where land and labour markets are not as well developed or when a public administration is not well established, the costs of administration and information gathering might rise to a much higher level, even surpassing the financial outlays [OUTLAYS FOR WHAT?]. In the context of developing countries where the pressure on land comes from agricultural purposes or from wood exports, these transaction costs should, theoretically, encompass policies required to slacken this pressure. This requirement could narrow the range of feasible low-cost carbon sequestration opportunities.

While some carbon sink cost studies have assumed that carbon sequestration project lands would be taken out of agricultural or timber production permanently (Nordhaus, 1991; Richards et al., 1993), most have allowed for some kind of derivative benefit in the form of forestry products such as pulpwood, timber, firewood, and biomass energy. Such derivative benefits raise several additional issues. First, harvesting of forestry products suggests the need to modify the flows of carbon associated with the project to reflect the removal of carbon from the site. This in turn raises the question of the rate of release of the carbon back to the atmosphere after harvest. Second, in a cost-effectiveness study, the costs of the project should be reduced to reflect the noncarbon benefits of the wood and agricultural products. Furthermore, to the extent that sequestration timber drives down the price of timber products, additional markets will develop for wood substitutes for energy-intensive materials such as concrete, aluminum, and steel, thus yielding further reductions in emissions.

As the discussion in Box 8.1 illustrates, harvesting can have a significant impact on the carbon benefits of a project. Further, the measure of that impact depends very much on the choice of summary statistics. At the same time, the economic benefits of timber harvesting can be significant. Studies that do not quantify either of these two effects will overstate both the costs of carbon sink projects and the carbon benefits. While the effect on carbon costs of ignoring timber harvesting is indeterminate, it is likely that inclusion of forestry products in the analysis would generally lower unit cost. Studies that include the effects of harvesting on carbon flows but do not incorporate its economic benefits will almost certainly overstate the unit costs of carbon sequestration. At the extreme, some forestry practices may pay for themselves in the form of forestry products and provide the carbon benefits as a costless (i.e., no-regrets) bonus-(often referred to as "no regrets" strategies). For example, Xu (1994) suggests that there may be negative costs associated with some carbon sequestration practices in China. Conversely, those studies that only consider the benefits of forestry products but do not adjust the carbon flows to reflect increased releases of carbon back to the atmosphere will understate the costs of carbon sequestration.

29

It is apparent that, as with energy efficiency, there are opportunities in the carbon sink area to achieve double dividends. These become evident under an analysis that provides full accounting for the wood and agricultural product benefits of projects as well as for the less tangible benefits such as habitat and watershed preservation, improved

local self-reliance, and soil erosion control. However, full accounting must also include those factors that tend to increase social costs relative to financial costs, such as the effect of removing nonmarginal quantities of land from agricultural production, transaction costs associated with establishing new land use patterns, administrative costs of implementing a large-scale carbon sequestration programme, and decreases in carbon benefits associated with timber harvesting. Perhaps the most important factor, one commonly ignored by carbon sink cost studies, is the potential for "leakage" of the carbon sequestration gains. If the selective subsidization of tree planting and other forestry practices creates a new supply of timber, owners of existing forests may try to avoid competition in the timber market by accelerating harvest of their stocks, decreasing the amount of postharvest replanting, and avoiding expansion of the holdings.

1. However, in most of the earlier modelling work, "long-term generally meant a time horizon of 10-20 years, not the 100-plus horizons used in many of the current climate change scenarios.

2. For a discussion of the wide range of views on future energy demand in the early 1980s, see Caputo (1984) and Thompson (1984).

3. For discussion of the sometimes vexed relationship between energy modellers and energy policymakers, see Robinson (1992).

4. See, for example, the various national case studies in Baumgartner and Midttun (1987).

5. Early work in this area is summarized in Little and Mirrless (1974) or Squirre and Van der Tak (1976). A good illustration of the interest in making a distinction between market prices and factor costs is given in the debates around the biomass ethanol program in Brazil, where the cost of ethanol is far lower if one utilize some form of social costs (Nastari, 1991).

6. Most discussions of the issue of "no-regrets" potential have centred around the first category of positive side effects described here: whether there exist "negative cost" measures such as some types of energy efficiency programmes. But whatever positive side-effects are included, the "no-regrets' concept should not be taken to imply that undertaking all such measures will guarantee no regrets with regard to the effects of climate change, if these effects ultimately are proven to be very significant. A better term might be "worth doing anyway'

measures.

7. See also the discussion in Chapter 7.

8. See section 6.7 of Chapter 6 for a discussion of some estimates of the magnitude of the environmental double dividend.

9. For a more general presentation of the different meanings of concepts such as "no regrets" and "double dividend," see Goulder (1994).

10. Of course it would also be possible to move from O to a point B' above and to the left of point B (increasing economic growth and also increasing emissions). This means that the economic surplus gained thanks to the removal of inefficiencies (i.c., moving from O to curve F) will be devoted to improving environmental quality only if there is a collective preference and political will to do so. It could also be possible to move to a point A' below and to the right of A (reducing both emissions and economic activity) if the surplus is devoted to very high investments with a low return and a very low efficiency in terms of environmental quality improvement. This could occur in the case of misallocation of efforts for a given level of concern for environmental qualtity.

11. Another counterargument acknowledges that a suboptimal baseline may be the most realistic assumption, even over the long run, but suggests that the cost of greenhouse gas abatement measures should be calculated net of the effect of any measures taken to move the economy towards the production frontier (e.g., the effect of fiscal distortions are assumed to be removed before calculating the cost of a carbon tax). By suggesting that no economic double dividends should be included in the cost of abatement, such arguments reduce the estimate of cost-effective abatement potential, but they also imply that the adoption of a more economically efficient baseline scenario may result in lower emissions for reasons unrelated to climate policies.

12. Of course, all this assumes a static production frontier. In reality, as a result of technological change and other factors, the frontier moves, usually to the right, over time. Bottom-up analyses often compare a point below the current production frontier with athe future production frontier. Bottom-up analysts typically also argue that

due to market imperfections, actual future production is likely to lie below the future production frontier. Since bottom-up analyses tend to caculate the cost savings due to the adoption of future technologies relative to existing ones, it is not surprising that they usually show net savings. Top-down analyses tend to focus on the costs of moving from the current to a future production frontier. Since that entails investment, it is not surprising that the costs are always positive.

13. The same debate surrounds the analysis of fiscal reforms linked to a carbon (or energy) tax. Some analysts consider that it is more legitimate to separate the gains from reducing existing fiscal distortions from the incremental impact of a carbon tax. Others consider that both such effects should be accredited to the carbon tax, since reducing fiscal distortions is associated with, and may be made more politically palatable by, the imposition of the carbon tax. This latter position, which can reduce or more than offset the costs of the carbon tax, is the position taken in most of the empirical modelling work to date; the former assumption is more typical of more theroretical analyses.

14. This is a logical identity and does not imply the independence of these three terms.

15. To give a simple example, if there were two countries in the world and the cost of emission reduction was always twice as expensive in country A as in B, then it would be cheaper to use up all the reduction potential in country B before reducing emissions in A.

16. In 1989, the range for Europe and North America was 20-35% (World Resources Institute, 1993).

17. Alternatively, countries might adopt strategies encouraging the development of very energy-efficient urban design.

18. For discussion and examples of such approaches see Jantsch, E. (1967); Harman, W. (1976); Gault, et al. (1987); Gal and Fric (1987); Glimel and Laestadius (1987); Godet (1986); and Robinson (1988).

19. See the country case studies in Baumgartner and Middtun (1987). See also Caputo (1984). For a discussion of the bases of disagreement, see Robinson (1982a)

20. For examples of more recent bottom-up work, see Goldemberg et al. (1988) and Johansson et al. (1993). 21. In Chapter 9 we provide the findings of some attempts to provide such comparisons.

22. Since the former baseline already incorporates such improvements, the cost per unit of further reduction is likely to be higher than it would be relative to a baseline without such improvements.

23. For a recent example, which characterizes individual bottom-up and top-down models, see Grubb et al. (1993).

24..For example, Environmental Protection Agency (1990); Edmonds, et al. (1986); Manne and Schrattenholzer (1993); and Johansson et al. (1993).

25. The term "backcasting" is also used in the economic modelling literature to refer to the process of simulating a model over historical time in order to compare such a historical "projection" with actual historical data. 26. For a discussion of the implications of backcasting approaches, see Gault, et al. (1987) and Robinson (1991b). 27. The SIC (Standard Industrial Classification) is an internationally recognized classification system for industrial activity. The two-digit SIC level is a very [highly] aggregated level that breaks industrial activity into less than

ten sectors.

28. Apparently, the only carbon sequestration cost study to employ an econometric approach (in contrast to the least-cost analysis of most other studies) is currently being prepared by Robert Stavins of the Kennedy School, Harvard University, USA. The econometric analysis incorporates not only data from direct financial cost, but accounts for behavioural considerations regarding how landowners respond to economic incentives.

29. Of course, as in the case of energy-related emissions, the assertion that opportunities exist to enhance sinks at a negative cost raises the obvious question of why these activities are not already being undertaken if their costs truly are negative.