Leakage refers to the indirect impact that a targeted LULUCF activity in a certain place at a certain time has on carbon storage at another place or time. In spite of the linguistic implication of the term "leakage" that the flows involved are small and abnormal (as in water dripping from a leaky pipe), leakage may also include carbon flows that are large and predictable. The term "leakage" has generally been used in the context of project-based accounting to refer to impacts outside the project boundary (see Section 5.2.3), but leakage can also occur across other types of system boundaries. For example, action to reduce logging in Annex I Parties to reduce emissions reported under Articles 3.3 or 3.4 could result in leakage of benefits if a resulting reduction in timber supply led to increased deforestation in non-Annex I Parties. This effect would become much larger if tropical countries with high current rates of deforestation were to join Annex I. Similarly, the benefits of reduced logging could leak within Annex I Parties as a result of induced increases in activities that are not covered in the Kyoto accounting system, such as logging in areas that remain forested and fall outside the remit of Articles 3.3 and 3.4. The potential for leakage need not be a bar to undertaking LULUCF activities; unless these effects are either prevented or their magnitude quantified and deducted from the carbon benefits attributed to the activity, however, credit will be awarded in excess of the true benefits, and net GHG emissions may be higher than targeted levels. Although "leakage" has a negative connotation, in some cases positive effects can occur outside of a project area-as when a demonstration effect from a mitigation project leads to replication of the activities beyond the project boundaries (see Section 5.2.3). Leakage is not unique to LULUCF-the subject of this Special Report; it also can occur in energy-sector mitigation.
Leakage can be induced through several different mechanisms, such as activity displacement, demand displacement, and investment crowding. Activity displacement could occur, for example, if a silvicultural plantation or a forest reserve were created at a given location and the people who were formerly living at the site were displaced and continued to clear forest elsewhere. This kind of leakage can occur across international borders. An example is provided by the logging ban in Thailand instituted in 1989. Much of the logging activity formerly occurring in Thailand moved to neighboring Mynamar (Burma) and Kampuchea (Cambodia) (Leungaramsri and Malapetch, 1992). A similar effect is likely to result from the logging ban begun in 1998 in 18 provinces of the Peoples' Republic of China; the demand is likely to be satisfied by increased logging in other countries throughout Asia and beyond.
In addition to leakage at identifiable sites, a diffuse form of leakage occurs through global markets. Demand displacement occurs when a forest protection or management project reduces the supply of a marketed product, resulting in increased logging elsewhere to satisfy the demand for that product. Leakage can also result from supply displacement. For example, plantations that have been subsidized as global warming response options may have their carbon benefits negated when wood products derived from them simply replace products that would otherwise have come from elsewhere, or when output from subsidized plantations causes the price of plantation-produced wood to fall and unsubsidized plantations elsewhere consequently are cut and replaced with pasture or other low-biomass land uses (Fearnside, 1995). Leakage may also occur because of an investment crowding effect in which the targeted investment project (e.g., reforestation) crowds out the demand for other beneficial investments (e.g., replanting after harvesting) that are not targeted by projects.
Changes in national or international policies can lead to leakage-for example, when a government changes policy to lower the country's overall emissions but the emissions are displaced to other countries. This type of leakage is only a concern if the LULUCF activities are displaced to a country that does not have a full inventory of its emissions and a national cap (i.e., Annex I countries). For example, U.S. analysts assessed a scenario of 21-percent reduction of national forest harvests over the 2000-2040 period, which would lead to increases in wood imports from other countries. The net impact would depend on the relative efficiency of harvest of the imported wood compared to that of U.S. forest stands that the wood would replace (Andrasko, 1997). If the logging were shifted to Canada, the emissions might be captured in Canada's national inventory and consequent mitigation commitments, but if the logging were shifted to a country with no cap, carbon would be emitted but not accounted for.
Leakage is one effect of ARD activities that cannot be avoided through choices of definitions. Any large-scale establishment of new forests will create off-site effects, especially if the new forests generate a commercial wood supply. Additional wood supply on the world market will reduce the price of wood compared to prices without this additional supply. This reduction in wood prices will reduce the profitability of establishing or continuing forestry operations elsewhere (Adams et al., 1993; IEA GHG R&D, 1999). Thus, reducing the establishment of other new forests would have a negative consequence, although where it occurs within Annex B countries it will be accounted for within those countries' adjustments to their assigned amounts. Any such effects in non-Annex B countries would not be accounted and are not readily prevented. If the lower wood price leads to reduced harvest from existing forests, on the other hand, it would be a beneficial side effect that further increases the atmospheric benefit. If Article 3.4 were to encourage increases in carbon stocks in existing forests, this effect could lead to lower harvest levels and counteract leakage effects from afforestation/reforestation.
As the scale of accounting increases, leakage errors should become less important. For instance, if accounting is based on observed changes in carbon stock levels at the national level, the data will implicitly capture leakage between sources within the nation. To the extent that national-level estimates omit certain LULUCF activities within the nation as well as indirect effects across nations, however, some leakage may still be a factor.
Program-level actions are generally much less prone to leakage than narrow projects that are tightly circumscribed in space, time, and subject matter. For example, broad policy initiatives are more likely to influence deforestation rates than are direct actions of limited scope. In tropical countries with large areas of remaining forest, reduction of deforestation has much greater potential climate benefits than other land-use change and forestry options (Fearnside, 1995). Reduction of deforestation also captures many more complementary benefits, such as maintaining biodiversity, watersheds, and water cycling. On the other hand, quantifying the direct effects of a program is much more difficult than quantifying the direct effects of more discrete activities. A probabilistic approach would be needed to compute the expected value of different options; under this approach, one would need to multiply the value associated with each outcome by the probability that the outcome will occur (Raiffa, 1968). For example, because of the great difference in potential benefits, investing in deforestation avoidance rather than relatively safe plantation silviculture options can be advantageous even in the face of a low probability of success for deforestation avoidance (Fearnside, 1999b). Optimal approaches are likely to include a mix of broad policy reforms and site-specific activities.
If accounting is to be adjusted for leakage, estimates would be needed of the magnitude of carbon benefits that are lost by each possible mechanism, and the carbon credit would have to be reduced accordingly. Because uncertainty in estimates of leakage magnitudes inevitably would be present, a further downward adjustment in carbon credit would be needed to assure a given certainty of achievement.
Global climate benefits may be reduced not only through leakage but also by other forms of project failure. The adjustments that would be needed to accurately represent net GHG benefits are similar regardless of the origin of the failure. Because the probability of success varies greatly among global warming response options, an adjustment of credit for these probabilities would be necessary to assure valid comparisons of the benefits for global climate associated with each option. Such adjustment is sometimes called "discounting," but we prefer to reserve this term for its traditional use as a time-preference weighting mechanism.
CDM projects that are undertaken to avoid tropical deforestation would face problems of minimizing, quantifying, and adjusting for leakage, as well as difficulties in establishing additionality and the possibility of reductions in credit resulting from uncertainties regarding the without-project baseline and the attribution of project effects. The large potential carbon and collateral benefits of avoiding tropical deforestation explain the high priority being given to achieving continued progress in addressing these matters. Many of these issues, including leakage, would cease to pose problems for crediting avoided deforestation, however, if tropical forest countries (other than Australia) were to join Annex I, thereby gaining access to emissions trading under Article 17, with the guarantee that deforestation reduction would be accounted for and potentially producing salable credits (Fearnside, 1999c).
Other reports in this collection