WPS8648 Policy Research Working Paper 8648 Forest Carbon Supply in Nepal Evidence from a Choice Experiment Sahan Dissanayake Randall A. Bluffstone E. Somanathan Harisharan Luintel N. S. Paudel Michael Toman Development Economics Development Research Group November 2018 Policy Research Working Paper 8648 Abstract This paper uses a choice experiment conducted in Nepal the core institution within which Reduced Emissions from during 2013 to estimate household-level willingness to Deforestation and Forest Degradation is implemented in participate in a village-level program under the Reduced Nepal and likely other countries. The study finds that aver- Emissions from Deforestation and Forest Degradation age and median values of payment required for agreement initiative requiring reductions in fuelwood collection, as to reduce fuelwood collection are substantially larger for a function of the price paid per unit of avoided carbon formal forest user groups than in informal communities. dioxide emissions. The analysis examines incentives to This reflects that formal groups likely already have fuelwood participate both in villages having formal community collection restrictions in place, whereas informal groups forest management, the core institution for implement- may de facto permit open access extraction. The analy- ing Reduced Emissions from Deforestation and Forest sis also suggests that households that are part of informal Degradation, and villages having only informal forest user groups react to Reduced Emissions from Deforestation and groups. Contrary to previous findings in the literature Forest Degradation very differently than households that about participation incentives, but in keeping with other are formal group members. Broadly speaking, “underpriv- recent studies of Reduced Emissions from Deforestation ileged” formal group member households, such as those and Forest Degradation pilots in Nepal, this study finds who are landless, female-headed, and poor, appear to be that relatively little emission reduction would take place at warier of fuelwood collection restrictions and thus require prices of $1.00 to $5.00 per ton of avoided carbon emis- higher payments than average respondents. This difference sions. Formal community forests will almost certainly be does not appear to carry over to informal group members.. This paper is a product of the Development Research Group, Development Economics. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/research. The authors may be contacted at sahan@pdx.edu. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Forest Carbon Supply in Nepal: Evidence from a Choice Experiment Sahan Dissanayake1 Randall A. Bluffstone E. Somanathan Harisharan Luintel N. S. Paudel Michael Toman JEL codes: Q23 Q54 Q56 Keywords: Nepal; Carbon Sequestration; Carbon Supply; Choice Experiment Acknowledgements The authors acknowledge the contributions of ForestAction Nepal to the collection of field data and Alec Kretchun for the preparation of the research site map. They also gratefully acknowledge the expert research assistance of Russell Drummond, without which this paper would be on a much shakier foundation. Bluffstone thanks the Fulbright Scholar Program and USEF-Nepal, which provided support for his input into the paper. Financial support to carry out the research was provided by the World Bank’s research budget. 1  Corresponding author sahan@pdx.edu. Affiliations: Dissanayake, Portland State University; Bluffstone, Portland State University; Somanathan, Indian Statistical Institute; Luintel, Portland State University and ForestAction Nepal; Paudel, ForestAction Nepal; Toman, World Bank. Development Research Group. 1. Introduction and Key Literature Meeting the Paris Agreement goal to contain the increase in average surface temperature to 1.5 to 2.0 degrees Celsius will imply significant costs, but if costs differ across the world economy, there is an opportunity to improve cost effectiveness by prioritizing reductions from those who can reduce net CO2e emissions the cheapest. In the main cost estimate reported in the 5th assessment, the IPCC indeed assumes that costs are divided up in such a way as to achieve a cost-effective allocation of CO2e emissions abatement effort (IPCC, 2014). Knowledge of marginal abatement costs – and especially where in the world economy low-cost abatement opportunities exist - is therefore critical for cost effectively allocating abatement effort and reducing total costs. Information about CO2e emissions abatement costs is particularly critical for designing payments ecosystem services (PES) programs, otherwise it is possible that compensation levels will be insufficient to spur abatement. One particularly important climate change-related PES program is Reduced Emissions from Deforestation and Forest Degradation (REDD+), which is a program by which The UN Framework Convention on Climate Change (FCCC) Annex 1 countries provide support to non-Annex 1 countries in exchange for measurable additional carbon sequestration (Angelsen, 2010; Bluffstone et al., 2013). What drives REDD+ is the notion that carbon sequestration in developing country forests is a class of particularly cost-effective climate investments; it therefore is cost-effective in a global sense for Annex 1 countries to “buy” forest carbon emission reductions from lower-income countries. Low-income countries emit relatively little carbon pollution but contribute to climate change through land use change and especially loss of forest biomass. Indeed, the majority of net deforestation and forest degradation occurs in tropical and sub-tropical areas, which is the location of virtually all low-income countries (Pan et al., 2011). Forest biomass loss from tropical land use change   2 is an important source of carbon emissions and an estimated net carbon source of 2.4  0.4 Gt per year (Pan et al., 2011). Net deforestation and forest degradation make up 12% to 20% of total greenhouse gas emissions (GHG), which is greater than all transport sources combined (Saatchi et al., 2011; van der Werf, 2009). The purpose of this paper is to estimate a nationally representative, household-level carbon supply function for rural Nepal. We model carbon supply as the marginal willingness to accept (WTA) payments for reducing fuelwood collections from common forests as part of a hypothetical REDD+ program. To estimate marginal willingness-to-accept, we use a choice experiment conducted in 2013 with 1,300 household heads or other household representatives. Choice experiments allow us to consider and decompose the effects of various aspects of our hypothetical PES program, where our particular interest in this paper is fuelwood reductions and elimination of open grazing; both these changes reduce emissions of carbon and increase sequestration. Using this experimental framework, we find relatively high median WTA values per ton of CO2e emissions avoided ($33 - $56) when we include best available information on avoided emissions. Both sets of estimates are consistent with recent literature from Nepal, but substantially above earlier estimates. We then examine factors that shift estimated carbon supply functions by estimating correlations between the marginal willingness to accept payments and a variety of socioeconomic variables. We find that respondents who are poor, members of female-headed households and have bigger families have higher WTA values. Those who own land, use biogas and believe they will benefit from REDD+ are estimated to require lower payments to reduce fuelwood collections. We also find that the opportunity costs of CO2 emissions reductions due to eliminating grazing are likely to be relatively high.   3 In most lower-income countries, such as Nepal, deforestation is less of an issue than degradation. Villagers collect forest products, such as fuelwood, forest fruits and vegetables, building materials and animal fodder, which reduces forest biomass. Fuelwood is a particularly important forest product (Cooke et al., 2008) and almost 3 billion people worldwide cook with biomass on a daily basis (Jeuland and Pattanayak, 2012). This dependence on fuelwood is expected to remain virtually constant through 2030 and decline only slightly by 2040 (IEA, 2006; 2014). Cooking with wood contributes to climate change via the release of carbon dioxide from non- sustainably harvested wood, plus other pollution emissions, including black carbon, carbon monoxide, and methane, even if the wood is sustainably harvested (Bailis et al., 2015). A number of these pollutants are particularly potent climate forcers, with black carbon having an estimated global warming potential (GWP) central estimate, which is 900 times more potent than CO2 (Bond et al. 2013).2 Cooking with wood using mainly inefficient technologies can therefore have particularly important effects on the climate. Though the fraction of people in Nepal who use liquefied petroleum gas (LPG) for cooking increased from 8 percent in 2001 to 17 percent in 2010 (Kanel et al., 2012), biomass makes up approximately 70% of the total energy used, 86% of households use fuelwood and 75% collect it themselves (Kandel et al., 2016). The poor are most dependent on fuelwood. About 95 percent of the poorest households use fuelwood as a source of household energy, but only 58 percent of the richest quintile use fuelwood (Kanel et al., 2012). With such a high dependence, reducing collection of forest products, such as fuelwood, to reduce greenhouse gas emissions imposes costs on those who rely on fuelwood for their cooking                                                              2 There is substantial uncertainty about the GWP of black carbon, mainly due to aerosol-cloud interactions. As in this paper several GHGs other than CO2 are examined, we use CO2 and CO2 equivalent (CO2e) interchangeably, with other gases converted to CO2e using their GWPs.   4 needs. These costs may come in the form of reduced convenience due to the need to conserve wood, increased monetary costs as households shift to higher cost, less GHG-intensive fuels like LPG (e.g. see Smith et al., 2000) or greater time costs as they plant trees on their own lands, because they cannot harvest enough wood from forests. An estimated 25% of developing country forests are managed by communities (Agrawal et al., 2008; World Bank, 2009) and about 15.5% of global forest area is under the formal control of communities (RRI, 2014). These observations make formal, informal, de jure and de facto community- controlled forests (Bluffstone et al, 2013) especially important sources of GHG emissions or carbon sequestration, particularly as most community forests are in low and lower- middle income countries where net forest biomass loss is occurring. Nepal has for many decades been a leader in formal community control of forests. It has over 19,000 formal, registered forest user groups that include over 35% of the population. As of 2015, in hill areas that are home to the majority of Nepal’s population, over 78% of households were formal community forest user group (CFUG) members.3. For example, in the hill district of Salyan (population approximately 250,000) in 2014 there were 558 CFUGs.4 These institutions have characteristics and practices that may differ from those outside the program. It is therefore important in any analysis to distinguish community forests (CFs) and non-community forests (non-CFs). These CFUGs in Nepal implement management structures in consultation with government officials and impose collection restrictions, including on fodder, firewood and timber. Communities                                                              3 National information available from the Ministry of Forest and Soil Conservation (MoFSC) Department of Forestry (DoF) Community Forestry Division website http://dof.gov.np/dof_community_forest_division/community_forestry_dof. Hill district information based on author calculations and data from the Community Forests Database (August 2015) available at http://dof.gov.np/image/data/Community_Forestry/Summary.pdf and the 2013 Nepal Statistical Yearbook (most recent available), which is available at http://cbs.gov.np/publications/statisticalyearbook_2013. All accessed August 16, 2016. 4 See http://www.dfosalyan.gov.np/eng/images/pdf/database/database_of_cfugs.pdf for details.   5 in our sample use a variety of mechanisms to control extractions of fuelwood, including restricting harvest technologies and number of harvesting household members (34%). They also allocate certain periods, often in winter (27%), holidays (10%) and specific days of the week, such as Saturday only (20%) for collection. Over two-thirds of non-CFs also take explicit steps to protect forests and approximately one-third even have written rules that govern extractions. Those without governance features are likely ungoverned and similar to open access. Actual estimates of prices required to incentivize carbon sequestration in low-income countries are very limited. Strassburg et al. (2009) use simulation methods to estimate the costs to reduce GHG emissions due to forest biomass loss in the 20 most forested developing countries. They find that a price of CO2e of $8.00 per ton would reduce global emissions by 90%. Similar simulation methods were used by Kinderman et al. (2008) and they find it would cost $10-$21 per ton of CO2e emissions avoided to achieve a 50% reduction in deforestation.5 These findings suggest that carbon sequestration can effectively compete with other mitigation approaches like renewable energy, energy conservation and fuel switching. At the same time, as discussed by Bluffstone et al. (2013), there is some controversy regarding the degree to which all local opportunity costs of carbon sequestration are included in such approaches (Gregorsen et al. 2011). In Nepal, Karky and Skutsch (2010) examine three REDD+ pilot sites and directly estimate the net benefits from three forest management scenarios, two of which include explicit mitigation within the context of a hypothetical REDD+ program. They estimate that households in all three sites benefit if the price per ton of CO2e is $5.00 and two of three benefit if the price is $1.00 per ton.                                                              5Costs to achieve a 10% reduction in deforestation were estimated to be $2 to $5 per ton of CO2e. This difference is due to increasing marginal costs of sequestration.   6 They estimate breakeven prices of $0.55 to $3.70 per ton of CO2 depending on the site and level of harvest reduction required. Marseni et al. (2014) analyze eight REDD+ pilot projects in two districts (Chitwan and Gorkha) in Nepal. They attempt to catalogue, and to the degree possible measure, the opportunity costs associated with accepting REDD+ payments to generate carbon credits. Costs are mainly in terms of foregone forest product collections (fuelwood, grass, fodder, timber, grazing), but also include time costs. They note that within the community forestry context, to participate in REDD+ there are large meeting and forest management time costs (e.g. fire control, planting thinning, pruning, cleaning, weeding). They note that the average payment of $1,734 across the eight sites does not cover estimated opportunity costs. They also note that a price of $10/ton of CO2 (at that time the EU Emissions Trading Scheme average price) would not even cover average meeting time opportunity costs. They do not estimate breakeven costs, but express doubt that REDD+ can be truly cost effective. Pandit et al. (2017) examine the economics of carbon sequestration in Nepal using household survey data from 47 CFUG REDD+ pilot sites in two of Nepal’s 75 districts. Half the sites are in the plains of Nepal (i.e. the Terai) and half in the hills. They estimate net benefits of the REDD+ program (as was done by Marseni et al., 2014), but also calculate average breakeven carbon prices. They find that if the opportunity cost of unharvested fuelwood is taken into account, at a carbon price of $10 per ton no CFUGs receive positive net benefits. Assuming that biomass would not have otherwise been set aside as an environmental “safeguard,” they find that the average Terai6 CFUG opportunity cost ranged from $50 to $74 per ton of CO2, depending on assumptions about the ratio of above ground (and therefore harvestable) to below ground biomass. In the hills, they estimate average costs                                                              6  The Terai is the plains region bordering India.   7 of $20 to $30 per ton of CO2. They conclude that opportunity costs are relatively high and significant regional heterogeneity in opportunity costs due to ecosystem productivity, opportunity costs of time, etc. may exist. Following this guidance, in our paper we also examine heterogeneous effects. 2. Methods and Data The goal of this paper is to estimate the supply of carbon sequestration in tons of CO2e from community-controlled forests as a function of hypothetical payments for reducing fuelwood harvests and grazing. Carbon supply is the behavioral response in terms of emissions reduced expected from offering a particular monetary incentive and because in rural Nepal most carbon emissions come from burning fuelwood, this paper primarily focuses on this potential source of reduced CO2 emissions. In our sample, 95% of households cook with fuelwood as their main fuel, sometimes in conjunction with other fuels and associated technologies. In rural areas of low-income countries like Nepal, grazing is estimated to be a significant source of above ground biomass reduction and therefore carbon emissions. Based on the results of several studies (e.g. Keller, 1998), IPCC (2000) recommends an estimate of 0.4 – 0.5 tons per hectare carbon gains (1.464 – 1.83 tons of CO2) due to grazing elimination in dryer areas. Though we are unable to provide very detailed estimates of carbon supply due to grazing, we offer estimates of the cost to incentivize the elimination of open grazing and the CO2 emissions reductions that can be expected from those payments. Less harvesting of fuelwood can potentially reduce carbon emissions if such actions result in increased efficiencies, replacing unsustainable harvest sources (e.g. open access forests) with sustainable sources, such as wood from trees planted on users’ own farms or use of biogas from existing animals. Of course, if households comply with an agreement to reduce fuelwood harvests from their own forests by simply harvesting fuelwood from another forest, then carbon sequestration   8 gains are reduced or eliminated due to such leakage. Furthermore, if open grazing elimination were to move cattle to nearby common lands, there would also be leakage. Understanding the likely extent of leakage is therefore critical for understanding the degree to which PES schemes, such as REDD+ actually achieve stated goals. In this paper, we are not able to conclusively address the possibility for leakage, but we do present descriptive results that suggest more stringent fuelwood collection limits would not cause households to primarily rely on alternative common forests. To estimate carbon supply functions, we use marginal effects derived from a choice experiment conducted in Nepal during April – June 2013. Alpizar et al. (2003), Hanley et al. (2001), Hensher et al. (2005), and Hoyos (2010) provide reviews of the choice experiment methodology. The purpose of this experiment is to examine preferences for REDD+ contracts in 1,300 households in 130 formal and informal forest user groups across Nepal.7 The sample was developed based on a nationally representative sample of 137 forests analyzed during an evaluation of the Nepal Community Forestry Program (MoFSC, 2013). We randomly chose 65 official community forests (CFs) from this sample and then matched it with 65 communities outside the program so our samples of CFs and non-CFs could be compared; half of respondents are in CFs and half in non-CFs. We limited our sample to communities in the middle hills and Terai regions of Nepal and omitted the high mountain region, which offers limited carbon sequestration potential. Figure 1 shows the sample distributed across Nepal.                                                              7 Comprehensive results from the analysis can be found in Dissanayake et al. (2015), but this paper did not explicitly use the results to analyze carbon supply. We therefore extend our earlier paper here.   9 Figure 1: Map of Research sites We start with a nationally representative CF sample and matched those areas with a sample of non-CF communities that have similar ecological and socio-economic characteristics. Using this methodology, we can draw inferences for rural areas across the country. This sample has been used for a variety of analysis, including examining carbon sequestration effects of CFs (Bluffstone et al., 2018), collective action drivers of carbon sequestration (Luintel et al., 2017) and equity effects of the CF program (Luintel et al., 2017), and, as was already mentioned, responses to hypothetical REDD+ contracts (Dissanayake et al., 2015), among others.   10 The context of our choice experiment is a set of hypothetical REDD+ contracts, where each contract has multiple attributes; the amount of the REDD+ payment, required fuelwood harvest reductions, required grazing closures, and the division of the payment between private households and the community. In this paper, we are interested in fuelwood and grazing ecosystem services and what level of compensation households would require to forego those services. Choice experiments are ideal for estimating the value of ecosystem services within such frameworks (Boxall et al. 1996, Louviere et al. 2000, Dissanayake and Ando 2014, Witkin 2015), because they allow the researcher to elicit marginal values for each attribute. The attributes were selected through focus group discussions in a total of 18 communities, half CFs and half non-CFs, including both hill and Terai regions.8 These attributes are presented in Table 1. As shown in the table, respondents who were not CF members had four attributes; contract payment denominated per household, percentage of the payment going to the household (as opposed to the community), required reduction in fuelwood and the required reduction in grazing. The grazing restriction attribute was not included in choice experiments for those who are members of CFs, because CFs virtually always have grazing restrictions. An important implication of having different choice experiments for CF member and nonmember households is that these samples must be analyzed independently; we therefore do not pool the data.                                                                8 The attributes were refined after examining the REDD+ literature and analyzing focus group results. Lancaster (1966) was one of the first to suggest that consumers get benefits from the characteristics of services rather than services themselves. Choice experiments can therefore be considered analogous to hedonic pricing, except for a stated preference setting. As is the case with all experimental methods, hypothetical bias and lack of external validity are potentially important concerns (e.g. see Carlsson, F. and P. Martinsson, 2001).   11 Table 1: Attributes and Levels in the Choice Experiment Attributes Description Levels REDD + payments (Rs. per Annual total REDD+ payment to  Rs. 1000 household per year) your community.  Rs. 2000  Rs. 3000  Rs. 4000  Rs. 5000 Portion of the REDD+ The portion of REDD+  100% community payment going to the payments that go to communities  50% community and household. for community projects and /or 50% household equally divided between  100% household households in your group. Reduction in amount of fuel Required fuelwood reduction  25% wood collected measured as a portion of your  50% current use.  75%  100% Only Households who are Not CF Members Grazing restrictions Open grazing is prohibited  Yes (only for non-CF  No households) Once an initial list of attributes was developed, we pretested the instrument with potential respondents. The final choice experiment contains background information about the REDD+ program, a description of the attributes and levels, seven choice sets, and a small demographic questionnaire. Appendix A provides the actual background information document used for the choice experiments. For each of the choice sets, the respondents chose between two alternatives and the status quo. Figure 2 presents an example of a choice set, with the first row showing the payment, the second presenting the distribution of the payment (100% household, 100% community and 50:50) and the last indicating the required fuelwood reduction. All attributes are depicted pictorially as well as verbally, because many respondents were illiterate. When conducting a choice experiment, the respondent makes choices over bundles of attributes and it is necessary to generate choice cards and identify the levels of each attribute that will   12 appear in each choice set. We follow standard practice in the choice modeling literature (Adamowicz et al. 1997, Adamowicz et al. 1998, Louviere et al. 2000) and create an efficient experimental design that allows both main effects and interaction effects to be estimated. The designs for the choice experiments were generated following Kuhfeld (2010)9 and achieve a 100% D-efficiency (Huber and Zwerina 1996, Carlsson and Martinsson 2003). The experimental design yielded 30 unique choice profiles for CF households and 60 unique choice profiles for non-CF households (Kuhfeld 2010, Vermeulen et al. 2008). We create a block design giving five unique experiments for CFs and 10 unique experiments for non-CFs, where each experiment has six choice questions. Carlsson et al. (2010) test for learning and ordering effects in CE surveys and show that dropping the first choice question can decrease the error variance of estimates. We therefore duplicate the first choice set, yielding a total of seven choices made by respondents, and drop the first choice before conducting the analyses to account for possible learning effects.                                                              9 The experiment design was conducted using the SAS experiment design macro (Kuhfeld 2010).   13 Figure 2 Example of One Choice Set for CF Member Households in Nepali Language The conditional logit model, a standard method for analyzing discrete choice experiments assumes that the respondents have homogeneous preferences, which is a strong and often invalid assumption, and also assumes the irrelevance of independent alternatives (IIA). Following the more recent literature, we therefore use a mixed multinomial logit model (MMNL)10 that incorporates heterogeneity of preferences and relaxes the IIA assumption (Hensher and Greene 2003, Carlsson et al. 2003, Dissanayake 2014). Assuming a linear random utility model, the utility gained by person q from alternative i in choice situation t is given by U qit   qi   q X qit   qit (1)                                                              10Also referred to as the mixed logit, hybrid logit, random parameter logit, and random coefficient logit model.   14 X qit  qi where is a vector of non-stochastic explanatory variables. The parameter represents an intrinsic preference for the alternative (also called the alternative specific constant, ASC). We include an alternative specific constant (ASC) term in the estimation to account for the fact that option A and option B are both REDD+ contract options that can be viewed as closer substitutes with each other than with the status quo option (Haaijer, et al. 2001; Blaeij et al. 2007). In this setting the unconditional choice probability for individual q is given by Pq ()   Lq (  ) f (  | )d  . (2) is given by f ( | ) , the true parameter of the distribution is  and q where the density of Lq (  q ) the conditional choice probability of individual q choosing alternative i in choice situation t is This choice probability allows for the coefficients to vary and thereby relaxes the assumption that each individual has identical preferences. We estimate the model using the mixlogit command in STATA (Hole 2007). Given that the mixlogit model identifies a distribution for each parameter, it is possible to use the results to generate individual level coefficients. We do so by following Revelt and Train (2000), Hensher and Green (2003), and Hole (2007) and calculate the likelihood of a respondent choosing a specific set of choices from a set of alternatives. It is important to note that the individual estimates we generate are the conditional means of the coefficient distribution for the respondents who made identical choices when faced with the same choice set (Campbell et al. 2006; Hole 2007). The expected q value of conditional on a given response pattern p and a set of alternatives a is given by   15 pqit   T I  exp( qi   q X qit q )      t 1 i 1  exp( qj   q X qjt )  f (  | ) d    jJ   E[  | pq , aq )  pqit   T I  exp( qi   q X qit q )    t 1 i1   exp( qj   q X qjt )  f ( | )d    jJ   (3) E [  | pq , a q ) The value of can be approximated using simulation and we use the mixlbeta command in STATA to derive the individual parameter estimates (Hole 2007). For the results included in the text of the paper, we assume the preferences for attributes are the payment vehicle and fuelwood reduction attributes are random and lognormally distributed. This assumption assures that willingness to accept compensation for fuelwood reductions is always positive. As robustness checks, we present results that assume attributes are normally distributed and that only fuelwood reduction (not the payment vehicle) is lognormally distributed. As the findings are not appreciably different from those presented in the paper, we only briefly mention these results in the text. These marginal effects are the willingness to accept payment (in Nepali Rupees) for a 1-percent reduction in fuelwood collections from common forests. As shown in Table 1, these are derived from four discrete percentage reduction possibilities. We use the percentage reduction metric, in contrast to a pure quantity-based measure, because it is relevant to all households regardless of their circumstances. It is also true, though, that different households have different family sizes, food consumption patterns, live at different altitudes, exist in environments of different qualities and characteristics, etc. They therefore will have different meanings of 1-percent reductions. We address this issue below. Choice experiments only produce a single estimated WTP or in our case, WTA, value for each household. To examine small reductions that are greater than 1%, we must assume homothetic   16 preferences for cash (in the form of REDD+ payments) and fuelwood collections up to the fuelwood collection reduction percentage required by the simulated program. This implies that if we increase the total payment by a given percentage, households will reduce their fuelwood collections by this same percentage. This assumption, which is necessitated by our use of the choice experiment methodology, therefore implies a scaling of the payment and response by the same factor. It also implies that each household’s willingness-to-accept per ton of fuelwood reductions is unresponsive to the level of the reduction program. Our carbon supply function estimates are therefore conservative (i.e. less reduction generated for a given payment), because in reality we would expect that individual households require higher prices to supply more carbon reductions; this is consistent with standard increasing marginal cost if greater levels of carbon sequestration are to be supplied, which we expect not only across households as we analyze here, but also within households. Because we assume that household-level opportunity cost functions are constant, we do not also assume those estimates apply to highly non-marginal fuelwood reductions. We therefore discuss our results in two ways. First, we present the proportion of households who are willing to participate in a hypothetical REDD+ fuelwood reduction program at various prices. We then simulate the quantity effects for a hypothetical nationwide program that reduces fuelwood collections by 25% from current levels, assuming constant household-level marginal costs. We chose this value after consultation with Nepalese forestry experts, because it would be viewed as significant for households, but would not be considered an overwhelming reduction. At the mean, this reduction would be about 75 kg per month or about two basket loads per household (average size about 6 persons). Reported responses, including fuel substitutions, associated with such reduction requirements are discussed below.   17 Though a percentage reduction will have relevance to all respondents, people will have different meanings of 1-percent reductions.11 We ideally would like to customize these meanings to each household. Our data indeed include information on average monthly fuelwood consumption and across our sample we find that the average fuelwood consumption is reported to be 298 kilograms per month. Only three-quarters of our sample respondents reported their fuelwood consumption, however, and we are concerned that those who did not provide information differ systematically from those who provided information. We also worry that consumption may differ from fuelwood collections, which is our interest in this paper. For example, some households may buy fuelwood, which would cause collections and consumption to differ. To create a standard meaning of a 1-percent reduction in fuelwood collections, we utilize a nationally representative estimate from the literature. We argue that this approach is appropriate, because our communities are randomly chosen from a nationally representative random sample, with half CFs and half non-CFs, half of which are in the hills and half in the Terai, which are the two most important and relevant stratification dimensions. Households within communities are also chosen randomly, so the moments of the sample should be representative of the population. Using these features of our sampling frame, we assume that a 1-percent fuelwood collection reduction in our sample on average has the same meaning as a 1-percent reduction in average, national, rural household fuelwood collections. We utilize the average fuelwood collection estimate of 2.40 tons per household per year from Somanathan and Bluffstone (2015), which is derived from the nationally-representative 2010/2011 Nepal Living Standards Survey (NLSS). This estimate is slightly lower than the 2.54 tons per household per year from Nepal et al. (2010), which used the 2003/2004 NLSS and is about 8% below                                                              11This is not unique to our analysis, but is inherent in all carbon forestry PES programs implemented in primarily subsistence agriculture settings.   18 the estimate of Baland et al. (2010), which uses the 1995/1996 survey. Our estimate is also 17% below those for Dolakha District by Kandel et al. (2016). We therefore see our estimate of a 1-percent decrease in fuelwood collections (in tons) as reasonable, but erring on the conservative side, because each 1% reduction implies less wood reduced than in our consumption data and in much of the literature. Only preservation of non-renewable biomass generates additional carbon savings and as we saw in the previous section, in our sample CFs implement a variety of restrictions to increase sustainability. We cannot know the sustainability implications of each household’s reduction in fuelwood collections or even the average portion of collections that are currently sustainable across our sample. To estimate the percentage of biomass harvests that are currently unsustainable, we use estimates from Bailis et al. (2015) and Drigo et al. (2014), who calculate nonrenewable biomass estimates by country and region within country using the Woodfuel Integrated Supply/Demand Overview Mapping (WISDOM) model. The estimate for Nepal is 52%, which is the estimate we employ in this paper, implying that slightly more than half of fuelwood harvest reductions generate additional carbon savings. Each ton of unsustainably harvested fuelwood that is not collected would have been burned to generate heat for cooking and perhaps heating. Not all the biomass in fuelwood is carbon and for our analysis we use the IPCC (2013) emissions factor (0.50) to convert fuelwood not burned to carbon not emitted. We then apply the standard factor of 3.67 tons of CO2 per ton of carbon to convert carbon to CO2, which is the metric for international carbon transactions. In Nepal, only about 4% of households who cook with fuelwood use improved biomass cookstoves (Nepal et al., 2011). Much of the rest use traditional, inefficient stoves, such as the Nepali chulo, which is made of mud and stone, and typically vents waste gases within the home. Such   19 traditional stoves not only emit CO2, but also other gases, including aerosols that are estimated to have negative global warming potentials (GWP) and black carbon, which is a short-lived, but potent greenhouse gas. Indeed, about 25% of black carbon emitted into the atmosphere comes from millions of small, typically traditional stoves throughout the developing world (Bond et al. 2013). Smith et al. (2000) indeed find that, depending on the timeline examined, the GWP of a meal cooked with biomass can be significantly higher than the GWP of a meal cooked using fossil fuels. We utilize calculations from Somanathan and Bluffstone (2015), based on emissions factors and estimated GWPs taken from IPCC (2013), Pandey et al. (2014) and Grieshop et al. (2011), combined with the estimated nonrenewable biomass harvest estimate from Drigo et al. (2014) and Bailis et al. (2015), to derive 1.6047 tons of CO2e saved per ton of fuelwood not harvested. These calculations are presented in Table 2. Table 2: Emission Factors and 100-Year Global Warming Potentials in Tons of CO2e. 100-year Global Emission Factor (g of Warming Potential Tons of CO2e saved per Pollutant Pollutant/Kg of wood) (tCO2e) ton of wood CO2 1358 0.52 0.7062 BC 0.7 900 0.63 OC 1.9 -46 -0.0874 SO2 0.1 -76 -0.0076 CO 76 1.8 0.1368 NMVOC 6.9 8.8 0.06072 CH4 4.9 28 0.1372 N2O 0.1 265 0.0265 TOTAL 1.6047 Sources: Emission factors: (Pandey et al. 2014), Table SI3. Global Warming Potentials: IPCC (2013), Appendix 8.A, except for SO2 which is from Grieshop et al. (2011). The GWP for CO2 is adjusted to account for an estimated 48% of all fuelwood that is sustainably harvested (Drigo et al., 2014 and Bailis et al., 2015).   20 We estimate participation at various CO2 prices based on our sample as well as a national rural CO2 supply function for a program that asks rural households to reduce fuelwood collections by 25%. We scale our experimental results using the 2011 estimated national rural population of 4.4 million rural households (CBS, 2012; 2016) and the 1.45 million rural households who are members of CFs (MoFSC, 2017). Finally, we are interested in potential carbon supply shifters. We therefore run OLS regressions of marginal WTA values on key household and community variables. We emphasize that the cross-section nature of our data does not allow us to convincingly infer causality, because it is possible that omitted variables simultaneously affect choice experiment responses and key household and community variables. Nevertheless, the experimental nature of our dependent variable virtually rules out reverse causality, though perhaps not all possible confounders and we do not seek to infer causality. We merely would like to know which respondent characteristics, technologies previously adopted, assets and perceptions of REDD+ are associated with higher WTA. Correlations are therefore sufficient for our purposes. 3. Results We begin our discussion with a brief overview of descriptive statistics, followed by estimates of household participation in programs to reduce fuelwood collections and then present national level supply functions for CO2 reductions from a program that requires all households to reduce fuelwood collections by 25%. We do this for CF members and those outside the CF program and then consider the costs of carbon sequestration due to open grazing closures in non-CF communities.12 Finally, we present our OLS analysis of carbon supply shifters.                                                              12In the appendix, for the assumed normally distributed models, we trim 12 of 650 observations for non-CFs and 2 of 650 for CFs that have WTA estimates that are above $1000 and below -$1000 for 1% (average 24 kg/year) reductions in   21 Overview of the Sample Table 3 below presents key descriptive statistics from our sample of 1,300 households. These are the variables we use as potential carbon supply function shifters. Respondents were mainly male, but with a significant minority (34%) of respondents being female. The sample includes a variety of ethnic groups, including Dalit (i.e. lowest) castes (16%) and 38% of households are classified as poor or ultra-poor. Virtually all households use fuelwood, but 9% also use biogas and 15% each use LPG or improved biomass cookstoves. Virtually all households own some land and 42% say that REDD+ will very likely or extremely likely benefit them personally. Table 3: Descriptive Statistics of Sample and Potential Carbon Supply Shifters Variable Mean Standard Deviation Member of CF (1 = yes, 0 otherwise) 0.50 0.50 Respondent age (years) 50.61 13.76 Respondent is female (1 = yes, 0 otherwise) 0.34 0.48 Female-headed households (1 = yes, 0 otherwise) 0.16 0.37 Household size (number of people) 5.96 2.52 Household is classified as poor or ultra-poor (1 = yes, 0 otherwise) 0.38 0.48 Dalit ethnic group (1 = yes, 0 otherwise) 0.16 0.37 Indigenous or Newar ethnic group (1 = yes, 0 otherwise) 0.41 0.49 Brahmin or Chetri ethnic group (1 = yes, 0 otherwise) 0.40 0.49 Madheshi ethnic group (1 = yes, 0 otherwise) 0.02 0.14 Respondent migrated to site from another location (1 = yes, 0 otherwise) 0.25 0.44 Uses biogas (1 = use, 0 otherwise) 0.09 0.29 Uses LPG (1 = use, 0 otherwise) 0.15 0.36 Uses improved biomass cooking stove (1 = use, 0 otherwise) 0.15 0.36 Uses firewood (1 = use, 0 otherwise) 0.95 0.22 Household owns land (1 = yes, 0 otherwise) 0.94 0.24 Walking distance from respondent’s house to road < 2 hours (1 = yes, 0.70 0.46 0 otherwise)                                                              fuelwood collections. The May 2013 exchange rate of the Nepali Rupee with the $US was RS. 85/$US. As no observations meet this criterion, the lognormal results are not trimmed.   22 Respondent says they are very likely or extremely likely to benefit 0.42 0.49 personally from REDD+ (1 = yes, 0 otherwise) Reported firewood used per month (kilograms) 298.87 394.84 Number of formal CFUGs (number of groups) 1.3 0.73 We now present illustrative information on the possibility for leakage as fuelwood collection restrictions are put in place. This information comes from our household survey. Respondents were directed to indicate all options that would apply, given restrictions of 25%, 50%, 75% and 100%. Figure 3 shows that faced with a 25% fuelwood collection restriction, over half of respondents would shift to fuels produced on their private lands. About 30% would include reductions in their energy consumption in their response portfolio, about 25% would gather from other forests and substantial portions would shift to biogas and LP gas or gather in CFs. Very limited portions of households would use crop residues, kerosene or dung (not shown). Figure 3 Alternative Energy Sources Identified for 25% Fuelwood Restrictions in Respondents’ Main Forests   23 Figure 4 shows identification of biogas and LP gas as alternative fuels for 25%, 50% and 100% fuelwood collection restrictions in respondents’ main forests. Other alternative energy sources used are roughly the same as in Figure 3 as restrictions increase. For example, 54% to 56% of respondents say they would rely on their own land for fuel and 25% - 28% say they will gather from government forests. In contrast to reliance on other fuels, which does not vary as hypothetical fuelwood collection restrictions increase, respondents report that biogas and LPG use will increase as fuelwood from main forests is more stringently regulated. When no fuelwood can be collected from respondents’ main common forests, indeed over 25% of respondents say they would include LPG in their fuel portfolios. Figure 4 Reported Additional Use of Biogas and LPG at Various Fuelwood Restrictions in Respondents’ Main Forests This indicative evidence suggests that some leakage will occur if a PES scheme to sequester additional carbon were put in place, because about a quarter of respondents say that they would gather additional wood from government forests and just under 20% say they would gather from community   24 forests. We do not know anything about the amounts they would gather as responses are prospective. It is also striking, though, that so many respondents also point to their own land as an alternative source of fuel. These extractions would presumably not represent leakage as households are likely to manage their biomass resources in their best interests and replant as appropriate to maintain flows. Cleaner fuels like biogas and LPG are also expected to be important fuel sources if REDD+ programs were to be instituted in rural Nepal. Carbon Supply by CF Member Households Figure 5 presents the estimated participation of CF member households in programs to reduce fuelwood collections at various CO2 prices, with CO2 emissions reductions credited as in table 2. Estimated prices to incentivize participation in the program range from $7.00 to $655 per ton of CO2, with a median of $48 and an average of $73. Assuming normally distributed payment vehicle and fuelwood reduction attributes, the median WTA is $56 with an average of $63. Approximately 25% of households are estimated to participate at a price of $30 per ton and 75% participate at a price of $93 ($27 and $86 assuming normal distributions). These results suggest that estimated costs of reducing fuelwood collections are relatively high, particularly considering that the 2015 global social cost of carbon is estimated at approximately $40 per ton (USIAWG, 2015).   25 Figure 5: CF Household REDD+ Program Participation at Various CO2 Prices Participation of CF Households in Program Requiring Fuelwood  Reductions at Various per‐ton  CO2 Prices* $700.00 $600.00 $500.00 $400.00 $300.00 $200.00 $100.00 $0.00 0.16% 2.95% 5.75% 8.54% 11.34% 14.13% 16.93% 19.72% 22.52% 25.31% 28.11% 30.90% 33.70% 36.49% 39.29% 42.08% 44.88% 47.67% 50.47% 53.26% 56.06% 58.85% 61.65% 64.44% 67.24% 70.03% 72.83% 75.62% 78.42% 81.21% 84.01% 86.80% 89.60% 92.39% 95.19% 97.98% 100.78% Participation in Program  * Assumes Lognormal Distribution of Payment and Fuelwood Reduction Attributes In Figure 6, we present our estimate of the CO2 supply function for a program that requires CF member households to reduce fuelwood collections by 25%. It is estimated that 1.45 million households are members of the CF Programme and it is therefore for this population that we estimate the supply of carbon. As discussed in our Methods and Data section, the moments of the price distribution are constant, because we estimate willingness to accept using a choice experiment, which only produces one marginal cost value for each respondent; this implies that the carbon supply price per ton does not increase when we evaluate a larger-than-1%-increase. For this reason, we only consider a program that requires a 25% reduction in fuelwood collections, which is likely to be at the upper end of “marginal,” but still significant enough to be monitorable. For reference, results for a 10% reduction program are included in the Appendix, but such a program seems too insignificant. At the median willingness to accept value of $48 per ton of CO2 reduced, we estimate that 695,877 additional tons will be sequestered by the population of 1.45 million households, with a total   26 revenue of $33.5 million or about $23 per household per year. At a price of $40, which is consistent with contemporary estimates of the social cost of carbon, approximately 570,142 additional tons are estimated to be sequestered by CF households, with total revenue of $22.8 million or about $16/household per year. At a price of $10 per ton, we estimate that very little carbon will be sequestered (less than 10,000 tons). Figure 6: CF CO2 Sequestration Supply at Various CO2 Prices National CF CO2 Supply Function  25% Fuelwood Reduction Program   $700.00 $600.00 $500.00 $400.00 $300.00 $200.00 $100.00 $0.00 2167.84 41188.96 80210.08 119231.20 158252.32 197273.45 236294.57 275315.69 314336.81 353357.93 392379.05 431400.17 470421.29 509442.41 548463.54 587484.66 626505.78 665526.90 704548.02 743569.14 782590.26 821611.38 860632.50 899653.63 938674.75 977695.87 1016716.99 1055738.11 1094759.23 1133780.35 1172801.47 1211822.59 1250843.72 1289864.84 1328885.96 1367907.08 1406928.20 Cumulative Tons of CO2 in National Rural Population * Assumes Lognormal Distribution of Payment and Fuelwood Reduction Attributes Carbon Supply by Households outside the Community Forestry Program We find that households who are outside the Nepal Community Forestry Program have lower willingness to accept fuelwood reductions (Figure 7). We estimate that the median cost per ton of CO2 sequestered by those outside the CF program is $41/ton and the average is $40 per ton ($33 and $40 assuming normal distribution of all attributes). For a given CO2 price, households outside the CF program are estimated to be willing to sequester more carbon than CF households. At the median,   27 the cost difference is approximately 40% and whereas CF households required a carbon price of $27 for 25% of households to participate, non-CF households are estimated to require only $16.40, or 41% less ($24 assuming lognormal distribution). Rather than requiring $86/ton for three-quarters of households to participate, those outside the CF program are estimated to demand only $55 or 36% less ($45 assuming lognormal). At $10 per ton, roughly 18% of households are willing to participate, compared with 9.6% for CF households (9% with lognormal distribution). Approximately 7% of respondents report negative willingness to accept values. Figure 7: Non-CF Household REDD+ Program Participation at Various CO2 Prices Participation of Non‐CF Households in Program Requiring Fuelwood  Reductions at Various per‐ton  CO2 Prices* $1,200.00 $1,000.00 $800.00 $600.00 $400.00 $200.00 $0.00 0.16% 2.80% 5.43% 8.07% 10.71% 13.35% 15.99% 18.63% 21.27% 23.91% 26.55% 29.19% 31.83% 34.47% 37.11% 39.75% 42.39% 45.03% 47.67% 50.31% 52.95% 55.59% 58.23% 60.87% 63.51% 66.15% 68.79% 71.43% 74.07% 76.71% 79.35% 81.99% 84.63% 87.27% 89.91% 92.55% 95.19% 97.83% 100.47% Participation in Program  * Assumes Lognormal Distribution of Payment and Fuelwood Reduction Attributes In addition to getting more reductions per monetary unit paid, there are also many more households outside the CF program than inside it (2.95 million households versus 1.45 million). Implementing a program to reduce fuelwood collections by 25% therefore generates much more sequestration. As shown in Figure 8, at the median WTA of $33 per ton of CO2, 1.43 million tons are estimated to be sequestered, raising revenues of $48 million. As this amount is spread over 2.95 million households, an average household receives just over $16 per year. At a price of $40/ton (consistent with the 2015 estimate of the global social cost of carbon), 1.67 million tons are estimated   28 to be sequestered and generate revenues of $67.2 million or $23 per household. At $10 per ton, roughly 500,000 tons are sequestered and generate about $5 million in revenues. Figure 8 Non-CF Household CO2 Sequestration Supply at Various CO2 Prices National Non‐CF CO2 Supply Function  25% Fuelwood Reduction Program   $1,200.00 $1,000.00 $800.00 $600.00 $400.00 $200.00 $0.00 2167.84 43356.80 84545.76 125734.72 166923.68 208112.65 249301.61 290490.57 331679.53 372868.49 414057.45 455246.41 496435.37 537624.34 578813.30 620002.26 661191.22 702380.18 743569.14 784758.10 825947.06 867136.02 908324.99 949513.95 990702.91 1031891.87 1073080.83 1114269.79 1155458.75 1196647.71 1237836.68 1279025.64 1320214.60 1361403.56 1402592.52 Cumulative Tons of CO2 in National Rural Population * Assumes Lognormal Distribution of Payment and Fuelwood Reduction Attributes We now provide indicative estimates of the costs of eliminating open grazing for non-CF households, with the goal of characterizing whether opportunity costs might be competitive with other mitigation options. To accurately apportion shares of grazing to each respondent is impossible. We therefore cannot estimate household-specific carbon supply functions for grazing elimination and our findings should therefore be viewed largely as suggestive of whether opportunity costs are high or low. In contrast to fuelwood reductions, for grazing, we assume normal distributions of grazing reduction and the payment vehicle, because assuming lognormal distributions generated 76 WTA values greater than $1,000; we consider these values, which represent 12% of our observations, as outliers. Assuming normal distributions, the range of WTA values by the NCF respondents is -$805   29 to $3,746 per year. The average WTA per year to forego open grazing is $46 and the median is $20. In making these calculations we drop the highest and lowest 3 observations as potential outliers (<1% of observations). Non-CF forest user communities on average have 296 households (maximum 1,881) and mean forest area per household is 0.75 hectares (maximum 9.25). We assume that respondent opportunity costs represented by WTA depend primarily on factors other than forest area, such as outside income-earning opportunities and non-forest fodder production/grazing possibilities and divide each household WTA by the mean forest per household of 0.75. This give us the respondent- specific WTA per hectare of forest, which is a potential estimate of the average opportunity cost should open grazing be eliminated. We then recalibrate these values to be per ton of CO2 sequestered rather than emitted, using the IPCC (2000) centroid value of 1.647 tons of CO2 per hectare. With the above assumptions, we find that the mean WTA is $37.24 per ton of CO2 and at the median, the estimated opportunity cost of grazing closure is $16.19. These costs are similar to the costs of reducing carbon via fuelwood collection restrictions in non-CF settings. Carbon Supply Shifters Table 4 presents an OLS regression of the potential carbon supply shifters in Table 2 on WTA of CF member households for reducing fuelwood collections; this approach allows us to consider heterogeneous effects and explain the distribution of WTA values in terms of potential carbon supply shifters. As our goal is to explain the distribution of WTA estimates rather than estimate the cost of carbon sequestration measured as CO2, we do not constrain WTA estimates to be positive. We therefore assume normal distributions. Estimates assuming lognormally distributed payment vehicle and fuelwood reduction attributes are presented in the Appendix.   30 We see that households that are poor, large, female-headed, use improved biomass cookstoves and are larger (likely indicating more children) require more compensation to be induced to participate in the REDD+ program that involves fuelwood reductions. Compared with the mean, female-headed households require roughly 30% more than average households and the same is true for households that use an improved biomass cookstove. Female respondents require roughly 15% more than the average and the effect is approximately the same for poor households. Each additional household member above the average adds about 4% to the WTA compared with the mean. Leaving aside the effect of having an improved cookstove as potentially endogenous, a household that is female-headed, poor, with 8 household members rather than the mean of 6 and the respondent is female, is therefore expected to have a WTA about 69% above the mean. Several variables have negative and statistically significant correlations with WTA values. If respondents are of Madheshi ethnicity, which means they are of Indian origin and likely live in the Terai, WTA is estimated to be much lower than the mean. Households who use biogas on average have estimated WTA values that are about 35% below the mean and the same is true for those who own land. All else equal, those who believe that REDD+ is very likely or extremely likely to benefit them personally on average have WTA values 19% below the mean. Table 4: OLS Regression of Potential Supply Shifters on Marginal WTA (in Nepali Rupees) of CF Member Households to Participate in REDD+ Programs Requiring Reduction in Fuelwood Collections Variable Regression Coefficient P Value Respondent age (years) -0.96 0.21 Respondent is female (1 = yes, 0 otherwise) 34.35* 0.10 Female-headed households (1 = yes, 0 otherwise) 63.71*** 0.01 Household size (number of people) 8.61** 0.03 Household is classified as poor or ultra-poor (1 = yes, 0 otherwise) 35.83* 0.09 Dalit ethnic group (1 = yes, 0 otherwise) -49.09 0.12 Indigenous or Newar ethnic group (1 = yes, 0 otherwise) -3.46 0.87 Madheshi ethnic group (1 = yes, 0 otherwise) -105.45*** 0.03   31 Respondent migrated to site from another location (1 = yes, 0 otherwise) -34.10 0.16 Uses biogas (1 = use, 0 otherwise) -72.04*** 0.01 Uses LPG (1 = use, 0 otherwise) -21.21 0.32 Uses improved biomass cooking stove (1 = use, 0 otherwise) 68.95* 0.08 Uses firewood (1 = use, 0 otherwise) -130.67 0.43 Household owns land (1 = yes, 0 otherwise) -79.26** 0.02 Walking distance from respondent’s house to road < 2 hours (1 = yes, 0 otherwise) 12.35 0.68 Respondent says they are very likely or extremely likely to benefit personally from REDD+ (1 = yes, 0 otherwise) -41.28** 0.03 Reported firewood used per month (kilograms) 0.01 0.40 Number in formal CFUGs (number of groups) -4.28 0.80 Constant 532.40*** 0.01 ***, **, * indicate significant at 1%, 5% and 10% levels. N= 597, R2=0.10, F=3.01, prob>F=0.0007 Assumes normally distributed payment vehicle and fuelwood reduction attributes Table 5 presents results of the same model using the subsample of respondents who are not members of CFs. We see that the model does not offer us nearly as many insights into heterogeneous effects and supply shifters as noted in Table 4 for CFs. The model is only marginally jointly significant and while the signs of the estimated regression coefficients are broadly similar to those in Table 4, only two estimated coefficients, household size and use of biogas, are significantly different from zero, with the same signs and rough magnitudes as in Table 4. These findings suggest that with the exception of these two variables, the shifters of WTA for reductions in fuelwood collections are fundamentally different in CFs and non-CFs; indeed, in non-CFs the hypothesized variables do not seem to matter for the placement of the carbon supply function. Table 5: OLS Regression of Potential Supply Shifters on Marginal WTA (in Nepali Rupees) of Non-CF Households to Participate in REDD+ Program Requiring Reduction in Fuelwood Collections Variable Regression Coefficient P Value Respondent age (years) 0.93 0.38 Respondent is female (1 = yes, 0 otherwise) -20.24 0.56 Female-headed households (1 = yes, 0 otherwise) 16.98 0.76 Household size (number of people) 12.24* 0.09 Household is classified as poor or ultra-poor (1 = yes, 0 otherwise) 47.78 0.17 Dalit ethnic group (1 = yes, 0 otherwise) -5.59 0.89   32 Indigenous or Newar ethnic group (1 = yes, 0 otherwise) 20.92 0.59 Madheshi ethnic group (1 = yes, 0 otherwise) 19.86 0.70 Respondent migrated to site from another location (1 = yes, 0 otherwise) -20.99 0.52 Uses biogas (1 = use, 0 otherwise) -73.20* 0.03 Uses LPG (1 = use, 0 otherwise) 48.95 0.35 Uses improved biomass cooking stove (1 = use, 0 otherwise) 11.30 0.83 Uses firewood (1 = use, 0 otherwise) 10.60 0.78 Household owns land (1 = yes, 0 otherwise) -34.11 0.54 Walking distance from respondent’s house to road < 2 hours (1 = yes, 0 otherwise) -22.47 0.55 Respondent says they are very likely or extremely likely to benefit personally from REDD+ (1 = yes, 0 otherwise) -9.94 0.62 Reported firewood used per month (kilograms) -0.04 0.39 Number in formal CFUGs (number of groups) -4.53 0.81 Constant 91.93 0.41 ***, **, * indicate significant at 1%, 5% and 10% levels. N= 608, R2=0.023, F=1.77, prob>F=0.05 Assumes normally distributed payment vehicle and fuelwood reduction attributes 46. Discussion and Conclusions In this paper, we use a choice experiment conducted in 2013 to estimate household-level willingness to participate in a REDD+ program that requires reductions in fuelwood collections, as a function of CO2 prices. We find that for both CF and non-CF households, robust participation occurs at prices that are higher than much of the literature and grazing reduction opportunity costs per ton of CO2 are similar to those for fuelwood reductions in non-CF settings. Rather than prices of $1.00 to $5.00 incentivizing participation, we find that relatively little carbon would be supplied at such prices. This basic finding is in line with two recent published papers focusing on REDD+ pilots in Nepal. These findings in combination with our results, which use very different techniques, suggest that optimism regarding carbon supply in community forestry settings could be somewhat misplaced. Our use of an experimental method to elicit WTA is novel in the recent literature, which has largely used bottom-up methods to estimate WTA within the context of REDD+.   33 CFs will almost certainly be the core institution within which REDD+ is implemented in Nepal. We find that average and median WTA values for CFs are substantially greater than for non- CFs, which is not surprising given that while non-CFs may be open access, CFs have typically already imposed fuelwood collection restrictions; REDD+ will just increase those restrictions to generate additional carbon sequestration. Our analysis of carbon supply shifters indeed suggests that non-CF households react to REDD+ very differently than households who are CF members. Broadly- speaking, “underprivileged” CF member households, such as those who are landless, female-headed and poor, appear to be warier of fuelwood collection restrictions within the context of REDD+. They therefore require higher payments, all else equal, than average respondents. Our choice experiment methodology has the important advantage that WTA values come from household members themselves, rather than being constructed based on assumed explicit costs. In this regard, our estimates may include a broader array of difficult-to-measure costs, such as inconvenience and transaction costs, that have been found in the literature to be significant and may be associated with - or suspected to be part of – REDD+. As we have noted, though, a downside of using the choice experiment methodology is that it can only provide us with one marginal (and average) WTA value for each household. We therefore can only construct our carbon supply functions across households and do not have the capacity to fully incorporate intra-household marginal costs of carbon supply. Because the estimates are for 1% decreases in fuelwood collections, they are likely underestimates of true marginal WTA for larger reductions in fuelwood collections (e.g. 25% as we have simulated). Furthermore, as is true for any experimental methodology, there is the potential that estimates might not be externally valid outside the experimental framework and even that the hypothetical nature of the estimation procedure may yield biased WTA estimates.   34 Future research on carbon supply from households should prioritize estimating carbon supply functions that incorporate both intra-household and inter-household marginal costs. Both are valid elements of carbon supply functions. We have included the latter, which we see as a contribution to a literature that has generally estimated at most a few points rather than anything like a “function.” People who are part of a formal collective action institution, the CF program, react very differently to REDD+ and perhaps other PES schemes than those without such affiliations. We see this point as important, because even in Nepal where the CF program is very large, it only covers about 35% of the rural population and those who are unaffiliated may be able to offer lower-cost carbon sequestration. Additional work on the sources and implications of differences is therefore likely to be very useful. References Adamowicz, W., Boxall, P., Williams, M. and Louviere, J. 1998. Stated preference approaches for measuring passive use values: choice experiments and contingent valuation. American Journal of Agricultural Economics. 80, 64-75. Adamowicz, W., Swait, J., Boxall, P., Louviere, J. and Williams, M. 1997. Perceptions versus objective measures of environmental quality in combined revealed and stated preference models of environmental valuation. Journal of Environmental Economics and Management. 32, 65-84. Agrawal, A., Chhatre, A., & Hardin, R. (2008). Changing Governance of the World's Forests. Science, 320, 1460-1462. Alpizar, F., F. Carlsson and P. Martinsson. 2003. Using choice experiments for non-market valuation, Economic Issues, 8, 83-110. Angelsen, A. 2010. The 3 REDD ‘Is’ Journal of Forest Economics 16 (2010) 253–256 Bailis, R, R. Drigo, A. Ghilardi, O. Masera. 2015. The Carbon Footprint of Traditional Woodfuels. Nature Climate Change 5: 266- 272. Baland, J-M, P. Bardhan, S. Das, D. Mookhjerjee and R. Sarkar. 2010. The Environmental Impact of Poverty: Evidence from Firewood Collection in Rural Nepal. Economic Development and Cultural Change. Vol. 59: 23-61. Blaeij, A.T. de, Nunes, P.A.L.D. and J.C.J.M. van den Bergh. 2007. ‘No-choice’options within a nested logit model: one model is insufficient. Applied Economics, 39: 1245-1252. Bluffstone, R. 2013. Economics of REDD+ and Community Forestry. Journal of Forest and Livelihood, 11(2), 69-74.   35 Bluffstone, R, E. Somanathan, P. Jha, Harisharan Luintel, R. Bista, M. Toman, N. Paudel, B. Adhikari. 2018. “Does Collective Action Sequester Carbon? Evidence from the Nepal Community Forestry Programme,” World Development 101: 133–141 Bluffstone, R., E. Robinson and P. Guthiga. 2013. “REDD+ and Community Controlled Forests in Low- Income Countries: Any Hope for a Linkage?” Ecological Economics, 87: 43-52. Bond, Tami C., Sarah J. Doherty, D. W. Fahey, P. M. Forster, T. Berntsen, B. J. DeAngelo, M. G. Flanner, et al. 2013. Bounding the Role of Black Carbon in the Climate System: A Scientific Assessment. Journal of Geophysical Research: Atmospheres 118 (11): 5380–5552. Boxall, P.C., Adamowicz, W.L., Swait, J., Williams, M. and Louviere, J., 1996.A comparison of stated preference methods for environmental valuation. Ecological Economics. 18, 243-253. Campbell, D., Hutchinson, W.G, Scarpa, R., 2006. Using mixed logit models to derive individual- specific WTP estimates for landscape improvements under agri-environmental schemes: evidence from the Rural Environment Protection Scheme in Ireland. In: Koundouri, P., Birol, E. (eds). Choice experiments informing European environmental policy. Edward Elgar, 58-81. Carlsson, F., Frykblom, P. and Liljenstolpe, C., 2003. Valuing wetland attributes: an application of choice experiments. Ecological Economics. 47, 95-103. Carlsson, F., Mørkbak, R. and Olsen, B. 2010. The first time is the hardest: A test of ordering effects in choice experiments. Working Papers in Economics 470. Carlsson, F. and P. Martinsson. 2003. Design techniques for choice experiments in health economics, Health Economics, 12, 281-94. Carter, J. and Gronow, J. 2005. Recent Experience in Collaborative Forest Management: A Review Paper. Occasional Paper No. 43. Indonesia: Centre for International Forestry Research. Central Bureau of Statistics (CBS). 2016. Statistical Yearbook Nepal. Central Bureau of Statistics, Kathmandu. Central Bureau of Statistics (CBS). 2012. National Population and Housing Census 2011. Central Bureau of Statistics, Kathmandu. Dangi, R. 2013. Progress Update on REDD+ Readiness. A Presentation Delivered in REDD+ Stakeholder Forum Meeting Held in 18 June 2013 in Kathmandu. Dissanayake, S. T. P. Jha, B. Adhikari, R. Bista, R. A. Bluffstone, H. Luintel, P. Martinsson, N. Paudel, E. Somanathan and M. Toman. 2015. Community Managed Forest Groups and Preferences for REDD+ Contract Attributes: A Choice Experiment Survey of Communities in Nepal, Policy Research Working Paper 7326, June 2015, The World Bank, Washington, DC. Dissanayake, S. T. M. and Ando, A. W. 2014. Valuing Grassland Restoration: Proximity to Substitutes and Trade-offs among Conservation Attributes. Land Economics, 90(2), 237-259. Drigo, R, R. Bailis, A. Ghilardi and O. Masera. 2014. Tier 1 Report: Pan-Tropical Demand and Sustainability, Yale-UNAM Project Report Geospatial Analysis and Modeling of Nonrenewable Biomas: WISDOM and Beyond, supported by the Global Alliance for Clean Cookstoves   36 Fox, J.M. 1993. Forest resources in a Nepali village in 1980 and 1990: the positive influence of population growth. Mountain Research and Development, 13(1), 89-98. Gregorsen, H, H. Lakany, A. Karsenty and A. White. 2011. Does the Opportunity Cost Approach Indicate the Real Cost of REDD+? Rights and Realities of Paying for REDD+ at www.cbfp.org/doc_rech_en.html?file=docs/research_docs/...REDD.pdf Washington, DC: Rights and Resources Initiative Grieshop, Andrew P., Julian D. Marshall, and Milind Kandlikar. 2011. Health and Climate Benefits of Cookstove Replacement Options. Energy Policy 39 (12): 7530–42. Guthman, J. 1997. Representing Crisis: The Theory of Himalayan Environmental Degradation and the Project of Development in Post-Rana Nepal. Development and Change, 28:45-69. Haaijer, R., Kamakura, W., and Wedel, M. 2001. The ‘no-choice’ alternative in conjoint choice experiments, International Journal of Market Research, 43, 93-106. Hanley, N., Mourato, S. and Wright, R.E. 2001. Choice Modelling Approaches: A Superior Alternative for Environmental Valuation? Journal of Economic Surveys. 15, 435-462. Hensher, D.A. and Greene, W.H. 2003. The mixed logit model: the state of practice. Transportation. 30, 133-176. Hensher, D.A., Rose, J.M. and Greene, W.H. 2005. Applied choice analysis: a primer. Cambridge UnivPr, . HMGN/MoLJ. 1993. Forest Act 1993. His Majesty's Government of Nepal and Ministry of Law and Justice. Hobley, M. 1996. Participatory forestry: The process of change in India and Nepal. Rural Development Forestry Network, Overseas Development Institute, London. Hole, A. R. 2007. Fitting mixed logit models by using maximum simulated likelihood. The Stata Journal 7, 388-401. Hoyos, D., 2010. The state of the art of environmental valuation with discrete choice experiments. Ecological Economics . 69, 1595-1603. Huber, J. and K. Zwerina. 1996. The importance of utility balance in efficient choice designs. Journal of Marketing Research 33, 307-317. IPCC, 2014: Summary for Policymakers. In: Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change [Edenhofer, O., R. Pichs-Madruga, Y. Sokona, E. Farahani, S. Kadner, K. Seyboth, A. Adler, I. Baum, S. Brunner, P. Eickemeier, B. Kriemann, J. Savolainen, S. Schlömer, C. von Stechow, T. Zwickel and J.C. Minx (eds.)]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. IPCC. 2013. Climate Change 2013: The Physical Basis Summary for Policymakers, Working Group 1 Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change. Geneva.   37 IPCC, 2000. Land Use, Land-Use Change and Forestry. Cambridge University Press, UK. http://www.ipcc.ch/ipccreports/sres/land_use/index.php?idp=177 Jeuland, M.A. and S.K. Pattanayak. 2012. Benefits and costs of improved cookstoves: Assessing the implications of variability in health, forest and climate impacts. Plos ONE 7(2): 1-15. Kandel, P., P. S. Chapagain, L. N. Sharma, O. R. Vetaas. 2016. Consumption Patterns of Fuelwood in Rural Households of Dolakha District, Nepal: Reflections from Community Forest User Groups. Small Scale Forestry Vol. 15: 481-495. Kanel, K. R., K. Shrestha, A. R. Tuladhar, M. R. Regmi. 2012. A Study on The Demand and Supply of Wood Products in Different Regions of Nepal. Prepared for REDD – Forestry Climate Change Cell, Department of Forestry, Ministry of Forests and Soil Conservation, Government of Nepal. Karky B. and M. Skutsch. 2010. The Cost of Carbon Abatement through Community Forest Management in Nepal Himalaya. Ecological Economics 69: 666-672. Keller, A. and R. Goldstein. 1998. Impact of Carbon Storage Through Restoration of Drylands on the Global Carbon Cycle. Environmental Management 22 (5): p757–766 Kindermann G, M. Obersteiner, B. Sohngen, J. Sathaye, K. Andrasko, E. Rametsteiner, B. Schlamadinger, S. Wunder and R. Beach. 2008. Global cost estimates of reducing carbon emissions through avoided deforestation. Proceedings of the National Academy of Sciences 105(30): 10302-10307. Kuhfeld, W.F. 2010. Marketing research methods in SAS experimental design, choice, conjoint, and graphical techniques. Lancaster, K.J., 1966. A new approach to consumer theory. The Journal of Political Economy 74, 132-157. Louviere, J.J., Hensher, D.A. and Swait, J.D. 2000. Stated choice methods: analysis and applications. Cambridge University Press. Luintel, H, R. Bluffstone and R. Scheller 2017. Collective Action Drivers of Carbon Storage Mimeo Luintel, H. R. Scheller, R. Bluffstone and B. Adhikari. 2017. The Effect of the Nepal Community Forestry Program on Equity in Benefit Sharing Journal of Environment and Development Vol. 26: 297–321 Marseni, T. N, P. R. Neupane, F. Lopez-Castro, T. Cadman. 2014. An Assessment of the Impacts of the REDD+ Pilot Project on Community Forests User Groups (CFUGs) and their Community Forests in Nepal. Journal of Environmental Management 136: 37-46. Ministry of Forests and Soil Conservation (MoFSC), Department of Forests. Community Forestry Division Website. http://dof.gov.np/dof_community_forest_division/community_forestry_dof Accessed May 1, 2017. Nepal, Mani, Apsara Nepal, and Kristine Grimsrud. 2011. “Unbelievable but Improved Cookstoves Are Not Helpful in Reducing Firewood Demand in Nepal.” Environment and Development Economics 16 (01): 1–23. doi:10.1017/S1355770X10000409. MoFSC. 2013. Review of 30 Years of Community Forestry in Nepal. Ministry of Forest and Soil Conservation , Government of Nepal.   38 Nepal, M., A. Nepal, Aspara and K. Grimsrud. 2011. “Unbelievable but Improved Cookstoves are not helpful in Reducing Firewood Demand in Nepal. Environment and Development Economics Vol. 16: 1– 23. Nurse, M. and Malla, Y. 2005. Advances of Community Forestry in Asia. Bangkok: Regional Community Forestry Training Center (RECOFTC). Ojha H, Timsina N and Khanal D.R. 2007. How are Forest Policy Decisions Made in Nepal? Journal of Forest and Livelihood 6(1): 1-17. Pan, Y, R. Birdsley, J. Fang, R. Houghton, P. Kauppi, W. Kurz, O. Phillips, A. Shvidenko, s. Lewis, J. Canadell, P. Ciais, R. Jackson, S. Pacala, A. D. McGuire, S. Piao, A. Rautiainen, S. Sitch, D. Hayes. 2011. A Large and Persistent Carbon Sink in the World’s Forests, Science 333:988-993. Pandey, Apoorva, Pankaj Sadavarte, Anand B. Rao, and Chandra Venkataraman. 2014. “Trends in Multi-Pollutant Emissions from a Technology-Linked Inventory for India: II. Residential, Agricultural and Informal Industry Sectors.” Atmospheric Environment 99: 341–52. Pandit, R, P. R. Neupane, B. H. Wagle. 2017. Economics of Carbon Sequestration in Community Forests: Evidence from REDD+ Piloting in Nepal, Journal of Forest Economics 26: 9-29. Regmi, M.C. 1984. The State and Economic Surplus: Production, Trade and Resource-Mobilisation in Early 19th Century Nepal. Varanasi: Nath Publishing House. Revelt D, Train K. 2000. Customer-Specific Taste Parameters and Mixed Logit: Households' Choice of Electricity Supplier, No E00-274, Economics Working Papers, University of California at Berkeley, http://EconPapers.repec.org/RePEc:ucb:calbwp:e00-274. Rights and Resources Initiative. 2014. What Future for Reform? Progress and Slowdown in Forest Tenure Reform since 2002. Washington DC. Saatchi, S.S, N. L. Harris, S. Brown, M. Lefsky, T. A. Mitchard, W. Salas, B. R. Zutt, W. Buermann, S. L. Lewis, S. Hagen, S. Petrova, L. White, M. Silman and A. Morel. 2011. Benchmark map of forest carbon stocks in tropical regions across three continents. Proceedings of the National Academies of Sciences 108(24): 9899-9904. Smith, K. R., Uma, R., Kishore, V.V.N., Zhang, J., Joshi, V. and Khalil, M.A.K. 2000 Greenhouse Implications of Household Stoves: An Analysis for India. Annual Review of Energy and the Environment, 25: 741–63 Somanathan, E. and R. Bluffstone. 2015. Biogas: Clean Energy Access with Low-Cost Mitigation of Climate Change. Environmental and Resource Economics Vol 62: 265-277. Springate-Baginski, O. and Blaikie, P. 2007. (Eds.). Forests, People and Power: The Political Ecology of Reform in South Asia. EarthScan Forestry Library. London and Sterling: Earthscan. Strassburg, B., R. K. Turner, B. Fisher, R. Schaeffer. and A. Lovett. 2009. Reducing Emissions from Deforestation – The “Combined Incentives” Mechanism and Empirical Simulations. Global Environmental Change 19: 265-278. Timilsina, N. P. and Paudel, N. S. 2003. State Versus Community: A Confusing Policy Discourse in Nepals Forest Management. Journal of Forest and Livelihood, 2(2), 8-16.   39 U.S. Interagency Working Group (USIAWG) on Social Cost of Carbon, United States Government. Technical Support Document: Technical Update of the Social Cost of Carbon for Regulatory Impact Analysis Under Executive Order 12866. May, 2013. Revised July 2015. Available at https://www.whitehouse.gov/sites/default/files/omb/inforeg/scc-tsd-final-july-2015.pdf Vermeulen, B., Goos, P. and Vandebroek, M., 2008.Models and optimal designs for conjoint choice experiments including a no-choice option. International Journal of Research in Marketing 25, 94-103. World Bank. (2009) Forests Sourcebook: Practical Guidance for Sustaining Forests in International Cooperation. World Bank: Washington, D.C.   40 APPENDIX The Nepal Community Forestry Program and REDD+ Participation Nepal experienced heavy deforestation and environmental degradation starting in the 1960s, which has been primarily attributed to the nationalization of forests (Ojha et al. 2007; Guthman 1997; Timilsina and Paudel 2003). The government and its development partners began to recognize the limitations of centralized management and as a result looked to decentralization and community- based forest management as alternatives. Community forest management has been pursued since the late 1970s and was formalized in the early 1990s. The introduction of the National Forestry Plan in 1976 first raised the possibility of ‘handing over’ forest management responsibilities to local governments (Fox 1993). The emergence of participatory discourse and increased international pressure (Hobley 1996) for devolution led to the enactment of the Decentralisation Act of 1982, which empowered local governments to manage local forests (Regmi 1984). While the government has enacted various policies and programs to support different community strategies over the past two decades, these share common directions in that they transfer management rights to locally formed groups and provide them greater tenure security and more benefits of forest management (Springate and Blaikie 2007; Carter and Gronow 2005; Nurse and Malla 2005). 13 Community forestry (CF) is the flagship program and is most significant in terms of scale and economic significance. As CFs are the most important forest collective action institution in Nepal, we only examine CFs and those whose primary forest is not registered within the CF program.                                                              13There are six different community forest management programs in Nepal: community forestry (CF), leasehold forestry, watershed management, collaborative forest management, conservation areas and buffer zone community forests (BZCF) around protected areas.     41 The Master Plan for the Forestry Sector of 1989 provides the foundation for CF governance. It was followed by the Forest Act of 1993, which provided a clear legal basis for CF, enabling the government to hand over areas of national forest to locally formed CFUGs. The provisions were later detailed in the Forest Regulations and the Community Forestry Operational Guidelines of 1995 that were updated in 2009. According to the Forest Act and associated regulations, the CFUGs are recognized as self-governing, independent, autonomous, perpetual and corporate institutions, so they can acquire, possess, transfer, or otherwise manage movable or immovable property (HMGN/MoLJ 1993: Article 43). The District Forest Officer, who is the main local-level forest officer, may hand over any part of a National Forest to user groups provided both parties agree to operational plans. The groups are entitled to receive all the benefits from the management of the forest. As of 2016 the CF program has expanded to over 19,000 CFUGs with the involvement of over 1.45 million households (almost 35% of the households) managing over 1.8 million ha of forests (MoFSC 2017). Almost 65% of the forests in the hills and 18% of Terai forests are CFs (MOFSC 2013). The REDD+ process in Nepal started in 2008 after the United Nations Framework Convention on Climate Change COP 13. Right after COP 13 the Government of Nepal submitted the REDD Readiness-Plan Idea Note (R-PIN) to the Forest Carbon Partnership Facility (FCPF) of the World Bank. Once the R-PIN was approved, the Ministry of Forest and Soil Conservation (MoFSC) established the REDD Forestry and Climate Change Cell (REDD Cell), an administrative unit to handle REDD+ activities. The development of the Readiness Preparation Proposal (RPP) took place during 2009/2010. There was keen interest and support of donors and NGOs during the RPP development process. The RPP was approved by the FCPF in October 2010. Since 2010 the government has been implementing various activities proposed by the RPP (Dangi, 2013). Apart from forming the national institutional structures for implementing REDD+, the government conducted several studies to support the National REDD+ Strategy, which was finalized in 2015.   42 Figure A.1: Status of REDD+ readiness in Nepal Adapted from Dangi 2013 Choice Experiment Preamble Read to Respondents Introduction to Climate Change and REDD+ I would like to ask you to participate in a brief survey to understand what you like and dislike about a possible agreement between your community and international organizations. This agreement would focus on your community forest [mention the name of the CFUG/non-CFUG here]. As you might know, the climate is changing. The climate of the earth on average is becoming warmer and weather patterns are changing. This climate change is caused by carbon pollution into the atmosphere from factories and vehicles mainly in the richer countries like Japan, United States of America and Europe [show and discuss the RECOFTC graphic on climate change]. As a result of international agreements that were first made about 20 years ago, these rich countries and others are responsible to reduce their carbon emissions. These countries are finding it difficult to reduce their emissions and the world climate has therefore continued to change. This climate change is considered a serious problem. To help with, or in addition to, the efforts to reduce the amount of carbon that the rich countries are emitting, an international program was created to use the abilities of forests to store carbon to help reduce climate change. As you may know, trees grow by combining solar energy, water and carbon from the atmosphere. Healthy forests therefore actually remove carbon from the atmosphere, which helps the climate [show and discuss RECOFTC graphic on carbon sequestration]. Money has been collected from richer countries for the purpose of reducing deforestation and forest degradation in low-income countries like Nepal. Using these funds it is expected that international organizations will pay money to governments, individuals and communities like yours to reduce deforestation, improve forest quality and capture carbon. This program is called REDD+ [show and discuss RECOFTC graphics on REDD+]. The program is voluntary and no communities or individuals in Nepal will be forced to participate.   43 Do you have any questions about what I’ve just said? Do you agree to participate? [Proceed if respondent agrees] Experiment Background There has been no decision to implement REDD+ in your area and to my knowledge there is no plan to do so. It may, however, come to Nepal and it is therefore very important to understand what you and others in your community who use and protect forests would like to see in such agreements. That is why we want to ask you for your views. The choice of whether to participate will be made by you and your fellow forest users. Though you and your neighbors may decide to participate in REDD+, there will be no coercion. If REDD+ were to come to Nepal, there will be an opportunity for Nepali communities to be paid money to capture carbon from the atmosphere in their forests. There would also be an opportunity for communities to enjoy other benefits from higher quality forests, such as more animals and plants, non-timber forest products and simply the chance to help and protect the forest environment. REDD+ agreements would be between international organizations interested in stopping climate change and the Government of Nepal. The Government would then make an agreement with your community, with active involvement of and some oversight by international organizations. The agreement will specify the responsibilities your community takes on, such as reductions in fuelwood collections and open grazing elimination (if appropriate). All these steps can improve forest quality and increase carbon sequestration. Progress will need to be monitored and verified every year. You may also need to make work and money contributions to your forest user group community in addition to what you are currently doing. The agreement will also specify the payment in rupees that will be made each year and will detail how those resources can be used. For example, resources coming to the community may be used for community development projects like children’s education, health and community recreation. They might also be used to fund household or individual projects administered by the community like support for income generation activities, installation of biogas digesters, purchase of tractors or use of improved seeds and fertilizers. Alternatively, resources (or some part) could be divided equally among households in your group. Each household might therefore receive an equal share of the annual REDD+ payment and those funds could be used as each household prefers. If you are part of a community forest user group (CFUG), this REDD+ agreement would be with the CFUG. If you have not established a CFUG, to participate in REDD+ and receive payments for increasing carbon in your forest you will need to establish a CFUG. As of now, there are no specific activities related to forest management that focus on REDD+. To participate in REDD+, your CFUG would need to develop or revise its forest management plan to increase carbon sequestration. Monitoring and verification would also need to be included in such plans and as I mentioned, a formal agreement would be developed. The government, probably through the District Forestry Office, with financial resources from international organizations, would provide training and financial support to help you develop these plans. Because international organizations are providing the REDD+ funds, there will be good and open record-keeping, which   44 will help control any potential mismanagement of community funds. The participation of such international organizations will also contribute to more equitable distributions of benefits among community members. We emphasize that the main responsibility for organizing the CFUG and its members to meet REDD+ requirements and distribute rewards will be with you and your neighbors. If you and your community would like to participate in REDD+, any conflicts or controversies within your community that block the making and implementation of a REDD+ agreement will need to be resolved. If you and your neighbors would like additional support, depending on the capacity, availability and goodwill in the District Forestry Office, help may be available with organizing your CFUG (if needed) or to improve its operation. We will now ask you to make 6 choices among possible REDD+ contracts. Each choice will have three options, one of which is the current situation with no REDD+. These options are described by the following attributes: Annual total REDD+ payment to your community. These amounts are presented as rupees per household (to calculate the total payment, multiply the per household amount by the number of households in your community) The portion of REDD+ payments that go to communities for community projects and /or equally divided between households in your group The word after the word “community” is the portion going to communities and the word after the word “households” is the portion to households like yours. REDD+ required fuelwood reduction measured as a portion of your current use Open grazing is prohibited or not (for non-CFUGs only) Do you have any questions?   45 Choice Sets Choice Set Non‐CFUG Households  Attribute in English Possible Levels by Level number Annual total REDD+ payment to your 1. Rs. 1000 community. 2. Rs. 2000 3. Rs. 3000 These amounts are presented as rupees per 4. Rs. 4000 household (to calculate the total payment, 5. Rs. 5000 multiply the per household amount by the number of households in your community) The portion of REDD+ payments that go to 1. Community None, communities for community projects and /or Households All equally divided between households in your group 2. Community Half, Households Half The word after the word “community” is the portion going to communities and the word 3. Community All, after the word “households” is the portion to Households None households like yours. Required fuelwood reduction measured as a 1. One-quarter portion of your current use 2. One-half 3. Three-Quarters 4. All Open grazing is prohibited 1. Yes 2. No The exchange rate at the time of the experiment was RS. 85/$US   46 Choice Set CFUG Households  Attribute in English Possible Levels by Level number Annual total REDD+ payment to your 1. Rs. 1000 community. 2. Rs. 2000 3. Rs. 3000 These amounts are presented as rupees per 4. Rs. 4000 household (to calculate the total payment, 5. Rs. 5000 multiply the per household amount by the number of households in your community) The portion of REDD+ payments that go to 1. Community None, communities for community projects and /or Households All equally divided between households in your group 2. Community Half, Households Half The word after the word “community” is the portion going to communities and the word 3. Community All, after the word “households” is the portion to Households None households like yours. Required fuelwood reduction measured as a 1. One-quarter portion of your current use 2. One-half 3. Three-Quarters 4. All The exchange rate at the time of the experiment was RS. 85/$US   47 Robustness Checks Participation in REDD+ Program and Estimated CO2 Sequestration Costs at Various Prices National CO2 Supply Function Assumed 10% Fuelwood Reduction Program Assumes Lognormally Distributed Payment and Fuelwood  Reduction Attributes.  $450.00 $400.00 $350.00 $300.00 $250.00 $200.00 $150.00 $100.00 $50.00 $0.00 867.14 16475.58 32084.03 47692.48 63300.93 78909.38 94517.83 110126.28 125734.72 141343.17 156951.62 172560.07 188168.52 203776.97 219385.41 234993.86 250602.31 266210.76 281819.21 297427.66 313036.10 328644.55 344253.00 359861.45 375469.90 391078.35 406686.80 422295.24 437903.69 453512.14 469120.59 484729.04 500337.49 515945.93 531554.38 547162.83 Cumulative Tons of CO2 in National Rural Population Participation of CF Households in Program Requiring Fuelwood  Reductions at Various per‐ton CO2 Prices.  Assumes Normal Distributions. $150.00 $100.00 $50.00 $0.00 0.16% 2.64% 5.12% 7.61% 10.09% 12.58% 15.06% 17.55% 20.03% 22.52% 25.00% 27.48% 29.97% 32.45% 34.94% 37.42% 39.91% 42.39% 44.88% 47.36% 49.84% 52.33% 54.81% 57.30% 59.78% 62.27% 64.75% 67.24% 69.72% 72.20% 74.69% 77.17% 79.66% 82.14% 84.63% 87.11% 89.60% ‐$50.00 ‐$100.00 Participation in Program  N=644. 2 Observations with estimated WTA < -$1000 trimmed as extreme values. No WTA values >$1000.   48   ‐$100.00 ‐$50.00 $0.00 $50.00 $100.00 $150.00 2167.84 36853.28 71538.72 106224.16 140909.60 175595.05 210280.49 244965.93 279651.37 314336.81 349022.25 383707.69 418393.13 453078.57 487764.01 522449.45 557134.90 591820.34 626505.78 661191.22 695876.66 730562.10 765247.54 799932.98 National CO2 Supply Function 834618.42 Assumes Normal Distributions.  869303.86 903989.31 938674.75 Cumulative Tons of CO2 in National Rural Population Assumed 25% Fuelwood Reduction Program.   973360.19 1008045.63 1042731.07 1077416.51 1112101.95 1146787.39 N=644. 2 Observations with estimated WTA < -$1000 trimmed as extreme values. No WTA values >$1000. 1181472.83 1216158.27 1250843.72 49 National CF Member CO2 Supply Function Assumed 10% Fuelwood Reduction Program.   Assumes Normal Distributions. $150.00 $100.00 $50.00 $0.00 867.14 14741.31 28615.49 42489.67 56363.84 70238.02 84112.19 97986.37 111860.55 125734.72 139608.90 153483.08 167357.25 181231.43 195105.61 208979.78 222853.96 236728.13 250602.31 264476.49 278350.66 292224.84 306099.02 319973.19 333847.37 347721.55 361595.72 375469.90 389344.08 403218.25 417092.43 430966.60 444840.78 458714.96 472589.13 486463.31 500337.49 ‐$50.00 ‐$100.00 Cumulative Tons of CO2 in National Rural Population N=644 2 Observations with estimated WTA < -$1000 trimmed as extreme values. No WTA values <$1000. Participation of CF Households in Program Requiring Fuelwood  Reductions at Various per‐ton  CO2 Prices Assumes Lognormal  Distribution of Payment Vehicle $300.00 $250.00 $200.00 $150.00 $100.00 $50.00 $0.00 0.16% 2.95% 5.75% 8.54% 11.34% 14.13% 16.93% 19.72% 22.52% 25.31% 28.11% 30.90% 33.70% 36.49% 39.29% 42.08% 44.88% 47.67% 50.47% 53.26% 56.06% 58.85% 61.65% 64.44% 67.24% 70.03% 72.83% 75.62% 78.42% 81.21% 84.01% 86.80% 89.60% 92.39% 95.19% 97.98% 100.78% ‐$50.00 ‐$100.00 ‐$150.00 ‐$200.00 ‐$250.00 Participation in Program    50   ‐$200.00 ‐$100.00 $0.00 $100.00 $200.00 $300.00 $400.00 $500.00 $600.00 $700.00 ‐$250.00 ‐$200.00 ‐$150.00 ‐$100.00 ‐$50.00 $0.00 $50.00 $100.00 $150.00 $200.00 $250.00 $300.00 0.16% 2167.84 2.95% 41188.96 5.75% 80210.08 8.54% 119231.20 11.34% 158252.32 14.13% 197273.45 16.93% 236294.57 19.72% 275315.69 22.52% 314336.81 25.31% 353357.93 28.11% 392379.05 30.90% 431400.17 33.70% 470421.29 36.49% 509442.41 39.29% 548463.54 42.08% 587484.66 44.88% 626505.78 47.67% 665526.90 50.47% 704548.02 53.26% 743569.14 56.06% 782590.26 58.85% 821611.38 61.65% 860632.50 Participation in Program  64.44% 899653.63 67.24% 938674.75 70.03% 977695.87 Distribution of Payment Vehicle 72.83% 1016716.99 Lognormally Distributed Payment Vehicle 25% Fuelwood Reduction Program  Cumulative Tons of CO2 in National Rural Population 75.62% 1055738.11 78.42% 1094759.23 81.21% 1133780.35 84.01% 1172801.47 Fuelwood Reductions at Various per‐ton  CO2 Prices with  Participation of Non‐CF Households in Program Requiring  86.80% 1211822.59 89.60% 1250843.72 National CF CO2 Supply Function ‐ Assumed Lognormal  92.39% 1289864.84 95.19% 1328885.96 97.98% 1367907.08 100.78% 1406928.20 51   ‐$200.00 ‐$100.00 $0.00 $100.00 $200.00 $300.00 $400.00 $500.00 $600.00 $700.00 2167.84 41188.96 80210.08 119231.20 158252.32 197273.45 236294.57 275315.69 314336.81 353357.93 392379.05 431400.17 470421.29 509442.41 548463.54 587484.66 626505.78 665526.90 704548.02 743569.14 782590.26 821611.38 860632.50 899653.63 938674.75 977695.87 1016716.99 Cumulative Tons of CO2 in National Rural Population 1055738.11 1094759.23 25% Fuelwood Reduction Program  Distribution of Payment Attribute WTA 1133780.35 1172801.47 1211822.59 1250843.72 1289864.84 1328885.96 National Non‐CF CO2 Supply Function ‐ Assumed Lognormal  1367907.08 1406928.20 52 OLS Regressions of Potential Supply Shifters on Marginal WTA (in Nepali Rupees) to Participate in REDD+ Program Requiring Reduction in Fuelwood Collections. Assumes Lognormally Distributed Payment Vehicle and Fuelwood Reduction Attributes OLS Regression of Potential Supply Shifters on Marginal WTA (in Nepali Rupees) of CF Member Households to Participate in REDD+ Programs Requiring Reduction in Fuelwood Collections Variable Regression Coefficient P Value Respondent age (years) -1.53** 0.03 Respondent is female (1 = yes, 0 otherwise) 7.44 0.77 Female-headed households (1 = yes, 0 otherwise) 59.72** 0.03 Household size (number of people) 12.53*** 0.01 Household is classified as poor or ultra-poor (1 = yes, 0 otherwise) 18.64 0.41 Dalit ethnic group (1 = yes, 0 otherwise) -32.12 0.34 Indigenous or Newar ethnic group (1 = yes, 0 otherwise) 0.47 0.99 Madheshi ethnic group (1 = yes, 0 otherwise) -59.49 0.26 Respondent migrated to site from another location (1 = yes, 0 otherwise) -45.76** 0.05 Uses biogas (1 = use, 0 otherwise) -36.32 0.39 Uses LPG (1 = use, 0 otherwise) -41.69* 0.07 Uses improved biomass cooking stove (1 = use, 0 otherwise) 72.14 0.11 Uses firewood (1 = use, 0 otherwise) -106.97 0.22 Household owns land (1 = yes, 0 otherwise) -143.40** 0.05 Walking distance from respondent’s house to road < 2 hours (1 = yes, 0 otherwise) 20.30 0.55 Respondent says they are very likely or extremely likely to benefit personally from REDD+ (1 = yes, 0 otherwise) -52.96*** 0.00 Reported firewood used per month (kilograms) 0.00 0.81 Number in formal CFUGs (number of groups) -20.52 0.25 Constant 690.69*** 0.00 Assumes lognormal payment and fuelwood reduction attributes ***, **, * indicate significant at 1%, 5% and 10% levels. N= 600, R2=0.10, F= 2.74, prob>F=0.0018   53 OLS Regression of Potential Supply Shifters on Marginal WTA (in Nepali Rupees) of Non-CF Households to Participate in REDD+ Program Requiring Reduction in Fuelwood Collections. Assumes Normally Distributed Payment Vehicle and Fuelwood Reduction Attributes Variable Regression Coefficient P Value * Respondent age (years) -1.31 0.06 Respondent is female (1 = yes, 0 otherwise) 0.55 0.98 Female-headed households (1 = yes, 0 otherwise) 12.97 0.62 Household size (number of people) 3.94 0.30 Household is classified as poor or ultra-poor (1 = yes, 0 otherwise) 2.71 0.89 Dalit ethnic group (1 = yes, 0 otherwise) -9.72 0.66 Indigenous or Newar ethnic group (1 = yes, 0 otherwise) -14.32 0.51 Madheshi ethnic group (1 = yes, 0 otherwise) -33.73 0.29 Respondent migrated to site from another location (1 = yes, 0 otherwise) -8.13 0.64 Uses biogas (1 = use, 0 otherwise) -42.74* 0.07 Uses LPG (1 = use, 0 otherwise) -34.32* 0.08 Uses improved biomass cooking stove (1 = use, 0 otherwise) -40.14 0.11 Uses firewood (1 = use, 0 otherwise) 64.88*** 0.01 Household owns land (1 = yes, 0 otherwise) -26.25 0.50 Walking distance from respondent’s house to road < 2 hours (1 = yes, 0 otherwise) -0.40 0.99 Respondent says they are very likely or extremely likely to benefit personally from REDD+ (1 = yes, 0 otherwise) -26.24** 0.02 Reported firewood used per month (kilograms) -0.01 0.72 Number in formal CFUGs (number of groups) 10.30 0.42 Constant 287.99*** 0.00 Assumes lognormal payment and fuelwood reduction attributes ***, **, * indicate significant at 1%, 5% and 10% levels. N= 618, R2=0.03, F=1.76, prob>F=0.05   54