JEQ Grow Your Career With ASA
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 23 June 2008
Published in J Environ Qual 37:1368-1375 (2008)
DOI: 10.2134/jeq2007.0288
© 2008 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sohngen, B.
Right arrow Articles by Andrasko, K.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sohngen, B.
Right arrow Articles by Andrasko, K.
Agricola
Right arrow Articles by Sohngen, B.
Right arrow Articles by Andrasko, K.
Related Collections
Right arrow Economics
Right arrow Ecosystem Management
Right arrow Carbon Sequestration
Right arrow Global Change
Right arrow Other Models

Avoided Deforestation as a Greenhouse Gas Mitigation Tool: Economic Issues

Brent Sohngena,*, Robert H. Beachb and Kenneth Andraskoc

a Dep. of Agricultural, Environmental, and Development Economics, The Ohio State Univ., 2120 Fyffe Rd., Columbus, OH 43210
b Food and Agricultural Policy Research Program, RTI International, 3040 Cornwallis Rd., Research Triangle Park, NC, 27709
c performed this work while at the USEPA Climate Change Div., and is now at the World Bank, Carbon Finance Unit, 1818 H St, NW Room MC3-835, Washington, DC 20433. The author's views do not represent those of the World Bank

* Corresponding author (sohngen.1{at}osu.edu).

Received for publication June 1, 2007.

   ABSTRACT
TOP
 NOTES
 ABSTRACT
INTRODUCTION
 Literature Review
 Methods
 Results
 Discussion and Conclusions
 REFERENCES
 
Tropical deforestation is a significant contributor to accumulation of greenhouse gases (GHGs) in the atmosphere. GHG emissions from deforestation in the tropics were in the range of 1 to 2 Pg C yr–1 for the 1990s, which is equivalent to as much as 25% of global anthropogenic GHG emissions. While there is growing interest in providing incentives to avoid deforestation and consequently reduce net carbon emissions, there is limited information available on the potential costs of these activities. This paper uses a global forestry and land use model to analyze the potential marginal costs of reducing net carbon emissions by avoiding deforestation in tropical countries. Our estimates suggest that about 0.1 Pg C yr–1 of emissions reductions could be obtained over the next 30 to 50 yr for $5 per Mg C, and about 1.6 Pg C yr–1 could be obtained over the same time frame for $100 per Mg C. In addition, the effects of carbon incentives on land use could be substantial. Relative to projected baseline conditions, we find that there would be around 3 million additional hectares (ha) of forestland in 2055 at $5 per Mg C and 422 million ha at $100 per Mg C. Estimates of reductions in area deforested, GHG mitigation potential, and annual land rental payments required are presented, all of which vary by region, carbon price paid, and time frame of mitigation.

Abbreviations: GHG, greenhouse gas • GTM, Global Timber Model • UNFCCC, United Nations Framework Convention on Climate Change


   INTRODUCTION
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
Literature Review
 Methods
 Results
 Discussion and Conclusions
 REFERENCES
 
TROPICAL deforestation is a major source of greenhouse gas (GHG) emissions, accounting for as much as 25% of global anthropogenic GHG emissions (Houghton, 2005). Temporary or partial forest removals for shifting cultivation and selective logging, as well as permanent forestland conversion to agricultural or other uses, contribute to releases of carbon stored in vegetation and soils to the atmosphere. Emissions depend on both the rate of deforestation and changes in carbon stock per hectare after deforestation, with changes in carbon stocks varying with land use, region, ecosystem, and use of the removed forest biomass. Burning results in immediate releases of forest carbon, whereas unburned organic matter releases carbon more slowly during the decay process. Carbon emissions from wood products may occur over decades, taking up to 100 yr or more for some products to have released all their carbon.

Although afforestation, reforestation, and reductions in deforestation have all been widely discussed as options to mitigate climate change, most of the research and policy discussion to date has focused on afforestation and reforestation. This is curious, given that a central estimate of about 1.6 Pg C yr–1 was emitted in the 1990s by deforestation in tropical countries (IPCC, 2007). One reason for the limited emphasis on deforestation projects rests with concerns about additionality, permanence, and leakage associated with projects focused on avoiding deforestation (see Schlamadinger et al. [2005]). Research on specific projects in countries experiencing deforestation, however, has suggested that, although these concerns are founded, it is possible to account for them when the carbon gains from specific projects are measured (e.g., Aukland et al., 2003; Sohngen and Brown, 2004; DeFries et al., 2006; Brown et al., 2007).

Given the large potential emission caused by deforestation, and recent research results suggesting that actions taken to avoid deforestation and conserve net carbon in forests could be measured and monitored, there is now renewed interest in considering avoided deforestation as a mitigation option. For example, at a side event to the ninth Conference of the Parties to the United Nations Framework Convention on Climate Change (COP9 of the UNFCCC), Santilli et al. (2003) introduced a new proposal to add avoided deforestation activities as eligible projects, which reopened debate about including avoided deforestation. During COP11, held in Montreal from 28 Nov. to 9 Dec. 2005, the Governments of Papua New Guinea and Costa Rica, on behalf of the Coalition for Rainforest Nations, proposed that parties to the UNFCCC address emissions from deforestation and create incentives for developing nations to manage these emissions. The COP11 decision proposed that parties to the UNFCCC be given an opportunity to provide their views on providing incentives for reducing deforestation before the 24th meeting of the United Nations Sessions of the Subsidiary Bodies, held in Bonn, Germany, in May 2006 (SBSTA 24). Twenty-one nations provided formal input by the time of the SBSTA 24 meeting, and an agreement was reached to continue considering the development of incentive mechanisms by which developing countries may reduce deforestation.

The concept of reducing deforestation has been widely discussed in the academic literature, but the idea of developing a program that gives countries incentives to reduce their deforestation has not been widely considered in international climate regimes. In addition, although a number of previous studies have examined the potential for avoided deforestation to play a role in GHG mitigation, relatively few authors have examined the associated costs of achieving different levels of emission reductions across multiple tropical regions. This paper thus examines the costs of avoiding deforestation. To accomplish this task, we use a global forestry market and land use model (GTM) originally developed by Sedjo and Lyon (1990) and subsequently expanded in Sohngen et al. (1999) and Sohngen and Mendelsohn (2003). GTM is used to simulate projected baseline emissions from deforestation processes occurring in tropical countries. Carbon incentives (e.g., prices) are then incorporated into the model, and marginal cost curves for avoiding deforestation are derived for regions of the world where the majority of tropical deforestation takes place. We begin below with a brief review of the current literature on deforestation and carbon emissions, and then describe the methods and results of our analysis.


   Literature Review
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Literature Review
Methods
 Results
 Discussion and Conclusions
 REFERENCES
 
The Food and Agriculture Organization (FAO) reported annual net forest cover losses (i.e., gross deforestation minus afforestation and reforestation) of around 8.9 million ha yr–1 in the 1990s, falling to 7.4 million ha yr–1 in the early 2000s (FAO, 2006). These losses amounted to a net loss in global forest cover of around 0.22% per year during the 1990s and 0.18% in the 2000s (Table 1 ). However, the global numbers mask substantial variation among the regions. In general, tropical regions are experiencing deforestation, and temperate regions are experiencing afforestation. Net forest loss in the tropical forests of South America, Central America, Southeast Asia, and Africa is estimated to have averaged about 11.5 million ha yr–1 since 1990, whereas forest cover increased in Europe, North America, and East Asia. Houghton (2003) suggests that deforestation rates were substantially higher in these same tropical regions during this period, around 15.8 million ha yr–1 in the 1990s. However, that study also suggests that afforestation occurred over large areas of land in East Asia, so that net forest cover change was a loss of around 12.1 million ha yr–1.


View this table:
[in this window]
[in a new window]

  		Table 1. Forest cover change, 1990–2005.{dagger}

 
The carbon consequences of these relatively large adjustments in forest cover are substantial. Not only is carbon lost to the atmosphere from net reductions in forest cover, but newly afforested or reforested lands store far less carbon per hectare (initially) than mature stands being deforested. In addition, the geographical variation in forest cover trends has important implications for carbon emissions because of the large differences in carbon stock per hectare across regions. In general, the tropical areas experiencing net deforestation have higher carbon stocks in forest biomass per hectare than temperate regions experiencing net afforestation. For instance, forests in North America have a weighted average of 117.8 Mg C ha–1, whereas Central America has 179.2 Mg C ha–1, and South America has 194.6 Mg C ha–1 (FAO, 2006).

One of the first studies examining carbon implications of forest cover changes globally, Dixon et al. (1994), suggests that the net effects may lead to emissions of up to 0.9 Pg C yr–1 for the entire world. DeFries et al. (2002), Potter et al. (2003), and Achard et al. (2002) similarly find that forests globally are a net source of emissions; DeFries et al. and Potter et al. both estimate a net global emission of 0.9 Pg C yr–1, consistent with Dixon et al. The global emissions estimate provided by Achard et al. is a bit higher at 1.1 Pg C yr–1. A study by Houghton (2003) indicates potentially far larger net emissions from deforestation of 2.2 Pg C yr–1 during the 1990s.

The results above are based on forest inventory data and changes in land uses observed through satellites or by other means. Alternative methods for calculating the flux between forests and the atmosphere have been developed in what are commonly called inversion models (see Ciais et al. [2000]). The results from inversion model studies have generally suggested that forests are smaller net sources and likely net sinks globally. For instance, using these techniques, Ciais et al. (2000) find that ecosystems globally sequester around 1.3 Pg C yr–1 (net of all deforestation, afforestation, and management processes). Gurney et al. (2002) find that deforestation in the tropics accounts for around 1.2 Pg C yr–1 of emissions, but these emissions are more than offset by gains in ecosystem carbon elsewhere. As a result, their study estimates that ecosystems, on average, sequester around 1.3 Pg C yr–1 globally. Similarly, Plattner et al. (2002) find that net global sequestration in ecosystems is around 0.7 Pg C yr–1.

Several studies have now examined the ecological potential to reduce deforestation in tropical regions, but there have been to date few global economic assessments of the potential for reductions in deforestation. Soares-Filho et al. (2006), for example, examine the potential for reducing deforestation in tropical areas of Brazil. They do not, however, estimate the costs of achieving these relatively large reductions in deforestation. Gullison et al. (2007) discuss global deforestation, but they base their discussion entirely on results presented in the recent report of the Intergovernmental Panel on Climate Change (IPCC, 2007). One recent economic study of forest carbon sequestration, Sohngen and Sedjo (2006), does suggest that a large proportion (73 to 88%) of carbon mitigation potential rests in the tropics, but they estimate only aggregate (net) land use change (lumping afforestation and deforestation together). Sathaye et al. (2006) similarly use a global economic model and also find that tropical regions can provide substantial carbon abatement services, but they did not calculate marginal costs associated with reductions in deforestation. Kindermann et al. (2006) use a supply model that calculates the cost of reducing emissions from deforestation within many tropical regions. Their model, however, does not link markets within or across these regions.


   Methods
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Literature Review
 Methods
Results
 Discussion and Conclusions
 REFERENCES
 
Future land use changes are modeled with GTM (see Sohngen et al. [2005] and Sohngen and Mendelsohn [2007] for recent model applications). GTM has been widely used for policy analysis in recent years, including analysis of regional carbon sequestration baselines (Sohngen and Sedjo, 2000), climate change impacts (Sohngen et al., 2001), and carbon sequestration analysis (Sohngen and Mendelsohn, 2003). The model is a dynamic optimization model that maximizes the net present value of consumers' surplus less costs of managing, harvesting, and holding (or renting) forests. A mathematical description of the model is presented in Sohngen and Mendelsohn (2007). Although additional price effects beyond those modeled here are possible (e.g., increases in agricultural production prices), the results provide first-cut estimates of how carbon prices could potentially influence levels of deforestation and afforestation globally. Future model developments should account for interactions with agricultural markets in different regions to analyze a broader set of policies.

A global demand function for timber logs is used to estimate consumer surplus in timber markets. Forests in 250 timber supply regions then feed this global demand. Age class distributions for forests were derived from local sources, where available, or assumed for regions without data on age classes. For temperate and boreal regions (most developed countries), age class distribution information was obtained from local sources. Tropical forests are assumed to be in old growth conditions, while age class distributions for subtropical plantations were derived from historical planting and harvesting rates.

Age class distributions and timber biomass growth functions were developed for each timber type. Cost functions for harvesting accessible and remote forests were developed from earlier estimates used in the study by Sohngen et al. (1999). Remote forests are those that have little infrastructure near them and consequently have high costs for timber extraction and transportation. Access costs in regions where data are not available are based on costs for similar forests in different regions of the world. Since all prices and costs in the model are denominated in 2000 U.S. dollars ($), the relative costs for harvesting or accessing forests are adjusted for differences in exchange rates.

In addition to accounting for the costs of harvesting and accessing forests, land opportunity costs are modeled with land supply functions. The land supply functions represent land moving from nonforest use to forest use in response to an increase in the (rental) value of forest use. Land supply elasticity is assumed to be 0.25 for all regions in the model, indicating that a 1% increase in forestland rental values will increase forestland area by 0.25% at initial land rents and forestland areas. Although the analysis contained in this paper does not present sensitivity analysis on this particular parameter, Sohngen and Mendelsohn (2007) found that a 50% increase or decrease in the elasticity estimate reduces or increases potential global sequestration by 20%. The results for total potential sequestration were found to be slightly more sensitive to assumptions about the elasticity of land supply in South America than the global average, but less sensitive in the other tropical regions examined in this paper (Central America, Southeast Asia, and Africa).

To simulate changes in land use over time, the land supply functions in tropical regions shift inward over time. This model does not solve forestry and agricultural land markets simultaneously, but it does simulate a path of agricultural expansion (or contraction) in all regions of the world by shifting the land supply functions. Shifting the land supply functions inward increases the opportunity costs of holding land in forests and therefore spurs additional conversion of land over time.

GTM accounts for the change in above- and below-ground vegetative carbon stock associated with shifting land from forestry to agriculture and from agriculture to forestry. The model also accounts for other types of management changes in forestry that influence carbon outcomes, such as changes in management intensity, changes in rotation ages, and changes in plantation forests, but this paper focuses on presenting the carbon outcomes related to deforestation and land use change in tropical regions where deforestation is largest. To generate marginal cost curves for carbon sequestration, the model is first used to generate a baseline. The baseline occurs when carbon prices in the model are set to zero. We then apply GTM to a series of carbon price scenarios where carbon prices are assumed to be constant for the entire 21st century at a level ranging from $5 per Mg C to $100 per Mg C ($1.36 per Mg CO2 to $27.25 per Mg CO2). The prices we examine represent different levels of overall goals for carbon abatement or mitigation (e.g., higher prices suggest more stringent targets). Reductions in deforestation and the marginal costs of the resulting emission reductions are then calculated by comparing the results of each positive carbon price scenario to the baseline. When implementing the carbon price scenarios, we assumed that additional carbon gained above the baseline is rented at an annual rental rate consistent with the carbon prices listed above. The annual rental rate is

Formula
where RC is the rental rate for carbon (the annual value paid per Mg C for holding carbon in the ecosystem), r is the interest rate, and PC is the price of carbon. The formula above assumes that the price of carbon remains constant, an assumption maintained throughout this analysis. If carbon prices were instead assumed to rise, the formula for calculating the rental value would need to be adjusted to account for increases in the price.

For the scenarios analyzed for this paper, all carbon gains relative to the baseline are assumed to be net carbon gains for the atmosphere. The specific management actions that lead to these net carbon gains in the model are land use change (i.e., reduced deforestation and afforestation of nonforested lands), increases in forest management intensity, increases in rotation ages, and wood product storage. Carbon gains in all regions of the world are paid the same amount. However, we only present results for four regions where deforestation is the largest. By providing these incentives, we account for the global nature of carbon policy. Payments are only paid for net carbon gains in this analysis, and carbon gains are paid exactly what they are worth for the time they are stored, so the carbon storage implied by these results does not need to be further corrected for additionality, permanence, and leakage concerns. This method of paying for carbon is consistent with a comprehensive approach for carbon sequestration and reduced emissions from avoided deforestation, where any increase in carbon stocks is credited, and any decrease debited, in carbon accounting. Other approaches to provide incentives for reducing emissions by avoiding deforestation are also plausible, but estimates of the resulting carbon gains would need to be corrected for the issues above, as discussed in Murray et al. (2007).


   Results
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Literature Review
 Methods
 Results
Discussion and Conclusions
 REFERENCES
 
We begin the results section by describing our baseline projections. All of the emission reduction scenarios are compared to this baseline scenario. Baseline deforestation in the tropical regions modeled is projected to be 13.1 million ha yr–1 over the period 2005 to 2015 (Table 2 ). On net, GTM projects a loss in forest cover of around 11.8 million ha yr–1 in tropical forest regions during this period. As a result, deforestation is projected to add around 1.5 Pg C yr–1 to the atmosphere over the next 10 yr and, when reforestation is considered, to lead to net losses of 1.4 Pg C yr–1. These estimates are roughly consistent with many of the estimates of tropical deforestation discussed above. Note that it is not only the net change in forestland that affects emissions, but also area deforested, as well as area afforested or reforested and carbon stock per hectare under different conditions.


View this table:
[in this window]
[in a new window]

  		Table 2. Projections of deforestation and net losses in forestland and carbon sequestration caused by land use change in tropical regions, 2005–2015.{dagger}

 
Over the next 50 yr, the model projects that baseline deforestation slows (Fig. 1 ). The rates of decline in deforestation are consistent with predictions that the demand for agricultural land will slow in the future. Reductions in the demand for agricultural land are driven by assumed increases in agricultural productivity of 2 to 3% per year (Nin et al., 2003) and a slowing of population growth, both of which would reduce the demand for land for agriculture.


Figure 1
View larger version (12K):
[in this window]
[in a new window]

  		Fig. 1. Projection of future land use change in tropical regions under baseline conditions. Source: Authors' calculations for this study.

 
Considering the change in deforestation rates associated with different carbon prices is useful because some countries may consider adopting indirect carbon policies that focus on altering the rate of deforestation rather than policies that target carbon explicitly. Assessing the relationship between the rate of deforestation and carbon prices can give policymakers a sense of what changes in the rate of deforestation would be feasible for different carbon prices. The results of the analysis indicate that average annual tropical deforestation rates could be reduced by 8.4 to 15.3% each year for a carbon price of $5 per Mg C (Table 3 ). The largest changes are projected to occur in Africa and Central America for this scenario. For higher carbon prices, not surprisingly, larger reductions in deforestation occur. At $100 per Mg C, the results suggest that deforestation can virtually be stopped. Central America and Africa obtain the largest reductions in the rate of deforestation for lower carbon prices, but all regions have approximately a 100% reduction when carbon prices are $100 per Mg C. Africa has the lowest land values and the largest total deforestation initially, so that carbon incentives have a fairly large effect there for low carbon prices. Central America has less total deforestation initially, so even fairly small changes in the rate of deforestation are large in percentage terms. Southeast Asia and South America have higher land rental values; consequently, it takes higher carbon prices to induce similar percentage reductions in the rate of deforestation.


View this table:
[in this window]
[in a new window]

  		Table 3. Reduction in average annual deforestation rate, 2005–2055.{dagger}

 
To illustrate changes in the annual amount of deforestation over the 50-yr time period (2005 to 2055), Fig. 2 shows the area estimated to be deforested in the baseline case and the five carbon price scenarios for South America. Reductions in deforestation occur initially and remain at a fairly consistent level for most of the period. At $100 per Mg C, as noted above, deforestation is stopped. Results are similar for the other tropical regions modeled.


Figure 2
View larger version (14K):
[in this window]
[in a new window]

  		Fig. 2. Effects of carbon price scenarios on the time path of deforestation in South America. Source: Authors' calculations for this study.

 
In the baseline, tropical deforestation is projected to lead to around 55.7 Pg of cumulative carbon loss over the period 2005 to 2055 (Table 4 ). For $5 per Mg C, this could be reduced to a loss of around 50.4 Pg, which is a gain of around 5.3 Pg C relative to the baseline by 2055. At higher prices, more carbon is saved. For $50 per Mg C, most of the losses are avoided by 2055. Note that, in the $50 per Mg C case, deforestation still occurs in all regions, but substantial areas of land that were deforested previously are converted back to forestland, so that the net losses from forests are fairly small over the time frame. For $100 per Mg C, forest areas rise substantially relative to the baseline, and around 76 Pg of additional carbon are stored.


View this table:
[in this window]
[in a new window]

  		Table 4. Summary of gains from carbon price scenarios.{dagger}

 
Figure 3 presents abatement cost curves for reducing emissions from deforestation for each of the tropical regions. At all carbon price levels, Southeast Asia offers the largest emissions reductions and Central America the smallest in absolute terms. Southeast Asia has relatively lower opportunity costs per hectare and higher carbon density per hectare on average, leading to its lower cost estimate. Costs are higher for Central America because that region generally has less land available for sequestration. Of course, this does not mean that there are not low-cost opportunities in Central America; however, it does suggest that in relative terms Central America has fewer opportunities than the other regions. Africa has similar marginal costs as Southeast Asia for lower levels of sequestration, although the curves diverge above 350 Tg C yr–1. Marginal costs for South America fall in between those for the other regions.


Figure 3
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 3. Abatement cost curves for avoided deforestation. Estimates are the gain in carbon sequestration relative to the baseline from reduced deforestation measured as the annual equivalent amount of gains from 2005–2055. Source: Authors' calculations for this study. Annual equivalent amounts are determined by first calculating the present value of carbon gains between the baseline and the carbon price scenario and then using standard techniques to calculate annual equivalent amounts over a 50-year time period with interest rate r=5%.

 
For the $5 per Mg C scenario, the rental values necessary to achieve these changes are estimated to range from $23 to $33 ha–1 yr–1 (Table 5 ). For the $100 per Mg C, they are 20 times larger, ranging from $466 to $659 ha–1 yr–1. Because of differences in carbon levels among the forest types in the region, the potential payments vary substantially.


View this table:
[in this window]
[in a new window]

  		Table 5. Average annual rental payments required to achieve the land use and carbon changes estimated under different carbon price scenarios ($ ha–1 yr–1).{dagger}

 
An important point to recognize with these payments is that the total cost of the program would be higher than these values alone if implemented as project-level activities, because the programs may also need to account for leakage (see Murray et al. [2004] and Sohngen and Brown [2004]). That is, to ensure that the carbon gains associated with reducing deforestation do not have leakage, potential programs should ensure that there are no carbon losses elsewhere as a direct result of the program aimed at reducing deforestation or address leakage in some other way.

Figure 4 shows the estimated reductions in deforestation that could be achieved in each of our tropical regions as a function of rental rates per hectare per year. The lowest marginal costs for reducing deforestation appear to lie in Africa, followed by South America, Southeast Asia, and Central America. Interestingly, the ordering of marginal costs for reducing deforestation differs from the ordering of the marginal costs of reducing carbon shown in Fig. 3. The largest difference exists with Southeast Asia. That region can sequester more carbon per hectare than other regions; thus, for less land use change, it achieves lower marginal costs for sequestration.


Figure 4
View larger version (13K):
[in this window]
[in a new window]

  		Fig. 4. Supply curves for reducing tropical deforestation, 2005–2055. Source: Authors' calculations for this study.

 

   Discussion and Conclusions
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Literature Review
 Methods
 Results
 Discussion and Conclusions
REFERENCES
 
This paper develops an economic analysis of the potential costs of reducing deforestation as a method to help mitigate climate change. Although the option of reducing deforestation to avoid carbon emissions has long been discussed, mechanisms that provide incentives to countries to develop programs to reduce the deforestation occurring within their boundaries have not yet evolved. To provide information that may be useful to policymakers considering such incentives, this paper employs an existing model to assess the potential costs and the potential carbon gains of reducing deforestation. Specifically, we used a global timber market and land use model that projects baseline carbon emissions from deforestation and other forestry-related land use activities. Carbon prices were then introduced into the model, and the resulting changes in deforestation and carbon are presented. To our knowledge, this paper is one of the first to consider how different carbon prices will affect potential levels of deforestation in tropical countries over time.

The results of the analysis presented in this paper indicate that there is large potential for reduced deforestation globally to help reduce the costs of reducing GHG emissions. For $100 per Mg C ($27.25 per Mg CO2), deforestation can potentially be virtually eliminated. Over 50 yr, this could mean a net cumulative gain of 76 Pg C relative to the baseline and 422 million additional hectares in forests. For lower prices of $5 per Mg C ($1.36 per Mg CO2), about 5 Pg C additional could be sequestered over 50 yr. The largest gains in carbon occur in Southeast Asia, which gains nearly 30 Pg C for $100 per Mg C, followed by South America, Africa, and Central America, which gain 22, 19, and 6 Pg C for $100 per Mg C, respectively. The effects of carbon incentives on land use could be fairly substantial. For $5 per Mg C, the model projects that by 2055 there would be around 3 million additional hectares of forestland in the four regions analyzed. For $100 per Mg C, the model projects that almost no deforestation occurs, and the four regions would have an additional 422 million hectares of forestland.

It is useful to put these results in a broader context. Over the last few years, the carbon market has become significantly larger, but most of the market trading occurs with energy projects and not with land use projects (LeCocq and Capoor, 2005). These results suggest that there is large potential to use reductions in deforestation as a climate mitigation tool, but we recognize that institutional infrastructure may not yet be available in all regions for trading carbon derived from reducing deforestation. Furthermore, the total costs of large programs would (not surprisingly) be large. For example, the $100 per Mg C carbon scenario implies substantial potential for carbon sequestration, but the total cost of this size program would be exceptionally large. Based on the average carbon per hectare in tropical forests today, policy programs or carbon markets would have to pay $465 to $660 ha–1 yr–1 to ensure that land does not convert to agriculture. Across the four regions considered above—Southeast Asia, South America, Africa, and Central America—the total costs of reducing deforestation would be $2.5 trillion at the $100 per Mg C price, suggesting very large overall costs.

It is also useful to put these results in context of other studies available. Two studies are of particular interest because they have done large global analyses. Sathaye et al. (2006) find that reducing deforestation can provide 34 Pg C by 2050 for $100 per Mg C on 454 million additional hectares of forestland. The results in the present analysis are larger, implying potentially 76 Pg C, although the total land use change is smaller (422 million additional hectares). These differences suggest that the model used in this study assumes more carbon is saved with each hectare preserved. The model used here also assumes more carbon emissions in the baseline associated with deforestation. Kindermann et al. (2006) find that reducing deforestation can lead to around 1.4 Pg C yr–1 between 2005 and 2025 for $100 per Mg C. In this analysis, we find that around 1.6 Pg C yr–1 can be preserved globally for $100 per Mg C by reducing deforestation.

The results in this study are surprisingly consistent with important noneconomic studies. Soares-Filho et al. (2006), for example, examine potential carbon emissions from deforestation in the Amazon Basin. They suggest that in the baseline up to 210 million hectares may be deforested over the next 50 yr. This estimate is larger than our estimate of around 136 million hectares of deforestation in all of South America. Based on this result and geographically detailed estimates of carbon losses from the forests that they simulate to actually be deforested, they find that 32 Pg C could be emitted over the next century, or 158 Mg C ha–1. Our estimate is that 17 Pg C will be lost over the same time period, or around 120 Mg C ha–1. One of their scenarios that protects land from deforestation increases total forest land in the region by 130 million hectares and preserves 17 Pg C from being emitted through deforestation by 2050. Although they do not present costs for their analysis, this is similar in scale to our $100 per Mg C scenario, which preserves 167 million hectares (reduced deforestation and afforestation combined here) and gains around 22 Pg C, or 132 Mg C ha–1. Our results appear to be consistent with the estimated potential for sequestration in the Amazon Basin in Soares-Filho et al. (2006), although our results imply that the carbon gains they suggest could cost as much as $75 to $100 per Mg C.

One limitation of the results presented here is that we have assumed that forests are treated as part of a comprehensive global approach for reducing GHG emissions. A comprehensive approach requires that all land be monitored and included in the program. Currently, the main approach for incorporating forestry into global climate policy is through efforts undertaken on individual projects, not through a comprehensive approach as modeled. Projects must account for monitoring, additionality, permanence, and leakage. Incorporating these factors into project design raises the transaction costs of carbon sequestration (Antinori and Sathaye, 2007). Thus, actually achieving the levels of carbon potential suggested above through the project-based approach would cost more than the estimates above indicate, though it is outside the scope of this study to determine how much more expensive the project-based approach would be.

The main focus of this paper has been to examine the marginal costs of carbon sequestration in a highly efficient program to reduce deforestation. In particular, we assume that policymakers can perfectly and costlessly distinguish—and pay for—actions that provide real and additional carbon from actions that do not. In reality, one would expect difficulties to arise in making these distinctions when implementing emissions abatement programs. Many policy decisions, such as those affecting the setting of a baseline, are likely to have a large influence on the total costs of a program. For example, generous baseline development could result in substantial payments for carbon that would otherwise have been freely stored in forests. The eventual development of methodological guidance for baseline development and monitoring could require remote sensing, data collection or ground-truthing methods, or geographic or temporal resolution, that significantly affect costs. Programmatic decisions on how to estimate and address transnational leakage could also raise costs. This paper has not addressed these issues, although the authors recognize their importance with respect to the total costs of any program developed to reduce emissions by avoiding deforestation.

The model used for the analysis does not fully account for all possible adjustments in land markets in tropical regions, nor does it account for country-level risks. The implications of not accounting for all land markets is that it is possible that these results underestimate marginal costs. As more land is devoted to forestry with carbon incentives, the price of agricultural products should increase, and to the extent these price increases are not fully captured by our rental functions, this would in turn raise the costs of further sequestration. For example, Sohngen and Sedjo (2006) consider two alternative assumptions about the demand for land from agriculture and show that higher demand from agriculture, not surprisingly, raises the costs of carbon sequestration in forestry. The analysis in this paper has been conducted with the higher demand scenario described in that paper. While sensitivity analysis like this is useful, it is clear that given the growing potential demands in agriculture provided by biofuel alternatives, developing modeling tools to account for the endogeneity in prices between agricultural and forestry markets is an important future research direction. In addition to developing a full specification of the agricultural market, it would also be useful to account for specific country-level risks (see, for example, Kindermann et al. [2006]). However, such analysis is beyond the scope of this paper.

In addition, the model used in this analysis is fairly aggregated with respect to land uses in tropical regions. In practice, there is much more spatial heterogeneity across the landscape, and actual carbon sequestration programs would need to take this heterogeneity into account when considering policies for reducing deforestation. Developing models that have more spatial resolution could help policymakers better target programs to specific regions or areas within the large, continental-scale results presented in this study. Nonetheless, when our results for South America are compared with the more spatially disaggregated results of Soares-Filho et al. (2006), the results are similar although the present analysis appears more conservative.


   ACKNOWLEDGMENTS
 
The authors appreciate comments made by participants in The Nature Conservancy Technical Advisory Panel meeting on 6 Mar. 2006, and participants in the workshop titled "Reducing Emissions from Deforestation in Developing Countries" at Bad Blumau, Austria, 10–12 May 2006. In addition, Brian Murray and Michael Obersteiner made many helpful comments. The authors remain responsible for any errors. This work was supported by the U.S. Environmental Protection Agency, Office of Atmospheric Programs, through RTI International under Contract GS-10F-0283K, TO# 14, and managed by Kenneth Andrasko. The views and opinions of the authors herein do not necessarily state or reflect those of the U.S. government or the U.S. Environmental Protection Agency. The authors also appreciate general support for model development provided by the Ohio State Univ. Climate, Water and Carbon program.


   NOTES
TOP
 NOTES
ABSTRACT
 INTRODUCTION
 Literature Review
 Methods
 Results
 Discussion and Conclusions
 REFERENCES
 
All rights reserved. No part of this periodical may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage and retrieval system, without permission in writing from the publisher.


   REFERENCES
TOP
 NOTES
 ABSTRACT
 INTRODUCTION
 Literature Review
 Methods
 Results
 Discussion and Conclusions
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Sohngen, B.
Right arrow Articles by Andrasko, K.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Sohngen, B.
Right arrow Articles by Andrasko, K.
Agricola
Right arrow Articles by Sohngen, B.
Right arrow Articles by Andrasko, K.
Related Collections
Right arrow Economics
Right arrow Ecosystem Management
Right arrow Carbon Sequestration
Right arrow Global Change
Right arrow Other Models

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome