Predicting the deforestation-trend under different carbon-prices
© Kindermann et al; licensee BioMed Central Ltd. 2006
Received: 13 October 2006
Accepted: 06 December 2006
Published: 06 December 2006
Global carbon stocks in forest biomass are decreasing by 1.1 Gt of carbon annually, owing to continued deforestation and forest degradation. Deforestation emissions are partly offset by forest expansion and increases in growing stock primarily in the extra-tropical north. Innovative financial mechanisms would be required to help reducing deforestation. Using a spatially explicit integrated biophysical and socio-economic land use model we estimated the impact of carbon price incentive schemes and payment modalities on deforestation. One payment modality is adding costs for carbon emission, the other is to pay incentives for keeping the forest carbon stock intact.
Baseline scenario calculations show that close to 200 mil ha or around 5% of todays forest area will be lost between 2006 and 2025, resulting in a release of additional 17.5 GtC. Today's forest cover will shrink by around 500 million hectares, which is 1/8 of the current forest cover, within the next 100 years. The accumulated carbon release during the next 100 years amounts to 45 GtC, which is 15% of the total carbon stored in forests today. Incentives of 6 US$/tC for vulnerable standing biomass payed every 5 year will bring deforestation down by 50%. This will cause costs of 34 billion US$/year. On the other hand a carbon tax of 12 $/tC harvested forest biomass will also cut deforestation by half. The tax income will, if enforced, decrease from 6 billion US$ in 2005 to 4.3 billion US$ in 2025 and 0.7 billion US$ in 2100 due to decreasing deforestation speed.
Avoiding deforestation requires financial mechanisms that make retention of forests economically competitive with the currently often preferred option to seek profits from other land uses. Incentive payments need to be at a very high level to be effective against deforestation. Taxes on the other hand will extract budgetary revenues from the regions which are already poor. A combination of incentives and taxes could turn out to be a viable solution for this problem. Increasing the value of forest land and thereby make it less easily prone to deforestation would act as a strong incentive to increase productivity of agricultural and fuelwood production, which could be supported by revenues generated by the deforestation tax.
Deforestation is considered the second largest source of greenhouse gas (GHG) emissions amounting to an estimated 2 gigatonnes of carbon (GtC) per annum over the last decade . It is a persistent problem. The UN Food and Agriculture Organization, in its recently released most comprehensive assessment of forests ever, puts deforestation at about 12.9 mil. ha per year . At the same time, forest planting, landscape restoration and natural expansion of forests reduce the net loss of forest area. Net change in forest area in the period 2000–2005 is estimated at -7.3 million hectares per year . This reduces the annual GHG emissions to an estimated 1.1 GtC. In comparison, 7.3 GtC were emitted in 2003 by using fossil energy sources .
Deforestation has been difficult to tackle by governments, as its drivers are complex and many land uses yield higher revenues than those from forested land. Some see climate policy as a new opportunity to effectively reduce a major source of greenhouse gases and biodiversity loss as well as to increase incomes of many people in rural areas whose livelihood depends on forests. The implementation of measures avoiding deforestation would require innovative financial mechanisms in the context of global climate policies. In this paper we study the potential magnitude of effects of different financial mechanisms to help reduce deforestation, using a modeling approach.
To estimate the impact of financial incentives, to reduce deforestation and assuming profit maximizing behavior, we calculate differences in net present value of different land uses using a spatially explicit integrated biophysical and socio-economic land use model. Key model parameters, such as agricultural land use and production, population growth, deforestation and forest product consumption rates were calibrated against historical rates. Land use changes are simulated in the model as a decision based on a difference between net present value of income from production on agricultural land versus net present value of income from forest products. Assuming fixed technology, the model calculates for each 0.5° grid cell the net present value difference between agricultural and forest land-uses in one-year time steps. When carbon market prices, transferred through a financial mechanism, balance out differences between the net present value of agricultural land and forest-related income, it is assumed, consistent with profit maximising behavior, that deforestation is avoided.
The net present value difference of forest versus other land uses can be balanced out through two mechanisms. One is to reduce the difference by adding costs to conversion through taxing emissions from deforestation, e. g. through a land clearance tax and wood sales taxes. The other is to enhance the value of the existing forest by financial support when keeping the forest carbon stock, to be paid in certain time intervals. In both cases the value of forest carbon stock would be pegged to carbon market prices. The modeling results for different hypothetical tax or subsidy levels show the potential magnitude of avoided deforestation through financial incentive or disincentive mechanisms. The model results are annual, spatially explicit estimates of the forest area and biomass development from 2000 to 2100, with particular focus on the period 2006 to 2025.
Results and discussion
Baseline deforestation 2000–2100 and effects of financial mechanisms aiming at cutting emissions in half
The assumption, that either only slash burn or all wood will be sold is unrealistic. Thus, a scenario where Latin America has 90% slash burn and 10% selling, Africa 50% slash burned and 50% selling and in the remaining area 10% slash burned and 90% selling, was examined. Under such scenario assumptions a carbon tax of 12 $/tC will cut deforestation in half. Also the assumption, that a carbon price will stay constant over time may not be close to reality but it can be used to see the long-term influence of a given carbon price.
We differentiate between the following cases:
Baseline: Introducing no carbon price.
Incentives: Introducing a carbon price which will be payed periodic for the carbon stored in the standing forest biomass.
All: Payments are done, without considering the effectiveness of the payment, in all regions.
Region: Payments are done in regions where the payments protect forest against deforestation.
Affected: Payments are done for forests where the payments protect them against deforestation.
Tax: Introducing a carbon price which has to be paid for releasing the stored carbon to the atmosphere.
Burn: All wood will be burned immediately.
Sell: All harvested wood will be sold.
Burn/Sell: A share of the wood will be burned and the other part sold.
Costs and revenues under different carbon prices
Regional effects of carbon prices on deforestation
Sources of deforestation in the model are expansion of agriculture and buildup areas as well as from unsustainable timber harvesting operations impairing sufficient reforestation. Deforestation results from many pressures, both local and international. While the more direct causes are rather well established as being agricultural expansion, infrastructure extension and wood extraction, indirect drivers of deforestation are made up of a complex web of interlinked and place-specific factors. There is large spatially differentiated heterogeneity of deforestation pressures. Within a forest-agriculture mosaic, forests are under high deforestation pressure unless they are on sites which are less suitable for agriculture (swamp, slope, altitude). Closed forests at the frontier to agriculture land are also under a high deforestation pressure while forest beyond this frontier are under low pressure as long as they are badly attainable. The model was build to capture such heterogeneity in deforestation pressures.
Avoiding deforestation requires financial mechanisms that make retention of forests economically competitive with the currently often preferred option to seek profits from other land uses. According to the model calculations, even relatively low carbon incentives of around 6 $/tC/5 year, paid for forest carbon stock retention or carbon taxes of 12 $/tC would suffice to effectively cut emissions from deforestation by half. Taxes revenues would bring about annual income of US$6 bn in 2005 to US$0.7 bn in 2100. The financial means required for incentives are estimated to range from US$3 bn to US$ 200 bn per year, depending on the design of the avoided deforestation policy. Our scenario, where incentives are payed in regions where deforestation will appear and the payment has an effect, estimates the necessary funds to cut emissions from deforestation in half in the magnitude of some US$ 33 bn per year, without including costs for transaction, observation and illegal logging protection. Increasing the value of forest land and thereby make it less easily prone to deforestation would act as a strong incentive to increase productivity of agricultural and fuelwood production.
The model is based mainly on the global afforestation model of  and calculates the net present value of forestry with equation (1 – 16) and the net present value of agriculture with equation (17 – 20). Main drivers for the net present value of forestry are income from carbon sequestration, wood increment, rotation period length, discount rates, planting costs and wood prices. Main drivers for the net present value of agriculture on current forest land are population density, agricultural suitability and risk adjusted discount rates.
These two values are compared against each other and deforestation is subsequently predicted to occur when the agricultural value exceeds the forest value by a certain margin. When the model comes to the result, that deforestation occurs, the speed of deforestation was constraint by estimates given by equation (24). The speed of deforestation is a function of sub-grid forest share, agricultural suitability, population density and economic wealth of the country.
All symbols in the following equations are explained in the section "Abbreviations".
Net present value of forestry
The net present value of forestry is determined by the planting costs, the harvestable wood volume, the wood-price and benefits from carbon sequestration.
For existing forests which are assumed to be under active managment the net present value of forestry given multiple rotations (F i ) over the simulation horizon is calculated from the net present value for one rotation (f i ) (equation 1). This is calculated by taking into account the planting costs (cp i ) at the begin of the rotation period and the income from selling the harvested wood (pw i ·V i ) at the end of the rotation period. Also the benefits from carbon sequestration are included denoted as (B i ).
The planting costs (eq. 3) are calculated by multiplying the planting costs of the reference country (cp ref ) with a price index (px i ) and a factor which describes the share of natural regeneration (pr i ). The ratio of plantation to natural regeneration is assumed to increase with increasing yield for the respective forests (eq. 4). The price index (eq. 5) is calculated using the purchasing power parity of the respective countries. The stumpage wood price (eq. 6) is calculated from the harvest cost free income range of wood in the reference country. This price is at the lower bound when the population density is low and the forest share is high and at the higher bound when the population density is high and the forest share is low. The price is also multiplied with a price index converting the price range from the reference country to the examined country. The population-density and forest-share was standardized between 1 and 10 by using equation (7) and equation (8) respectively.
The harvested volume (V i ) is calculated by multiplying the mean annual increment (MAI i ) with the rotation period length (R i ) accounting for harvesting losses (eq. 9).
The rotation period length (eq. 10) depends on the yield. Fast growing stands have a short and slow growing sites a long rotation length. In this study the rotation length is in the range between 5 and 140 years.
The mean annual increment (eq. 11) is calculated by multiplying the estimated carbon uptake (ω i ) and a transformation factor which brings the carbon weight to a wood volume (C 2W i ). The carbon uptake (ω i ) is calculated by multiplying the net primary production (NPP i ) with a factor describing the share of carbon uptake from the net primary production (eq. 12).
The benefits of carbon sequestration (eq. 13) are calculated by discounting the annual income from additional carbon sequestration and subtracting the expenses incurred from harvesting operations and silvicultural production. At the end of a rotation period the harvested carbon is still stored in harvested wood products and will come back to atmosphere with a delay. This is considered in the factor (θ i ) which shares the harvested wood volume to short and long living products(eq. 14).
The effective carbon price represents the benefit which will directly go to the forest owner. In equation (16) a factor describing the percentage of the transaction cost free carbon price is used. A factor leak i is calculated as the average of the percentile rank from "political stability", "government effectiveness" and "control of corruption" .
f i = -cp i + pw i ·V i + B i (2)
cp i = cp ref ·pr i ·px i (3)
SNFs = 1 + (1 - Fs) * 9 (8)
V i = MAI i ·R i ·(1 - HL i ) (9)
MAI i = ω i ·C 2W (11)
ω i = NPP i ·CU (12)
frac slp = 1 - frac llp (15)
epc i = pc i ·leak i (16)
Net present value of agriculture
The net present value of agriculture (A i ) is calculated with a two-factor Cobb-Douglas production function (equation 17). It depends on the agriculture suitability and the population density. A high agriculture suitability and a high population density causes high agricultural values. The value ranges between a given minimum and a maximum land price. The parameters a i and γ i determine the relative importance of the agriculture suitability and the population density and ν i determines the price level for land. The agriculture suitability and the population density are normalized between 1 and 10.
γ i = α i (20)
Decision of deforestation
The deforestation decision is expressed by equation (21). It compares the agricultural and forestry net present values corrected by values for deforestation and carbon sequestration. For the deforestation decision the amount of removed biomass from the forest is an important variable. The agricultural value needed for deforestation increases with the amount of timber sales and its concomitant flow to the HWP pool. On the other hand the agriculture value will be decreased by the amount of released carbon to the atmosphere. This mechanism is expressed by a deforestation value (DV i , eq. 22). The model also allows for compensation of ancillary benefits from forests. This additional income is modeled either as a periodical income or a one time payment and will increase the forestry value by (IP i ). If it is a periodic payment it has to be discounted, which has been done in equation (23).
There exist several ways of how financial transfers can be handled. Two mechanisms are realized in equation (21). One is to pay the forest owner to avert from the deforestation, the other is to introduce a carbon price that the forest owner gets money by storing carbon and paying for releasing it. The introduction of a carbon price focuses the money transfer to the regions where a change in biomass takes place. Payments to avoid emissions from deforestation can be transfered to cover all of the globe's forests, target to large "deforestation regions" or individual grids.
Coefficients for equation (25) – Deforestation speed
The deforestation rates (Ft dec ) were taken from , where the forest area from 1990, 2000 and 2005 for each country was given. For the estimation of the model parameters the area difference between 1990 and 2005 was used to infer the deforestation rate. All values which showed an increase of the forest area have been set to 0, because the model should only predict the deforestation. Countries with an increasing forest area have a deforestation rate of 0. It should be mentioned that the change rate is based on the total land area in the grid i and not on the current forest area.
By using c2/F s the model can only be used on grid's where there is some share of forest. This makes sense, because on places where there is no forest, no deforestation can appear. The model will only be usable on grids where forests occur. Therefore, for parameterization, the average agricultural suitability and the population density of a country are also only taken from grids which indicate forest cover.
Development of forest share
After calculating the deforestation rate, the forest share has to be updated each year with equation (27) assuring that the forest share stays within the permissible range of 0–1.
fsx i, year = Fsi, year - 1- F i, dec (28)
Aboveground carbon in forest biomass
The model describes the area covered by forests on a certain grid. It can also describe the forest biomass if the average biomass on a grid is known and the assumption was made, that the biomass in forests on the grid is proportional to the forest area.
For this reason a global carbon map of aboveground carbon in forest biomass, was created, based on country values from . By dividing the given total carbon, for each country, with the forest area of the country, the average biomass per hectare can be calculated. Now the assumption was made, that the stocking biomass per hectare on sites with a higher productivity is higher than on sites with a low productivity. Not for every country with forests  gives values of the stocking biomass. So a regression, describing the relation between tC/ha and NPP, was calculated and the biomass of grids of missing countries have been estimated to obtain a complete global forest biomass map.
In the simulations the effect of different carbon-prices and/or incentives, for keeping forest, have been tested. The simulation period started in the year 2000 and ends in 2100. The decision, whether deforestation takes place or not and how fast it goes on, was done in one year time steps. Scenario drivers, available on coarser time resolution (e.g. population density), have been interpolated linearly between the given years.
Outputs of the simulations are trajectoria of forest cover, changes in carbon stocks of forests, and financial resources required to cut emissions from deforestation under varying scenario assumptions.
Spatial dataset available on a 0.5° × 0.5° grid
1990 – 2015
1990 – 2100
1990 – 2100
2010 – 2080
2010 – 2080
Country level values
Fraction of long living products
Decay rate long
Decay rate short
Factor carbon uptake
Frequency of incentives payment
tC to m3
Maximum rotation interval
Minimum rotation interval
Carbon price incentives
Minimum Land price
Maximum Land price
Minimum wood price
Maximum wood price
- α i :
Importance of agriculture
- γ i :
Importance of population
- ν i :
Land price level = minimum land price of reference country × price index (px i ) [$/ha]
- ω i :
Carbon uptake per year [tC/year/ha]
- θ i :
Fraction of carbon benefits in products 
- A i :
Net present value of agriculture [$/ha]
- AgS i :
Agricultural suitability [0–1]
- b i :
Baseline, how much carbon uptake will be if there is no forest, e.g. 0.1 
- BMP i :
Biomass in Products [tC/ha]
- BM i :
Aboveground living wood biomass [tC/ha]
- B i :
Present value of carbon benefits [$/ha]
- Bul :
Share of buildup land 
- C 2W:
Conversion factor form 1t Carbon to 1m3 wood [m3/tC]
- cp i :
Planting costs [$/ha]
- cp ref :
Planting costs reference country [$/ha]
- CU :
Carbon uptake, share of NPP stored in wood 
- Crl :
Share of crop land 
- dec llp :
Decay rate of long living products e.g. 0.03 
- dec slp :
Decay rate of short living products e.g. 0.5 
- DV i :
Deforestation Value [$/ha]
- epc i :
Effectiv carbon price [$/tC]
- f i :
Net present value of forestry for one rotation period [$/ha]
- F i :
Net present value of forestry [$/ha]
- F s :
Actual share of forest [0–1]
- F dec :
Decrease of the forest share
- fr i :
Frequency of incentives money payment [Years]
- frac llp :
Fraction of long living products e.g. 0.5 [0–1]
- frac sb :
Fraction of slash burned area e.g. 0.9 [0–1]
- frac slp :
Fraction of short living products e.g. 0.5 [0–1]
- Fs :
Forest area share [0–1]
- Fs year :
Forest share of a certain year 
- fsx year :
Theoretical forest share of a certain year 
- Ft dec :
Theoretical decrease of the forest share
- GDP :
Gross domestic product [$1995/Person]
- H i :
Hurdle e.g. 1.5 
- HL i :
Harvesting losses e.g. 0.2 
- i :
- leak i :
Factor of money which will in real reach the forest 
- IP i :
Incentive payment [$/ha]
- MAI i :
Mean annual wood volume increment [m3/ha]
- NPP i :
Net primary production [tC/ha/year]
- pc i :
Carbon price [$/tC]
- pca i :
Incentives carbon price [$/tC/fr i ]
- Pd i :
Population density [People/km2]
- PL max :
Maximal land price of reference country × price index (px i ) [$/ha]
- PL min :
Minimal land price of reference country × price index (px i ) [$/ha]
- PPP i :
Purchasing power parity [$]
- PPP ref :
Purchasing power parity of reference country [$]
- pr i :
Ratio of area planted [0–1]
- pw i :
Stumpage wood price [$/m3]
- pw max :
Maximum revenue of wood, e.g. 35$/fm [$/fm]
- Pw min :
Minimum revenue of wood, e.g. 5$/fm [$/fm]
- px i :
Price index 
- r :
Discount rate [e.g. 0.05]
- R i :
Rotation interval length [years]
- SAgS i :
Standardized agricultural suitability [1-10]
- SFs :
Standardized not forest area share [1-10]
- SPd :
Standardized population density [1-10]
- V i :
Harvest wood volume [m3]
- x i :
Theoretical decrease of the forest share if Fs i > 0 ∧AgS i > 0
We acknowledge the support by the Greenhouse Gas Initiative (GGI) project, an institute-wide collaborative effort within IIASA. The interdisciplinary research effort within GGI links all the major research programs of IIASA that deal with research areas related to climate change, including population, energy, technology, and forestry, as well as LUCs and agriculture. GGI's research includes both basic and applied, policy-relevant research that aims to assess conditions, uncertainties, impacts, and policy frameworks for addressing climate stabilization, from both near-term and long-term perspectives. Support from the EU FP 6 Project Integrated Sink Enhancement Assessment (INSEA, SSPI-CT-2003/503614 with DG RTD) is gratefully acknowledged.
We graceful thank the reviewers for their reports.
- IPCC International Panel on Climate Change, ed Robert T Watson and Core Writing Team:Climate Change 2001: Synthesis Report. A Contribution of Working Groups I, II, and III to the Third Assessment Report of the Integovernmental Panel on Climate Change. Cambridge University Press, Cambridge, United Kingdom, and New York, NY, USA; 2001. [http://www.ipcc.ch/pub/syreng.htm]Google Scholar
- FAO: Global Forest Resources Assessment 2005, Progress towards sustainable forest management. Volume 147. FAO Forestry Paper. Rome: ood and Agriculture Organization of the United Nations; 2005.Google Scholar
- Marland G, Boden T, Andres R: Global, Regional, and National CO2 Emissions. In Trends: A Compendium of Data on Global Change. Carbon Dioxide Information Analysis Center, Oak Ridge National Laboratory, U.S. Department of Energy, Oak Ridge, Tenn., U.S.A.; 2006.Google Scholar
- Benítez PC, Obersteiner M: Site identification for carbon sequestration in Latin America: A grid–based economic approach. Forest Policy and Economics 2006, 8: 636–651. 10.1016/j.forpol.2004.12.003View ArticleGoogle Scholar
- Kaufmann D, Kraay A, Mastruzzi M:Governance Matters IV: Governance Indicators for 1996–2004, World Bank Policy Research Working Paper Series No. 3630. World Bank. 2005. [http://ssrn.com/abstract=718081]Google Scholar
- R Development Core Team:R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria; 2005. [ISBN 3–900051–07–0]. [http://www.R-project.org] [ISBN 3-900051-07-0].Google Scholar
- World Bank: World Development Indicators. World Bank. 2005.Google Scholar
- Benítez P, McCallum I, Obersteiner M, Yamagata Y: Global Supply for Carbon Sequestration: Identifying Least-Cost Afforestation Sites Under Country Risk Consideration. Tech. rep., International Institute for Applied System Analysis; 2004.Google Scholar
- Obersteiner M, Alexandrov G, Benítez PC, McCallum I, Kraxner F, Riahi K, Rokityanskiy D, Yamagata Y: Global Supply of Biomass for Energy and Carbon Sequestration from Afforestation/Reforestation Activities. Mitigation and Adaptation Strategies for Global Change 2006, 1381–2386.Google Scholar
- Alexandrov GA, Yamagata Y, Oikawa T: Towards a Model for Projecting Net Ecosystem Production of the World Forests. Ecological Modelling 1999, 123: 183–191. 10.1016/S0304-3800(99)00128-3View ArticleGoogle Scholar
- JRC:The Global Land Cover Map for the Year 2000. GLC2000 database. European Commision Joint Research Centre; 2003. [http://www-gvm.jrc.it/glc2000]Google Scholar
- CIESIN: Center for International Earth Science Information Network (CIESIN), Columbia University; and Centro Internacional de Agricultura Tropical (CIAT). Gridded Population of the World Version 3 (GPWv3): National Boundaries. Palisades, NY: Socioeconomic Data and Applications Center (SEDAC), Columbia University.2005. [http://sedac.ciesin.columbia.edu/gpw]Google Scholar
- CIESIN: Center for International Earth Science Information Network, Columbia University; and Centro Internacional de Agricultura Tropical (CIAT). Gridded Population of the World Version 3 (GPWv3): Population Density Grids. Palisades, NY: Socioeconomic Data and Applications Center (SEDAC), Columbia University.2005. [http://sedac.ciesin.columbia.edu/gpw]Google Scholar
- Grubler A, Nakicenovic N, Riahi K, Wagner F, Fischer G, Keppo I, Obersteiner M, O'Neill B, Rao S, Tubiello F: Integrated assessment of uncertainties in greenhouse gas emissions and their mitigation: Introduction and overview. Technological Forecasting and Social Change, in press.
- Tubiello FN, Fischer G: Reducing climate change impacts on agriculture: Global and regional effects of mitigation, 2000–2080. Technological Forecasting and Social Change, in press.
- WDPA Consortium: World Database on Protected Areas. Copyright World Conservation Union (IUCN) and UNEP-World Conservation Monitoring Centre (UNEP-WCMC). 2004.Google Scholar
- Ramankutty N, Foley JA, Norman J, McSweeney K: The global distribution of cultivable lands: current patterns and sensitivity to possible climate change. Global Ecology & Biogeography 2002,11(5):377–392. 10.1046/j.1466-822x.2002.00294.xView ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.