Cover Image
close this bookCarbon Counts: Estimating Climate Change Mitigation in Forestry Projects (WRI, 1997, 32 pages)
close this folderII. Leakage
View the document(introduction...)
View the documentCauses of Leakage
View the documentLeakage Index
View the documentApplication of the Index


Leakage is defined here as the unexpected loss of estimated net carbon sequestered. In some cases leakage may be positive - that is, more net carbon reductions were achieved than was expected, but it is the loss of greenhouse gases that most concern us. Leakage can be the result of incorrectly estimating the project's impact or of unexpected effects such as a population increase in the project region, during the course of a project. Therefore, correctly estimating a project's impact and designing projects that avoid leakage are critical.

Figure 1 illustrates the land-use dynamics and potential leakage over time in a project region characterized by subsistence agriculture, where population growth and demand for agricultural land drive deforestation. As population grows, marginal upland is brought into agricultural production. As this land eventually degrades into pasture, people cultivate increasingly higher, and more easily degraded slopes. Each of the three scenarios depicted in Figure 1 features permanent agriculture on the lowland areas and a combination of cattle grazing and farming in the upland areas. The project goals are to move from scenario 1 (the baseline) to scenario 3, and sequester carbon by relieving pressure on the forest through the introduction of higher productivity agroforestry and tree plantations, which would provide fuelwood. However, as shown by scenario 2, leakage occurs. Agroforestry is more productive than the pasture and agriculture it replaces, but the movement of cattle to upland areas results in forest conversion and degradation, albeit at a slower pace than indicated in the baseline. Ideally, leakage can be anticipated, or mitigated, by the measures listed in scenario 3-halting encroachment, allowing sustainable use of forests, and incorporating cattle into the woodlots via silvopastoral systems.

In response in part to uncertainties regarding correct net carbon calculation, investors are focussing on simple afforestation projects, because their impacts appear to be easier to quantify than those of other types of forestry carbon sequestration projects. In fact, tree planting is the only forestry activity the U.S. Department of Energy defines as a "standard" rather than "reporter-designed" project, meaning that enough credible data has been assembled to estimate the project's carbon benefits. (6) Of the greenhouse gas mitigation projects reported to the Department of Energy, tree planting projects are the most popular. (7) The carbon storage portfolio sponsored by the Netherlands' state utility consists entirely of tree planting on degraded lands. (8)

The popularity of tree planting may also be explained in part by the notion that growing biomass accumulates carbon while mature forests are merely stable. However, these "stable", mature forests have usually accumulated much more carbon in biomass than tree plantations are likely to, as Table 1 illustrates. The biomass accumulation of the mature forests in the northwestern United States, Malaysia's Sabah region, and Cameroon represents only carbon in living biomass; the total carbon of tree plantations with rotations varying by species, represents projected accumulations of carbon in living biomass and forest products after 300 years. (9)

Table 1 compellingly illustrates that tree plantations are less effective than natural forests at storing carbon, even when carbon storage in wood products are included in the estimate. Furthermore, a recent study of a tropical rain forest in Brazil indicates that undisturbed forests are continually sequestering carbon. The study estimated that the forest sequestered one metric ton of carbon per hectare per year. (10) This finding underscores the greenhouse gas benefits of projects that prevent deforestation.

It is important to note that when the authors in this publication refer to projects that prevent deforestation, they are not necessarily referring to forest preservation projects that set aside forestland in a park or protected area, which can have significant leakage problems. Rather they are referring to those projects that successfully prevent deforestation by addressing underlying land-use dynamics and demands on resources in and around the project site. Satisfying demands for these resources can, and often does, involve tree planting, as noted later in the case studies.

Figure 1. - Expanding Agricultural Frontier

1. Baseline

2. With Project (leakage)

3. With Project (no leakage)

Table 1. Comparison of Potential Biomass

Forest Type

Tons of Carbon per Hectare

Mature Douglas Fir, Northwestern United States

611 (1)

Mature Closed Forest, Sabah, Malaysia

348 (2)

Mature Primary Moist Forest, Cameroon

279 (3)

Industrial Black Locust Plantation, Europe


Industrial Slash Pine Plantation, Brazil


Afforestation of Tropical Wasteland, Borneo


Industrial Poplar Plantation, Europe


1. Mark E. Harmon, William K. Ferrell, and Jerry F. Franklin, "Effects on Carbon Storage of Conversion of Old-Growth Forests to Young Forests," Science, 9 February 1990, 699.

2. Francis E. Putz and Michelle Pinard, "Reduced Impact Logging as a Carbon Offset Method," Conservation Biology, vol. 7 no. 4, December 1993, 755-757.

3. Sandra Brown, Andrew J.R. Gillespie, and Ariel E. Lugo, "Biomass Estimation Methods for Tropical Forests with Applications to Forest Inventory Data," Forest Science, vol. 35, no. 4, December 1989, 895.

The urge to pursue simple projects in an effort to avoid difficult issues is understandable but unnecessary. It is possible to evaluate projects, identify conditions likely to result in leakage, determine factors contributing to this problem, and recommend actions to avoid or at a minimum account for it.

The analysis of specific carbon sequestration projects indicates that leakage can and should be incorporated into project design and that projects addressing the drivers of land use-change will maximize project benefits, reduce risks and costs, and minimize the potential for leakage.

Causes of Leakage

A project may fail to meet its carbon sequestration target because of unforeseen circumstances that are beyond the control of project participants; improperly defined key parameters, such as time horizon, project boundaries, or baselines; or inappropriate project design and activities in light of the land-use patterns in and around the project site.

Unforeseen Circumstances

Extreme weather, political instability, climate change, pests, disease, or fire are unforeseen circumstances that project designers and implementors cannot control. Such circumstances can be dealt with at two levels: the individual project level and a broader multi-project, national, or international level. Project failure at the macro level can be guarded against in at least three ways: creating a diverse portfolio of projects that hedge against individual project failures, pinning carbon reduction credits to national baselines, or creating insurance funds that would provide funding for a replacement project should one fail. (11) Each of these options would mitigate the consequences of an individual project failure.

Of the three options, the portfolio strategy is the only one currently being used. To sequester nearly 100 million tons of carbon, FACE, (the financing arm of the Netherlands' state utility), has set up projects in seven countries. FACE has sufficiently diversified so that the failure of one project will not seriously hamper its carbon mitigation effort. Utilities that do not have the resources to invest in more than one forestry project can still reduce the risks of an individual project failure by working in concert. In the United States, for example, 40 utilities have jointly invested in the nonprofit UtiliTree Carbon Company to provide more than $2 million for five or six forestry-based projects. (12)

The portfolio concept can be applied not just to the number but also to the types of projects. Because natural forests are usually adapted to naturally occurring events typical to their region, projects that maintain these forests may therefore be more resilient than projects focussing solely on tree plantations, which tend to be more susceptible to drought, fire, and disease. (13)

Our project review indicated some guidelines for dealing with these and other unforeseen circumstances at the individual project level. One guideline is to provide benefits to the people living in the project area Projects that fail to provide local benefits can solidify political opposition and increase the projects' risk of failure.

Conversely, projects that provide such benefits can create incentives for local people to overcome risks and even expand projects' impacts.

The CARE/Guatemala project, which increased fuelwood availability and agricultural productivity by providing trees through CARE-sponsored tree nurseries, has persisted during years of political strife and uncertainty because it involved local people as stakeholders in its success. Moreover, these people have adopted the project's techniques in areas beyond its boundaries by setting up their own tree nurseries, thereby potentially increasing the amount of carbon sequestered (sometimes called positive leakage) and providing other ancillary benefits.

Other projects have used different methods to enhance their success. The RUSAFOR project, for example, contains both a stick and a carrot. RUSAFOR has built loss provisions into its contract with the Russian Forest Service, stipulating that the Service will replant trees if the original ones do not survive. Also, the project splits any future carbon credits evenly between the United States and Russia.

In conclusion, the mere existence of risk from unforeseen circumstances is not a sufficiently compelling reason to exclude forestry and land-use projects from a joint implementation program. The strategies outlined above can reduce the risk of a project's failure to an acceptable level. Furthermore, risk is not endemic to forestry projects, but is also a concern for energy-based, greenhouse gas offset projects. To date, no forestry or land-use projects to sequester carbon have failed.

Improperly Defined Key Parameters

One of the challenges of avoiding leakage is to correctly define a project's three key parameters. One is the baseline - what would happen in the project's absence. It is the foundation for estimating the project's net carbon sequestration benefits. The second parameter is the appropriate project lifetime. And the third parameter is the project's boundaries which may differ from the project's physical boundaries, the area where carbon sequestration activities are directly implemented. The impact of these activities may extend beyond the physical boundaries. The Leakage Index, presented in a later section, describes the various drivers of land-use change and project components that help to define a project's boundaries. These three parameters are particularly important in helping project designers more accurately estimate carbon benefits, discover potential sources of leakage, and design projects to minimize or eliminate this problem. (14)


Net carbon sequestration is difficult to estimate, in part because of the conjectural nature of baseline projections. The case studies suggest a few guidelines for constructing reasonable projections.

First, estimates of what would happen in the project's absence should be based on existing forest trends and an identifiable cause. In the Krkonose project, in the Czech Republic, such estimates were aided by the existence of historic data on the Krkonose forest's decline and a clear understanding of the decline's cause - acid rain resulting from emissions from nearby power plants. The CARFIX project in Costa Rica used LANDSAT data collected during a 5-year interval to refine its baseline projections.

Second, barriers to positive change should be analyzed to strengthen the case for a likely baseline scenario. The point of this action is to determine if the social benefits offered by forestry projects could be realized without outside intervention. Although agroforestry or sustainable forestry projects generally provide environmental and social benefits, small landowners, who typically have little access to capital, may be unable to wait for these benefits because they need to support themselves today. (15) Both the original CARE/Guatemala and CARFIX projects provided interim income until the projects' benefits could be realized. Furthermore, without such external funding, there would be no incentives to produce nonmarket goods such as sequestered carbon.

Third, reforestation projects should include evidence of barriers to natural forest regeneration such as intensive cropping, pollution, soil degradation, perverse forest policies, pests, or fire. Because forests eventually rebound from disruption, project planners must make a compelling case that the area would not regenerate naturally. Baseline assumptions remain a thorny issue for reforestation and afforestation projects.

Project Lifetime

Picking an appropriate lifetime for a project is also an important parameter in developing an accurate estimate of expected future net carbon benefits. Two issues are associated with choosing an appropriate time horizon. The first is deciding which time horizon most accurately measures the effects of project activities, and therefore most accurately estimates the net reduction of greenhouse gases. Different time horizons will yield different net estimates. The second is ensuring that the project continues long enough to mitigate the global warming potential of greenhouse gas emissions but not so long that the project becomes unreasonable in terms of risk and monitoring. The following three options address each of the above issues to varying degrees.

Under the first option, referred to as "lifetime of an emitting activity," the project would be linked to the lifetime of a power plant's operations or other emitting activities that the project was designed to mitigate. The CARE/Guatemala project, for example was linked to the 35-year lifetime of the AES-power plant it was offsetting. However, a 35-year time horizon may be surpassed in duration by the entire lifetime of certain carbon forestry projects, such as RUSAFOR, which is situated in a slow-growing boreal zone. (16) One advantage of such a time horizon is that the carbon emitter, who is typically the project investor, can play a monitoring role, as is the case with CARE/Guatemala.

Under the second option, referred to as "lifetime of a project," only the carbon sequestered for the exact duration of project activities would be estimated. In this case the time horizon would vary significantly from project to project as, for example, from 3 years for the logging project in Malaysia to 60 years for RUSAFOR. However, only counting carbon for the duration of project activities, which do not include monitoring, carries the risk of miscalculating project impacts. As Figure 2 illustrates, the potential differences in net carbon sequestration estimates can be large, depending on the time period used for calculation. If estimates are calculated only for the first 5 years, the difference between the estimate for conventionally logged land and that for reduced-impact logged land will be greater than under a longer time frame. Also, if the conventionally logged land regenerates, this difference will decrease.

Under the third option, referred to as, "lifetime of carbon dioxide in the atmosphere," the project's lifetime is estimated to be 50-200 years, the length of time that carbon dioxide persists in the atmosphere. (17) However, assuming continued carbon sequestration over a time horizon reaching the upper end of this range is unreasonable given the impossibility of accounting for uncertainties about events and activities that far into the future.

Figure 2. Reduced Impact Logging (RIL) Carbon

For the purpose of estimating a project's net carbon mitigation, the time frame should neither mask future land-use trends nor render parameter estimates meaningless. It is necessary to balance a reasonable time frame for future projections against the need to sequester greenhouse gas emissions that persist for hundreds of years. Given a changing landscape, using too long a time horizon renders parameter estimates undependable, reducing the reliability of a project's carbon estimates.

In terms of tracking carbon, a project's lifetime should last until the time at which the carbon dioxide could have cycled out of the atmosphere, a minimum of 50 years. In this way, the project is most likely to have reduced the atmospheric concentration of greenhouse gases. Ideally, the project will be designed to provide local benefits that will ensure the continuation of project activities.


Properly defining project boundaries is absolutely critical to avoid leakage. Choosing an appropriate project boundary for making carbon sequestration estimates requires determining the spatial relationship between the demand causing land-use change, and the supply source. Project activities can have impacts at a project level, a local/regional level, or a global level.

Small pilot projects on land with little or no competing uses need only consider the area of direct project activities because the project's impact is unlikely to extend beyond its immediate boundaries. Projects in Russia and the Czech Republic fall into this category. Krkonose is situated in a national park, where there is no danger of encroachment, so there are no alternate land uses to displace. FACE is funding efforts on 15,000 hectares, and only the dynamics of those hectares need to be considered.

More often, competing land uses will mean that a project's impact extends beyond its own borders to the local area or region. The Olafo and CARE projects in Guatemala, and CARFIX are in this category. In dynamic settings where factors such as population growth, agricultural productivity, fuelwood needs, and concerns about deforestation interact, the project's impacts will extend beyond the areas of direct intervention. Developing an agroforestry project in a region of subsistence agriculture, requires more than planting X number of trees on X number of hectares and calculating the carbon sequestered. A project developer must also consider what type of land is converted to agroforestry. If it is agricultural land, will agroforestry increase productivity? If not, will farmers need to clear more land than was previously thought necessary?

For some projects, notably those involving logging or agricultural production for export markets, the demand and supply dynamic will be essentially global in that the projects cannot control demand and cannot affect the quantity consumed. If the Malaysia logging project reduces output, the world market will consume the same amount of timber; it just won't come from the Malaysian project site.

Inappropriate Project Design

The spatial relationship between supply and demand not only determines the appropriate project boundaries but also structures the process of identifying and avoiding leakage during the project design and evaluation phase. Potential types of leakage are activity shifting, market effects, and project construction effects. (18) Unlike unforeseen circumstances, these manifestations can largely be anticipated and avoided. Some or all are listed in the guidelines issued by the U.S. Initiative on Joint Implementation and by the U.S. Department of Energy for reporting greenhouse gas offset projects.

Activity shifting occurs when the activity causing carbon loss in the project area is displaced to another location. As an example, consider a project that buys out farmers and preserves the former agricultural land and surrounding forest. In response, farmers may resume agricultural activities in a neighboring forest rather than move to a city or take up another occupation. If farmers resume such activities elsewhere, the project has merely displaced the source of carbon emissions. Similarly, the Malaysia logging project which reduced damage from logging, may increase timber harvests elsewhere by decreasing the project site's short-term timber output.

Market effects occur when demand is unmet because a project reduces supply or because it unexpectedly increases demand. A United Nations Development Programme proposal for a carbon sequestration and biodiversity project in an arid area of the Sudan refers to such effects. According to the proposal, increased fuelwood and other resulting improvements might encourage immigration, increasing pressure on the new fuelwood source and undercutting net carbon savings. (19)

The case of Carton de Colombia, a pulpwood producer, provides a useful example of market effects, even though it was not a carbon storage project, but resembles such a project. Carton de Colombia, aimed to maintain a sustained yield through natural regeneration and by minimizing damage to residual trees in a lowland tropical forest on Colombia's Pacific Coast. (20) However, the job opportunities created by the pulpwood producer triggered an influx of colonists, who could not all be absorbed and employed. As a result, timber poaching and conversion of recently harvested areas to agriculture nullified potential project gains.

Construction effects occur when a project increases the energy-intensiveness of an activity by, for example, mechanizing agriculture, or introducing new emissions-producing activities by requiring major infrastructural developments such as large dam construction. With the exception of biomass projects for energy production, it is unlikely that a forestry project requiring such a large construction effort would be cost-effective enough to implement for carbon sequestration benefits alone. This study did not examine any such projects.

Of the three potential types of leakage, thus far only two-activity shifting and market effects-have emerged from our case studies as a concern. They are both related to unmet demand. Activity shifting will only occur if project boundaries are not configured to include relevant demands. Market effects are associated with a demand shift in which the project itself provokes an increase in demand and upsets original assumptions and hence carbon sequestration estimates.

The next two sections present a conceptual framework, the Leakage Index, that helps determine when leakage is likely to be an issue and uses case studies to illustrate the index.

Leakage Index

A rough guide to leakage potential is shown in Table 2, the Leakage Index. Estimates of leakage potential were based primarily on an analysis of the type of demand for a resource (for example, agricultural land, fuel, or timber for either local or export consumption); market boundaries (local, regional, national, or global); and the extent to which the project satisfies the demand for a resource.

Column one of the Leakage Index presents the main drivers of land-use change and the deforestation resulting from demand for agricultural land, fuelwood, and timber. The underlying concept is that decreasing output or access to needed resources prevents a project from meeting its carbon sequestration goals. The extent of the unmet demand determines the magnitude of leakage caused by project activities.

Figure 1 depicts an example of this dynamic. If upland forest had been strictly protected, and no agricultural extension and tree planting had been undertaken, demand for agricultural land and fuelwood would be unmet. This unmet demand would result in leakage as people attempted to intensify production on pasture land, further degrading the central, sloped areas, as they encroached on the protected area for fuel, or as they moved out of the area to cultivate unprotected forest.

Determining the primary drivers of land-use change requires an understanding of the project area and the human activities there, which in turn determine the extent of project boundaries. Although CARFIX proponents used GIS estimates indicating a 5 percent rate of deforestation, they needed to determine whether the deforestation was driven by demand for timber or for agricultural land. In this case, timber demand proved to be a secondary issue.

The second column of the Leakage Index further delineates the nature of demand, which could be either for local use, subsistence use, or export. This delineation will help define the project boundary and the amount of leverage the project can exert on demand. If demand for a resource is local or regional, the project can possibly offer substitute resources. However, the project will have little or no leverage if the demand is for a large regional area or a global market.

Column three of the Leakage Index lists likely project components based on those employed in carbon sequestration projects to date, and column four lists conditions under which these components become vulnerable to leakage. As these columns indicate, a project that reduces access to resources without offering alternatives will likely result in leakage, as people within the project area will move elsewhere to find other sources.

For a project to successfully sequester carbon, it must either expand or have a neutral impact on output of a resource. Alternatively, the project could provide a substitute resource. CARFIX, for example, addressed agricultural land demand by substituting income from sustainable forestry and carbon sequestration.

Column 5 of the Leakage Index offers an assessment of a project's potential for leakage: moderate or high, during the short or long term. Because the Index is qualitative, there is no strict interpretation for these designations. These designations are based on the availability of strategies to avoid leakage and the likely magnitude of leakage. A moderate designation means that the amount of leakage as well as its presence or absence, is dependent on individual site conditions. A high designation means that, unless there are mitigation strategies, leakage will occur. Where timber is the primary resource demanded, leakage may be of short- or long-term duration, as proponents of sustainable forestry projects argue that in the long term project sites are more productive than their conventionally logged counterparts.

The final column lists possible strategies for avoiding leakage. Each of these strategies has been implemented in ongoing carbon sequestration projects or proposed for such projects. In most cases, determination of appropriate strategies will depend on examination of forces leading to land-use change, which will be addressed through the project's key activities. However, in some cases, redesigning a project and adding activities may be too costly, or may not be feasible because of the project's location. If redesigning a project is impossible, potential carbon sequestration benefits must be recalculated to reveal the project's soundness. An example of refiguring carbon estimates for a timber project is offered in the Timber Demand section following the Leakage Index.

In summary, carbon projects must carefully consider impacts on surrounding areas. Does the project help meet local needs for income, fuel, and food, or does it lockup resources? If a project preserves a forest for its carbon sequestration benefits without regard to local needs, it will either shift demand for land or fuelwood to adjacent areas or will deprive the local population, ultimately engendering local opposition to the project and others like it in the future.

Table 2. Leakage Index

Primary Drivers

Market Boundaries

Project Components

Conditions Signaling Leakage

Potential Net Effect


Agricultural Land

Subsistence for local use

Increased agricultural productivity through green cover crop cultivation, agroforestry, soil conservation practices, or other measures

Increase output but free resources for development on adjacent lands

Moderate leakage

Protect adjacent forests;
Implement Sustainable forestry;
Introduce ecotourism

Forest preservation

Decrease agricultural output

High leakage

Create alternative income source;
Add agricultural productivity component

Local, regional, or global export

Increased agricultural productivity

Free resources for development on adjacent lands

Moderate leakage

Protect adjacent forests;
Implement sustainable forestry;
Introduce ecotourism

Forest preservation

Decrease agricultural output

High leakage depending on where activity shifts

Create alternative income source such as sustainable forestry


Local use or regional market


Common property resource;
Offsite market demand

Moderate leakage potential

Employ transferable technology






Local use

Sustainable forestry (Reduced impact logging, Natural forest management)

Decrease short-term timber output

Short-term leakage

Reestimate project impacts over short-term;
Develop alternative timber sources such as plantations on marginal land

Decrease long-term timber output

Leakage throughout project life. (High)

Reestimate project impacts;
Develop alternative timber sources such as plantations on marginal land

Forest preservation

Decrease or halt timber output

High degree of leakage

Develop alternative timber sources such as plantations on marginal land;
Introduce sustainable harvest in buffer areas


Sustainable forestry (Reduced impact logging. Natural forest management)

Decrease short-term timber output

Short-term leakage

Reestimate project impacts over short-term

Decrease long-term timber output

Long-term leakage

Reestimate long-term project impacts

Forest preservation

Decrease or halt timber output


Develop alternative timber sources such as plantations on marginal land

Application of the Index

The case studies below illustrate how the drivers of land-use change have (or have not) been addressed. Successes and failures are compared in the following sections to show how project evaluators and designers can avoid leakage by project design, or identify those projects in which leakage appears to be unavoidable.

Agricultural Land Demand

The Olafo and CARFIX projects are likely to stabilize the agricultural frontier and minimize or avoid leakage. Both projects satisfy demands causing land-use change.

The boundary of the Olafo project is regional/local. In the project area, land-use change is driven by the conversion of forest to agricultural land for subsistence. (21) Because the project decreases the amount of land required for agriculture it appears to have low or no leakage potential.

Figures 3 and 4 illustrate the difference in agricultural land required with and without the project. Figure 3 shows the "With Project" scenario, which reduces the number of hectares required for agricultural production by decreasing the amount of fallow time required. Because less agricultural land is required, the pressure to convert the forest to agricultural uses diminishes. Concurrently, natural forest management gives standing forests value, thus creating incentives for local people to protect them.

The use of green cover crops and other techniques of the Olafo project are easily adaptable in other areas. Therefore, the project will not shift resource demand elsewhere or cause immigration. Market effects and activity shifting will be avoided because the project benefits are not concentrated but instead can be applied where they are needed. The analysis may understate these benefits.

Figure 4 shows a potential shortage of agricultural land without the project because available agricultural land is scarcer than required agricultural land. This shortage indicates that farmers may move beyond the project area in search of new agricultural land, thus increasing carbon emissions outside the project area under the baseline scenario.

Both the shortage and the resulting increase in carbon emissions elsewhere can be approximated. By year 40, for example, the difference between required and available agricultural land is about 500 hectares. If the project region is surrounded by forest, the analyst may reasonably assume that an additional 500 hectares of this forest would have been converted to agricultural uses in the absence of the project. At an average biomass of 400 tons per hectare, an additional 200,000 tons of carbon would have been sequestered by the project. The same calculation can be made if an agricultural land shortage exists either under the "With" or "Without Project" scenarios.

Logging and the subsequent expansion of the agricultural frontier for cattle production drives deforestation in the CARFIX region. The key to slowing deforestation and avoiding leakage, therefore, is to provide an alternative income source to cattle production.

Figure 3. Olafo Required and Available Agricultural Land

Figure 4. Olafo Required and Available Agricultural Land

CARFIX proposes to generate income from selling carbon offsets and sustainably harvested timber. Annual payments will be advanced to the landowners in the years preceding harvests. Income substitution should be successful because landowners are voluntary participants in the project and have thus presumably decided that timber and the supplementary income provided by CARFIX can replace income from cattle production.

FUNDECOR, the agency implementing CARFIX, contends that conversion to cattle grazing produces little marketable timber. Because the project will increase timber output, there will be no unmet timber demand. (22) However, some activity shifting could result from reducing cattle production. Whether cattle production is for export or for local or subsistence consumption is unclear, so fully assessing the leakage from decreasing such production is difficult. The low leakage scenario would reflect essentially subsistence consumption, as increasing farmers' income would offset the need for raising cattle. If the cattle are for export, leakage would depend on the location of the alternative production region and whether cattle operations can be intensified, or if additional land must be cleared to accommodate them.

Some activity shifting could also occur. Although CARFIX does not displace farmers as landowners, the project does call for silviculture which may be less labor intensive than cattle production. The danger is that between harvests farmers will cultivate additional land or otherwise expand their activities in ways that increase carbon emissions.

Carbon projects dealing with demand for agricultural land on the edges of standing forests must find a way to give those forests value. In these areas, agriculture and forestry can be used as substitute income sources and alternate land uses. The trick is to stabilize the agricultural frontier, encourage sustainable forestry as a way to earn income, and avoid a one-shot "mining" operation and conversion to agricultural land.

Fuelwood Demand

The Sudanese and CARE/Guatemala carbon sequestration projects illustrate both the problems associated with meeting fuelwood demand and their solutions. The Sudanese project shows high potential for leakage, whereas the CARE/Guatemala project indicates a low potential for leakage.

The Sudanese project aims to ease a serious fuelwood shortage in an area where fuelwood collection is resulting in deforestation and rangeland degradation. The project's goals are to increase agricultural productivity by irrigating gardens, and improving fuelwood resources by planting windbreaks and woodlots. But these improvements, that are concentrated in a small area, may encourage settling of nomadic people or livestock herders. Such a population influx, as noted above, would deplete fuelwood resources and erode carbon sequestration gains as a result of market effects. Leakage will be avoided if the project is able to expand its scale, either by enlarging its boundaries and areas of activities or by employing transferable technologies.

To reach the relevant areas, CARE/Guatemala expanded its scale to include most of the upland regions. Like the Sudanese project region, the CARE/Guatemala project region was facing probable fuelwood shortages, which the project sought to address through woodlots and agroforestry. Baseline projections showed that the initial conversion of forest land to agriculture would provide ample fuelwood. However, once land conversion slows, fuelwood collection begins to degrade the remaining forest. CARE/Guatemala's conversion of degraded land to woodlots and permanent agriculture to agroforestry increased fuelwood supply, thus meeting most fuelwood needs. Moreover, CARE established tree nurseries, run by local farmers, which later became self-sufficient.

In this case, the methods of increasing fuelwood availability and agricultural productivity were widely reproducible, and the project sponsor had the mobility to work at the country level, ensuring that project benefits could be diffused where needed.

Timber Demand

In general, projects that constrain the supply of timber will not reduce net carbon emissions as much as anticipated, because they will lead to activity shifting. A model of world timber supply indicates that restrictions on timber supply in one region will likely lead to timber harvests in a different region. (23) Because timber is internationally traded and can be supplied by many parts of the world, reductions in output in one place can be replaced by increases in another area.

However, the model's finding is based on decreasing output in North America, where there may be fewer market imperfections. The response to reduced output may be different in areas where government subsidies and poorly negotiated timber concessions result in incentives to over harvest. (24) In these areas, projects that constrain timber supply will not shift harvests elsewhere. Because the harvests are induced through government policies, logging companies will not necessarily be able to find such profitable opportunities in other areas.

The Malaysia logging project aimed to reduce damage from logging while maintaining the same timber output as conventionally logged areas. Because harvests are not the result of gross market imperfections and because the extracted timber was for a global market, maintaining timber output was the key to avoiding leakage. However, output from the project site reportedly has been less than conventionally logged areas. The shortfall will likely be made up by harvests elsewhere in the region, another part of the concession, or possibly from another part of the world, thus signaling leakage.

Evidence of leakage indicates that carbon should be recalculated but does not necessarily mean that the project is without merit. That determination depends on whether the project's net carbon savings and other benefits make it competitive with similar projects.

To recalculate the original net carbon estimate, the project evaluator needs to determine approximately how much area would have to be logged to compensate for the decrease in output. The Malaysian project decreased timber output by 49 cubic meters per hectare on 450 hectares, making the total shortfall 22,050 cubic meters. Logging approximately 145 hectares conventionally (at 152 cubic meters per hectare) or 214 hectares using reduced-impact methods (at 103 cubic meters per hectare) would make up for the reduction. (25)

Because reduced-impact logging techniques reduce biomass loss by 50 percent per hectare, as long as the number of hectares logged to compensate for the timber shortfall is less than 50 percent of total hectares, the project still results in net carbon savings, although at a greater cost per ton. (26) In addition, reduced-impact and sustainable harvesting techniques increase long-term forest productivity by protecting young trees. The timber concessionaire may need to compensate for lower yields in the near term but will be rewarded with higher yields in the future while avoiding costs associated with enrichment plantings. (27)

If timber demand is the primary driver of carbon loss and the timber is for an export market, maintaining output is imperative. If timber is for a local market, the project may have an opportunity to provide an alternate timber source, or an alternate income source from nontimber products; increased environmental services; or, in the future-added value from sustainably harvested timber.