Scott Barrett - Political Economy of The Kyoto Protocol

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 14, NO.
POLITICAL ECONOMY OF THE KYOTO
Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

PROTOCOL
SCOTT BARRETT
London Business School1
The Kyoto Protocol, negotiated in December 1997, is the first international treaty to limit emissions of green-
house gases. But Kyoto does not mark the conclusion to international cooperation on climate change. It is
really just a beginning. This paper shows that, in the aggregate, the benefits of undertaking the Kyoto reduc-
tions should exceed the corresponding costs—provided these are achieved cost-effectively. But, although Kyoto
seeks to promote cost-effectiveness, it may yet prove very costly. Moreover, the agreement may not even achieve
the reductions that it promises, either because emissions will relocate to the countries that are not required to
stay within Kyoto-prescribed ceilings or because ‘‘paper’ trades will be promoted by the protocol's mecha-
nisms. More fundamentally, Kyoto does not deter non-compliance, and it only weakly deters non-participation.
These flaws need to be mended, but the nature of the problem makes that an especially difficult task.
I. INTRODUCTION must be met. And though global emissions will

continue to rise, even if the protocol is implemented
The Kyoto Protocol, negotiated in December 1997, to the letter, the reduction from a business-as-usual
is a climate change treaty with a difference. Unlike emissions bench-mark may be close to being opti-
the Framework Convention on Climate Change that mal in the sense of balancing the global marginal
preceded it, the Kyoto Protocol incorporates targets costs and benefits of abatement. Assuming full
and timetables—that is, ceilings on the emissions of participation and cost-effective implementation, a
greenhouse gases and dates by which these ceilings recent study by the Clinton Administration esti-
1
This paper began to form in my mind at the NBER–Yale Global Change Workshop in Snowmass, Colorado, where I was able
to learn from, and put my questions to, some of the leading economists working on this issue. I am grateful to all the participants
for sharing their ideas, and especially to Charles Kolstad, William Nordhaus, and Robert Stavins for inviting me to participate in
the workshop. I am also grateful to Wilfred Beckerman, Dieter Helm, Tim Jenkinson, Chris Riley, Stephen Smith, Robert Stavins,
Peter Zapfel, and an anonymous referee for helpful comments on an earlier draft. David Pearce also provided helpful comments
at a seminar presentation.
20 © 1998 OXFORD UNIVERSITY PRESS AND THE OXFORD REVIEW OF ECONOMIC POLICY LIMITED
S. Barrett
mated the marginal cost of meeting the Kyoto agreement will never enter into force. In July 1997,
targets to range from $14 to $23/ton (Clinton Ad- the United States Senate voted 95–0 in favour of a
ministration, 1998). Most estimates of the global non-binding resolution urging the President of the
marginal damage of greenhouse gas emissions are United States not to negotiate an agreement that
of a similar magnitude (see IPCC, 1996, ch. 6), and required that only the industrial countries reduce
so it would seem that the Kyoto Protocol is a near their greenhouse-gas emissions or that would result
ideal outcome for the world. in serious harm to the US economy, where by
‘serious harm’ the Senate meant, in the words of
But this is only if the assumptions behind the Clinton Senator Robert Byrd, a co-author of the resolution,
Administration’s estimates are correct: that partici- ‘capital flight and a loss of jobs in the United States’.
pation in the agreement will be full and implementa- This is important because the Senate must ratify (by
tion cost-effective. A number of features of the a two-thirds majority) any treaty that is to be binding
Protocol will promote cost-effective implementa- on the United States, and an effective climate-

tion, including provisions for trading in the entitle- change treaty is sure to require US participation.
ments to emit greenhouse gases. However, though There are many reasons for this, but perhaps the
the details of the flexible mechanisms incorporated most obvious is that United States is the world’s
in Kyoto have yet to be worked out, implementation largest emitter. Its emissions are about 50 per cent
may turn out to be very costly, not least because higher than the entire emissions of the European
participation is unlikely to be full. The marginal cost Union.
of implementing Kyoto could be ten times the
estimates noted above. Moreover, the reduction in The Clinton Administration previously endorsed the
emissions effected by Kyoto could be less than the principle that the industrial countries should reduce
amount promised because of ‘trade leakage’. In- their emissions first, and could not easily reverse out
deed, since leakage will be greater the greater are of this promise in Kyoto. And nor was Europe keen
the between-country differences in marginal costs, on relaxing the so-called Berlin Mandate. So the
the same forces causing costs to be higher will agreement reached in Kyoto clashed with the Sen-
cause benefits to be lower. The Kyoto Protocol may ate’s recommendation that developing countries
turn out not be such a good deal after all. reduce their emissions (whether implementation of
the Protocol will be costly to the United States is a
Worse, the agreement may not even be sustainable, question requiring some analysis, and I shall return
and not just because high implementation costs to it later). Of course, the Senate could have been
could impel the parties to renegotiate the treaty. For bluffing, perhaps in the hope that its resolution would
another potential problem stalks Kyoto: compliance give President Clinton an edge in the Kyoto negotia-
enforcement and free-rider deterrence. The Proto- tions. Indeed, during the debate on the resolution,
col defers discussion of enforcement to a future Senator Byrd said that the resolution would ‘add
meeting of the parties, but it is sensible to ask: what strength to our US negotiating team’. But just after
would happen if in, say, 10 years’ time, one of the the negotiations ended in Kyoto, a number of sena-
parties to the agreement announces that it will not be tors asked that the treaty come to the Senate floor
able to comply with it? Or suppose, instead, that a for ratification so that they could reject it. President
party announces that it will withdraw from the Clinton has since said that he would not send the
agreement, because the costs of meeting it are too treaty to the Senate without ‘meaningful participa-
steep. What will prevent such a withdrawal? The tion from key developing countries’.
treaty, at least in its present form, offers little
protection from such deviations. And this is not just If the United States does not ratify the treaty, it is
a problem for the future. Countries can reason possible that the agreement will still come into force.
backwards. If future deviations cannot be pre- To enter into law, and therefore to become binding
vented, why should a country invest in abatement on the countries that are parties to it (but not other
measures today? countries), the Protocol must be ratified by at least
55 countries, responsible for at least 55 per cent of
Even this may not be the worst of the Protocol’s the total carbon-dioxide emissions of the so-called
problems. It is possible, maybe even likely, that the ‘Annex I’ countries (the industrial countries listed in
21
OXFORD REVIEW OF ECONOMIC POLICY, VOL. 14, NO. 4
Table 1
Status of the Kyoto Protocol
Annex I CO2 emissions Share of Annex I Kyoto target Projected emissions

countries 1990 1990 emissions 2008–12 2000
(gigagrams) (%) (% relative to 1990
or alternative base year)
United States 4,957,022 36.00 93 104

European Union* 3,288,667 24.05 92 103
Austria* 59,200 0.43 92 111
Belgium* 114,410 0.84 92 n.a.
Denmark* 52,025 0.38 92 103

Finland* 53,900 0.39 92 131
France* 366,536 2.68 92 109
Germany* 1,014,155 7.42 92 90
Greece* 82,100 0.60 92 115
Ireland* 30,719 0.22 92 120
Italy* 428,941 3.14 92 113
Luxembourg* 11,343 0.08 92 67
Netherlands* 167,600 1.23 92 92
Portugal* 42,148 0.31 92 129
Spain* 227,322 1.66 92 122
Sweden* 61,256 0.45 92 104
UK* 577,012 4.22 92 102
Australia* 288,965 2.11 108 115
Canada* 462,643 3.38 94 110
Iceland 2,172 0.02 110 105
Japan* 1,155,000 8.45 94 104
New Zealand* 25,476 0.19 100 116
Norway* 35,514 0.26 101 111
Switzerland* 45,070 0.33 92 97
Liechtenstein 208 n.a. 92 118
Monaco n.a. n.a. 92 n.a.
Economies in Transition 3,364,259 24.60 103 81
Alternative base year 3,531,476 — 98 77
Bulgaria* 1990 82,990 0.61 107 84
1988 96,878 — 92 72
Czech Republic 165,792 1.21 92 82
Estonia 37,797 0.28 92 54
Hungary 1990 71,673 0.52 110 96
1985–7 83,676 — 94 82
Latvia 22,976 0.17 92 74
Lithuania* n.a. n.a. 92 n.a.
Poland* 1990 414,930 3.03 108 96
1988 478,880 — 94 83
Romania 1990 171,103 1.25 107 n.a.
1989 198,479 – 92 n.a.
Russian Federation 2,388,720 17.47 100 83
Ukraine n.a. n.a. 100 n.a.
Slovakia 58,278 0.43 92 84
22
S. Barrett
Table 1 (continued)
Annex I CO2 emissions Share of Annex I Kyoto target Projected emissions

countries 1990 1990 emissions 2008–12 2000
(gigagrams) (%) (% relative to 1990
or alternative base year)
Croatia n.a. n.a. 95 n.a.

Slovenia n.a. n.a. 92 n.a.
Total 1990 13,675,067 100 95 98
Total base 13,842,284 — 94 97
Notes: Two Annex I countries (Belarus and Turkey) are excluded from the table, as they are not included

in Annex B of the Kyoto Protocol. Four other countries (Liechtenstein, Monaco, Croatia, and Slovenia) are
included in Annex B but not in Annex I. *Indicates that the country is a signatory to the Kyoto Protocol,
as of 23 October 1998. CO2 emissions exclude land-use change and forestry.
Source: All data are from the web page of the Climate Change Secretariat, http://www.unfcc.de.
the original Framework Convention) in 1990. As trading arrangements (explained later in the paper).
shown in Table 1, the United States accounts for It thus seems likely that many Annex I countries will
only 36 per cent of Annex I emissions. So, if enough await US ratification before serving the Kyoto
of the countries making up the balance of Annex I Protocol up to their own parliaments. This means
emissions ratify the agreement, Kyoto will still enter that, if the USA does not ratify the agreement, then
the canon of climate law. it may not enter into force.2
As of October 1998, 59 countries had signed the Why should countries negotiate a treaty that could
Protocol, including the 15 member states of the leave them worse off, or that may never even enter into
European Union and nine other Annex I countries international law? The scenario seems unlikely, but it
(signatories are identified in Table 1 by an asterisk). is entirely in keeping with the history of climate-change
These signatories make up just over 42 per cent of policy. As described in section II, countries have
total Annex I emissions, and so the minimum partici- previously announced their intention to keep within
pation required by the treaty would seem to be self-imposed emission ceilings—and then failed to
within easy reach. But putting a signature on a treaty meet them. Moreover, the Kyoto Protocol is not
does not obligate a country to ratify and, as of unique in the annals of international cooperation.
October 1998, only one country has ratified the The Law of the Sea Convention, negotiated in 1982,
Kyoto Protocol (though this in itself signifies nothing did not enter into force until 1994—and participation
as the treaty was only recently negotiated): the small by the major maritime powers, including the United
island state, Fiji. States and United Kingdom, had to await negotia-
tion of a side agreement which effectively rewrote
Ratification by the current signatories is not inevita- key provisions in the original treaty.
ble. If the USA does not ratify the agreement, the
other Annex I countries will benefit less from Of course, predicting whether Kyoto will endure, or
participating; these countries will have to undertake whether it will achieve much if implemented, de-
the emission reductions prescribed by the treaty pends on many details. It depends, especially, on
(and shown in column 3 of Table 1) without the assumptions about how the important concepts in
benefit of substantial US abatement. It is even the agreement will be interpreted, about the institu-
possible that non-participation by the USA will tions that will be developed to support it, about the
increase the cost to these countries of keeping costs of taking action, and about the future evolution
within their Kyoto limits, because of the treaty’s of the treaty. All these details are uncertain. They
2
Just as it is hard to imagine a Gulf War coalition forming without the support of the United States, so it is hard to see how an
effective climate-change regime could develop without American backing.
23
are discussed in my analysis of the agreement in endured. It is perhaps the most important feature of
section III. the Kyoto Protocol.
In the long run, whether or not Kyoto enters into law In the same year that this conference was held, the
will not matter very much. If the Protocol fails to Intergovernmental Panel on Climate Change (IPCC)
become law, countries will attempt to renegotiate was formed, at the request of the UN General
the agreement. If Kyoto does enter into law but later Assembly. The IPCC was asked to report on what
collapses for whatever reason, a new agreement was known and not known about climate change, on
can always be negotiated. Even if Kyoto suc- the potential impacts of climate change, and on what
ceeds—if it enters into law and is implemented to could be done to forestall and adapt to climate
the last detail—a string of amendments will need to change. The IPCC’s first assessment report, pub-
be negotiated, to say what must be done after 2012. lished in 1990, concluded that ‘emissions resulting
Kyoto is really just the start of a long process, and from human activities are substantially increasing

it must be remembered that climate change is a very the atmospheric concentrations of the greenhouse
long-run problem. What will matter most in the gases . . . [and] will enhance the greenhouse effect,
future is whether countries perceive that substantial resulting on average in an additional warming of the
mitigation is justified, and whether the international Earth’s surface’ (IPCC, 1990, p. 1). The report
system can muster the cooperation needed to sus- calculated that ‘the long-lived gases [including CO2]
tain this effort. I turn to these fundamental issues in would require immediate reductions in emissions
section IV. The final section of the paper pulls these from human activities of over 60 per cent to stabilize
different analyses together and revisits the theme of their concentrations at today’s levels’, and it pre-
this introduction. dicted that, under the ‘Business-as-Usual’ scenario,
global mean temperature would rise by between
0.2oC and 0.5oC, and mean global sea level would
II. GETTING TO KYOTO rise by between 3 and 10cm, per decade during the
next century. Rather ominously, the IPCC noted
(i) Preliminaries that ‘the complexity of the system means that we
cannot rule out surprises’.
The so-called greenhouse gases include not only
carbon dioxide (CO2), but also methane, nitrous
(ii) Unilateral Pledges
oxide, fluorocarbons (including hydrofluorocarbons
and perfluorocarbons), tropospheric ozone (precur- Following publication of the IPCC’s 1990 report, a
sors of which include nitrogen oxides, non-methane number of OECD countries announced intentions to
hydrocarbons, and carbon monoxide), and sulphur reduce their CO2 emissions.4 Some pledged to meet
hexafluoride.3 However, CO2 accounts for the bulk the Toronto target (Austria, Denmark, Italy, Lux-
of aggregate warming potential and, mainly for this embourg; New Zealand pledged that it would do so
reason, the policy debate has focused on the extent by 2000 rather than 2005). Some set a goal of
to which emissions of this gas should be limited. In stabilizing their CO2 emissions at the 1989 level by
1988, a semi-political conference held in Toronto 2000 (Norway) or at the 1990 level by 2000 (Finland,
recommended that, as a first step, CO2 emissions Switzerland, United Kingdom) or to reduce emis-
should be reduced 20 per cent from the 1988 level sions 3–5 per cent by 2000 (The Netherlands).
by 2005. This so-called ‘Toronto target’ was arbi- Germany, helped by unification, set the most ambi-
trary, but the idea that countries should commit to tious target: to reduce CO2 emissions 25–30 per
meeting a target for emission reduction (as opposed cent from the 1987 level by 2005. Australia pledged
to, say, a carbon tax or a technology standard) has to reduce its emissions of all greenhouse gases not
3
Other halocarbons, including chlorofluorocarbons (CFCs) and hydrochlorofluorocarbons (HCFCs), are also potentially
important from the policy perspective, but are being controlled by the Montreal Protocol and its associated amendments. Moreover,
it is now known that the direct warming effect of these gases is partly offset by a cooling effect caused by the reduction in stratospheric
ozone.
4
The International Energy Agency (1992) has compiled a comprehensive listing of climate-change policies, and I am drawing
here from this report.
24
S. Barrett
controlled under the Montreal Protocol (that is, meeting it. A collective policy for meeting the target
excluding CFCs and HCFCs), while other countries was needed. The European Commission proposed
(Canada and the United States among them) set a meeting the target by means of an energy conserva-
target of stabilizing the emissions of all greenhouse tion programme coupled with a fiscal measure, a
gases, including those covered by the Montreal carbon tax. The tax, which was to be set at a rate
Protocol. France and Japan pledged to stabilize their equivalent to $3 per barrel of oil, rising over time to
CO2 emissions at the 1990 level by 2000 but only on $10 per barrel, would probably have been enough to
a per-capita basis (allowing emissions to increase as meet the stabilization target (see Barrett, 1992). But
population increased). Spain, a relatively poor OECD in May 1992, shortly before the Rio Earth Summit
member, set the goal of limiting its growth in CO2 convened, the Community announced a number of
emissions to 25 per cent. Finally, some countries modifications to the original tax proposal.
merely promised to play a part in achieving a
collective target. In October 1990 the European The first of these was to supplement the carbon tax

Community (EC) announced its intention to stabilize with an energy tax (the combined tax would be
Community-wide emissions at the 1990 level by equivalent to the per-barrel tax noted above). Os-
2000, a target to which all its member states were tensibly, the intention was not just to reduce carbon-
collectively bound. Members of the European Free dioxide emissions but also to conserve energy. But
Trade Association, including Iceland and Sweden, the real reason was to dilute the advantage that a
were in turn bound by a separate agreement jointly pure carbon tax would give nuclear energy and
to meet the EC target. countries with high shares of nuclear electricity
generation (such as Belgium and France). A second
All this may give the impression that much was modification was to exempt the main energy-using
being done, but the reality was different. Few industries from having to pay the tax. This was to
countries put into place policies that would contrib- stop these industries from suffering a ‘competitive
ute to their targets being met, and there seemed little disadvantage’, relative to non-EC countries. The
need to do so. For some of these targets were final modification was to make implementation of
intended merely as goals, while others were condi- the EC tax conditional on other OECD countries
tional on other countries taking similar action (this (especially the United States and Japan) adopting
was true of Britain, for example, under the Thatcher the same tax. As the chances of this were nil, this
government). Though New Zealand set for itself an meant that Europe was not prepared to implement
ambitious goal of reducing its CO2 emissions 20 per the policy needed to achieve its own target.
cent from the 1990 level by 2000, it simultaneously
insisted that any policy adopted should have a net (iii) To Rio
benefit for New Zealand. Several countries claimed
to be ‘committed’ to achieving a particular emission The Community’s policy was being mapped out just
ceiling, but none truly was committed. If a country as negotiations on the Framework Convention on
learned later that its interests would be badly served Climate Change were coming to a close. Through-
by meeting its target, then there would be nothing to out these negotiations, Europe tried to persuade the
stop it from failing to meet it. Indeed, it would be hard United States to fix a date for stabilizing its CO2
to argue that a country would even be morally bound emissions. The United States refused, however, and
to meet a target which it had set for itself, especially the final text of the Framework Convention, which
when other countries were at the same time failing was signed by over 150 countries at the Rio Earth
to meet their targets. Summit in June 1992, did not commit any signatories
to meeting specific targets and timetables (contrary
The EC’s climate policy was especially important, to reports one often reads in the newspapers).
partly because of the Community’s relatively large Article 4 says that developed country parties recog-
share of global emissions and partly because of the nize ‘that the return by the end of the present decade
way the Community’s target was framed. When the to earlier levels of anthropogenic emissions of car-
target was agreed in 1990, no decision was made as bon dioxide and other greenhouse gases’ would be
to how it would be met, and as it was a collective desirable. It also says that these parties should
target, no country was individually responsible for devise policies ‘with the aim of returning individually
25
or jointly to their 1990 levels of these anthropogenic and 2020. These quantitative ceilings were to be
emissions’. But in contrast to the Kyoto Protocol, no included in a new protocol that might be ready for
country was required by the Framework Conven- signing by the end of 1997. Importantly, developing
tion to meet any particular target by any particular countries were not expected to limit their emissions.
date. Indeed, it was precisely for this reason that this It was this differential treatment of industrialized
agreement was ratified by so many countries and and developing nations in the so-called ‘Berlin man-
came into force so quickly (in December 1993). date’ that the US Senate later objected to and that
ultimately came to be embodied in the Kyoto Proto-
(iv) After Rio col.
The IPCC revised its earlier predictions in 1995, It is as well to recall, however, that at this time most
partly to take account of the effect of aerosols on countries had still not devised, let alone imple-
radiative forcing. Aerosols are tiny airborne parti- mented, effective policies for meeting the targets

cles, released when fossil fuels are burned, and they had set unilaterally years before. Some coun-
result in a local cooling effect (unlike some green- tries, including Norway and Finland, conceded that
house gases, which can persist in the atmosphere they did not expect to meet their targets (Grubb,
for decades, even centuries, aerosols have an at- 1995), despite having imposed hefty carbon taxes.
mospheric lifetime of about a week). Once aerosols The few countries that did expect to meet their
were included in the climate models, the IPCC targets were only able to do so for reasons of
predicted more modest change: an increase in global fortuitous circumstance (in Britain, the ‘dash for
mean temperature of about 0.14–0.28oC per dec- gas’; in Germany, unification), not determined policy.
ade, as compared with an increase of 0.16–0.36oC Most importantly, the European Union (EU) signally
per decade when the effects of greenhouse gases failed to devise a policy sure of meeting its ‘commit-
alone were considered. Aerosols were also pre- ment’ to stabilize emissions at the 1990 level. In a
dicted to limit mean sea level rise to about 2–8cm per letter to the chairman of the European Parliament’s
decade. environment committee, leaked on the eve of the
Berlin conference, Jacques Santer, the President of
Though inclusion of aerosols lowered slightly the the European Commission, conceded that ‘a single
predicted consequence of climate change, it also tax . . . applicable in all member states [was] no
increased confidence in the estimates of climate longer conceivable’.5 At the same time, the Com-
models. When aerosols were included, the pre- mission had not developed an alternative collective
dicted changes accorded better with the historical policy for meeting the EU’s target. Evidence sup-
record. Partly for this reason, the IPCC (1995, p. 22) plied to the European Commission suggested that at
was able to warn that, ‘the balance of evidence most three of the EU’s 15 member states would
suggests that there is a discernible human influence stabilize their own CO2 emissions at the 1990 level
on global climate’. Still, even today the science of by 2000.6 And, yet, when Europe’s diplomats headed
climate change is riddled with uncertainties: about for Kyoto, they were hoping to tighten up on the
the extent and timing of climate change; about earlier targets, to secure an agreement that would
regional variations; about whether small changes in reduce emissions (of the three main gases, carbon
atmospheric concentrations could, beyond some dioxide, methane, and nitrous oxide) 7.5 per cent by
point, trigger a discontinuous change in some impor- 2005 and 15 per cent by 2010.
tant climate feature.
III. ANALYSIS OF THE KYOTO

(v) From Berlin to Kyoto
AGREEMENT
At the first Conference of the Parties to the Frame-
work Convention, held in Berlin in 1995, the indus- As noted in the Introduction, the Kyoto Protocol
trialized parties agreed to negotiate emission limits specifies maximum emission levels for the so-called
within specified time frames, such as 2005, 2010, Annex I countries (see Table 1), and dates by which
5
The European, 17–23 March 1995, p. 1.
6
Ibid.
26
S. Barrett
these ceilings (calculated relative to 1990 emission perhaps much more. Shifting more abatement would
levels) must be met. Just as significant, the emis- save even more money (though of course the
sions of developing countries are entirely uncon- marginal cost saving will fall as more abatement is
strained by the protocol. These twin features fulfil shifted). Total costs will, of course, be minimized
the promise made at the First Conference of the where the marginal cost of abatement is every-
Parties in Berlin in 1995, and in this sense made where equal.
Kyoto a success. Ultimately, however, whether
Kyoto succeeds will depend on how it becomes As noted earlier, estimates prepared by the Clinton
implemented, and especially on whether implemen- Administration suggest that a cost-effective agree-
tation can be made cost-effective. ment—that is, an agreement which reduced global
emissions by the same amount as required by the
A variety of so-called ‘flexible mechanisms’ are Kyoto Protocol, but which did so by distributing the
built into Kyoto, and they have the potential of burden of abatement such that marginal costs were

supporting a cost-effective final allocation of cli- everywhere equal—could lower marginal costs to
mate-change mitigation. It is hard to say, however, around $14–23 per ton, about one-tenth the level
to what extent this potential will be realized. The that would be needed to implement the individual
data are sketchy in places and analyses of some emission ceilings in the protocol. This is a huge
features of the agreement have not yet been under- difference, and one that is reflected also in other
taken. Much will also depend on how the flexible studies. For example, Nordhaus and Boyer (1998)
mechanisms take shape and how countries devise estimate the marginal cost of a cost-effective Kyoto
their own policies. We can say something about Protocol to be $11/ton in 2010. Manne and Richels
certain bench-mark cases, and problem areas can (1998) obtain a much higher figure—$70/ton in
be pointed out. But that is about as far as our analysis 2010—but one that is still low in comparison with
can go. It happens, however, that this is enough to their estimate of marginal costs when the Kyoto
support the warning that introduced this paper. Protocol targets are met exactly.
(i) Cost Implications of the Emission Limits Estimates of reductions in total costs are of a similar
relative magnitude. According to the Clinton Ad-
Suppose that the limits negotiated in Kyoto were ministration’s analysis, the total cost to the USA of
met exactly, with no potential for arbitrage across implementing Kyoto could be just $7–12 billion per
countries. That is, suppose that the EU kept its year, if the agreement is implemented cost-effec-
emissions to 92 per cent of its 1990 level, that the tively, but perhaps ten times as large otherwise.
USA limited its emissions to 93 per cent of its 1990 Manne and Richels (1998) predict that cost-effec-
level, that China emitted as much as it pleased, and tive implementation of Kyoto would cost the USA
so on. Then the marginal cost of climate-change around $20 billion or 0.25 per cent of GDP in 2010,
mitigation would vary from country to country. It but perhaps four times as much if implemented
would be zero in China, where emissions were without trading. (Would this cause ‘serious harm’ to
unconstrained (and growing rapidly), and high in the US economy? Ask the Senate.) Nordhaus and
Europe and the United States. How high? Accord- Boyer (1998) estimate that the total cost of imple-
ing to one study (Nordhaus and Boyer, 1998), the menting Kyoto without trading (in present value
marginal cost of implementing the individual targets terms) would be about seven times the cost-effec-
in the protocol could be $125 per ton of carbon by tive level.
around 2010. Another study (Manne and Richels,
1998) predicts that marginal costs could be $240 per But this is to compare extremes. As detailed below,
ton of carbon in 2010. This difference in marginal the Kyoto Protocol offers a number of mechanisms
cost ($0 in the developing countries compared to intended to lower total implementation costs. As
$125 or $240 in the OECD countries) in turn implies also explained, these mechanisms will not work
that the total cost of achieving any given emission perfectly, and so will not mimic the cost-effective
ceiling will be excessive. Shifting just one ton of outcome. The costs of implementing Kyoto are
abatement from the OECD countries to the devel- likely to lie somewhere between the bench-marks
oping countries would save the world at least $100, given above.
27
Note, however, that we cannot even be sure of this. pheric concentrations of CO2 at much lower costs
For example, the estimates of marginal and total than reducing emissions of greenhouse gases result-
costs given above assume that domestic implemen- ing from industrial activity.’ At the margin, assuming
tation by every nation is cost-effective—that the that only abatement of gross emissions is under-
marginal costs of abatement are everywhere equal taken, this must surely be right. However, Stavins
within each country. This is unlikely to happen. It is (1998b) finds that the marginal cost of carbon
certainly not a feature of most environmental poli- sequestration rises steeply—more steeply than mar-
cies that have been adopted in the past. The carbon ginal gross abatement costs for the United States.
taxes adopted by most Nordic countries, for exam- So the aggregate cost savings from carbon seques-
ple, vary by sector, with households having to pay tration may not be all that large.
more than industry (partly out of a concern for trade
leakage). Until we know the policies that countries Measurement problems are also bound to be rife.
will develop to meet their targets—and these have The Protocol insists that the changes in net emis-

not been spelled out yet—we will not know how sions be ‘measured as verifiable changes in carbon
costly it will be to meet the Kyoto targets. stocks’, but such changes cannot be measured with
the same precision as the carbon emissions resulting
(ii) Flexible and Market Mechanisms from fossil fuel consumption. And how is one to
interpret whether an action constitutes a ‘direct
Net emissions targets human-induced land-use change’? Would the re-
The extent of climate change will depend on atmos- cent fires in Indonesia count? Settling these matters
pheric concentrations (though with a lag) of green- is a subject of ongoing negotiations.
house gases, and changes in these concentrations
depend on the removal of CO2 from the atmosphere Comprehensive emissions targets
as well as gross emissions. CO2 removal depends in The Protocol’s net emission limits apply to a bundle
turn on land use: growing trees absorb carbon from of greenhouse gases and not just carbon dioxide.
the atmosphere; the standing forest stores carbon (if The other gases include methane, nitrous oxide,
burned, trees release carbon back into the atmos- hydrofluorocarbons, perfluorocarbons, and sulphur
phere). So if trees are planted and the standing hexafluoride.8 These are bundled up with carbon
forest is prevented from being burned, concentra- dioxide into an aggregate measure, with the weights
tions will fall (all else being equal), and these activi- attached to individual gases reflecting their ‘global
ties should be encouraged just as emissions are warming potentials’. Abatement of a ton of nitrous
discouraged. The emission limits specified in the oxide, for example, is equivalent to abatement of
Kyoto Protocol do this. They allow deductions for around 315 tons of carbon.
‘removals by sinks resulting from direct human-
induced land-use change and forestry activities, This ‘comprehensive approach’ to climate change
limited to afforestation, reforestation and deforesta- mitigation was championed by the Bush Adminis-
tion since 1990’. In other words, the Kyoto targets tration, and is to be welcomed. In allowing for trade-
limit net emissions.7 offs between different types of gases, the total cost
to climate-change mitigation will be lowered.
Will including carbon sinks reduce the costs of
meeting the Kyoto limits substantially? The Clinton By how much will this mechanism lower costs? I
Administration’s (1998, p. 24) analysis suggests that have not seen any estimates, though in a statement
it could. ‘Promoting afforestation and reforesta- before the US House of Representatives,9 Janet
tion,’ the report maintains, ‘may reduce atmos- Yellen of the Council of Economic Advisers noted
7
To be precise, the Protocol allows sinks to play a role in capping emissions. It does not include carbon sinks in the emissions
baseline, with one exception. If a country’s carbon sinks were a net source of greenhouse-gas emissions in 1990, then its net emissions
from sinks must be incorporated into the baseline.
8
Note that the European Union and Japan sought to limit just three gases. It was the USA that insisted on including all six gases.
9
Janet Yellen, Statement before the US House of Representatives Committee on Government Reform and Oversight,
Subcommittee on National Economic Growth, Natural Resources, and Regulatory Affairs, 19 May 1998, http://www.state.gov/
www/policy_remarks/1998/980519_yellen_climate.html.
28
S. Barrett
that ‘a strategy of reducing non-CO2 greenhouse Whether savings like these will ever be realized will
gas emissions by a greater percent than CO2 emis- depend on how the institutions supporting trading
sions could lower emissions permit prices (that is, develop. If the trading arrangements allow a market
marginal costs) by as much as 10 per cent’. to develop which provides ready price discovery
and low transactions costs, then the bulk of these
‘Banking’ gains will be realized. Otherwise just a fraction,
Kyoto does not require that the emission ceilings perhaps a small fraction, of these gains will be
shown in Table 1 be met every year; it requires only pocketed.
that they be met by each Annex I party on average
over the 5-year period, 2008–12. Moreover, parties Europe has thus far been suspicious of the concept,
are allowed to carry forward additional reductions believing it to be a ploy for letting the United States
to a future control period. That is, if a country evade its responsibilities. This is a gross misunder-
reduces its emissions by more than required in the standing of the problem. As noted before, where

first control period (2008–12), it can ‘bank’ or carry abatement takes place is of no relevance to the
forward the surplus to the next control period. climate. Absolutely nothing can be gained by mak-
Finally, certified emission reductions, carried out ing the United States or any other country pay more
under the Clean Development Mechanism (CDM; than is necessary for abatement. Indeed, it is not
see below) from 2000 to 2007 can also be carried even obvious that the USA would gain dispropor-
forward to the first commitment period, 2008–12. tionately from trading. Calculations by McKibbin et
This allows Annex I countries to benefit from taking al. (1998) show that Europe would gain more from
early action through the CDM. trading than the United States.
These provisions could be helpful, but they do not go ‘Hot air’ trading
far enough. In particular, Kyoto does not allow One reason that trading among the Annex I coun-
parties to shift emission reductions toward the fu- tries would lower marginal and total costs is that the
ture—that is, to ‘borrow’ future emission reduc- economies in transition are allowed by Article 3 to
tions. Of course, if abatement is shifted forward, the choose an alternative base year to 1990 (subject to
benefits of the abatement in present value terms will some restrictions). As shown in Table 1, Bulgaria
fall. But costs may fall much more. If abatement is has chosen 1988 as a base year; Hungary, the
rushed (and it will be under the Kyoto timetable), average of 1985–7; Poland, 1988; and Romania,
some of the existing capital stock will have to be 1989. The effect is to create a surplus of emission
scrapped before its useful life is up. It would be entitlements that may not be exhausted by economic
cheaper if emission reductions could be effected by growth in these countries, even by 2010. Russia
incremental investments. Manne and Richels (1998) must retain its 1990 base year, but it will still have a
estimate substantial savings to a gradual transition to huge surplus by the year 2000, if the projections
the Kyoto targets, with marginal costs being re- shown in Table 1 prove correct (unfortunately,
duced by a factor of ten or more in 2010. projections to 2008–12 are not available).
Emissions trading As long as these emission ceilings do not bite,

Perhaps the most important flexible mechanism in marginal abatement costs in the economies in tran-
the Kyoto agreement is the provision for trading sition will be zero without Annex I trading. Trading,
among the Annex I countries. According to the however, will lower costs for all the Annex I
Clinton Administration’s (1998) analysis, this provi- countries for two reasons: first, by redistributing
sion could lower the marginal cost of implementing abatement within the Annex I group of countries,
Kyoto by 72 per cent, and lower the total cost of such that marginal costs are everywhere equal; and
implementation by 57 per cent, compared with the second, by relaxing the total constraint on Annex I
bench-mark of meeting the national targets unilater- emissions.
ally. Nordhaus and Boyer (1998) obtain a similar
result. By their calculations, the present value total To see the importance of this second effect, con-
cost of implementing Kyoto would be reduced 45 sider the consequences of trading within a US–
per cent by Annex I trading. Russian umbrella. In the year 2000, the estimates in
29
Table 1 suggest that emissions in the USA could not agreement would have increased, whereas the Eu-
exceed 0.93×4,957,022 = 4,610,030 gigagrams with- ropean economies in transition would still have had
out trading. If the estimates in the table are to be an incentive to participate, so long as their incentive
believed, Russia will easily stay within its limits, compatibility constraints were satisfied.
emitting only 0.83×2,388,720 = 1,982,638 gigagrams
of CO2 in 2000. Hence, without trading, total emis- Joint implementation
sions for both countries would not exceed 4,610,030 The Kyoto Protocol also allows ‘joint implementa-
+ 1,982,638 = 6,592,668 gigagrams. But Russia is tion’ (JI) trades among the Annex I countries.
allowed to emit up to 2,388,720 gigagrams of CO2 in These are bilateral project-based, rather than mar-
2000. So total allowed emissions for both countries ket-based, trades, in which one country receives
under a trading regime are 4,610,030 + 2,388,720 = ‘emission reduction units’ for undertaking projects
6,998,750 gigagrams. Trading thus eases the total in another country that reduce net emissions.
constraint on the two countries by 6,998,750 –

6,592,668 = 406,082 gigagrams in 2000. For reasons JI trades must be individually negotiated, and so will
that should be obvious, this difference in aggregate entail transactions costs. These costs will likely be
emissions between the trading and no-trading cases high because of the elusive nature of the commodity
is sometimes referred to as ‘hot air’. being traded. JI projects must provide ‘a reduction
in emissions by sources, or an enhancement of
As noted at the bottom of Table 1, the Kyoto removals by sinks, that is additional to any that
emission constraints are expected to bind in the would otherwise occur’ (emphasis added). Calcu-
aggregate, even by the year 2000. Annex I emis- lating this additional reduction will not be easy,
sions are projected to be 97 per cent of the adjusted because of course one is not able to observe the
base-year emissions, whereas Kyoto requires that emissions profile that would have been realized had
they be 94 per cent of this level. However, the ‘hot the trade not taken place. This must instead be
air’ released by trading does ease the aggregate inferred. Costly analyses will thus need to be under-
emissions constraint for Annex I emissions. Annex taken. Experience with the emissions trading pro-
I trading lowers costs partly by lowering total abate- gramme in the United States suggests that where
ment. transactions costs are high, bilateral trading will be
limited.
Note, however, that though the ‘hot air’ provision
appears to be a loophole, had it not been created— Clean development mechanism
had the economies in transition been given tighter The JI concept is extended to include non-Annex I
emissions constraints—it is likely that the other countries through the CDM. This allows Annex I
Annex I countries would have insisted that their own countries to meet their emission ceilings by under-
emission constraints be relaxed. For in reducing the taking projects in developing countries that provide
amount of ‘hot air’, the costs to the other Annex I ‘additional’ and ‘certified’ emission reductions. The
countries of fulfilling their commitments would CDM is potentially of huge significance, for it
increase. When seen in the context of the negotia- provides the only means within the Kyoto frame-
tions, a bigger problem with the ‘hot air’ provision work of shifting abatement toward the non-Annex
may be that it gave away something for nothing. I countries.
Of course, the economies in transition could be But the CDM has a number of problems. One is that
justified in putting their resources somewhere other it is not obvious whether the CDM would be limited
than in climate-change mitigation (many of these to emission reductions or whether it can include
countries are poorer than some non-Annex I coun- sequestration projects. The provisions for JI explic-
tries). But the other Annex I countries have given itly allow sequestration projects to be included, but
these economies more than was needed to make the CDM article is silent on this question.
their participation incentive compatible. This is not
just a matter of redistributing the gains from coop- An even more important difference is that one of the
eration. Had less been given away, the incentives parties to a CDM transaction will not have its
for the other Annex I countries to participate in the emissions capped. Potentially, therefore, the CDM
30
S. Barrett
could produce only ‘paper’ emission reductions. Most poor countries would have every incentive to
Moreover, as Stavins (1998a) warns, it is likely that walk away from an agreement that required them to
the least beneficial CDM projects will be adversely dig into their pockets, and few people would blame
selected by this mechanism. Indeed, the problem is them for doing so. But if developing countries had
doubly worrying. Not only do developing countries agreed to be bound by targets, then they would be
have incentives to offer projects that would have able to trade with the Annex I countries and—
been undertaken anyway, but the Annex I countries subject to appropriate choice of their emission ceil-
have incentives also to select these projects, if they ings—be virtually sure of being better off. An earlier
can be acquired at lower cost (this is just another draft of the Protocol allowed developing countries to
manifestation of the free-rider problem). choose, at any time and on a voluntary basis, a level
of emissions control that was appropriate to their
It will therefore be a matter of interest not only to the circumstances, but the provision was subsequently
parties engaging in a CDM transaction but also to all expunged, apparently at the insistence of China and

other parties whether a transaction really will pro- India (see Jacoby et al., 1998). Since inclusion of
vide ‘reductions in emissions that are additional to developing countries in some manner is vital, the
any that would occur in the absence of the certified matter is sure to be on the agenda of future meetings
project activity’. And it is for this reason, in contrast of the parties.
to the JI provisions, that the emission reductions
resulting from a CDM transaction must be ‘certified ‘Supplemental’ trading
by operational entities to be designated by the A further problem is that JI, CDM, and emissions-
Conference of the Parties’. trading transactions are intended to be ‘supplemen-
tal’ to domestic actions, a constraint reaffirmed by
Though necessary, certification will be costly, and the G8 group of countries meeting in April 1998.
the countries carrying out CDM trades will have to According to a Financial Times article (6 April
pay for certification (as noted in the Protocol, ‘a 1998) on the G8 summit, the virtue in this con-
share of the proceeds from certified project activi- straint is that it will prevent the leading industrial
ties [will be] used to cover administrative expenses’). nations (plus Russia) from being able ‘to evade
Moreover, Kyoto insists that a share of the proceeds painful domestic reductions in greenhouse gas emis-
from CDM trades also be used ‘to assist developing sions’.
country Parties that are particularly vulnerable to
the adverse effects of climate change to meet the This is a twisted logic. It cannot be good for the
costs of adaptation’. This sounds like a tax. If CDM environment. If anything, the restriction on trading,
transactions are taxed, and if transactions costs are in elevating between-country differences in mar-
high, the volume of CDM trades will be very low. ginal costs, will harm the environment by magnifying
the leakage problem. And it cannot be sure to make
There is no way of knowing by how much costs will developing countries any better off either.
be reduced by the CDM, not least because the
important details have yet to be negotiated. In her Whether this constraint will ever bite, however, is
statement to the US House of Representatives, another unknown, for the parties have not defined
however, Janet Yellen offered a guess: what ‘supplemental’ means. If the notion is inter-
preted as being qualitative, then it will easily be
The CDM cannot realistically be expected to yield all the satisfied, for even with unconstrained trading every
gains of binding targets for developing countries, but it
Annex I country will undertake some abatement at
might shave costs by roughly another 20 to 25 per cent
from the reduced costs that result from trading among
home. More serious would be an arbitrary, quanti-
Annex I countries. tative limit on trading. Unfortunately, there is some
support for such a cap, especially in Europe. The
As suggested by this statement, CDM transactions European Parliament adopted a resolution in Sep-
costs could have been reduced considerably had the tember 1998 calling for ‘an agreement to have a
Kyoto diplomats succeeded in negotiating emission quantitative ceiling on the use of flexibility mecha-
limits for the developing countries. The issue is not nisms to ensure that the majority of emissions
whether these countries should pay to participate. reductions are met domestically’.
31
Table 2
European Union Burden-sharing Agreement
Member state National target

(%)
Austria –13
Belgium –7.5
Denmark –21
Finland 0
France 0
Germany –21
Greece +25

Ireland +13
Italy –6.5
Luxembourg –28
Netherlands –6
Portugal +27
Spain +15
Sweden +4
United Kingdom –12.5
Total EU –8
‘Bubbles’ and ‘umbrellas’ brella’ group of trading countries under this article,
Article 4 of the Protocol allows parties to negotiate and it is likely that international trading will begin in
a side agreement, in which they pledge to fulfil their this way.
Kyoto ceilings jointly. This provision was important
in that it made it possible for the European Union to (iii) Non-permanent Emission Caps
negotiate on behalf of its 15 member states in Kyoto.
The emission ceiling shown in Table 1 for the Another concern about the emission limits in the
European Union is thus an aggregate ceiling. The Kyoto Protocol is that they are not permanent (as
European side agreement, establishing emission are the limits in the Montreal Protocol and the US
ceilings for individual member states, was negoti- sulphur-dioxide trading programme, for example).
ated in September 1998 and resulted in the burden- Emission limits for subsequent control periods will
sharing agreement shown in Table 2. be established by future conferences and codified in
future amendments; negotiations of the second
Under the terms of the Kyoto agreement, Europe is round of limits (that is, those that apply beginning in
thus treated as a ‘bubble’ (in the jargon of the US 2013) are required to begin by 2005, but Kyoto has
emissions-trading programme). As long as the total nothing more to say about these limits.
target for Europe is achieved, each member state is
considered also to be in compliance. However, This matters because many actions to reduce emis-
should the total target not be met, each member sions involve investments with very long lifetimes.
state is held individually accountable for meeting the Whether these investments will be worth making
targets it accepted in the side agreement. will depend on the magnitude of future limits. If one
believes that future limits will be very tight, then
Note that the concept need not be confined to long-term carbon-saving investments will appear
Europe. A number of countries (Australia, Canada, more attractive today. If one believes that future
Japan, New Zealand, Russia, Ukraine, and the limits will be slack, then costly carbon-saving invest-
United States) have discussed setting up an ‘um- ments will not pass the required hurdle.
32
S. Barrett
Strategy may also intrude. If a country invested But there is a deeper reason, too. For suppose that
more in abatement than needed just to meet its such a level could be identified. Then, if parties to the
target in the 2008–12 period, then this may only agreement pledged to ensure that this level was not
increase the emission reduction that it would have to exceeded, every party would have a strong incen-
meet in the next period. The reason is that, once the tive to withdraw from the agreement (or not to
costs of the investment have been sunk, the costs to accede to it in the first place). The reason is that, if
this country of reducing its emissions in the next a party withdrew and increased its emissions, the
period will be lower; its bargaining position will remaining parties would have to reduce their emis-
therefore have been compromised. Turning this sions to ensure that the aggregate concentration
argument around, a country might be able to nego- target continued to be met. In a sense, the with-
tiate an easier target for the next control period if it drawal would be rewarded. Similarly, if a country
invested less in reducing its abatement costs in the acceded to the agreement, the burden of meeting
first control period. the aggregate target would be spread more widely,

and, as a consequence, the original signatories
But it is easy to overstate this problem. Suppose would presumably be allowed to reduce their abate-
Kyoto had imposed permanent emission ceilings. ment levels—at the expense of the additional party
Then a different problem would arise: the parties to having to increase its abatement. Accession would
the protocol would question the credibility of the essentially be punished. An aggregate target thus
ceilings, knowing that the limits could always be exacerbates any incentives that may already exist
renegotiated. If the countries believed that the for countries to free ride. That Kyoto does not
future ceilings were too tight, they would ‘under- specify an aggregate target is a virtue.
invest’ in abatement. Of course, once they had done
so, the costs of meeting the original limits would be So, how should the targets reflect damages? Obvi-
higher, and the case for lowering these ceilings ously, if the concern were with limiting total dam-
would therefore be strengthened. The belief that the ages, then the effect would be the same as just
initial limits were ‘too tight’ would be self-fulfilling. described. However, suppose parties to the agree-
ment were concerned only with maximizing their
(iv) Arbitrary Emission Limits own collective pay-off (the difference between
their total benefit and cost of mitigation). Then the
Nordhaus and Boyer (1998, p. 17) question Kyoto’s incentives would be better aligned. If a country
choice of emission limits, noting that they do not withdrew from the agreement, the remaining parties
relate to ‘a particular goal for concentrations, tem- would reduce their abatement (since the aggregate
perature, or damages’. The targets certainly should marginal damage for the parties to the agreement
take account of damages (see especially section V); would fall with the withdrawal); the withdrawal
at the very least they should provide a benefit would be punished. If a country acceded, the coun-
(measured in terms of the damages avoided by the tries that were already parties to the agreement
mitigation) that exceeds the cost of meeting the would increase their abatement (since the aggre-
targets. But they should not take direct account of gate marginal damage for parties would increase);
concentrations or temperature (even though these the accession would be rewarded.10
will be linked to damages).
(v) Quantities vs Prices
One reason for this is that it is very hard to say by
how much emissions should be limited. For example, Setting quantitative targets may seem to be the
though the Framework Convention requires that obvious remedy, and it has been at the forefront of
concentrations be stabilized at ‘a level that would negotiations ever since the Toronto conference. But
prevent dangerous anthropogenic interference with it has problems.11 One problem is that the link
the climate system’, no one knows what this level is. between actions and outcomes, as measured in
10
This is the basic mechanism underlying the self-enforcing agreements studied in Barrett (1994).
11
Hahn (1998) summarizes a number of alternative prescriptions. See also Nordhaus (1998).
33
emissions relative to an historical base year, is Annex I countries reduce their emissions, compara-
tenuous. Carbon-dioxide emissions were 7 per cent tive advantage in the greenhouse-gas-intensive in-
lower in Britain in 1995 compared with 1990, even dustries will shift towards the non-Annex I coun-
though Britain has not adopted a radical policy for tries. This trade effect will be reinforced by the
reducing emissions. Similarly, emissions in Ger- workings of the energy market; as demand for the
many fell 12 per cent between 1990 and 1995. carbon-intensive fuels in the Annex I countries falls,
Emissions in Bulgaria, the Czech Republic, Estonia, world prices for these fuels will fall, and consump-
Hungary, Latvia, and Slovakia fell by even more— tion in the non-Annex I countries will therefore
by up to 50 per cent over this same period, without increase. Consequently, emissions outside the An-
any of these countries adopting radical climate nex I countries will increase; the environmental
change mitigation policies. By contrast, emissions in benefits of the agreement will be reduced. Poten-
all the countries that imposed carbon taxes in the tially, if leakage is strong enough, the agreement
early 1990s (Denmark, Finland, The Netherlands, would only succeed in redistributing global emis-

Norway, and Sweden) were 4–15 per cent higher sions. The effort to negotiate and implement the
in 1995 than in 1990. agreement would have been wasted.
Another problem is uncertainty. There is, of course, How significant a problem is ‘leakage’? The Clinton
great uncertainty about the magnitude of climate- Administration (1998, p. 72) maintains that, with
change damages. But there is uncertainty also about cost-effective implementation, the Protocol ‘would
the costs of climate-change mitigation, and in a likely have little impact on competitiveness’. Maybe
seminal paper Weitzman (1974) showed that the so. But if implementation is not cost-effective—and
latter kind of uncertainty can have important impli- as I have already explained it could be far from this
cations for the choice of policy instrument (emission mark—then the consequences could be different.
limit versus carbon tax). If a quantitative limit were Bernstein et al. (1998) find that leakage could be
fixed, marginal costs would be uncertain. If a tax significant: for every 100 tons of carbon abated by
were fixed, emission reductions would be uncertain. the Annex I countries, non-Annex I emissions could
Weitzman showed that the tax is superior if the rise 5–10 tons (global emissions would thus fall by
marginal cost curve is steep relative to the marginal only 90–95 tons). Manne and Richels (1998) and
benefit curve. Essentially, the tax ensures that Nordhaus and Boyer (1998) also predict significant
marginal costs and benefits do not differ by much. levels of leakage.
Pizer (1998) has calculated that taxes would be These levels may not appear high, but they will
much more efficient than quantity limits for climate- certainly be politically visible.12 Leakage would
change mitigation (in his simulations, the net benefits damage particular industries, and these will surely
to using the tax are five times the estimate for a lobby for protection. The Senate resolution drew
quantity control). A combination of policies can do attention to the problem, and the proposed EC
even better (Roberts and Spence, 1976), though carbon tax was modified partly to take account of
Pizer (1998) finds that a hybrid policy is unlikely to the concerns voiced by the energy-intensive indus-
improve much on the pure tax scheme in the case of tries about a possible loss in ‘competitiveness’. It is
climate change. The essential point is that, even if no surprise that unilateral carbon taxes within coun-
the Kyoto targets were met cost-effectively, an tries vary by sector, with industry—and especially
alternative policy that leaned more in the direction of the energy-intensive export industries—always pay-
controlling marginal costs directly (carbon taxes) ing the lowest amount. When the EU burden-
would be even better. sharing rule was being negotiated, a number of
countries (Austria, Denmark, The Netherlands,
(vi) Leakage Spain, and Finland) wanted to make meeting the
national targets conditional on the introduction of
Because participation in the Kyoto Protocol is EU-wide emissions-control measures. These coun-
not full, there is a potential for ‘leakage’. As the tries were concerned that, as they reduced emis-
12
Previous studies have shown that leakage could be more substantial (IPCC, 1996, ch. 11).
34
S. Barrett
sions, perhaps by imposing steep carbon taxes at However, the facts are open to a different interpre-
home, output in the sectors most highly taxed would tation: that
shift elsewhere within the Union. The Danish min-
ister said that, though he accepted that Denmark’s both the high rate of compliance and relative absence of
–21 per cent target was unconditional, Denmark enforcement threats are due not so much to the irrel-
evance of enforcement as to the fact that states are
would only be able to achieve –17 per cent without
avoiding deep cooperation—and the benefits it holds
EU-wide measures being adopted.13 whenever a prisoners’ dilemma situation exists—because
they are unwilling or unable to pay the costs of enforce-
This links up with a point made in the Introduction: ment (Downs et al., 1996, p. 387).
that concerns about leakage provide another reason
for wanting to encourage trading. In reducing the This last interpretation may seem cynical and un-
between-country difference in marginal costs, trad- convincing. After all, as we have seen, Kyoto does
ing reduces leakage. Trading therefore lowers costs strive to sustain ‘deep’ cooperation—a treaty that

and increases benefits. imposes a cost measured as a fraction of GDP can
hardly be described as ‘shallow’. But, then again,
Kyoto has not even entered into force yet, let alone
IV. COMPLIANCE ENFORCEMENT been implemented. So we cannot really choose
AND FREE-RIDER DETERRENCE between these different theories.
Assume the best: that enough countries ratify Kyoto Indeed, it would not even be sensible to choose
that it comes into force and that the flexible mecha- between them because neither quite gets to the
nisms in Kyoto allow abatement to be cost-effec- heart of the matter. The Chayeses consider the
tive. Then we can ask: Will the parties to Kyoto need to enforce compliance as being independent of
actually comply with the agreement? Will they stay the need to deter free-riding—something that they
within the limits prescribed by Tables 1 and 2? dismiss as being of little practical importance. Downs
et al., by contrast, conflate the two problems.
It is a remarkable fact that non-compliance with Compliance enforcement and free-rider deterrence
international agreements is extremely rare. And, are related problems and should be analysed jointly.
when it does occur, the reason is usually that the
deviant was for some reason unable to comply, It is important to note that customary law does not
rather than that it chose not to comply. require that states be parties to a treaty. Sovereignty
means that countries are free to choose to partici-
But why do parties comply? One reason is that they pate in a treaty or not as they please (Barrett, 1990).
are expected to by customary of international law. So if free-riding is to be deterred—if participation in
And it is obvious why custom demands compliance. a treaty is to be full—then some kind of treaty-based
If states could not be relied upon to act as they said mechanism must provide the right incentive. It must
they would act, then what would be the point of correct for the harmful incentives that otherwise
entering into agreements? condemn countries to the fate of the famous prison-
ers’ dilemma.
But does this mean that compliance is not a prob-
lem? If it does, then it should not matter that the Suppose that an agreement exists, that it consists of
Kyoto Protocol does not (yet) include any provisions a certain number of parties, and that it requires that
for punishing non-compliance. As Chayes and these parties undertake some action. The required
Chayes (1995, pp. 32–3) note, the authority to action (climate-change mitigation) is costly to the
impose sanctions ‘is rarely granted by treaty, rarely parties that undertake it, but provides a benefit that
used when granted, and likely to be ineffective when is shared by parties and non-parties alike (climate-
used’. So Kyoto’s failure to enforce compliance by change mitigation is a public good). So each party
sanctions may be an irrelevance. will have an incentive to withdraw from the agree-
13
‘EU States Agree Kyoto Emissions Limits’, ENDS Environment Daily, 17 June 1998, http://www.ends.co.uk/subscribers/
envdaily/articles/98061701.html.
35
ment, for in doing so each can gain more from what it would do if it withdrew from the agreement.
avoiding steep mitigation costs than it loses from its Hence, if every signatory is deterred from with-
own small slice of greenhouse-gas abatement. drawing, each also is deterred from not complying.
The binding constraint on international cooperation
If a party is to be deterred from withdrawing— is free-rider deterrence, not compliance enforce-
which it is entitled to do under international law14— ment. Once free-riding can be deterred, compliance
then it will need to be punished for withdrawing, and can be enforced free of charge.
punished severely. It will be up to the other parties
to the agreement to impose the punishment, but they The example of the Montreal Protocol is relevant
may be reluctant to do so. The reason is that it is very here. This agreement, which is phasing out the use
hard to punish a deviant without also harming one- of ozone-depleting chemicals world wide, is among
self. For example, suppose the punishment is that, in the great successes of international cooperation. It
the event of one country withdrawing (and there- is also often held up to be a model for future

fore cutting its abatement substantially), the other agreements. Like Kyoto, the Montreal Protocol did
parties reduce their mitigation. Then the countries not initially incorporate a mechanism for punishing
called upon to impose the punishment will be shoot- non-compliance; choice of such a mechanism was
ing themselves in the foot, so to speak. The punish- to be deferred to a future meeting of the parties. So
ment may not be credible. failure by Kyoto to include a mechanism for en-
forcement might seem not to matter. However,
Let us suppose, however, that a credible punishment there is a big difference between the two treaties.
can be found to deter some level of free-riding. Then The Montreal Protocol did offer incentives for
it can be shown that the same punishment can be countries to participate in the form of a trade
relied upon to enforce compliance (deter non-com- sanction between parties and non-parties in the
pliance); see Barrett (1998a). The reason is intui- substances controlled by the treaty and in products
tive. Suppose a party contemplates ‘cheating’ on containing these substances. And this device has
the agreement, perhaps by reducing its emissions by succeeded in making participation in the Montreal
less than required by the agreement. To be deterred Protocol virtually full.15 It has also been invoked to
from cheating, it must face a punishment, and the enforce compliance with the agreement.16 When
punishment must be sufficiently severe that the seen in this light, compliance enforcement is a
country is made better off by not cheating. The problem for Kyoto because the agreement does not
larger the deviation from compliance, the larger employ a mechanism to deter free-riding.
must be the punishment which deters non-compli-
ance. But the larger the required punishment, the Actually, the minimum participation clause may
larger will be the harm self-inflicted on the countries provide some assistance in deterring free-riding.
asked to impose it. If a punishment becomes too You can think of it this way. Suppose more than 55
large it will cease to be credible and non-compliance countries have ratified the treaty, and Annex I
will not be deterred. participation falls just a tiny bit short of the 55 per
cent minimum required for entry into force. Then, if
Recall, however, that I have assumed that there one more Annex I country ratifies, and so makes the
exists a credible punishment that can deter (further) minimum participation clause bind on all parties, it
withdrawals from the agreement. The worst harm will have a non-marginal effect on the behaviour of
that a signatory could do by not complying would be others—the other Annex I parties will now have to
for it to choose an emission profile that matched fulfil their obligations under the treaty. This might
14
The Kyoto Protocol allows a party to withdraw 3 years after the Protocol has entered into force for a party, upon giving 1
year’s notice.
15
A provision was also made for controlling trade in products made using these substances, but this was never implemented.
16
The biggest challenge to the Montreal Protocol came when Russia declared that it would not be able to comply by 1996. The
Implementation Committee threatened to invoke sanctions—and the combination of this threat and the sweetener of financial
assistance was enough to compel Russia into preparing a plan for eventual compliance. The carrot of financial assistance was
justified, by the way, since the original Montreal Protocol was negotiated by the Soviet Union in 1987, before its collapse. See
Barrett (1998b).
36
S. Barrett
just provide the incentive for the marginal ratifica- ing the agreement. To compound these problems,
tion, and push the treaty over the minimum partici- delay in implementing Kyoto will raise the costs of
pation threshold.17 sticking to the Kyoto timetable, and so increase the
incentives not to stick to this timetable.
However, this trick is not sure to work—and even
if it did succeed, it provides absolutely no incentive The solution to all these problems may seem obvi-
for successive accessions to the treaty. To see this, ous: invoke the kind of sanctions used by the Mon-
notice that the next country to ratify will not alter the treal Protocol. However, production of every good
behaviour of the existing parties one little bit. So why has implications for greenhouse-gas emissions.
should it accede? The Kyoto Protocol does not Should all trade between parties and non-parties be
provide any incentives for more than the minimum banned? The threat to do so would almost certainly
of participation. This is in sharp contrast to Mon- not be credible. Should trade in a select range of
treal, which provides ample incentives for full par- products be banned? That might be credible, but it

ticipation. might also threaten the stability of the multilateral
trading arrangements. The answers are not obvi-
Let us suppose, however, that Kyoto’s minimum ous. But perhaps the questions should be asked (I
participation level is met and that the agreement was told that the subject never came up in Kyoto).
enters into force. Could full implementation then be
relied upon? The answer is not obvious. Suppose
just one country foresees that it will fail to comply. V. SUMMARY
Then it could withdraw from the agreement, upon
giving sufficient notice, and so avoid having to If there is one lesson to draw from this analysis it is
deviate from the custom of compliance. Of course, this: the Kyoto Protocol must produce for its parties
its withdrawal would be penalized if it brought about a favourable benefit–cost ratio or else it will either
the collapse of the agreement, as required by the never enter into law or it will collapse.
minimum participation clause. But the other parties
may not want the agreement to collapse, even taking As I noted in the opening paragraph of this paper, the
as given this country’s withdrawal, perhaps be- overall reductions in emissions contained within
cause, having previously sunk money into abate- Kyoto probably could provide a benefit–cost ratio
ment investments, the cost of sticking with the for the world in excess of one. However, actually
agreement would be low. But if this is true—if a realizing this potential gain will not be easy. The
country cannot expect to be punished for deviating, overall level of abatement prescribed by Kyoto
then every party would have an incentive not to try would have to be achieved cost-effectively—and
very hard to comply with the agreement. this will require that abatement be undertaken in
non-Annex I countries. Participation by the non-
A more likely scenario is that a number of countries Annex I countries could potentially be achieved
will wait to undertake substantial investments in through the Clean Development Mechanism, but
abatement until others have already done so. The this would be sufficient only under the most favour-
risk is that, with everyone behaving in this way, the able of assumptions. It seems more likely that
policies and investments needed to implement Kyoto emission caps would also need to be negotiated for
will not be made. The Protocol seems to have the developing countries. Let me repeat here that
anticipated this problem, for it requires that every this does not imply that the non-Annex I countries
Annex I party demonstrate progress in achieving its would need to pay for this abatement themselves.
target by 2005. But this will not suffice. If enough of The reason for broadening participation is not to
these parties have made little progress, then none redistribute costs so much as to lower the total bill.
can be singled out for having acted unusually. There is an important precedent for this. The Mon-
Anyway, if no penalties can be applied, a lack of treal Protocol capped emissions of ozone-depleting
progress by all parties, or a large enough number of substances in developing countries, and these coun-
parties, would only provide a reason for renegotiat- tries did not have to pay to stay within these limits;
17
This is what I call a ‘linchpin’ equilibrium. See Barrett (1998a).
37
the ‘incremental costs’ of their compliance were riding and non-compliance—has not yet come into
paid for by the industrialized countries. view, but it may prove the harder climb.
Achieving a favourable benefit–cost ratio implies

not just that costs must be kept low, but also that POST SCRIPT
benefits must be kept high. As noted earlier, lower-
ing implementation costs will actually raise benefits After this paper was written, the parties to the
by lowering leakage. But there is another problem: Framework Convention met in Buenos Aires (in
one way of lowering costs is to approve CDM November 1998). The issues raised in this paper
transactions that may not ultimately yield reductions were not resolved at this meeting, but a Plan of
in net emissions (so-called paper trades). Shaving Action was agreed, with deadlines for finalizing the
costs in this way would ultimately ruin the agree- Protocol’s flexible mechanisms. For the first time,
ment. This is yet another reason why Kyoto should the issue of how compliance should be enforced

be revised to include emission caps for the develop- was raised, though to my knowledge no mechanism
ing countries. for enforcing compliance was proposed. At the
meeting, the United States became the 60th country
If these requirements can be met (and that is a big to sign the Kyoto agreement. Another small island
if), then the US Senate’s objections would fall away, state, Antigua and Barbuda, became the second to
and the Kyoto Protocol could then enter into force. ratify it. Argentina, which hosted the meeting, an-
The problems of non-compliance and free-riding nounced its intention to adopt an emission limit
would at the same time be eased. If the costs of voluntarily, and Kazakhstan said that it would join
participation were lowered (and the benefits in- the group of Annex I countries and accept, in the
creased), then the incentives to deviate in these words of the press release, a ‘legally binding target’
ways would be reduced. (adding more ‘hot air’?).
However, these incentives to deviate would not be These developments are to be welcomed, but the
eliminated by cost-effective abatement. Achieving fundamental problems raised in this paper remain.
a favourable benefit–cost ratio is only a necessary The press release issued at the start of the Buenos
condition for achieving global cooperation; it is not Aires talks concluded by noting that the agreement
sufficient (Barrett, 1994, 1998a). And it is not would not become legally binding until the minimum
obvious how the required sanctions could be made participation requirements had been met. ‘It is
credible. So Kyoto has two mountains to climb. The hoped,’ the statement reads, ‘that this will happen in
first—achieving a favourable benefit–cost ratio—is 2001.’ It is regrettable that we cannot anticipate
challenge enough. The second—deterring free- with more confidence an event of such importance.
REFERENCES
Barrett, S. (1990), ‘The Problem of Global Environmental Cooperation’, Oxford Review of Economic Policy, 6(1), 68–
79.
— (1992), ‘Reaching a CO2 Emission Limitation Agreement for the Community: Implications for Equity and Cost-
Effectiveness’, European Economy, Special Edition No. 1, 3–24.
— (1994), ‘Self-Enforcing International Environmental Agreements’, Oxford Economic Papers, 46, 878–94.
— (1998a), ‘A Theory of Full International Cooperation’, Journal of Theoretical Politics, forthcoming.
— (1998b), ‘Montreal vs. Kyoto: International Cooperation and the Global Environment’, in I. Kaul, I. Grunberg,
and M. A. Stern (eds), Global Public Goods: International Cooperation in the 21st Century, New York, Oxford
University Press, forthcoming.
Bernstein, P. M., Montgomery, W. D., and Rutherford, T. F. (1998), ‘Trade Impacts of Climate Policies: The MS-MRT
Model’, paper prepared for the Yale–NBER Workshop on International Trade and Climate Policy, Snowmass,
CO, 12 August.
Chayes, A. and Chayes, A. H. (1995), The New Sovereignty, Cambridge, MA, Harvard University Press.
38
S. Barrett
Clinton Administration (1998), ‘The Kyoto Protocol and the President’s Policies to Address Climate Change:
Administration Economic Analysis’, White House, Washington, DC, July.
Downs, G. W., Rocke, D. M., and Barsoom, P. N. (1996), ‘Is the Good News About Compliance Good News About
Cooperation?’, International Organization, 50, 379–406.
Grubb, M. (1995), ‘The Berlin Climate Conference: Outcome and Implications’, Briefing Paper No. 21, London, Royal
Institute of International Affairs.
Hahn, R. W. (1998), The Economics and Politics of Climate Change, Washington, DC, American Enterprise Institute
for Public Policy Research.
International Energy Agency (1992), Climate Change Policy Initiatives, Paris, OECD.
IPCC (1990), Climate Change: The IPCC Scientific Assessment.
— (1995), IPCC Second Assessment: Climate Change 1995, WMO and UNEP.
— (1996), Climate Change 1995: Economic and Social Dimensions of Climate Change, Cambridge, Cambridge
University Press.
Jacoby, H. D., Prinn, R. G., and Schmalensee, R. (1998), ‘Kyoto’s Unfinished Business’, Foreign Affairs, July/August.

McKibbin, W. J., Shackleton, R., and Wilcoxen, P. J. (1998), ‘What to Expect from an International System of Tradable
Permits for Carbon Emissions’, mimeo.
Manne, A. S., and Richels, R. G. (1998), ‘The Kyoto Protocol: A Cost-Effective Strategy for Meeting Environmental
Objectives?’, mimeo.
Nordhaus, W. D. (1998), ‘Is the Kyoto Protocol a Dead Duck? Are There Any Live Ducks Around? Comparison of
Alternative Global Tradable Emissions Regimes’, mimeo, Department of Economics, Yale University.
— Boyer, J. G. (1998), ‘Requiem for Kyoto: An Economic Analysis of the Kyoto Protocol’, paper prepared for the
Energy Modeling Forum meeting, Snowmass, Colorado, 10–11 August.
Pizer, W. A. (1998), ‘Prices vs. Quantities Revisited: The Case of Climate Change’, Resources for the Future Discussion
Paper 98–02.
Roberts, M. J., and Spence, M. (1976), ‘Effluent Charges and Licenses Under Uncertainty’, Journal of Environmental
Economics and Management, 5, 193–208.
Stavins, R. N. (1998a), ‘What Can We Learn from the Grand Policy Experiment? Lessons from SO2 Allowance Trading’,
Journal of Economic Perspectives, 12, 69–88.
— (1998b), ‘The Costs of Carbon Sequestration: A Revealed-Preference Approach’, American Economic Review,
forthcoming.
Weitzman, M. L. (1974), ‘Prices vs. Quantities’, Review of Economic Studies, 41, 477–91.
39

Scott Barrett - Political Economy of The Kyoto Protocol

Uploaded by

Copyright:

Available Formats

Scott Barrett - Political Economy of The Kyoto Protocol

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Scott Barrett - Political Economy of The Kyoto Protocol

Uploaded by

Copyright:

Available Formats

OXFORD REVIEW OF ECONOMIC POLICY, VOL. 14, NO.

POLITICAL ECONOMY OF THE KYOTO

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

I. INTRODUCTION must be met. And though global emissions will

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Annex I CO2 emissions Share of Annex I Kyoto target Projected emissions

United States 4,957,022 36.00 93 104

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Annex I CO2 emissions Share of Annex I Kyoto target Projected emissions

Croatia n.a. n.a. 95 n.a.

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

III. ANALYSIS OF THE KYOTO

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Emissions trading As long as these emission ceilings do not bite,

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Member state National target

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Achieving a favourable benefit–cost ratio implies

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

Downloaded from http://oxrep.oxfordjournals.org/ at Ondokuz Mayis University on November 10, 2014

You might also like