garnaut, Roger Guesnerie, Geoffrey species, and plants in a variety of complex ways,
Heal, Daniel Hawellek, Claude Henry, Cameron Hepburn,
Paul Joskow, Jean-Pierre Landau, James Mirrlees, Ernesto
most notably via water in some shape or form:
Moniz, Steven Pacala, Nicola Patmore, Vicky Pope, Laura storms, floods, droughts, sea-level rise. These
Ralston, Mattia Romani, John Schellnhuber, Matthew changes will potentially transform the physical
Skellern, Robert Socolow, Martin Weitzman, Dimitri and human geography of the planet, affecting
Zenghelis, and all of those who worked on and guided the where and how we live our lives. Each of these
Stern Review team. The views expressed here are mine and
do not necessarily reflect the judgements or positions of
five links involves considerable uncertainty. The
those who kindly provided advice, or of the London School absorption-stock accumulation, climate-sen?
of Economics, or of the UK Government, for whom I was sitivity, and warming-climate change links all
working while leading the Stern Review on the Economics involve time lags.
of Climate Change. This is dedicated to my close friend,
distinctive and distinguished economist and fine man,
The key issues in terms of impacts are not
Andrew Glyn, who died on December 22, 2007, and whose simply or mainly about global warming as
funeral took place in Oxford, UK, on the same day as the such?they concern climate change more
Ely Lecture, January 4, 2008. broadly. Understanding these changes requires
1
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
2 AEA PAPERS AND PROCEEDINGS MAY 2008
specific analysis of how climate will be affected is different from our usual examples in four key
regionally. Levels and variabilities of rain? ways: (a) it is global in its origins and impacts;
fall depend on the functioning of weather and (b) some of the effects are very long term and
climate for the world as a whole. As discussed governed by a flow-stock process; (c) there is
below, temperature increases of 4-5?C on aver? a great deal of uncertainty in most steps of the
age for the world would involve radical and scientific chain; and (d) the effects are potentially
dangerous changes for the whole planet, with very large and many may be irreversible. Thus,
widely differing, often extreme, local impacts. it follows that the economic analysis must place
Further, the challenge, in large measure, is one at its core: (i) the economics of risk and uncer?
of dealing with the consequences of change tainty; (ii) the links between economics and eth?
and not only of comparing long-run equilibria. ics (there are major potential policy trade-offs
Under business as usual (BAU), over the next both within and between generations), as well as
two centuries we are likely to see change at a notions of responsibilities and rights in relation
rate that is fast-forward in historical time and to others and the environment; and (iii) the role
on a scale that the world has not seen for tens of of international economic policy. Further, the
millions of years. potential magnitude of impacts means that, for
This very brief and oversimplified descrip? much of the analysis, we have to compare strate?
tion of the science carries key lessons for eco? gies that can have radically different develop?
nomics. The scientific evidence on the potential ment paths for the world. We cannot, therefore,
risks is now overwhelming, as demonstrated in rely only on the methods of marginal analysis.
the recent Intergoverrnmental Panel on Climate Here, I attempt to sketch briefly an analysis that
Change (IPCC) Fourth Assessment Report, or brings these three parts of economics to center
AR4 (IPCC 2007). Like most of those here today, stage. It is rather surprising, indeed worrying,
I am not a climate scientist. As economists, our that much previous analysis of practical policy
task is to take the science, particularly its analy? has relegated some or all of these three key
sis of risks, and think about its implications for pieces of economics to the sidelines.
policy. Only by taking the extraordinary posi?
tion that the scientific evidence shows that the The Structure of the Argument.?The struc?
risks are definitely negligible should econo? ture of the argument on stabilization is crucial,
mists advocate doing nothing now. The science and we begin by setting that out before going
clearly shows that the probability and frequency into analytical detail. The choice of a stabili?
of floods, storms, droughts, and so on, is likely zation target shapes much of the rest of policy
to continue to grow with cumulative emissions, analysis and discussion, because it carries strong
and that the magnitude of some of these impacts implications for the permissible flow of emis?
could be catastrophic. sions, and thus for emissions reductions targets.
While an understanding of the greenhouse The reduction targets, in turn, shape the pricing
effect dates from the nineteenth century,1 in and technology policies.
the last decade, and particularly in the last few Understanding the risks from different strate?
years, the science has fortunately started to give gies is basic to an understanding of policy. Many
us greater guidance on some of the possible articulated policies for risk reduction work in
probability distributions linking emissions and terms of targets, usually expressed in terms of
stocks to possible warming and climate change, emission flows, stabilization levels, or average
thus allowing us to bring to the table analytical temperature increases. The last of these has
tools on economic policy toward risk. the advantage that it is (apparently) easier for
The brief description of the science above tells the general public to understand. The problem
us that GHG emissions are an externality which is that this apparent ease conceals crucial ele?
ments that matter greatly to social and economic
outcomes?it is the effects on storms, floods,
1 Joseph Fourier recognized in the 1820s (Fourier 1827) droughts, and sea-level rise that are of particular
that the atmosphere was trapping heat; three decades later,
importance, and a heavy focus on temperature
John Tyndall (1861) identified the types of gases responsi?
ble for the trapping; and at the end of the century, Svante can obscure this. Further, and crucially, tem?
Arrhenius (1896) gave calculations of the possible effects perature outcomes are highly stochastic and
of doubling GHGs. cannot be targeted directly. Emissions can be
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 3
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
4 AEA PAPERS AND PROCEEDINGS MAY 2008
in October/November 2006 (see http://www. can be run many times for different possible
sternreview.org.uk, including Postscript) and parameter choices. Such exercises yield Monte
in book form (Stern 2007) a year ago, but goes Carlo estimates of probability distributions of
beyond it in many important ways?in relation outcomes. A discussion of various methods and
to subsequent policy discussions, new evidence models may be found in Malte Meinshausen
and analysis, and discussions in the economics (2006) and in Chapter 1 of the Stern Review.
literature. Figure 1 and Table 1 are drawn from the mod?
There are four further parts to this paper. The els of the UK's Hadley Centre. The work of the
second part focuses on risks and how to reduce Hadley Center was a particular focus of models
them, and on costs of abatement. The third part for the Stern Review for a number of reasons.
examines formal modelling and damage assess? First, it is one of the world's finest climate sci?
ment. The fourth part examines policy, and in ence groups, with a very large computing capac?
particular the role of different policy instru? ity. Second, it was close by and the staff were
ments. The final part outlines what I see as the extremely accessible and helpful. Third, its
central elements of a global deal or framework probability distributions are fairly cautious, bal?
for collaborative policy and discusses how that anced, and "middle of the road" (Meinshausen
deal can be built and sustained. 2006); this judgement is sustained by a com?
parison of their results with the subsequently
I. Stabilization of Stocks of Greenhouse Gases I: published AR4 (IPCC 2007).
Risks and Costs Figure 1 and Table 1 present estimated proba?
bilities for eventual temperature increases (which
A. Risks and Targets take time to be established) relative to preindus
trial times (around 1850), were the world to sta?
The relation between the stock of GHGs in bilize at the given concentration of GHGs in the
the atmosphere and the resulting temperature atmosphere measured in ppm C02e. Figure 1
increase is at the heart of any risk analysis. The portrays 90 percent confidence intervals?
preceding link in the chain, the way the carbon the solid horizontal bars?for temperature
cycle governs the process relating emissions increases. The lower bound (fifth percentile)
to changes in stocks, and the subsequent link, is derived from the IPCC Third Assessment
from global average temperature to regional and Report, or TAR (Tom Wigley and Sarah Raper
local climate change, are full of risk as well. But 2001; IPCC 2001a, b)3 and the upper bound is
the stock-temperature relationship is the clear? from the Hadley Center (Hadley Center 2005;
est way to begin, as it anchors everything else. James M. Murphy et al. 2004). The dotted bars
Broadly conceived, it is about "climate sensitiv? cover the range of the 11 studies examined by
ity"?in terms of modelling, this is indicated Meinshausen (2006). The bar for 550ppm C02e
by the expected eventual temperature increase (with a 90 percent interval of 1.5?C to 5.3?C)
from a doubling of GHG stocks.2 approximately represents the possible range for
There are now a number of general circula? "climate sensitivity."
tion models (GCMs?also known as global cli? Concentrations are currently around 430ppm
mate models) that have been built to describe C02e (Stern Review, Figure 1.1 (Stern 2007,
the links from emissions to climate change. 5)?Kyoto GHGs), and are rising at around
The large ones work with a very large number 2.5ppm C02e per annum. This rate appears to be
of geographic cells, consume computer time accelerating, particularly as a result of the very
extremely heavily, and can be run only on some rapid growth of emissions in China. On fairly
of the world's biggest computers. Nevertheless, conservative estimates (International Energy
particularly if combined with appropriate link?
ing to a large number of other machines, they
3 The TAR was without probabilities but Wigley and
Raper produced distributions based on it. The Stern Review
2 Climate modellers tend to define "doubling" in rela? blended the TAR and Hadley because the former was
tion to preindustrial times. The relationship from stock to based on international discussion, but the latter was more
temperature increase is approximately logarithmic, so that recent. The Stern Review used lower climate sensitivities
doubling from other stock levels would be likely to yield a than Hadley, although the IPCC's more recent AR4 (IPCC
similar increase. 2007) is closer to those used by Hadley.
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 5
95%
5% 400ppnJCO2e
490ppmCO2e
550ppmCO?e
esoppm ico*
?*_ ?
:
*--M
750ppmCC|?e
Source: Stern Review, Table 1.1 (Stern 2007, 16); Meinshausen 2006; Wigley and Raper
2001; Murphy et al. 2004.
Stabilization level
(in ppm C02e) 2?C 3?C 4?C 5?C 6?C 7?C
450 78 18 3 1 0 0
500 96 44 11 3 1 0
550 99 69 24 7 2 1
650 100 94 58 24 9 4
750 100 99 82 47 22 9
Source: Stern Review Box 8.1 (Stern 2007, 220) with some added information.
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
6 AEA PAPERS AND PROCEEDINGS MAY 2008
We do not really know what the world would from "cross-sectional" (Robert Mendelsohn
look like at 5?C above preindustrial times. The et al. 2000, 557) comparisons of regions with
most recent warm period was around 3 million current temperature differences of around
years ago when the world experienced tempera? 5?C?comparisons between, say, Massachusetts
tures 2-3?C higher than today (Eystein Jansen and Florida miss the point. Nor, given the non
et al. 2007, 440). Humans (dating from around linearities involved, can we extrapolate from
100,000 years or so) have not experienced any? lower temperature increases (say 2?C) concern?
thing that high. Around 10,000-12,000 years ing which there is more evidence. Most people
ago, temperatures were around 5?C lower than contemplating 5?C increases and upward would
today, and ice sheets came down to latitudes just surely attach a very substantial weight on keep?
north of London and just south of New York. ing the probability of such outcomes down.
As the ice melted and sea levels rose, England From this perspective, an examination of
separated from the continent, rerouting much of Table 1 suggests that 550ppm C02e is an upper
the river flow. These magnitudes of temperature limit to the stabilization levels that should be
changes transform the planet. contemplated. This level is nevertheless rather
At an increase of 5?C, most of the world's ice dangerous, with a 7 percent probability of being
and snow would disappear, including major ice above 5?C and a 24 percent probability of being
sheets and, probably, the snows and glaciers of above 4?C. The move to 650ppm C02e gives
the Himalayas. This would eventually lead to a leap in probability of being above 4?C to 58
sea-level rises of 10 meters or more, and would percent, and of being above 5?C to 24 percent.
thoroughly disrupt the flows of the major rivers Further, we should remember that the Hadley
from the Himalayas, which serve countries com? Center probabilities are moderately conserva?
prising around half of the world's population. tive?one highly computationally intensive
There would be severe torrents in the rainy sea? Monte Carlo estimate of climate sensitivity
son and dry rivers in the dry season. The world found a 4.2 percent probability of temperatures
would probably lose more than half its species. exceeding 8?C (David Stainforth et al. 2005). A
Storms, floods, and droughts would probably be concentration in the region of 550ppm C02e is
much more intense than they are today. clearly itself a fairly dangerous place to be and
Further tipping points could be passed, which the danger posed by even higher concentrations
together with accentuated positive feedbacks looks unambiguously unacceptable. For this
could lead to "runaway" further temperature reason, I find it remarkable that some econo?
increase. The last time temperature was in the mists continue to argue that stabilization levels
region of 5?C above preindustrial times was in around 650ppm C02e or even higher are pref?
the Eocene period around 35-55 million years erable to 550ppm, or even optimal (William D.
ago. Swampy forests covered much of the world Nordhaus 2007a, 166; Mendelsohn 2007, 95). It
and there were alligators near the North Pole. is important to be clear that the "climate policy
Such changes would fundamentally alter where ramp" (Nordhaus 2007b, 687) advocated by
and how different species, including humans, some economists involves a real possibility of
could live. Human life would probably become devastating climatic changes.
difficult or impossible in many regions that are In thinking about targets for stabilization,
currently heavily populated, thus necessitating we have to think about more than the even?
large population movements, possibly or prob? tual stocks. We must also consider where we
ably on a huge scale. History tells us that large start; costs of stabilization; and possibilities of
movements of population often bring major con? reversal, or backing out, if we subsequently find
flict. And many of the changes would take place ourselves in or approaching very dangerous ter?
over 100-200 years rather than thousands or ritory. The costs of stabilization depend strongly
millions of years. on where we start. Starting at 430ppm C02e, sta?
While there is no way that we can be precise bilizing at 550ppm C02e or below would likely
about the magnitude of the effects associated cost around 1 percent of world GDP with good
with temperature increases of this size, it does policy and timely decision making (see Section
seem reasonable to suppose that they would, IB); for stabilization at 450ppm C02e, it might
in all likelihood, be disastrous. We cannot cost 3 or 4 times as much (possibly more). With
obtain plausible predictions by extrapolating bad policy, costs could be still higher. Note that
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 7
B. Costs
the comparison of costs between 450ppm andof Abatement and Prices of G
550ppm C02e illustrates the cost of delay5?
Towould
waiting for 30 years before strong action this point, our discussion of targ
take us to around 530ppm C02e, from focused on those for the stabilization of
which
point the cost of stabilizing at 550ppm We mustC02enow ask about implications fo
would likely be similar to stabilizing sions paths and how much, with good
at 450ppm
they would cost. We have already anti
C02e starting from now. Under most reasonable
assumptions on growth and discounting, part of the broad answer?around 1 per
a flow
of 1 percent of GDP for 50-100 years world GDP per annum to get below
starting
now would be seen as much less costly C02e?but
than awe must look at the argume
flow for a similar period of 4 percent little
or so more
of detail.
GDP, starting 30 years later. Figure 2 illustrates possible paths for
It can be argued that, at some future lization
point, at 550ppm C02e (thin line), 5
C02e (dotted)
we might be able to turn to geoengineering, for and 450ppm C02e (dot-d
the solidtoline is BAU. There are many pa
example, firing particles into the atmosphere
keep out solar energy, analogous to thestabilization
effect of at a given level?see, for ex
Sternare,
major volcanic eruptions in the past. There Review Figure 8.2 (Stern 2007, 2
however, substantial dangers associated all of with
them are a similar shape to those sh
initiating other effects we do not understand.
a path peaks later it has to fall faster). An
We might well be replacing one severe carbon
risk with cycle weakens, the cuts would hav
another; however, extreme circumstances could
larger to achieve stabilization at a give
require an extreme response. And there see Stern
are dif? Review Figure 8.1 (Stern 200
ficult issues of global governance?would Broadly it be
speaking, however, a path stabili
550ppm C02e
right for just one country, or group of countries, to or below will have to show
do this? It seems much more sensible, sions peaking in the next 20 years. Fo
at accept?
stabilization levels, the peak will have
able cost, to avoid getting into this position.
The above is basically the risk-management
sooner. The magnitudes of the implied
economics of climate change. For antions expendi?
between 2000 and 2050 are around
ture of around 1 percent (between -1 centpercent
for 550ppm C02e, 50 percent for
and 3 percent) of world GDP (see Section C02e, IB),
and 70 percent for 450ppm C02
we could keep concentration levels well below
relative to BAU are indicated in the figur
550ppm C02e and ideally below 500ppm C02e.
Figure 3 shows that, to achieve these
While leaving the world vulnerable, this emissions,
would it will be necessary to take
avoid the reckless risks implied by the acrosshigher
the board and not in just two or th
stabilization concentrations (e.g., 650ppm
torsC02e)
such as power and transport. For th
advocated by some economists. Thinking as aabout
whole, energy emissions represent
the information basis for this argument two-thirdsalso of the total, and nonenergy a
one-third.
points to caution. If (as is unlikely) the risks of Land use change, mainly def
high concentrations turn out to be lowtion and and
we degradation of forests, accou
have taken action, we would still have nearly
purchased 20 percent of the total. Given t
a cleaner, more biodiverse, and more attractive
world economy is likely to be perhaps
world, at modest cost. If our actionstimes are weak
bigger in mid-century than it is now
and the central scientific estimates arelutecorrect,
cuts of around 50 percent would
we will be in very dangerous circumstances
cuts of 80-85 percent in emissions per
output.
from which it may be impossible, or very Further, since emissions from so
costly,
to recover. tors (in particular agriculture) will be diff
cut back to anything like this extent, an
richer countries should make much bigge
portional reductions than poor countr
Section IV), richer countries will need
5 There would be some negatives (moreclose-to-zero
inflexible emissions in power (electric
transport
equipment in place) and some positives (more technical by 2050. Close-to-zero emiss
knowledge). power are indeed possible and this would
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
8 AEA PAPERS AND PROCEEDINGS MAY 2008
100
10 ^ - ? -,^.^^70gg?Mr;-;?
-v
2000 2010 2020 2030 2040 2050 2060 2070 2080 2090 2100
ENERGY
EMISSIONS Industry (14%)
Other energy
related (5%)
Waste (3%)
Agriculture
Transport (14%)
(14%)
NONENERGY
EMISSIONS
(8%) Land use
(18%)
Total emissions in 2000:42 GtCO,e
close-to-zero emissions for much of transport. way of life to the extent of that brought by elec?
This would, however, require radical changes tricity, rail, automobiles, or the Internet.
to the source and use of energy, including much On the path for stabilization there would
greater energy efficiency. Achieving the neces? be different options for cutting emissions that
sary reductions would also require an end to would be more prominent at different times. In
deforestation. The totality of such reductions the earlier periods, there would be greater scope
would, however, not result in a radical change in for energy efficiency and halting deforestation,
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 9
social
and with technical progress there will be, andcost of carbon (SCC) reflecting
future damage of an incremental emission. T
already are, strong roles for different technolo?
gies in power and transport. levels quoted here for the MAC are consis
Various different options for abatement with ranges for the SCC indicated in the S
were
Review along an abatement path for 550p
discussed in Chapter 9 of the Stern Review.6
McKinsey has recently carried out C02e stabilization.
a more
However, the SCC is very slippery num
detailed study (Per-Anders Enkvist, Tomas
Naucl?r, and Jerker Rosander 2007)?see cally since it is so sensitive to assumptio
Figure
about
4. There are several important lessons from thismodel structure, including future em
sion
type of curve. First, there are many options paths, carbon cycles, climate sensitiv
for
future
reducing emissions that have negative cost; that technologies, and ethical approache
is, they save money. Second, there is avaluation
whole over the centuries to come. The
at time
range of options and each should be explored in t is the expectation of the integral9 o
detail?for example, the costs associated from
witht onward of:
combating deforestation in the McKinsey curve
are, in my view, far too high.7 Third, the?emis?
the marginal social utility of consumption
r (embodying ethical values and a partic
sions savings from any one option will depend
path)
on what it replaces. Fourth, given the broad
range of options, policy is very important?bad
X the impact on consumption at r of all rel?
policy will lead to the uptake of more expensive
evant preceding temperature changes (and
options. Fifth, technical progress will be impor?
tant and should be promoted so that the range resultant climate change)
of options is widened and costs are reduced.
Finally, and of special importance, startingXnow
the impact on a relevant temperature
increase of increases in preceding carbon
in a strong way and with clear signals will allow
more time for planned choices, discovery stocks
of
options, and exploration of the renewal periods
X the impact on all relevant stocks of an
and timings for equipment. This is the meas?
increase in carbon emissions at t, where
ured, lower-cost approach. Going more slowly
and then moving in haste when and if the sci? "impact" in the above is to be interpreted
ence is confirmed still more strongly, is likelyas to
a partial derivative.
be the expensive option.
Very importantly for policy, this type ofGiven
fig?this sensitivity, it is remarkable how
carelessly the SCC is often quoted?it is quite
ure gives us an understanding of where carbon
prices (or GHG prices more generally)common,
shouldfor example, for people to quote an
SCC
be. By 2030, cuts at the world level would without even referring to a reference emis?
have
to be of the order of 20 Gt C02e (see Figure
sions 2)
path, to say nothing of all the other rel?
evant assumptions that matter greatly.
for stabilization at 550ppm C02e. This suggests
a C02 price of around ?30 per ton.8 Thus, the SCC is a very weak foundation for
A fairly clear idea of where the carbonpolicy.
priceThe target approach and the calculation
of the associated MAC is more attractive from
should be from the point of view of necessary
the point of view both of policy toward risk and
abatement is of great help both to policymakers
of clarity of conclusions. It is also important,
and to investors. It also provides the opportu?
however, to check prices derived from the MAC
nity to check against estimates of the marginal
against SCC calculations and to keep policy
under revision, as further information and dis?
6 Illustrative MAC curves were provided in the Stern covery arrives. Some notion of the SCC is also
Review, Figures 9.1 (Stern 2007, 243) and 9.2 (Stern 2007, useful in examining the emissions savings from,
249).
7 Erin C. Myers (2007, 9-12) reviews the literature and
highlights the outlier status of the McKinsey deforestation
estimate; see also the discussion in Section IV. 9 This sketch of the calculation assumes the simple
8 This is not the place to speculate about euro-dollar objective of the maximization of the integral of expected
exchange rates over two or three decades. utility.
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
10 AEA PAPERS AND PROCEEDINGS MA Y 2008
203
Coal-to- Avoid
Cost of abatement COS; ?a?*?
EUR/tOQ*
* -27QtonCQ,ebelow40EUfVton(-46%vs.
BAU)
* -7 Gton of negative and zero cost opportunities
* Fragmentation of opportunities
for example, transport programs or policies. If The Stern Review (Chapter 10) also looked
the MAC and SCC were thought to be in very at top-down macro modelling of costs of
different ball parks for an anticipated path, it emissions reductions (see also Terry Barker,
would suggest strongly that policy revision is Mahvash Saeed Qureshi, and Jonathan K?hler
necessary. 2006). Both the bottom-up (Chapter 9) and
Prices for abatement should be at a level that the top-down (Chapter 10) studies produced
supports carbon capture and storage (CCS) for numbers in similar ranges?around 1 percent
coal. Without CCS for coal it will be difficult of world GDP. There is, of course, consider?
(and more costly) to achieve the necessary cuts, able uncertainty. Bad policy or delayed deci?
given that many countries will rely heavily on sions could give higher numbers. Stronger
coal for power generation for the next 30 or technical progress could give lower numbers.
40 years (IEA 2006; 2007). China and India Assumptions about substitutability between dif?
(Expert Committee on Integrated Energy Policy ferent goods and options matter, too. Since the
(ECIEP) 2006), for example, will be using coal Stern Review was published, there have been
for around 80 percent of their electricity plants a number of new studies, both bottom-up and
for the next 30 years or so?for the simple rea? top-down. Significant examples of the former
sons that coal is cheap and available within their are those from McKinsey (Enkvist, Naucl?r,
own borders; that they are familiar with the and Rosander 2007) and the IEA (2007), both
technologies; and that the plants can be erected of which indicated costs either in the region we
quickly. Speed is of great importance for them, suggested, or somewhat lower. Similar conclu?
as the costs of electricity outages are very high. sions are drawn in the AR4 (IPCC 2007).
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T ELY LECTURE 11
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
12 AEA PAPERS AND PROCEEDINGS MAY 2008
methods of CBA are derived. This is not at all Let us start with the definition of a discount
to use a different theory. On the contrary, it is to rate in policy evaluations. It is simply the propor?
maintain the theory and to avoid a gross misap? tionate rate of fall of the value of the numeraire
plication of the special (i.e., marginal) case. used in the policy evaluation. In the simple case,
The centrality of nonmarginal changes and with aggregate consumption as the numeraire,
of risk means either using the risk-analysis we have a social discount factor or SDF, ?(t),
approach of Section IA, or using aggregate mod? which measures the social value of a unit of con?
elling with a social welfare function to compare sumption at time t relative to a unit at time zero.
consequences. Both have their role, but for the The social discount rate, or SDR, is then -?/?.
reasons given I would see the former as the main A number of general conclusions follow
plank of the argument. The latter has a valuable immediately from these basic definitions. First,
supplementary role which we now investigate. the SDF and the SDR depend on a given refer?
In setting out a social welfare function to ence path for future growth in consumption and
evaluate damages and costs, the valuation of will be different for different paths. Second, the
consequences on different dimensions?social, discount rate will vary over time. Third, with
health, conflict, and so on?will be extremely uncertainty, there will be a different discount
difficult. I do not go into these issues. I focus rate for each possible sequence of outcomes.
on one issue that has, understandably, received Fourth, there will be a different discount rate
considerable attention in discussion of the Stern for different choices of numeraire. In imperfect
Review?how to value benefits accruing to dif? economies, the social value of a unit of private
ferent people at different times. There are una? consumption may be different from the social
voidable ethical issues. They are the subject of value of a unit of private investment, which may
Section IIB. In Section IIC we examine some of be different from the social value of a unit of
the challenges, results, and sensitivities of for? public investment. And the rates of changes of
mal modelling, and comment on new evidence these values may be different too.
and discussions concerning the Stern Review's A further key element for understanding dis?
damage estimates after one year. count rates is the notion of optimality of invest?
ments and decisions. For each capital good, if
B. Ethics resources can be allocated without constraint
between consuming the good in question and its
Discounting.?Much of the discussion of use in accumulation, we have, for that good, the
ethics in relation to the Stern Review has been result that the social rate of return on investment
focused on discounting. Sometimes, simplistic (the marginal productivity of this type of good
approaches to discounting conceal or obscure at shadow prices), the SRI, should be equal to
the underlying structural and ethical logic by the SDR in terms of that good (i.e., with that
shoehorning the issues into a simple discount good as numeraire). This is intuitively clear and
rate specified entirely externally to the problem. in optimal growth theory is a standard first
However, careful use of theory and concepts is order condition. But where there are constraints
crucial. Some have argued that "the discount on this optimization, as there usually will be
rate of the Stern model" is too low in relation to in imperfect economies, this condition that the
market rates of return. This argument has gener? SRI equal the SDR is not generally applicable.
ally been thoroughly confused for a whole set of Dr?ze and Stern (1987, 1990), for example, show
reasons. It arises from inappropriate application how opportunity costs, and thus shadow prices
of a marginal method to a strongly nonmarginal and shadow rates of return, depend on which
context, failure to apply modern public econom? alternative use a unit of resource comes from.
ics, ignorance of the multi-good nature of this Further, in such economies it will not generally
problem, and, in some cases, ignorance of the be true that the private rate of return on invest?
difference between a social discount rate and ment (PRI) will be equal to the SRI. And simi?
a pure time discount rate. Given this pervasive larly, private discount rates (PDRs) can diverge
confusion about the basic theory of discounting, from SDRs. Such divergences can arise from all
it seems worthwhile to clarify briefly the logic forms of market imperfections, including exter?
of discounting as applied to climate change and nalities. In this case we have the additional com?
relate it to some simple empirical data. plication that key players, future generations,
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 13
havegeneral
are not directly represented. Thus, in the a reading on the concept at issue here, the
case:
social discount rates for the key goods. Thus,
observations on the PRIs and PDRs have only
PDR^SDR^SRI^PRI. limited usefulness. And note that the problems
that prevent the equalities in this chain, such
Before looking into discount rates along aas missing markets, unrepresented consumers,
given path, we should remind ourselves that the
imperfect information, uncertainty, production,
most basic mistake here is to use a marginal and consumption externalities are all absolutely
concept (discount rates) around a current pathcentral for policy toward the problem of climate
for strategic choices and comparisons among change. We come back again to a basic con?
paths. Policy on climate change means choosingclusion: the notions of ethics, with the choice
among paths with very different growth patterns of paths, together determine endogenously the
for a whole collection of capital goods, includ? discount rates. There is no market-determined
ing those relating to natural endowments. Thus,rate that we can read off to sidestep an ethical
discussion.
it is simply wrong to look at rates as currently
observed, or in historical terms, which refer to It must surely, then, be clear that it is a seri?
existing paths. A choice among paths means ous mistake to argue that the SDR should be
also choosing the implied set of discount rates
anchored by importing one of the many private
associated with the paths (Stern 2007, 27-31;rates of return on the markets (or a rate from
for more on this issue see Cameron Hepburn government manuals, or a rate from outside
2006). This is simply another way of expressing empirical studies). Yet it is a mistake that many
the old idea that the shadow prices or marginalin the literature have made. Nordhaus (2007b,
values depend on where you are. It is absolutely690) and Martin L. Weitzman (2007b), for
fundamental here for this very nonmarginal setexample, substitute a market investment return
of choices to recognize that the social discountof 6 percent for the SDR, thus producing a rela?
rates are endogenous, not exogenous. They are tively high 6 percent rate of discount on future
determined by ethical values, which have to beconsumption. This mistakenly equates the
PRI to the SRI and the SRI to the SDR. Such
discussed explicitly, and by the paths that result
from climate change and investment choices. an approach is entirely inappropriate given
Let us suppose, however, that we go past thisthe type of nonmarginal choices at issue and
problem and look at discount rates around the a risk structure of the problem, and in light
given path, or path of choice. What can we learn
of developments in modern public economics,
from observed rates in markets? Rates at whichwhich encompasses social cost-benefit analysis
households can borrow and lend, usually for and which takes account of many imperfections
periods of no longer than three or four decades,in the economy, including unrepresented con?
give a reading on their private discount rates or
sumers, imperfect information, the absence of
PDRs (assuming they equate their discount ratefirst-best taxes, and so on.
with their market rate, with some appropriate If, despite these difficulties, we nevertheless
treatment of uncertainty). But as this borrowing
insist on looking to markets for a benchmark rate
of discount, then what do we find? In the United
and lending takes place through private deci?
sions made by individuals acting in a market,
Kingdom and United States, we find (relatively)
this does not necessarily answer the relevant
"riskless," indexed lending rates on government
question in the context of climate change deci?bonds centered around 1.5 percent over very long
sions by a society?namely, how do we, acting periods. For private very long-run rates of return
together, evaluate our responsibilities to future
on equities, we find rates centered around 6 or 7
generations over very long periods? percent (Rajnish Mehra and Edward C. Prescott
2003, 892; Kenneth J. Arrow et al. 2004, 156;
Rates of return on investment generally reflect
private rates of return narrowly measured. They Sree Kochugovindan and Roland Nilsson 2007a,
take no account of externalities, which are of
64; 2007b, 71). Given that it is social discount
the essence for this discussion. Thus, even if
rates that are at issue, and also that actions to
we think we can observe some private rates ofreduce carbon are likely to be financed via the
discount for some households, and some pri? diversion of resources from consumption (via
vate rates of return for some firms, we do not
pricing) rather than from investment, it is the
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
14 AEA PAPERS AND PROCEEDINGS MAY2008
long-run riskless rates associated with consumer (2004), and also the Stern Review (Stern 2007,
decisions that have more relevance than those 60). Environmental services are also likely to
for the investment-related equities. Thus, even be income elastic, which will further reduce the
if one were to endorse the approach of import? implied SDR.
ing a discount rate from markets, when one uses Finally, we underline an unhappily common
the rate of return closer (but not equivalent) to mistake?namely confusing the pure time dis?
the relevant concept?the risk-free rate?it is count rate (PTDR) with the SDR. With a very
far from clear that one would obtain a rate of simple single good structure and consumption
discount on future consumption as high as the 6 at time t having social value u(c)e~8\ we have
percent advocated by Nordhaus (2007b, 690). the SDF, ?, as u'(c)e~8t.13 Its proportionate rate
Weitzman (2007c) has recently produced an of fall (the SDR) is T](c/c) + 5, where 77 is the
interesting insight into the difference between elasticity of the social marginal utility of con?
the riskless rate and equity returns in terms of sumption with respect to consumption.14 Often
perceived high weights in the downside tail of 77 is taken to be a constant. In this very simple
equity returns?the implication being that the case, we can now see the difference between the
perceived equivalent return on equities, allow? SDR and the PTDR. The PTDR is the rate of
ing for risk, is close to the lower riskless rates. fall of the value of a unit of consumption, simply
In this context Weitzman (2007a, b), has also because it is in the future, quite separately from
suggested encapsulating risk and uncertainty in the levels of consumption enjoyed at the time.
some contexts into discount rates. In my view, Here, the PTDR is S. For example with 8 = 0,
however, it is far more transparent to treat risk r] = 1.5, and tic ? 2.5 percent, we have a social
directly through the approach to social welfare discount rate of 3.75 percent, in excess of the UK
under uncertainty than to squash it into a single government's test discount rate (Her Majesty's
parameter that tries to reduce the problem to one Treasury 2003), notwithstanding a PTDR of
of certainty. zero. It is 77 and the growth rate that capture the
Suppose, however, that we persisted with idea that we should discount the consumption
the argument that it is better to invest at 6-7 of future generations on the basis that they are
percent and then spend money on overcom? likely to be richer than ourselves. This reason
ing the problems of climate change later rather for discounting is, and should be, part of most
than spending money now on these problems. models, including those of the Stern Review. We
The multi-good nature of the problem, together shall show in the next subsection that the cost,
with the irreversibilities from GHG accumula? in terms of climate changes, of weak or delayed
tion and climate change, tell us that we would action in the formal models is much greater than
be making an additional mistake. The price of that of timely and stronger action, in terms of
environmental goods will likely have gone up abatement expenditure, over a range of param?
very sharply, so that our returns from the stand? eter values for 77.
ard types of investment will buy us much less in A S of 2 percent (3 percent)?as endorsed by
reducing environmental damage than resources many commentators such as Nordhaus (2007b)
allocated now (see also Section I on the costs
of delay).12 This reflects the result that if envi?
13 The SDF is the marginal utility of consumption
ronmental services are declining as stocks of
time t (and we normalize the SDF to one for / = 0). I
the environment are depleted, then the SDR consider a changing population N(t), and replace u{c
with that good as numeraire will be negative. Nu(c) where c is C/Nand C is total consumption at tim
On this, see the interesting work by Michael the partial derivative with respect to C is u'(c).
Hoel and Thomas Sterner (2007), Sterner and 14 Unfortunately, some, including Nordhaus (200
and Weitzman (2007b), have been tempted to think th
U. Martin Persson (2007) and Roger Guesnerie value for the PTDR can be "backed out" from this exp
sion by equating the SDR with some market rate of retu
For example, with a market investment return of 6 perc
12 The issue is still more complex in this context, as consumption growth of 2 percent, and 77 = 2, one "in
delays in action result in environmental damage along the that 5 = 2 percent. Thus, the fallacy that the SDR can
way, as well as increasing the cost of achieving a given sta? anchored by some market rate of return leads to a se
bilization level. On balance, the extra intertemporal com? fallacy, namely that society's PTDR can be "revealed"
plexity is likely to strengthen this paper's argument in this market behavior (instead of requiring explicit specific
case. on ethical grounds).
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 15
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
16 AEA PAPERS AND PROCEEDINGS MAY 2008
not settle, discussions about value judgements? It is striking that there are some, such as
for further exploration see Dietz, Hepburn, and Nordhaus (2007b) and Weitzman (2007b), who
Stern (2008). In using such evidence, we must appear to argue for high 77 (equal to 2 or 3) in
constantly bear in mind two key issues. First, we intertemporal analysis yet do not bring out how
must ask about the relevance of individual deci? this is potentially inconsistent with standard
sions for the societal decisions about the prob? cost-benefit analysis treatments of intragen
lem at hand?here social decisions by the world erational distribution (which effectively assume
community now, bearing in mind consequences 77 = 0) or with some intratemporal tax and trans?
for future generations. And, second, if we infer fer policies.
values from decisions, we must ask whether we
have modelled well the decision processes, the Intertemporal Distribution: In discussions
objectives, and the perceived structure of the of 77 in an intertemporal framework, there has
problem as seen by the decision maker. been much focus on implied saving rates. Some
Anthony B. Atkinson and Andrea Brandolini (Partha Dasgupta 2007, 6; Nordhaus 2007b,
(2007) have produced an interesting set of 694-96), following arguments in Kenneth J.
examples on empirical income distributions and Arrow (1995, 12-17), have criticized the rela?
actual transfer schemes in relation to welfare tively high weight placed by the Stern Review on
weights.16 They conclude that constancy of 77 the consumption of future generations (whether
across a range of increases is difficult to "square via 77 or 8) by arguing that the Review's param?
with" the way that many transfer schemes occur eter choices can, in certain scenarios, imply
in practice; in addition, there are many exam? implausibly high optimal savings rates. As is
ples where policies appear inconsistent with 77 clearly explained in the Review (Stern 2007,
greater than one. For example, given the cur? 54), with 6 = 0, output proportional to capital,
rent income distribution in the United States, and no technical progress, the optimal savings
an 77 of two would imply that a redistribution rate is I/77. With 77 close to one, this would lead
from the fifth-richest decile to the second-poor? to very high optimal savings rates. At the same
est decile would be welfare-improving even if time, the Review also states clearly (Stern 2007,
only 7 percent of the transfer reached the recipi? 54) that this result is highly dependent on model
ent; for a transfer from the richest decile to the assumptions.
second-poorest, virtually any redistribution Brad DeLong, in a short blog entry (DeLong
would be welfare-improving regardless of loss 2006), points out this flaw in the Dasgupta
along the way, so long as the recipient received Nordhaus position and argues that techni?
some benefit (Atkinson and Brandolini 2007, cal progress would greatly reduce the optimal
14). Of course, interpretation of actual intratem? savings rate. Mirrlees and Stern (1972) presented
poral tax and transfer schemes will depend on a more fully developed argument. Using a
many assumptions about the structure of incen? standard one-good, infinite-horizon Ramsey
tives17 and policymaking procedures. Perhaps growth model, constant returns to scale, and a
people think that tax-transfer disincentives are Cobb-Douglas production function, they show
very strong and they oppose transfers for these that under one specification?with constant
reasons. Or notions of rights and duties may population, a competitive share of capital equal
influence them. The upshot is that empirical to 0.375, and 3 percent exogenous technologi?
estimates of implied welfare weights can give a cal progress?the optimal consumption path
wide range of 77, including 77 below one and even for 77 = 2 and 8 = 0 involves a savings rate, s,
as little as zero. between 0.19 and 0.29 (or 0.23 if constrained to
a constant s). This is far below the 0.5 that would
be optimal with 77 = 2 and 8 ? 0 in the simpler
16 The welfare weight on an individual with consump?
tion c is taken here as the social marginal utility of con? case of output proportional to capital and no
sumption at that level. To keep things simple, we assume technical progress.
that this depends only on the individual's consumption and Just as with intragenerational values, the
not on her preceding consumption or the consumption of
others. approach of the "inverse optimum" or implied
17 See, for example, Stern (1976), who shows how sensi?
social values does not take us very far in this
tive tax calculations are to assumptions about substitutabil context. We cannot really interpret actual sav?
ity between goods and leisure. ing decisions as revealing the collective view
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 17
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
18 AEA PAPERS AND PROCEEDINGS MAY 2008
this respect, they are still less credible. Those temperature, via climate change. Of the ethical
of us schooled in the optimal tax and optimal elements, the following are crucial: the type of
growth analysis of the 1960s, 1970s, and 1980s ethical values considered (including the role of
learned just how sensitive model results can rights and obligations); the type of outcomes
be to simple structural assumptions, such as introduced into evaluation functions (including
the form of preferences, production, or techni? separate goods or services such as environment,
cal progress, even before parameter values are health, and standard elements of consumption);
introduced (Atkinson and Joseph Stiglitz 1976, the functional forms used to capture evaluations;
1980; Angus Deaton and Stern 1986). and the parameters within those functional
The models portrayed here should be seen forms, including those covering intra- and
as helpful supplements exploring some serious intergenerational values. The ethical discussion
logical and modelling issues related to the esti? should not be shoehorned into a narrow focus on
mation of damages from BAU and their com? just one or two parameters such as 77 and 8; the
parison with alternative paths. We shall see, not ethical issues and their interactions with a model
surprisingly, that the key assumptions influenc? structure designed to reflect a range of uncer?
ing damage estimates concern risk and ethics. It tainties are much broader and deeper.
is surprising, however, that these two issues did
not occupy until recently the absolutely central Stern Review Damages and Sensitivity.?The
position that the logic of the analysis demands. Stern Review base case had damages from BAU
The result is that?given the recent evidence on relative to no climate change of around 10 per?
emissions, carbon cycles, and climate change cent of consumption per annum measured in
sensitivity?most of the studies prior to a year terms of the Balanced Growth Equivalent, or
or two ago grossly underestimated damages BGE (see Mirrlees and Stern 1972). Here, the
from BAU. BGE for any given path is calculated from the
The PAGE20 model was chosen for the work of expected social utility integral of that path by
the Stern Review first, because, in contrast with asking "what initial consumption level, growing
a large majority of preceding work, it places risk at a given growth rate and without uncertainty,
and uncertainty at center stage. It provides for would give this expected social utility inte?
a Monte Carlo analysis of explicit distributions gral?" The difference between the BGEs with
of a large number of parameter values. Second, and without climate change can be thought of as
Chris Hope, its originator, chose the parameters the premium, in terms of a percentage of annual
and their distributions to straddle a range of cli? consumption, that society might be willing to
mate models, IAMs, and economic models in pay to do away with the risk and uncertain?
the literature. Third, Chris Hope kindly made ties associated with dangerous climate change.
the model available and was very generous with Essentially, the BAU provides a calibration in
his advice. The model was described extensively terms of consumption (useful since "expected
in Chapter 6 of the Stern Review as well as by integrated utils" are hard to interpret) for the
Hope (2006a, b) and Dietz et al. (2007a, b). expected utility integral: it summarizes an aver?
Key assumptions on the form of the models age over time, space, and possible outcomes.
and of the parameters in these models may be Table 2 presents some of the results of the
grouped into two broad headings: the structural PAGE model. The parameter 77 was discussed
elements that shape the estimated consequences in Section IIB and is the elasticity of the social
of different kinds of emissions strategies, and marginal utility of consumption where the inte?
the ethical elements that shape the evaluations grand for expected social utility is the sum over
of different outcomes. Of the structural elements / of Niu{Ci/N^)e~0\ and where Q and N? are
in this approach, four are crucial: the emission consumption and population in region /. In the
flows; the functioning of the carbon cycle link? model, y is the exponent of a power function
ing flows to stocks; the climate sensitivity link? linking temperature T to damage through the
ing stocks to temperature; and the damages from function AT7 (Stern 2007, 660?the damages
vary by region). Table 2 provides BGE differ?
ences (in percent) across paths without and with
20 PAGE 2002, Policy Analysis of the Greenhouse Effect climate change, with a 5-95 percent confidence
2002 Integrated Assessment Model, see Hope (2006a, b). interval in brackets. We think of increases in y
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 19
as capturing increases in the structural risks,21 costs of the damages from climate change that
and of increases in 77 as capturing increases in would thereby be avoided. While the measure?
aversion to inequality and risk. ment of estimated damages may vary, this key
Intuitively, we can think of y as combining lesson is robust to parameter changes.
both the relation between temperature and dam? In this type of modeling, results are highly
ages, and the distribution of temperatures aris? sensitive to assumptions on both structural risks
ing from a certain emission path. These are, of and ethics, suggesting that great care should be
course, distinct effects, but both an increase in exercised in choosing the key parameters. We
y and a broader distribution for the temperature can illustrate the importance of these two issues
(in particular more weight in the upper tail, in terms of both computations in the model and of
either from a weakening in the carbon cycle or general results. Replacing all random variables
from higher climate sensitivity) has the effect in the PAGE model by their modes brings down
of producing a higher probability of large dam? the central case of damages from BAU from 10
ages. The effects are treated separately in the 11 percent to 3-4 percent.23 Thus, it is wrong
Review (Chapter 6 and the Technical Annex to to argue, as Dasgupta (2007) and Nordhaus
Postscript), where many more sensitivity results (2007b) have, that the Chapter 6 results of the
are given. These two processes (damages and Stern Review arise solely from assumptions
temperature distributions) can and should be related to ethics, in particular the use of 17 =
modelled separately, but here we keep the dis? 1 and, at least in the view of Nordhaus, a low
cussion and presentation as simple as possible. 8. Both risks and ethics are crucial to any seri?
While we shall discuss results in terms of ous assessment of policy toward climate change
the sensitivity of estimated damages with and and, in particular, assessment of damages from
without climate change, we must emphasize that BAU.
stabilization at 550ppm C02e removes around A formal result is provided in Box 1, which
90 percent22 of the damages (Stern 2007, 333), shows that for any given set of structural risks
so that we are essentially comparing two strate? and a utility function, pure time discounting (a
gies, namely BAU and stability below 550ppm key element in the ethics) can be set so that the
C02e. A key broad lesson from this type of estimated damages are as small as we please.
modelling is that the costs of stabilizing below Further, for any given pure time discounting,
550ppm C02e are generally far lower than the risks and utility can be set such that damages
are as big as we please.
Recently, in a series of papers (Weitzman
21 To keep things simple, the results in the table have y
fixed?that is, nonstochastic. The Monte Carlo probabili? 2007a, b), Marty Weitzman has argued that
ties are therefore generated by the variations in the many when we consider how the various different
other parameters. In the Postscript to the Review (Stern
2007, 658-71), stochastic y is presented. The base case of
y fixed and equal to 2 in Table 2 corresponds closely to the
base case for stochastic y in Chapter 6 of the Review and 23 See Dietz et al. (2007a, c). The drop from replacing
the Technical Annex to Postscript. all random variables by their means is smaller but still
22 Measured in terms of BGE. substantial.
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
20 AEA PAPERS AND PROCEEDINGS MAY 2008
? Write expected utility integral as ?Qg(t)f(t) dt, where g(t) = E[u(c)}, and u is
the welfare difference without and with climate change; f(t) is the pure time dis?
count factor. g(t) will depend on model structure, policies/path, and shape of u (c).
It is possible that g(t) is infinite for some finite T (see Weitzman 2007a).
? For any given g(t), we can construct/(i) so that J*g(t)f(t) dt < e for any e >
0, i.e., there are arbitrarily small losses from climate change. An example is/(i) =
{\/g(t))e~?tmth?> Vs.
? For any given/(i), we can construct g (t) so that there are infinite losses from cli?
Boxl
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 21
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
22 AEA PAPERS AND PROCEEDINGS MAY 2008
FUND 2.8
DICE/RI?E -
PAGE2002(100%)
Meinstaisen(90%)
Sources: Toi 2002; Nordhaus and Boyer 2000; Hope 2006a; 2006b; IPCC 2007; Meinshausen
2006.
is not so much why the Stern Review's model? (b) The carbon cycle is likely to weaken as
ling obtained high damages under BAU, as why a result of, for example, the possible col?
the earlier literature made assumptions that give lapse of the Amazon forest at temperature
such low results. increases of above 3-4?C, or the decreas?
ing absorptive capacity of the oceans.
D. Damages and Sensitivity, Further, a thawing of the permafrost is
One Year On from the Review likely to result in strong methane release.
Looking back, I think the Review was (c) The climate sensitivity assumed in the
too cautious on all four of the key structural Review is likely to be conservative (as
elements: (a) emissions growth, (b) carbon cycle, argued in Section I).
(c) climate sensitivity, and (d) damages from a
given temperature. (d) The damages from given temperature
increases assumed in the Stern Review
(a) The work of Ross Garnaut and his seem very low. The Review's mean dam?
Commission, working for the new age loss (based on estimates in the eco?
Australian government on climate change, nomic literature) from 5?C was around
is revisiting the emissions scenarios in 5 percent of GDP (Stern 2007, 180). As
the IPCC Special Report on Emissions argued in Section I, a temperature increase
Scenarios or SRES (IPCC 2000). In its of 5?C would most likely result in massive
Chapter 6 model (Stern 2007, 173-88), the movements of population and large-scale
Stern Review used the second highest of conflict.
the four scenarios (called A2). Garnaut is
now suggesting that the highest of the four, Considering these structural factors together,
A1F1, is likely to be the best description of the modelling of the Stern Review probably
BAU (Garnaut 2007). Key among the rea? underestimated significantly the risks of high
sons is the growth rates of the developing damages from BAU, perhaps by 50 percent
world, particularly China and India, and or more if one compares the first two rows of
their continued strong emphasis on coal Table 2. Much of the earlier literature grossly
(ECIEP 2006). underestimated the risks.
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 23
NonJhaus, output
Toi, output
ToLequliy
0 1 2 3 4 5
Increase in global mea
Figure 6. The Modelled D
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
24 AEA PAPERS AND PROCEEDINGS MAY 2008
international focus, to promote collaboration, less so than others and allow the balance of taxes
take account of equity, and reduce global costs. to adjust away from other taxes such as income
Careful analytical investigation by econo? or VAT; alternative uses of revenue are possible
mists of policies on climate change involves the too, including those related to climate change.
whole range of the tools of our trade, including We should beware, though, of arguments about
the economics of risk and uncertainty, innova? double dividends: environmental taxes have
tion and technology, development and growth, dead-weight losses in addition to their benefi?
international trade and investment, financial cial effects in addressing externalities. Taxes on
markets, legal issues, ethics and welfare, as well GHGs would require measurement of GHGs,
as public and environmental economics. It will just as in trading, but taxes on petroleum prod?
no doubt require the development of further ucts, coal, or other fossil fuels can act as fairly
analytical methods. And it necessitates close good approximations, avoiding direct emission
collaboration with scientists and other social measurement, which can be relatively costly to
scientists. small enterprises.
Our focus here in this very brief discussion As discussed in Section IA, where the world
of policy will be on price-oriented mechanisms is perfect other than in relation to the tax in
and on technology, but we should also note a question, quantity controls and price measure?
sixth key element that is often overlooked in dis? ments can have dual and essentially identi?
cussions of economic policy, namely how pref? cal effects. Where there is risk, uncertainty,
erences change as a result of public discussion. and imperfections in this market and in other
This was an integral part of John Stuart Mill's parts of the economy, there will be price uncer?
(Mill 1972 [1861], 262) perception of democracy tainty, quantity uncertainty, or both, depend?
and policy formation (see also the discussion in ing on the policies chosen and the nature of the
Chapter 9 of Stern, Jean-Jacques Dethier, and F. uncertainty. Both price certainty and quantity
Halsey Rogers 2005). In this context, it involves certainty are important: firms would like clear
a change in public understanding of responsi? and simple price signals for decision making;
ble behavior. Thus, people will spend time on quantity overshooting on emissions is danger?
separating out different elements of waste for ous. With learning and readjustment of pol?
recycling, or they will drive more carefully, not icy (although not so frequently as to confuse
only because there may be a financial incen? structures and issues), the difference in effects
tive for recycling or penalties for bad driving, between a tax-orientated policy and a quan?
but also because they have a view of responsible tity/carbon-trading policy may not be so large.
behavior. Given where we start, however, in my view the
Pricing an externality can be done in a number danger of overshooting emissions targets is of
of ways. First, there is carbon taxation; second, great significance.
carbon trading on the basis of trade in rights Tradable quotas, the second method of estab?
to emit which are allocated or auctioned; and lishing a price for GHGs, have the advantage
third, implicit pricing via regulations and stan? of providing greater certainty about quantities
dards which insist on constraints on actions or of emissions than taxes. The European Union
technologies which involve extra cost but which Emissions Trading Scheme (EUETS) has shown
imply reductions in emissions. Each of the three that a big part of the economy can be covered
has different advantages and disadvantages and (currently around one half of European emis?
all three are likely to be used. Understanding the sions) with relatively low administrative bur?
pros and cons, where the different mechanisms dens by focusing on major emitting industries,
can and should be used, and how to deal with such as power.
problems of overlaps, are all very important By starting with allocations that are not paid
issues. We have the space to look briefly only at for and moving to auctions, trading can build
a few of the relevant considerations. acceptance by industry because it allows for a
Taxes have the advantage of being imple less dramatic adjustment. Free allocations based
mentable by individual governments without on historical emissions do have important prob?
international agreement. All taxes are conten? lems, however: they are likely to slow adjustment
tious but those on recognized "bads" such as since immediate profit pressures are lower; they
tobacco, alcohol, or carbon emissions may be can give competitive advantages to incumbent
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 25
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
26 AEA PAPERS AND PROCEEDINGS MAY 2008
10
1985
1
??
o
O)
O)
D
?
LU
O Electricity from
o * biomass (-85%)
0.1
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 27
0.18%
Public support for action will more a framework in which each country, or
be founded
group of countries,
not only on recognition of the magnitude of can assess its own respon?
sibilitiesthat
the problem, but also on the realization and targets
it with some knowledge of
is possible to construct collaborative
where thepolicies
rest of the world is going and how it
can interact.It is a
that are effective, efficient, and equitable.
great responsibility of economists toOn targets?a
help key element of effectiveness,
design
those policies. And they must doorsoaction on an appropriate scale?we should
urgently?
be clear how
the international discussion is moving far the international discussion
quickly
has already
and key decisions will be taken over themoved.
nextThe G8-G5 summit chaired
few years. by Germany in Heiligendamm in June 2007
The following is my own attempt to describe declared a world target of 50 percent reductions
the outline of a possible global deal based on by 2050. As sometimes happens in international
the preceding analysis and on my own intensive communiqu?s, not all details (such as base date
experience over the last two years of involve? and levels of agreement among attendees) were
ment in public discussion, taking account of the clear; but it was a significant marker nonethe?
recent UNFCCC meetings at Bali last month. less. And it is broadly consistent with the type
Let us begin with overall reductions targets and of stabilization range, around 500ppm C02e
the allocation of responsibilities across coun? for example, discussed in Section I. In what
tries. Our earlier discussion of trading, technol? follows, unless otherwise stated, emissions
ogies, and deforestation will then allow us to see reductions will be measured from 1990, cover?
quickly the broad structure of a global deal. Let ing all GHGs (in the six-gas Kyoto sense) and
us be clear* at the outset that this should not be emissions sources. The Heiligendamm 50 per?
seen in the overly formal way of a WTO discus? cent target is for the world as whole and it is
sion, founded in legal structures, with compli? generally agreed (see below) that, in the spirit
ance driven by sanctions, and where no one is of the Kyoto language of "common but differ?
bound until the full deal is agreed. This is much entiated treatment," the richer countries should
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
28 AEA PAPERS AND PROCEEDINGS MAY 2008
take responsibility for reductions bigger than Policies that could support this constituted the
the average. In what follows, we shall think of subject of Section III and should be at the heart
rich country reductions as including those dis? of a global deal.
charged by purchases on international markets. Let us investigate equity in a little more detail.
At Bali in December 2007, three coun? The history of flows and their relation to future
tries, Costa Rica, New Zealand, and Norway, stabilization targets should, in my view, be
declared targets of 100 percent reductions by central to a discussion of equity. All too often,
2050, i.e., "going carbon-neutral." The latter equity is seen solely or largely in terms of the
two are highly likely to need international pur? relative level of future flows (for example, per
chases to get there. Note, too, that reductions of capita convergence by 2050). A few numbers and
more than 100 percent are possible?many in a little basic arithmetic will help to understand
developing countries would regard targets for the issues. Currently, global emission flows are
rich countries above 100 percent as appropriate, around 40-45 Gt C02e. With a world popula?
given past history?and that such reductions tion of around 6 billion, that means average
that would almost inevitably involve interna? global per capita emissions are around 7 tons.
tional purchase. Given that the world population in 2050 will be
California has a target of 80 percent reduc? around 9 billion, in order to achieve 50 percent
tions by 2050. France has its "Facteur Quatre": reductions (i.e., an aggregate flow of around 20
dividing by 4, or 75 percent reductions, by 2050 Gt C02e) by then, per capita emissions will have
(Stern 2007, 516). The United Kingdom has a to be 2-2.5 tons. And since around 8 billion of
60 percent target but the Prime Minister Gordon these people will be in currently poor countries,
Brown indicated in November 2007 that this those countries will have to be in that range28
could be raised to 80 percent (Brown 2007). even if emissions in currently rich countries
Australia, under the new government elected were to fall to zero. It is clear from this basic
at the end of November 2007, has now signed arithmetic that any effective global deal must
Kyoto and has a target of 60 percent (Kevin have the currently poor countries at its center.
Rudd 2007); 80 percent is under considera? From the point of view of equity, the num?
tion after the Garnaut Review is published next bers are stark. The currently rich countries are
summer. responsible for around 70 percent of the existing
Targets for 2050 seem far away but the stock, and are continuing to contribute substan?
long lifetime of many investments means that tially more to stock increases than develop?
early decisions are needed to reach them. ing countries. The United States, Canada, and
Intermediate targets are also being set. At the Australia each emit over 20 tons of C02e (i.e.,
European Spring Council, 20-30 percent targets from all GHGs) per capita, Europe and Japan
were set for 2020; Germany has set 40 percent over 10 tons, China more than 5 tons, India
targets by 2020. The European Council also set around 2 tons, and most of sub-Saharan Africa
targets for renewables and CCS for 2020 and much less than 1 ton. Recent per capita C02
beyond, but it is the overall emissions targets, emissions (i.e., omitting other GHGs) for some
and their achievement, which are crucial. How countries are illustrated in Figure 9.
they are achieved country by country will vary In the lower part of this graph are three big,
and must take account of economic as well as fast-growing developing countries. China is
environmental, social, and political consid? growing especially quickly. Even with fairly
erations. At Bali, many were pressing for rich conservative estimates, it is likely that, under
countries to accept 25-40 percent cuts by 2020. BAU, China will reach current European per
That is indeed in the right range for rich country capita emissions levels within 20-25 years. With
cuts of 80 percent by 2050 and is now at least its very large population, over this time China
an initial 2020 benchmark. Overall, in discus? under BAU will emit cumulatively more than
sions of global and rich country targets, ranges the USA and Europe combined over the last 100
consistent with the criteria of effectiveness and
equity are now the basic benchmarks, and many
key commitments have been made. Delivery
on targets at reasonable cost?essentially effi? 28 In this context, I am referring to absolute emissions
ciency?is, of course, crucial and a challenge. originating in the country rather than who pays.
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 29
26
1990 1891 1982 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
years. That is one indication of the urgency of C02e as a reservoir sized 270ppm C02e, which
finding a global response quickly. the world will get close to exhausting over the
But let us keep focused on equity. With 80200 years between 1850 and 2050. If we start
percent reductions by 2050, Europe and Japanthe clock in the late 1980s or now, it would be a
would be around the required two-ton globalreservoir sized around 140ppm C02e or 120ppm
average level. An 80 percent reduction by the C02e, respectively.
United States, Australia, and Canada by 2050 From this perspective, equalizing the per cap?
would leave them around four tons, twice theita flows of emissions?or the size of the glass
required average level. Thus, a 50 percent over? drawn per person per year from the reservoir?
all reduction and an 80 percent rich country by 2050, shortly before it is dry, is a very weak
reduction would still leave average rich countrynotion of equity. It takes no account of all the
flows above the world average in 2050. guzzling that took place by the better-off over
Turning to stocks, let us think about the paththe preceding 50-200 years (depending on when
from some initial level to a stock stabilization (towe start the clock). There is a very big difference
be specific, suppose that level is 550ppm C02e), between a stock and a flow notion of equity. An
and about who consumes what along the way. We 80 percent reduction of flows by rich countries
can think of the initial level as 280ppm C02e,by 2050, in the context of a 50 percent reduction
corresponding to preindustrial times (aroundoverall, is not a target for which rich countries
1850); or we could start 20 years ago (aroundshould congratulate themselves warmly as dem?
390ppm C02e), when the problems of climateonstrating a splendidly powerful commitment
change began to receive strong policy attention;to equity. And the contract-and-converge argu?
or we could start now (around 430ppm C02e).ment for some common flow level, or for using
One perspective on equity would be to see thesuch a level as the eventual basis of trading, on
difference between 280ppm C02e and 550ppmthe asserted grounds that there are "equal rights
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
30 AEA PAPERS AND PROCEEDINGS MAY 2008
to emit or pollute," does not seem to me to have and pace required. They feel the inequities of
special claim on our attention.29 Rather, the tar? the situation and phenomena acutely. Just when,
get of equalizing by 2050 (allowing for trade) they argue, they are beginning to overcome pov?
may be seen as being a fairly pragmatic one, on erty, in part by rapid growth, they should not be
which it might be possible to get agreement, and asked to slow down. Financing, together with
one that, while only weakly equitable, is a lot technology demonstration and transfer, will
less inequitable than some other possibilities, be needed to convince them that moving to a
such as less stringent targets for rich countries. low-carbon growth path is not the same thing as
If we take any particular good, it will gener? moving to a low growth path.
ally be true that rich people consume more than The third element refers to the short- and
poor people. That is simply an expression of medium-term approaches to trading between
their being richer. In the case of the reservoir, rich and poor countries. The current system, the
or the "contents of the atmosphere," it is hard Clean Development Mechanism (CDM), was
to think of an argument as to why rich people established by Kyoto and operates at the level
should have more of this shared resource than of a project in a poor country (a so called non
poor people. They are not exchanging their labor Annex 1 country in the Kyoto Protocol). If a firm
for somebody else's and they are not consuming in a rich country (an Annex 1 country) is part of
the proceeds of their own land, or some natu? a trading scheme (such as the EUETS) which
ral resource that lies beneath it. I do not have recognizes the CDM, then that firm can buy
any special "correct" answer to the challenge of an emissions reduction achieved by the project,
understanding equity here, but it is a challenge subject to the project using technologies or
we cannot avoid discussing. Any global deal will approaches from an admissible list. The amount
have to involve some implicit or explicit under? of the notional reduction comes from compar?
standing over the sharing of this "reservoir." ing the project with a counterfactual?what the
The key elements of the global deal have, with entity doing the project might otherwise have
one exception, now been raised and discussed. done. Approval of a project goes through the
Let me express the deal or framework in terms poor country authorities and a special institu?
of two groups of three headings, the first cover? tional structure, currently in Bonn. The system
ing targets and trade and the second covering is slow, cumbersome, and very "micro."
programs for which public funding is likely to Trading on the scale required to reach the
be required. This set of six policies or programs type of targets discussed (see Table 3) requires
is the international part of a deal. The domestic a much simpler, "wholesale" system.30 At the
policies will vary across countries, using dif? same time, to get agreement with poor coun?
ferent combinations of policy instruments and tries, it will have to continue to be "one-sided,"
technologies as discussed in Section III. The six as in the CDM, i.e., you can gain from innova?
elements of a global deal are expressed in bullet tion, but are not penalized for BAU. Wholesale
point form in Tables 3 and 4. measures can include technological benchmarks
The first element of the first group covers the such as employing CCS (currently excluded
targets. The global target was explained and jus? from CDM), or sectoral benchmarks such as
tified in Section I and the distribution of targets getting below a certain amount of C02 per ton
above in this section. The second, the impor? of cement. As one-sided trading measures, the
tance of emissions trading, was emphasized in benchmarks could be set ambitiously.
Section III: the justification for a major focus on After these trading mechanisms have been in
GHG trading in policy lies in its promotion of place (with associated technology sharing) for
both efficiency and collaboration. Unless financ? a while, developing countries will be able to
ing flows for the extra costs of reducing emis? have confidence that a trading system can work
sions are available to poor countries, they are on an appropriate scale. Then it would be rea?
extremely unlikely to join the effort on the scale sonable to ask them to accept targets consistent
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 31
with overall global goals in the context of a The public funding requirements are grouped
strong set of goals by rich countries. If we look in three elements in Table 4. Each would require
for targets from poor countries now, the only a paper in itself for appropriate treatment,
ones that would be accepted would be far too and we can give only headlines. Deforestation
loose and would knock the bottom out of inter? accounts for up to 20 percent of current emis?
national trading, i.e., collapse the price. And in sions; the numbers are not easy to specify pre?
the future these loose targets would be likely to cisely?probably 5-8 Gt C02e per annum. These
form a base-line for subsequent discussion. That flows could be roughly halved, in my view, for
is why a staged approach is essential if currently around $5 per ton of C02, taking into account
poor countries are to accept participation in opportunity costs of land and the institutional,
responsible global stabilization so that by 2050 administrative, and enforcement measures nec?
their emissions average around 2 tons per capita. essary. Some have estimated higher costs (e.g.,
Recall that this is a half or a third of China's McKinsey?Enkvist, Naucl?r, and Rosander
current level. It is very unlikely to be possible 2007) but there appear to be large amounts of
to find financial flows on the scale required to "initial" reductions available at lower costs, par?
incentivize appropriate action from the public ticularly if programs are large-scale and coordi?
sector of rich countries. Witness the difficulty nated across countries (for further discussion see
in getting resources for Overseas Development Myers 2007; Daniel Nepstad et al. 2007; Niels
Assistance (ODA), which will be strained still Anger and Jayant Sathaye 2007). This would
further by the challenge of adaptation (see help to avoid reduced deforestation in country
below). The trading system provides for private A simply displacing activity and thus increasing
flows. deforestation in country B. Public sector flows
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
32 AEA PAPERS AND PROCEEDINGS MAY 2008
can be combined with private sector flows as the United Nations Development Programme
avoided deforestation is brought into the carbon (UNDP) estimated additional costs for develop?
trading process so that all countries are given ing countries of around $85 billion per annum
incentives. Indeed, one of the responsibilities of by 2015 (UNDP 2007, 15). And they will pre?
the publicly funded program would be to work sumably rise after that.
toward trading. Such extra financing will be hard to find. It
The second element in this second group, the may be compared with the $150-200 billion per
demonstration and sharing of technologies, is annum extra that would arise if the Organisation
urgent; financial resources must be made avail? for Economic Co-operation and Development
able and institutional arrangements designed. (OECD) countries moved to 0.7 percent GDP in
This is an important area for economic research. ODA by 2015, as many of them have promised.
One problem of particular urgency, for reasons The ODA promises of the 2002 UN International
described above, is the demonstration of CCS Conference on Financing for Development in
for coal. There are no current plants using CCS Monterrey, Mexico, in connection with the
for coal-fired generation on a commercial scale. Millennium Development Goals, and of the
From 2015 or 2020 on, the world will need most 2005 UK-chaired G8 Gleneagles summit on
of its new coal-fired electricity generation plants Africa, and preceding EU commitments in
to be operating with CCS if it is to have any July, were powerfully argued and justified at the
chance of realizing its targets. If CCS cannot time. They took little account of climate change.
work on the necessary scale, then we need to If that aspect is added, as it should be given
know soon and follow alternative strategies. At the magnitude of the challenge, and combined
present, however, it does look promising. There with the historical responsibilities for stocks of
is geological work to be done to identify storage GHGs and the implied consequences for poor
capacity, and careful legal and regulatory work countries, then the argument for 0.7 percent, in
to be done to allocate risk and responsibility. my view, becomes overwhelming. The Stern
Geology and coal vary greatly across the world Review left the argument at that point, although
and many demonstrations of commercial-scale a case could have been made for increasing the
plants are necessary. Feed-in subsidies, world? ODA targets.
wide, of around $5 billion per annum could sup? The framework I have now described does, in
port 30+ such plants over the next 7-8 years my view, meet the criteria of: effectiveness?it
and cover a broad range of examples.31 is on the right scale; efficiency?it relies heavily
There should also be support for many other on markets and market-orientated innovation;
technologies. We do not know what the most effi? and equity?it does at least give some specifi?
cient clean technologies will be in the future, and city to the "common but differentiated responsi?
the answers are likely to vary with location. CCS bility" already accepted internationally. It builds
is emphasized here simply because we can be on existing commitments and some aspects of
fairly confident that BAU will involve a great deal the current discussions in international fora. It is
of coal for electricity over the next 20-30 years. also designed to give some realistic opportunity
Perhaps it will be a medium-term technology and for the major developing countries to become
be replaced by others over the longer term. strongly involved, as they must if serious targets
Finally, in the global deal, I would emphasize are to be agreed and achieved.
an element that has not been discussed here and It is a framework that could allow all coun?
that will be of great importance. Even with very tries to move quickly along what they see to be
responsible policies, the world is likely to see a responsible path. What is very striking here
an additional 1-2?C of warming over and above is how broadly basic understanding has already
the 0.8?C it has already experienced. Adaptation been established. Country by country, we see
will be necessary worldwide and will be par? targets being erected and measures being set
ticularly difficult for poor countries. Recently, by individual countries recognizing their own
responsibilities as they see international agree?
ment being built. People seem to understand
31 Own calculations using, for example, the McKinsey
cost curve, and working with power stations of a few hun?
the arguments for action and collaboration on
dred megawatts. I am grateful to Dennis Anderson for his climate change much more readily than they
advice. do for international trade. But I do not want to
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 33
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
34 AEA PAPERS AND PROCEEDINGS MAY 2008
Arrow, Kenneth, Partha Dasgupta, Lawrence Dietz, Simon, Chris Hope, Nicholas Stern,
Goulder, Gretchen Daily, Paul Ehrlich, Geof? and Dimitri Zenghelis. 2007a. "Reflections
frey Heal, Simon Levin, et al. 2004. "Are We on the Stern Review (1): A Robust Case
Consuming Too Much?" Journal of Economic for Strong Action to Reduce the Risks of
Perspectives, 18(3): 147-72. Climate Change." World Economics, 8(1):
Atkinson, Anthony B., and Andrea Brandolini. 121-68.
2007. "On Analysing the World Distribution Dietz, Simon, Dennis Anderson, Nicholas Stern,
of Income." Unpublished. Chris Taylor, and Dimitri Zenghelis. 2007b.
Atkinson, Anthony B., and Joseph Stiglitz. 1976. "Right for the Right Reasons: A Final Rejoin?
"The Design of Tax Structure: Direct versus der on the Stern Review." World Economics,
Indirect Taxation." Journal of Public Econom? 8(2): 229-58.
ics, 6(1-2): 55-75. Dietz, Simon, Chris Hope, and Nicola Patmore.
Atkinson, Anthony B., and Joseph Stiglitz. 1980. 2007c. "Some Economics of 'Dangerous'
Lectures on Public Economics. London: Climate Change: Reflections on the Stern
McGraw-Hill Companies. Review." Global Environmental Change,
Barker, Terry, Mahvash S. Qureshi, and Jonathan 17(3-4): 311-25.
K?hler. 2006. "The Costs of Greenhouse Dr?ze, Jean, and Nicholas Stern. 1987. "The
Gas Mitigation with Induced Technological Theory of Cost-Benefit Analysis." In Hand?
Change: A Meta-Analysis of Estimates in book of Public Economics, Vol. 2, ed. Alan
the Literature." Tyndall Centre for Climate J. Auerbach, and Martin S. Feldstein, 909-89.
Change Research Working Paper 89. Amsterdam, NH: Elsevier.
Beckerman, Wilfred, and Cameron Hepburn. Dr?ze, Jean, and Nicholas Stern. 1990. "Policy
2007. "Ethics of the Discount Rate in the Stern Reform, Shadow Prices, and Market Prices."
Review on the Economics of Climate Change." Journal of Public Economics, 42(1): 1-45.
World Economics, 8(1): 187-210. Enkvist, Per-Anders, Tomas Naucl?r, and Jerker
Brown, Gordon. 2007. "Speech on Climate Rosander. 2007. "A Cost Curve for Greenhouse
Change." Speech, World Wildlife Fund, Gas Reduction." The McKinsey Quarterly, 1:
London, November 19, 2007. http://www. 35-45.
numberlO.gov.uk/output/Pagel3791.asp. Expert Committee on Integrated Energy Policy.
Carbon Dioxide Information Analysis Center. 2006. Integrated Energy Policy: Report of the
2007. United States Department of Energy. Expert Committee. Government of India Plan?
http://cdiac.ornl.gov. Date accessed Janu? ning Commission. New Delhi, August.
ary 25, 2008. Fankhauser, Samuel, Friedel Sehlleier, and Nicho?
Cline, William. 2007. "Comments on the Stern las Stern. 2007. "Climate Change, Innovation,
Review." In Yale Symposium on the Stern and Jobs." Unpublished.
Review, Chapter 6, 78-86. New Haven, CT: Fourier, Joseph. 1827. "M?moire sur les tem?
Yale Center for the Study of Globalization. p?ratures du globe terrestre et des espaces
www.ycsg.yale.edu/climate/stern.html. plan?taires." M?moires de l'acad?mie royale
Dasgupta, Partha. 2007. "Commentary: The des sciences, 7: 569-604.
Stern Review's Economics of Climate Garnaut, Ross. 2007. "Will Climate Change
Change." National Institute Economic Review, Bring an End to the Platinum Age?" Paper pre?
199(1): 4-7. sented at the inaugural S. T. Lee Lecture on
Deaton, Angus S., and Nicholas Stern. 1986. Asia & The Pacific, Australian National Uni?
"Optimally Uniform Commodity Taxes, Taste versity, Canberra, November 29.
Differences and Lump-Sum Grants " Econom? Guesnerie, Roger. 2004. "Calcul ?conomique et
ics Letters, 20(3): 263-6. d?veloppement durable." Revue ?conomique,
DeLong, Brad. 2006. "Partha Dasgupta Makes a 55(3): 363-82.
Mistake in His Critique of the Stern Review." Hadley Centre for Climate Prediction and
http://delong.typepad.com/sdj/2006/ll/ Research. 2005. "Stabilising Climate to Avoid
partha_dasgupta.html. Dangerous Climate Change?A Summary
Dietz, Simon, Cameron Hepburn, and Nicholas of Relevant Research at the Hadley Centre."
Stern. 2008. "Economics, Ethics, and Climate The Met Office, http://www.metoffice.gov.uk/
Change." Unpublished. research/hadleycentre/pubs/brochures.
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 35
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
36 AEA PAPERS AND PROCEEDINGS MAY 2008
Murphy, James M., David M. H. Sexton, David N. Australia, Canberra, December 3, 2007. http://
Barnett, Gareth S. Jones, Mark J. Webb, Mat? www.pm.gov.au/media/Release/2007/media_
thew Collins, and David A. Stainforth. 2004. release_0003.cfm.
"Quantification of Modelling Uncertainties in Smith, Joel B., Hans Joachim Schellnhuber, and
a Large Ensemble of Climate Change Simula? M. Monirul Qader Mirza. 2001. "Vulnerabil?
tions " Nature, 430(7001): 768-72. ity to Climate Change and Reasons for Con?
Myers, Erin C. 2007. "Policies to Reduce Emis? cern: A Synthesis." In Climate Change 2001:
sions from Deforestation and Degradation Impacts, Adaptation and Vulnerability, ed.
(REDD) in Tropical Forests: An Examina? James J. McCarthy, Osvaldo F. Canziani,
tion of the Issues Facing the Incorporation of Neil A. Leary, David J. Dokken, and Kasey S.
REDD into Market-Based Climate Policies." White, 913-68. Cambridge, UK: Cambridge
Resources for the Future Discussion Paper University Press.
07-50. Solow, Robert M. 1974. "The Economics of
Nepstad, Daniel, Britaldo Soares-Filho, Frank Resources or the Resources of Economics."
Merry, Paulo Moutinho, Hermann Oliveira American Economic Review, 64(2): 1-14.
Rodrigues, Maria Bowman, Steve Schwartz Stainforth, D., T. Aina, C. Christensen, M. Col?
man, Oriana Almeida, and Sergio Rivero. 2007. lins, N. Faull, D. J. Frame, J. A. Kettleborough,
"The Costs and Benefits of Reducing Car? et al. 2005. "Uncertainty in Predictions of the
bon Emissions from Deforestation and Forest Climate Response to Rising Levels of Green?
Degradation in the Brazilian Amazon." Paper house Gases." Nature, 433(7024): 403-6.
presented at the United Nations Framework Stern, Nicholas. 1976. "On the Specification of
Convention on Climate Change Conference Models of Optimum Income Taxation." Jour?
of the Parties, Thirteenth Session, Bali, Indo? nal of Public Economics, 6(1-2): 123-62.
nesia, December 3-14, 2007. Stern, Nicholas. 1977. "The Marginal Valuation
Newbery, David. 1992. "Long Term Discount of Income." In Studies in Modern Economic
Rates for the Forest Enterprise." Unpub? Analysis: The Proceedings of the Association
lished. of University Teachers of Economics, Edin?
Nordhaus, William D. 2007a. "The Challenge of burgh 1976, ed. M. J. Artis and A. R. Nobay,
Global Warming: Economic Models and Envi? 209-54. Oxford: Basil Blackwell.
ronmental Policy." http://nordhaus.econ.yale. Stern, Nicholas. 2007. The Economics of Climate
edu/dice_mssj)72407_all.pdf. Change: The Stern Review. Cambridge, UK:
Nordhaus, William D. 2007b. "A Review of the Cambridge University Press.
Stern Review on the Economics of Climate Stern, Nicholas, Jean-Jacques Dethier, and F.
Change." Journal of Economic Literature, Halsey Rogers. 2005. Growth and Empower?
45(3): 686-702. ment: Making Development Happen. Cam?
Nordhaus, William D., and Joseph G. Boyer. bridge, MA: MIT Press.
2000. Warming the World: Economic Mod? Sterner, Thomas, and U. Martin Persson. 2007.
els of Global Warming. Cambridge, MA: MIT "An Even Sterner Review: Introducing Rel?
Press. ative Prices into the Discounting Debate."
Okun, Arthur M. 1975. Equality and Efficiency: Resources for the Future Discussion Paper
The Big Tradeoff Washington, DC: Brookings 07-37.
Institution Press. Toi, Richard S. J. 2002. "Estimates of the Dam?
Pearce, David, and David Ulph. 1995. "A Social age Costs of Climate Change, Part II: Dynamic
Discount Rate for the United Kingdom." Cen? Estimates." Environmental and Resource Eco?
tre for Social and Economic Research on the nomics, 21(2): 135-60.
Global Environment Global Environmental lyndall, John. 1861. "On the Absorption and
Change Working Paper GEC-1995-01. Radiation of Heat by Gases and Vapours."
Pigou, Arthur C. 1932. The Economics of Wel? Philosophical Magazine, 22: 169-94,273-85.
fare. 4th ed. London: Macmillan. United Nations Development Programme. 2007.
Ramsey, F. P. 1928. "A Mathematical Theory of Human Development Report 2007/2008
Saving." Economic Journal, 38(152): 543-59. Fighting Climate Change: Human Solidar?
Rudd, Kevin. 2007. "Ratifying the Kyoto Pro? ity in a Divided World. New York: Palgrave
tocol." Media release, Prime Minister of Macmillan.
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms
VOL. 98 NO. 2 RICHARD T. ELY LECTURE 37
This content downloaded from 193.55.102.233 on Mon, 21 Jan 2019 09:15:48 UTC
All use subject to https://about.jstor.org/terms