A first cost benefit analysis of action to reduce deforestation

Emissions from deforestation are globally significant. Understanding the benefits and costs of reducing deforestation and forest degradation would allow us to plan a pervasive monitoring, verification and reporting infrastructure as part of a Planetary Skin. The areas most vulnerable to deforestation are mainly concentrated in tropical countries and the bulk of emissions arise when the land is converted to agricultural production (Stern, 2006). While emissions from deforestation contribute significantly to levels of greenhouse gases, there is potential for them to be cut considerably. No new technology is needed to facilitate action and cutbacks could be made relatively quickly. It has subsequently been claimed that curbing deforestation is a highly cost-effective way of reducing greenhouse gas emissions (Stern, 2006). In this paper, we assess the validity of this claim. Using a newly-developed version of the PAGE2002 model, CCPAD, we show the costs and benefits of taking action to reduce emissions from deforestation and forest degradation (REDD). Our results indicate that introducing 50% REDD in 2010 would cost $1.7 trillion (90%CI: $0.7 $3.3 trillion) and create a mean drop in impacts of $5.3 trillion (90%CI: $0.6 $17 trillion), bringing a mean net benefit of $3.7 trillion (90%CI: $-0.7 $14.3 trillion). This positive mean net benefit indicates that REDD actions are viable and worthwhile. We also examine several policy alternatives to assess the costs and benefits of introducing REDD actions after 2010, at varying scales, and on top of aggressive abatement elsewhere. Our results clearly indicate that REDD actions bring higher benefits the earlier, and more aggressively, they are applied. Introduction Emissions from deforestation are very significant globally. These emissions could potentially be cut significantly fairly quickly – no new technology has to be developed. The areas of globally significant forest most vulnerable to deforestation are mainly concentrated in tropical countries. The bulk of emissions from deforestation arise when the land is converted to agricultural production. (Stern, 2006) To deal with climate change we need to make the connection between the global problem – global impacts and the response in terms of limiting global emissions – and the local level where the costs and benefits are experienced and where implementation, design and investment solutions need to be carefully applied. Bridging the two requires a complex “middle layer” in which the global institutional architecture needs to be carefully designed to provide a trusted, credible, verifiable and fungible system of global carbon emission reductions (UNFCCC, 2008). The success of this vital intermediary set of vehicles will rely on the pervasive monitoring and accrediting of emissions at the local level and the broadening and deepening of carbon markets at the global level. The process must draw together a wide array of actors across sectors, institutions and regions in a system that is effective at managing the risks of extreme events and is efficient in minimizing the costs of action by inducing abatement where it is cheapest. In the process it will generate large financial flows between buyers and issuers of carbon credits, with as much as $50-100bn a year flowing to developing countries by 2020 (Office of Climate Change, 2007) This ‘middle layer’ can be thought of as an essential global utility. It is required to provide a massive deployment of trusted and reputable open-standards based sensing, authentication, certification, and monitoring capabilities with transparent access, auditability and low transaction costs. These new capabilities would also need to be widely distributed and networked globally so as to be able to incorporate the large but highly fragmented mitigation opportunities across markets and sectors. Extending this concept further we can envision these sensing and monitoring capabilities extended into the critical areas of deforestation, water, biodiversity and food productivity, risk management of nuclear waste and carbon sequestration and storage to mention just a few – to form a Planetary Skin. Devastation of the world’s rainforests is occurring at the rate of one England per year (FAO, 2005) and is the second largest contributor to global greenhouse gas emissions after the power sector (Houghton, 2003). The destruction of this invaluable planetary resource is driven by a number of factors: The links between food, fuel, fibre and forests which have contributed to recent severe food price inflation. This has arisen from land competition between food and fuel, a competition artificially created by short-sighted biofuel policies as well as increases in demand for meat caused by increasing wealth in China, India and elsewhere in the developing world. At the same time we face regional economic development challenges for the 1.6 billion poor living off the forests’ natural resources (Chomitz et al, 2006) The market and governance failures at play here involve many organisations at many levels, from global companies and public institutions to the forest communities themselves. But all conspire to make forests more valuable dead than alive. Reversing this deadly trend will require mechanisms to manage deforestation hotspots by establishing forest property rights so that carbon sequestration services can be priced and marketed. Forest carbon stocks will need to command a predictable price that can compete with food, fibre, fuel and cattle grazing. To do this it is critically important to understand the drivers of benefits and costs and the associated uncertainties of actionable policies to drastically reduce tropical deforestation and degradation. For this a model is required. Background to the Cisco Cambridge Policy Analysis of Deforestation (CCPAD) Model In an earlier paper, we used PAGE2002, a probabilistic integrated assessment model used in the Stern review (Stern, 2006), to calculate the impacts of BAU deforestation on climate change (Hope, 2008). In this paper we use a new version of the model to calculate the costs and benefits of taking action to reduce emissions from deforestation and forest degradation (REDD). We first look at the costs and benefits of 50% REDD starting in 2010 and then compare those costs to two policy alternatives to illustrate the benefits of taking immediate and vigorous action. Changes to PAGE2002 The standard PAGE2002 model doesn’t allow for declining emission cutbacks in later years (Hope, 2006). Any cutbacks made are assumed to last for ever. As the maximum cutback for REDD could occur as early as 2010, the standard PAGE2002 model would estimate the cost of REDD without taking later REDD policies into account. This is not reasonable, as it is clear that later REDD actions will also bring costs. The logic for the standard treatment of abatement in PAGE2002 is that it involves capital expenditure that cannot be recouped. – In addition, standard abatement tends to increase with time, so the issue of declining cutbacks rarely arises. The type of activity necessary for REDD will involve annual payments for as long as deforestation is avoided. So, if the amount of REDD declines, the payments will also decline (Richards and Jenkins, 2007, box 3). This requires a change in the model to remove the constraint that cutbacks cannot decline. This is done by replacing the equation representing the cutbacks for gas g, in analysis year i and region r: [41] { } r g i r g i r g i r g i ZC CB CB , , , , , , 1 , , ER , max − = − % with a new equation representing the cutbacks in Gtonne instead of % for CO2 (gas 1): [41] ( ) { } 1000 / E ) 100 / ) ER ( ( , 0 max , 1 , 0 , 1 , , 1 , , 1 , 1 , 1 , r r i r i i i r i r i ZC Ylo Yhi CB CB ⋅ − ⋅ − + = − Gt The cost equation is also changed to reflect the new units. The cost of prevention for CO2 (gas 1), in analysis year i and region r is [44] { 1000 )} ( )) ( , 1 , 1 , , 1 , 0 , , 1 , 1 , , 1 , 1 , 1 , 1 , ( ⋅ − ⋅ + + ⋅ = ≤ r r i r r r i r r r r i CB PC CB MAX CH MAX CL MAX CL if $M The relevant variables are defined in table 1. Table 1 variables in the PAGE2002 model affected by the change Variable Description Unit E Emissions Mtonne ER Emissions compared to base year % Yhi End of analysis period year Ylo Start of analysis period year ZC Zero cost emissions compared to base year % CB Cutbacks in emissions compared to base year Gtonne CL Costs of cheap preventative action $M/Mtonne CH Additional costs of expensive preventative action $M/Mtonne MAX Cheap cutbacks compared to base year % PC Preventative costs $M This change correctly accounts for the costs of the REDD emission reductions for as long as they are implemented. The new model is called the Cambridge Cisco Policy Analysis of Deforestation (CCPAD) model, and is identical to PAGE2002 apart from this change. Inputs Deforestation estimates in the main part of this paper are taken from Houghton, 2003. Unit costs of REDD are taken from Kindermann et al, 2008, converted to the annual payment form used in the CCPAD model. It is difficult to see any ‘break points’ that help to define the ‘lower cost’ and ‘extra cost’ ranges as required by the CCPAD model. This means that a large amount of judgement is required to interpret the data in this paper and produce the input values shown in table 2 which are used in the CCPAD runs. All parameter values are independent triangular probability distributions. Taking the most likely values for illustration, the first 30 GtCO2 of emission reductions in Latin America will cost $0.15 per tonne of CO2 per year, and any reductions beyond this will cost $0.60 ($0.15 + $0.45) per tonne of CO2 per year. REDD costs in Asia are 60% of this (but with a larger range, possibly up to twice as costly), while costs in Africa are 80% of those in Latin America. All costs are in year 2000 $US. Table 2 Unit costs of REDD actions min most likely max CO2 low cost in Latin America 0.06 0.15 0.3 $ per tCO2/yr CO2 added cost in Latin America 0.3 0.45 0.75 $ per tCO2/yr Asia Preventative costs factor 0.4 0.6 2 Africa Preventative costs factor 0.6 0.8 1 L America low cost CO2 range 15 30 50 GtCO2 Africa low cost CO2 range 40 50 60 GtCO2 China low cost CO2 range 4 8 16 GtCO2 India low cost CO2 range 6 12 24 GtCO2 Source: based on Kindermann et al, 2008 Other emissions are taken from IPCC scenario A2, and all other parameters are as in Hope, 2008. In later sections, results are also calculated with alternative estimates of deforestation and other emissions, to check the sensitivity of any policy recommendations to alternative assumptions. Results REDD starting in 2010 at 50% of deforestation Figure 1 shows the difference in global annual emissions of CO2, with and without the 50% REDD policy starting in 2010. Figure 2 shows that the mean costs of 50% REDD are $39 bn per year in 2030 (90%CI: $24 – 57 bn per year) and rise to $121 bn per year by 2100 ( 90%CI: $85 $166 bn per year). Figure 1 Difference in CO2 emissions of 50% REDD starting in 2010 by year