and also within the group of developing countries. Some of this variation, however, reflects differences in how the underlying variables have been measured.

Because reductions in national energy intensities are related to structural changes in national economies, the growth of the secondary sectors in developing countries may give a biased view of their energy efficiency improvement results. In most industrial countries, in contrast, a trend towards "dematerialization" (i.e., a shift away from the highly energy-intensive secondary towards the less energyintensive tertiary sector) has favoured lower energy intensities.

There is a broad consensus in the literature in favour of efficiency improvement, because it is seen as directly beneficial irrespective of any impacts on greenhouse warming and because it has significant scope for negative net cost (i.e., no-regret) applications.

Because the end use phase is the least efficient part of an energy system, improvements in this area would produce the greatest benefits. The potential for improvements in production seem especially promising, especially in the power production, transportation, steel and cement production, and residential sectors. [THIS IS A BIT CONFUSING. IS POWER PRODUCTION AN END USE? WHAT DOES THE TERM "PRODUCTION" COVER?] The potential in the developing countries is roughly similar in magnitude to that in industrialized countries. By contrast, energy conservation may be achieved somewhat more easily in the industrialized countries.

Optimism about the scope for no-regret options with respect to energy efficiency varies considerably and depends to a large extent on the discount rate that is employed. Revealed consumer discount rates for household investments can be very high indeed. Similarly, in developing countries a lack of access to information and limited human capacity [WHAT IS MEANT BY HUMAN CAPACITY?] and financial resources may cause the revealed time preference to be much higher than commercial interest rates.

The potential for energy savings is estimated at 10-40% for production and 10-50% for residential use. However, to achieve such results, institutional and information factors are crucial. So too is the degree to which the option may help in deriving other environmental benefits.

With respect to fossil fuel switching, relatively little information about costs is available, although it is recognized that fossil fuels will remain the dominant energy source for several decades yet. Estimates of the costs of switching vary to

a large extent, depending on the type of measure, the fraction of natural gas lost to the atmosphere from leakage during production and distribution, and the opportunity costs of the option (which depend to a large extent on the availability of, for instance, coal reserves).

These opportunity costs may be particularly large in populous countries with massive coal reserves, such as China and India. In fact, in developing countries growth may even result in a transition from less carbon-intensive biomass to more carbon-intensive fossil fuels.

Renewable energy technologies may be sustainable with respect to energy inputs but may not always be socially and environmentally benign in other respects. This is particularly so in the case of large-scale applications (for example, of major hydro or biomass projects) in developing countries.

The technical potential of the renewable options not currently utilized varies from 50% for biomass to 75% for hydro to several thousand per cent for wind. Many renewable technologies, however, tend to be site-specific (i.e., their application is limited to a finite number of specific sites). Other problems include potential environmental risks, technological readiness, and cost-effectiveness.

.Though some renewable options are almost mature, others are still in the demonstration stage. Practicable potentials therefore vary to a large extent, although much will depend on the costs of the various options.

Cost estimates diverge widely, mainly due to the time horizon adopted, the discount rate chosen, and the capacity and useful lifetime assumed. Moreover, costs are strongly influenced by site-specificity, variability of supply, and the form of final energy delivered. Other aspects that influence cost behaviours are learning effects, economies of scale, and the need for immediate storage or transport of the energy generated.

The promise of renewables lies mainly in their large potential and modest price on the spot. These factors are particularly relevant for developing countries, which, by using local renewables, could reduce their dependence on imported fossil fuels. Local communities could benefit significantly from small-scale applications and their net positive side effects.

In view of these considerations, the future role of renewables is hard to predict precisely; the share of renewables in the 2020 energy mix will, however, probably not exceed 25%.

Nuclear energy technology is long past the demonstration stage, but the issue of the safe storage of nuclear waste remains unresolved. Because of their long design and construction time (10-15 years) and the enormous investment costs of nuclear power plants, the nuclear option is also rather inflexible.

In view of the waste disposal problem and the consequent lack of public support, the share of nuclear energy in total energy use is expected to increase only to a limited extent during the coming decades.

Capture and disposal have potential in cases where a switch from coal to other fossil fuels is difficult for one reason or another. Some technologies already exist; others are being developed.

The disposal option is ultimately limited not only for technical reasons but also because not all forms of disposal can permanently prevent the reentry of carbon into the atmosphere. This is irrespective of the way in which disposal would take place [THIS SENTENCE APPEARS TO CONTRADICT THE PREVIOUS SENTENCE, WHICH IMPLIES THAT SOME FORMS OF DISPOSAL WOULD PROVIDE PERMANENT SEQUESTRATION.] The practicability of this option is still a matter of discussion, because in some types of disposal (e.g., in aquifers or oceans) environmental uncertainties are unknownimpacts are uncertain.

The scope of forestry options is determined by the large expected potential, modest costs, low risk, and positive side effects. However, there is still a large amount of uncertainty with respect to the net carbon release from deforestation and land use changes on the one hand and the long-term carbon absorption capacity of afforestation efforts on the other. Basically, forestry measures, like removal options, are to be seen as an intermediate response policy.

Uncertainties in assessments of the global potential for halting or slowing deforestation and for reforestation are linked to the extent of human encroachment into the forests, the area available for forestry measures, and the annual and cumulative carbon uptake per hectare.

Mitigation policies using forests are generally considered relatively costeffective, especially if applied in developing countries. With the costs of afforestation, much depends on whether one assumes that the forests can be exploited sustainably or, instead, should be left alone to mature and on the

Halting or slowing deforestation is probably one of the most urgent and cost-effective options. However, social, political, and infrastructural barriers may restrict this option as well as the scope of reforestation.

Estimates of cost-effectiveness of forestry measures depend strongly on whether one takes a static or dynamic point of view. There is a clear tendency to focus increasingly on cost functions rather than point estimates; the former approach seems clearly more relevant in the case of large areas. Moreover, the cost assessment methodology has been increasingly refined (for example, by the inclusion of discounting procedures). Cost estimates, which are now probably more realistic, tend to fall within a range of $30-$60/tC for large annual uptakes.

With respect to methane, the emission data available reveal wide discrepancies between various regions. Information about methane leakage and distribution is also rather scanty, and some of it is unreliable. The same applies to information about the costs of methane control options.

Information about the cost functions of the various mitigation options is still weak. That is because the functions are not only time-specific but also region- and context-specific. This [WHAT DOES "THIS" REFER TO?] also relates to the puzzling fact [WHY PUZZLING?] that the scope for no-regret options seems to be significant, especially in developing countries. This apparent scope is most likely due to the high actual time preference rates, lack of information, and [limitations of?] human capacity. All this and the different assessment perspectives mentioned earlier may explain why virtually no studies exist in which the optimal mix of options is designed on the basis of their underlying cost functions and feasibility.

The few studies of this kind that have been done provide only tentative results but do indicate - given present knowledge about the cost functions of the various options that the pure application of the cost minimization principle would require a significant share (probably more than half) of the emission reduction targets to be achieved via the application of options outside the OECD area. In addition, in terms of the size of the emission reduction, energy conservation and efficiency improvements and the forestry option seem to provide the largest potential from a cost minimization point of view. The potential of the forestry option is widely debated, however, because of the limitations of net absorption in time and because much depends on forest exploitation and local acceptance.

To illustrate the former [i.e. THE POINT ABOUT APPLICATION OF OPTIONS OUTSIDE THE OECD?], the result of a (LP-based [LP=?]) cost minimization simulation using the available cost-function information in terms of eest-functions-disaggregated by region is presented in Table 1 [IS THIS A NEW TABLE FOR THIS SUMMARY?] for a predetermined emission reduction target of 2.4 GtC. In view of the tentative and uncertain character of the underlying data, the outcomes can only be seen as an illustration of what an optimal policy mix might be (recognizing that marginal costs per option per region generally tend to increase to the point where they eventually become prohibitive). Obviously technological or political breakthroughs may significantly affect the optimal mix.

Adaptation options can be surveyed in many ways. One is to consider what should be adapted to and how it should be done. No systematic cost data on the various adaptation options are available, although information about land protection costs against flooding and sea level rise is rapidly increasing. Many efforts are now underway, however, to reduce the vulnerability of agricultural production to climate change through adaptation policies. Especially in developing countries there is an urgent need for both more information and a better infrastructure for the actual implementation of adaptation techniques.

Finally, the point has to be made that when it comes to the introduction and application of the various options, the developing countries occupy a special position. That is because the application and acceptance of these options often crucially depends on the international transfer of technologies as well as the countries' own local institutions and abilities to build their human capacity. Therefore, the conditions needed to ensure the success of these processes, such as joint implementation and technology transfers from developed to developing countries, deserve a high priority on the academic research agenda.