Economic Model to Forecast Future Rates of Deforestation and forest Degradation in Nepal

Transcription

1 Economic Model o Forecas Fuure Raes of Deforesaion and fores Degradaion in Nepal REDD Implemenaion Cenre Minisry of Foress and Soil Conservaion Babar Mahal, Kahmandu, Nepal Ocober 19, 2015

2 Table of Conens Table of Conens Execuive Summary Inroducion Analysis Mehods Dynamic Opimizaion Model Source of model inpus Developmen Scenarios Modeled Resuls and Discussion Compuable General Equilibrium model Model Daabase The Closure Scenario Analysis Resuls and Discussion Conclusions and Policy Implicaions References Appendix 1: Land Use Model Appendix 2: Land Use Model - Regional modeled land use resuls beween Appendix 3: CGE Calculaions underlying he simulaions Appendix 4: CGE Daa Sources and Mehods Used in Updaed and Exended 2007 SAM Appendix 5: Revenue from Fores Desrucion Appendix 6: Lis of paricipans of naional sakeholder workshops Page 2 of 66

3 1 Execuive Summary This repor describes he developmen of wo imporan models for analyzing he impacs of REDD policy in Nepal: A dynamic land use model and a Compuable General Equilibrium Model (CGE). The dynamic land use model provides projecions of disribuion of land use change and fores managemen change. The CGE model provides projecions of he demand for wood and agriculural producs as well as he resuling demand for land. The models are designed o esimae he effecs of REDD policy spaially wihin Nepal and o illusrae how he resuling financial flows and land use changes will inerac wihin he economy. Boh models are developed wih Nepal specific mos recen available daa. The dynamic land use model is a dynamic opimizaion approach ha models land use in hree caegories namely foress, crops, and livesock. Paymens are made direcly wihin he model for carbon and he model re-allocaes land among uses and regions o opimally sore carbon and produce crop, livesock, and fores oupus. The CGE model is a sandard approach ha relies on a recenly developed social accouning marix for he counry of Nepal. The dynamic land use model suggess ha foresland is increasing modesly in he baseline case, by around 2000 hecares per year over he nex 60 years. The change is no he same in all regions. The Mid-hill region experiences more afforesaion over he nex 60 years, while Terai and Churia experience modes deforesaion of around 1800 hecares per year in oal as land convers o croplands. Some of he increase in croplands in Terai and Churia is derived from pasureland, albei a small amoun. When carbon prices are added o he dynamic land use model, he model suggess ha for around $10 per on CO2, deforesaion will largely be sopped. Ineresingly, deforesaion coninues o occur in Churia in all of he scenarios examined, suggesing he relaively high value of crop producion here. The larges changes in land-use occur in he leas producive area, he Mid-Hill region. The CGE model is simulaed wih a REDD policy ha induces 2617 addiional hecares each year o be mainained in agriculural producion for a paymen of 2761 million Nepalese Rupees each year. Under his policy, he effecs are iniially posiive for Nepal, bu over ime, he removal of land from agriculural producion reduces rae of growh in overall economic aciviy in Nepal alhough he overall such effecs are small. Page 3 of 66

4 2 Inroducion Reducing Emissions from Deforesaion and fores Degradaion (REDD) is evolving as a means o reduce fores secor carbon emissions hrough fores managemen and enhanced fores governance in foresry and relaed secors. The World Bank s Fores Carbon Parnership Faciliy (FCPF) is assising Nepal o develop and apply sraegies o address he drivers of deforesaion and fores degradaion. This sudy is an inpu o assess he sraegic opions and developing a model o forecas fuure raes of deforesaion and fores degradaion. A REDD scheme exchanges paymen, o a developing counry, for a promise o reduce fores clearing and/or fores degradaion. The paymen and he resricions on fores use will affec various secors of he economy, and he effecs will vary over ime. Thus a muli-period, muli-secoral economic model is needed o flesh ou hese effecs in deail. The objecive of his projec is o develop a dynamic land use model and a compuable general equilibrium economic simulaion model o projec he fuure land use in Nepal and assess he carbon paymen impacs on land use changes and he poenial of carbon sequesraion in Nepal. Generally, any carbon policies ha resul in posiive carbon prices would aler land rens and land uses across fores, crop, and pasure land. The repor provides some general background on REDD schemes and on foress in Nepal. The models used o projec land use and aendan economic impacs are hen described along wih he resuls of several alernae policy scenarios. Two models are developed, a dynamic land use model of he enire counry and a Compuable General Equilibrium model (CGE model) focused on he Terai and Churia areas of he counry. Carbon axes, REDD schemes and foress in Nepal From 1880 unil 2012 world emperaures rose 1 by 0.8 C. During he las 30 years a consensus has developed ha his emperaure rise (a) is largely due o CO 2 emissions caused by humans, and (b) may cause serious problems. The CO 2 damage, which affecs all regions, is due o he accumulaed level of amospheric CO 2 (no he curren emision rae) and may 2 persis for decades, even if emissions cease. The main emission sources are burning of fossil fuels and deforesaion. Deforesaion accouns for perhaps 6 17% of all emissions worldwide (Werf e al, 2009), and accouns for a much larger share of emissions in some counries. Deforesaion emissions arise because a fores is a carbon sore (even hough a maure fores absorbs lile ne CO 2 ). Conversion of fores o, say, cropland, releases his carbon as CO 2. 1 source: NASA-GISS Global mean land-ocean emperaure 2 Esimaes of he half-life of amospheric CO 2 range from 5 o 200 years (IPCC, 2001). Page 4 of 66

5 To reduce amospheric CO 2, emissions axes have been proposed -- bu no widely adoped. If applied uniformly, such axes would raise enormous revenue -- mos of i from richer counries. Poliically accepable schemes are usually far more limied in scope. For example: They may apply o one counry only, or a group of counries. 'Grandfahering' arrangemens may exemp from axaion emissions below a sipulaed (ofen hisorical) level. Whole secors, such as agriculure, may be exemp. Ofen expor-oriened secors are exemp. A variaion on he ax scheme is a paymen no o pollue. For example, a poorer counry may be paid o replace older, dirier, power saions wih more efficien modern generaors. The paymen would reflec he emissions saved by he upgrade. REDD (Reducing Emissions from Deforesaion and fores Degradaion in developing counries) schemes fall wihin his caegory. Firs, he schemes focus on foress only. A counry mus esablish a 'reference level' of fores-relaed emissions ha do or migh occur in he absence of a REDD paymen. REDD paymens will be made in proporion o he exen ha fuure emissions fall below his reference level. There are many pracical difficulies wih such schemes. Esablishmen of he reference levels, and subsequen emissions monioring impose complex requiremens. A basic REDD scheme does no reward previous progress in reducing deforesaion or fores degradaion. There is no paymen for fores sewardship ha merely mainains he saus quo. Thus inernaional discussions coninue, which may resul in agreemen abou modified REDD schemes. Neverheless he CGE modelling focusses on a basic REDD scheme, where paymens from abroad are in proporion o he reducion in emissions below a reference level. As explained below, he CGE analysis (a) focus only on he Terai/Churia par of Nepal; and (b) base he reference level on he average rae of deforesaion during recen decades. Foress in Nepal 3 Fores sill covers much of he non-alpine regions of Nepal, alhough much has been cleared in he mos culivable plains areas. Many foress are owned and managed ulimaely by he sae; alhough managemen has been delegaed in some places o he local level. There is selecive logging of old/maure rees bu raher lile acive fores managemen (such as planing, hinning, and opimized iming of cuing). Ofen, he sock of rees is older han would be deemed opimal. Illegal logging also occurs. 3 The brief survey in his and he nex secion draws on maerial found a he excelllen Page 5 of 66

6 As well as providing sawlogs, foress and heir producs suppor Nepalese agriculure and poorer people. Firewood, used for cooking and heaing, is gahered from he ground, from low branches and from coppiced sumps, filling perhaps half of Nepal's energy needs. Much firewood gahering is illegal or unconrolled. Forage for livesock is gahered similarly. Livesock may graze in he fores; heir dung and lier from he fores floor are used o ferilize fields. All of hese aciviies are ofen pursued above a susainable level, leading o fores hinning and degradaion. Figure 1. Physiographic Zones in Nepal (aken from FRA/DFRS, 2014) The Hills area of Nepal (Figure 1) has hisorically suffered from poor communicaions, povery and over-populaion. Deforesaion and fores degradaion (caused by over-harvesing of firewood and fodder) are long-sanding problems. Ye since he 1970s considerable progress has been made o comba hese problems, especially hrough Communiy Foresry schemes, whereby local communiies are encouraged o manage and proec local fores resources. Two Hills-specific facors underlying he success of such schemes are ha (a) i is usually possible o associae paricular fores areas wih paricular villages; and (b) villages end o be fairly ehnically homogenous. These facors assis in creaing and enforcing local fores managemen plans (which are embedded in wider schemes for communiy developmen). Deforesaion and fores degradaion have been slowed, and someimes reversed. Anoher facor helping o preserve he Hills foress is oumigraion o he Terai area (see nex subsecion). Alhough he Communiy Foresry schemes were very successful, i is expeced ha here is lile scope o grealy expand he schemes (personal communicaion). Consequenly he CGE modelling of his sudy assumes ha a REDD scheme will no affec Hills fores managemen. Page 6 of 66

7 While mosly hills and mounains, Nepal includes also a ferile low-lying plains region along is souhern border, called he Terai (Figure 1). The Churia or Siwalik is generally composed of he firs range of foohills. Hisorically he Terai/Churia region was foresed, wih he iconic Sal ree as he dominan species. High prevalence of malaria resuled hisorically in low populaion levels, alhough expor logging ook place from he 1920s. In he 1950s malaria was eradicaed using DDT. The governmen encouraged fores clearing and migraion from he overcrowded hills. Migrans from India arrived oo. Today much of he Terai fores has been convered o cropland, and deforesaion coninues. The Terai/Churia region now accouns for mos of Nepal's agriculure and GDP, and abou half he populaion. Deforesaion causes CO 2 emissions (see below) bu he loss of he unique Terai/Churia habia has equally raised inernaional concerns. For various reasons Communiy Foresry schemes have been less successful in he Terai han in he hills (Nagendra e. al., 2005). Beer roads, he proximiy of he Indian border, and he high value of Sal imber make i harder o proec agains illegal logging. Compared o he hills, i is less easy o assign paricular fores areas o paricular communiies, and local populaions (ofen migrans from elsewhere) are more divided by ehniciy and case. Finally, he greaer feriliy of he Terai increases he pressure o conver foress o cropland. Deforesaion and fores degradaion Fores clearing causes emissions because he above-ground carbon conen of fores is greaer han han of he replacemen use (eg, crops). Ye fores degradaion is also noed as a cause of emissions. The degraded fores has fewer, smaller rees, and sequesers less carbon. The calculaions presened laer in his repor ignore emissions from degradaion; ha is, hey are assumed o be unchanged by a REDD scheme. We use Figure 2 o explain his assumpion. The op panel A of Figure 2 graphs he 2015 above ground carbon per hecare of he Terai/Churia, wih he leas-carbon land (crops) a he lef, and he mos-carbon land (healhy fores) a he righ. In beween is an area of degraded fores, wih a range of carbon conens. We may imagine, as he diagram suggess, ha he degraded fores forms a fringe beween cropland and healhy foress. The area under he graph is he oal amoun of sequesered carbon (shown as zero for crops). Panel B of Figure 2 shows he siuaion in 2045 wih no REDD scheme. Crop area has increased, and fores area has fallen by he same amoun. The area of (and carbon sored in) degraded fores does no change. So emissions can be esimaed merely by measuring he change in fores area. In effec, he degraded fores fringe simply moves righ over ime. To an observer on he ground, i could well appear as if degradaion were he main cause of emissions. Imagine a fores residen who lived iniially a he border beween healhy and degraded fores (shown by a house symbol). Over he period he would see his fores surroundings seadily degraded by grazing, lier removal and excessive harvesing of firewood and forage -- all Page 7 of 66

8 leading o emissions. There would be lile above-ground carbon around his house in so ha emissions from he final conversion o cropland migh be small. Panel C depics an alernae scenario for where fores area has again shrunk, bu crop area has no expanded. Insead, he area of degraded land has grown. This scenario is also possible, bu is no simulaed here. A: Terai/Churia land use 2015 above ground carbon per hecare crop area degraded fores area healhy fores area B: Terai/Churia land use 2045 above ground carbon per hecare crop area degraded fores area healhy fores area C: Alernae Terai/Churia land use 2045 above ground carbon per hecare crop area Figure 2. Deforesaion and fores Degradaion degraded fores area healhy fores area In he REDD policy scenario presened in CGE modelling we assume ha Terai deforesaion ceases from 2015, so ha Panel A also depics he 2045 siuaion. Thus, each par of he fringe area mus reain over ime he same carbon conen. By conras, in he base scenario, each spo in he fringe Page 8 of 66

9 area seadily loses carbon. In oher words, while he base scenario assumes over-use of he fringe, he policy scenario scenario assumes susainable use of he fringe. The susainabiliy migh be achieved by more careful managemen or by simply harvesing less firewood and forage -- which would have coss. Our REDD policy simulaion ignores mos of hese coss, so ha he coss of a REDD scheme are somewha underesimaed. 4 Is REDD a cheap way for richer counries o purchase emissions credis? The argumen is someimes made ha emission conrol by prevening fores clearing and degradaion may be much cheaper for he world as a whole han oher approaches, such as burying CO2 or using renewable energy. For example, afer he Brazilian fores has been plundered for imber and charcoal, he cleared land may be used as beef pasure which raises lile revenue. In such a case, a small REDD paymen migh be enough o preserve he fores. On he oher hand, a large REDD paymen migh be insufficien o preven Indonesian rainfores from being convered o profiable oilpalm planaions. Clearly, wihin Nepal local condiions vary grealy, so ha some places resemble Brazil, ohers Indonesia. There migh be paricular areas where a localized REDD scheme was aracive even wih a low carbon price. The daa used in his sudy for CGE modelling does no allow for such fine-grained geographical deail. Broad averages are used. Hence we overlook he possibiliy of some local "low-hanging frui". On he oher hand we assume a raher generous REDD paymen ($US 50 per onne of CO2). 4 Appendix 5 explores his aspec furher, building on useful commens by Dr Hom Pan, Senior Advisor, Naional Planning Commission, Kahmandu, Nepal. Page 9 of 66

10 3 Analysis Mehods This analysis is conduced in wo sages. Firs a dynamic opimizaion model of land use in Nepal is developed. This model maximizes he sum of he ne presen value of marke welfare in he foresry, crop, and livesock secors in Nepal. This model and he resuling analysis are developed for he enire counry. Second, a Compuable General Equilibrium (CGE) analysis of wo mos deforesed regions (i.e. Terai and Churia) in he counry is developed. 3.1 Dynamic Opimizaion Model The model in his sudy is based on he global-level land use model described in Choi e al. (2011). Tha model has been modified exensively o examine fores, crop, and pasure land use wihin Nepal for his sudy. The regions analyzed in his sudy are five physiographic zones, and include he enire area of Nepal (Terai, Churia, Mid-Hills, Mid-Mounain, and High-Mounain). For analysis purposes, he Mid-Mounain and High-Mounain physiographic zones are combined ogeher and presened as he Mounain zone in his repor 5. A number of imporan inpu variables were missing or unavailable when consrucing he model, so a number of key inpu variables have been derived from oher sources, or assumed. These include informaion on imber growh and yield, he age disribuion of ree species, land rens, and agriculural inpus in physiographic zones. Despie hese daa limiaion, he model provides robus resuls and can be easily accommodaed wih addiional daa if new daa becomes available for Nepal. The model opimally chooses inpus o producion, such as land, capial, and labor, in order o produce oupus like wood producs, crops, and livesock. The model can be described mahemaically as a consrained opimizaion problem (see Appendix 1). The key daa inpus required for he model include: Toal land area in foresry, crop, and livesock secor (000 ha) Capial inpus for crop and livesock Labor inpus for crop and livesock Produciviy facor for capial, labor, and land composie Land rens for foresry, crop, and livesock Timber yield (growh) funcion The model opimizes producion and land use for 150 years wih 10-year ime seps beginning in The reason for running he model over such a long ime period is o minimize he effecs of 5 Land use changes in High-Mounain region is insignifican compare o he oher regions and here are lile aciviies in crop producion. Toal agriculural producion in Mid Mounain and High Mounain region is abou 6% ou of oal producion in Nepal (SINA, 2014). Therefore, combining he wo Mounain zones will no have large impacs on he analysis resuls. Page 10 of 66

11 he erminal condiions on curren policy resuls. 6 We are ineresed only in he resuls for he firs 60 years, and presen hose below. The annual discoun rae is 5%. The model is solved using a General Algebraic Modeling Sysem (GAMS) program wih he CONOPT solver (Rosenhal, 2015). One of he criical issues of his analysis is o assess wheher carbon policies could be used o reduce deforesaion. To simulae hese ypes of policies wih he model, we incorporae carbon prices ino he objecive funcion. For our model, we ren carbon sored in he fores and pay for carbon ha is sored in wood producs for he ime i is sored here. This provides an esimaion of he implicaions of an efficien carbon policy. Oher policies could be explored as well, alhough we expec ha hey would be more expensive. For addiional deails on he model, see Appendix 1. We have conduced he analysis for hree developmen scenarios, a base case and wo alernaive demand scenarios. For he foresry secor, demand is assumed o increase relaively slowly in he baseline, following recen hisorical rends. Crop demands and livesock demands are assumed o increase more rapidly. 1. Base Scenario: crop demand assumed o increase a 5% annually for he firs 6 decades (i.e. unil 2070); fores produc demand rises a 3% annually for he firs 6 decades, and livesock produc demand increases a 4% annually for he firs 6 decade. Afer 6 decades all demands are held consan. 2. High Demand Scenario (HDS): crop demand assumed o increase wice as quickly as in he Base Scenario, 10% annually, and sabilizes. Fores and livesock raes remain he same as he Base Scenario. 3. Low Demand Scenario (LDS): crop demand assumed o say consan a he iniial level in 2010 while oher produc demands rise as he Base Scenario (see Appendix 2 for he elasiciy values). When carbon paymens are inroduced, deforesaion will decline and afforesaion will increase. The scenario wih higher (Iower) crop demand will require more (less) cropland usage, and should be more (less) cosly o reduce he deforesaion wih carbon paymens Source of model inpus This analysis is based on he mos recen available land use daa (FRA/DFRS, 2014) for all 5 physiographic zones (Terai, Churia, Mid-Hill, Mid Mounain, and High Mounain) in Nepal. In each physiographic zone, here are 5 regions (Easern, Cenral, Wesern, Mid-Wesern, and Far-Wesern) (Figure 3). This FRA daa provides our iniial land use disribuion of fores, grassland, and agriculural land use for 2010 (Table 1). 6 In dynamic opimizaion models, long ime horizons for soluions minimize he impac of selecing erminal condiions on he resuls of ineres, i.e., he policy projecions over he nex 50 years. Page 11 of 66

12 A Regions in FRA daa B Regions in Census daa Figure 3. Comparison of regional definiions beween FRA and Census Page 12 of 66

13 Table 1. Land use area (000 hecares) Physiographic zone Terai Churia Mid-Hill Mounains Region Crop Grass Fores Easern Cenral Wesern Mid-Wesern Far-Wesern Easern Cenral Wesern Mid-Wesern Far-Wesern Easern Cenral Wesern Mid-Wesern Far-Wesern Easern Cenral Wesern Mid-Wesern Far-Wesern Source: FRA/DFRS (2014) We performed a daa approximaion process for agriculural inpus, agriculural oupus, and parameers for crop and livesock producion funcion because here is discrepancy in regional definiions beween he FRA land use daa and he agriculural census daa (SINA, 2014 & CBS, 2014) which provides agriculural labor inpus and oupus (Table 3). Land rens from GTAP (2015) also are composed of a differen regional definiion (based on 18 Agro Ecological Zones) so we approximae he land ren daa as well. While here are 5 physiographic zones in FRA daa, he agriculural census daa (SINA, 2014 & CBS, 2014) is divided ino 3 Ecological zones (Eco-Terai, Eco-Hill, Eco-Mounain) wih he same regions (Easern, Cenral, Wesern, Mid-Wesern, and Far-Wesern) (See comparison of maps in figure 3). We divided agriculural census daa in each Ecological zone ino each physiographic zone by applying proporion of labor uses for inpus and proporion of crop oupus for land ren and oupu producion funcion. Page 13 of 66

14 Table 2. Summary of model inpus and daa source Model Inpu Source Toal land area in foresry, crop, and livesock secor FRA/DFRS (2014) (000 ha) Capial inpus for crop and livesock GTAP, SINA (2014), CBS (2014) Labor inpus for crop and livesock GTAP, SINA (2014), CBS (2014) Produciviy facor for capial, labor, and land Ludena e al (2007) composie Land rens for foresry, crop, and livesock in each GTAP, SINA (2014), CBS (2014) disric Timber yield (growh) funcion DFRS (1993) & DFRS (1999) The process for geing physiographic zone daa from census daa is performed wih he following seps. 1. Calculae he proporion of labor inpus and crop oupu in 15 regional Ecological zones (5 regions X 3 Ecological zones) (Table 3). This calculaion provides he proporion of labor use and oal crop oupu in Nepal. For example, Easern Terai region produces 14% of oal crop producion in Calculae he proporion of physiographic zone in each regional Ecological zone (Table 4). Since FRA daa is based on GIS daa, combining FRA GIS daa wih disric map, he proporion of cropland for each physiographic zone in each disric could be calculaed 7. By doing his o all he regions, we can obain he oal physiographic zone in each Ecodevelopmen zone. For example, Easern Eco-Terai region is composed of 95% Terai physiological zone and 5 % Churia physiographic zone. 3. Muliplying each proporion of labor usage and oupu in each Ecological zone (Table 3) o physiographic zone raio (Table 4) generaes he labor inpu and crop oupu daa for each regional physiographic zone (5 regions X 5 physiographic zones). 7 The underlying assumpion of his procedure is ha crop inpus and oupu are proporion o cropland area. I migh no reflec regional produciviy differences in each physiographic zone originaing from soil produciviy or ecological condiions and i could be a srong assumpion bu his is he only available agriculural daa for physiographic zone a his poin. Page 14 of 66

15 Table 3. Proporion of labor use and oal crop oupus from census Regional Labor Oupu Eco-developmen zone proporion proporion Easern Terai 15% 14% Easern Hills 8% 8% Easern Mounain 2% 2% Easern Toal 25% 25% Cenral Terai 19% 24% Cenral Hills 10% 10% Cenral Mounain 2% 2% Cenral Toal 30% 36% Wesern Terai 10% 11% Wesern Hills 11% 9% Wesern Mounain 0.2% 0.1% Wesern Toal 21% 20% Mid-wesern Terai 8% 6% Mid-wesern Hills 5% 4% Mid-wesern Mounain 1% 1% Mid-Wesern Toal 14% 11% Far-wesern Terai 6% 5% Far-wesern Hills 2% 2% Far-wesern Mounain 1% 1% Far-Wesern Toal 9% 8% Grand Toal 100% 100% Source: SINA (2014) & CBS (2014) Table 4. Percen of Physiographic zone in each Ecological zone calculaed wih cropland area Easern Terai Churia Mid-Hill Mid- Mounain High- Mounain Toal Eco Terai 95% 5% 100% Eco Hills 1% 8% 88% 2% 100% Eco Mounain 0% 0% 51% 48% 1% 100% Cenral Terai Churia Mid-Hill Mid- Mounain High- Mounain Toal Eco Terai 82% 16% 1% 100% Eco Hills 0% 17% 80% 3% 100% Eco Mounain 60% 40% 100% Page 15 of 66

16 Wesern Terai Churia Mid-Hill Mid- Mounain High- Mounain Toal Eco Terai 77% 17% 6% 100% Eco Hills 3% 82% 15% 100% Eco Mounain 20% 80% 100% Mid-Wesern Terai Churia Mid-Hill Mid- Mounain High- Mounain Toal Eco Terai 48% 45% 8% 100% Eco Hills 8% 73% 19% 100% Eco Mounain 100% 100% Far-Wesern Terai Churia Mid-Hill Mid- Mounain High- Mounain Toal Eco Terai 87% 12% 100% Eco Hills 3% 92% 4% 100% Eco Mounain 25% 75% 100% Toal labor and capial inpu daa for physiographic zones is calculaed using GTAP daa ha provides he oal usage in Nepal. Alhough census daa provides labor inpus (CBS, 2014), i doesn have capial inpu while GTAP provides he boh. To have consisen daa for inpus, we used GTAP labor and capial inpu daa for he whole Nepal and applied he regional informaion from he census daa (CBS, 2014). In addiion, livesock secor inpu daa is no available from he census daa so we apply he same raio used for he crop secor wih oal inpu for livesock from GTAP. Similar as labor inpu calculaion process, land ren in physiographical zone is obained using GTAP land inpu daa and above seps (Table 3 and Table 4). While labor usage proporion is used for labor inpu daa, oupu proporion is used for ren daa calculaion (Table 3). Land inpu in US dollar erm is available from GTAP land use daa for Agro Ecological Zones (AEZ) in Nepal. The AEZ disribuion, however, is no correlaed wih he disrics used in his model, so we sum land use inpus from GTAP across he AEZs o derive oal land ren for each secor in Nepal. We hen disaggregae his proporionally o he regions used in his model. Table 5 presens he resuling land rens, labor inpus, and capial inpus used in his sudy. Page 16 of 66

17 Table 5. Calculaed daa Regional land ren (US million dollars) Physiographic zone Terai Churia Mid-Hill Mounain Region Crop Grass Fores Easern Cenral Wesern Mid-Wesern Far-Wesern Easern Cenral Wesern Mid-Wesern Far-Wesern Easern Cenral Wesern Mid-Wesern Far-Wesern Easern Cenral Wesern Mid-Wesern Far-Wesern Toal in 3 regions Nepal Toal in GTAP GTAP land use daa version 8. The land ren, labor and capial inpu daa are Value of Firms' purchases of endowmens a Marke prices (VFM). Page 17 of 66

18 A) Regional labor and capial inpus (US million dollars) Crop Livesock Labor Capial Labor Capial Terai Churia Mid-Hill Mounain* Nepal oal *Mounain includes Mid Mounain and High Mounain physiographic zones. Timber yield is calibraed using he global imber model parameers as a saring poin and available volume daa obained from in Nepal (DFRS, 1993 & DFRS, 1999). Since here is no available imber age informaion, we assume ha he maximum imber age is 110 years. We also assume ha all he imber in Nepal is a maximum age. Timber yield (growh) funcion defined as V exp( / age), age is imber age, α and β is parameer. Adjusing he parameer α and β, we approximae he volume a age 110 o mach he available daa from DFRS (1993) and DFRS (1999). Figure 4 presens he imber yield by age in each disric and Table 6 presens he calibraed yield funcion parameers and Table 7 shows he carbon conen (CO 2 e) per hecare by imber ages. For he presenaion purpose, we show resuls in 20 year bins, bu he model uses decadal daa. Page 18 of 66

19 Figure 4. Toal aboveground biomass yield (growh) in each region Page 19 of 66

20 Table 6. Calibraed parameers of imber yield funcion. Terai imber yield Parameer (α) Parameer (β) Easern Cenral Wesern Mid-Wesern Far-Wesern Churia imber yield Parameer (α) Parameer (β) Easern Cenral Wesern Mid-Wesern Far-Wesern Mid-Hill imber yield Parameer (α) Parameer (β) Easern Cenral Wesern Mid-Wesern Far-Wesern Mounain imber yield Parameer (α) Parameer (β) Easern Cenral Wesern Mid-Wesern Far-Wesern * Yield funcion defined as V exp( / age), age is imber age, α and β is parameer. Page 20 of 66

21 Table 7. Carbon amoun per hecare by imber ages (on CO 2 e/ha)* Timber ages (years) Easern Cenral Terai Mid-Wesern Wesern Far-Wesern Easern Cenral Churia Mid-Wesern Wesern Far-Wesern Easern Cenral Mid-Hill Mid-Wesern Wesern Far-Wesern Easern Cenral Mounain Mid-Wesern Wesern Far-Wesern *Carbon numbers in his able are only for he above ground carbon Developmen Scenarios Modeled To address he uncerainy in fuure demand for agriculural oupus and wood producs, he model was run under hree demand scenarios. Income and populaion boh are drivers of fuure demand for wood producs. We sar wih a se of baseline assumpions abou fuure demand based on hisorical raes of change. This is our "base scenario." In all cases, we assume ha demand growh sabilizes afer 60 years and is consan. The high demand scenario hen assumes ha he demand for crops increases more rapidly over ime. This could occur if income and populaion increase wihin Nepal or if expors increase dramaically. The low demand scenario holds he crop demand consan while allow livesock and fores demands o increase. Page 21 of 66

22 For he carbon analysis, we calculae he effecs of a series of carbon paymen scenarios, ranges from $1/on CO 2 e ($5/on C) o $41/on CO 2 e ($150/on C). By applying his, we can derive he supply (marginal cos) curve of carbon from foresland in 5 physiographic zones. We assume ha he carbon price is consan over ime. We apply he series of carbon paymen scenarios on each 3 differen food demand scenarios Resuls and Discussion Under Base Scenario, he model projecs ha here will be over 115 housand new hecares of foresland over he nex 60 years (Figure 5 and Appendix 2). A he same ime, less land is used for crops and more land is used for pasure. Alhough overall land use rends in Nepal sugges an increase in foress, he model projecs ha deforesaion occurs in he Terai and Churia regions, accouning 53 housand hecares and 50 housand hecares respecively (Table 8). In boh regions, deforesed hecares are mosly shifed ino cropland uses. Churia region is projeced o ge 51 housand of new cropland and Terai region is projeced o gain 56 housand hecares of new cropland. The census daa sugges ha hese regions are he mos producive agriculural regions in Nepal, accouning more han 60% of crop producion (Table 3). The resuls show he opposie rend in Mid-Hill region, wih 218 housand hecares of afforesaion mainly from cropland conversion. 9 I is absoluely possible o apply oher carbon price schemes such as IPCC s Assessmen Repor resuls. The reason for applying his mehod is o focus on he poenial of carbon supply in Nepal. Moreover, given he various uncerainies in parameers and daa, his simplifies he analysis and provides more sraighforward implicaions of carbon sequesraion poenial. Page 22 of 66

23 Figure 5. Toal land use resuls under Base Scenario (000 ha) Table 8 presens oal land use changes under 3 differen demand scenarios for he year 2070 relaive o Wih higher crop demand, more cropland is required o mee he higher demand. As a resul, over 300 housand addiional hecares compare o he base scenario are convered o cropland (Increasing 133 housand hecares under High Crop demand versus decreasing 169 Page 23 of 66

24 housand hecares under he Base). Mos of his increased cropland shifs from foresland, 250 housand hecares, alhough some also comes from pasures (around 53 housand hecares). Under he Low Demand Scenario, model projecs ha here will be major land use shif away from cropland o oher uses. Lower crop demand assumpion requires less land use in crop and i resuls in 637 housand hecares of cropland shif o oher land uses. Among hese shifs, 496 housand hecares are ransferred o foresland by 2070, an addiional 381 housand hecares compared o he Base scenario. Compared o fores and cropland, pasure does no change as much. Under he High Demand Scenario, i is projeced ha here will be around 50 housand hecares of loss compared o he Base Scenario and an addiional 87 housand hecares under Low Crop Demand. While his may seem surprising, he low demand scenario significanly reduces he demand in he secor ha requires mos of he land use change, he crop secor. Table 8. Modeled oal land use changes under differen demand scenarios beween 2010 and 2070 in Nepal (000 hecares) Fores Crop Pasure Base High Crop Demand Low Crop Demand Terai Churia Mid-Hill Mounain Toal Terai Churia Mid-Hill Mounain Toal Terai Churia Mid-Hill Mounain Toal Carbon paymen reduces he deforesaion and increases he afforesaion as he carbon price increases (Table 9). Under he Base Scenario, he $11 per on CO 2 e paymen increases foresland area over 1.5 million hecares, and a $41 per on CO 2 e carbon paymen, foresland area increases 3.6 million hecares by Higher crop demand requires more cropland usage and i is hus more cosly o reduce he deforesaion wih carbon paymen. Lower crop demand scenario requires less land use in crop hus foresland is projeced o increase he mos across he differen carbon prices. Page 24 of 66

25 For he mos par, addiional land in foress under he carbon price scenarios is derived from crop land. The high carbon price scenarios imply large changes in land use. For example, under he Base Case, cropland declines by abou 2/3 rds by Impacs on pasureland are smaller, bu ha is mainly because less land is used overall for pasure. Table 9. Modeled land changes for 60 years under 3 differen modeled demand scenarios wih carbon paymen (Appendix 5 shows deailed regional resuls) $0/ on CO 2 e $3/ on CO 2 e $5/ on CO 2 e Foresland (000 ha) $11/ on CO 2 e $27/ on CO 2 e $41/ on CO 2 e Base High Crop Demand Low Crop Demand Cropland (000 ha) Base High Crop Demand Low Crop Demand Pasureland (000 ha) Base High Crop Demand Low Crop Demand The sum of carbon gain above he Baseline ($0 per on CO 2 e) is calculaed o he annual equivalen of carbon gain using 5% discoun rae (Table 10). I ranges from 32 housand ons per year o 3.2 million ons per year under Base Scenario depending on he carbon price. As expeced, he carbon gain is he larges under Low crop Demand Scenario, up o over 3.4 million ons per year. In Figure 6, marginal cos curve (carbon supply) is ploed for he 3 Scenarios, basically presening Table 10. I is eviden ha he marginal cos curve (or carbon sequesraion supply curve) under Low crop Demand Scenario is he lowes (cheapes). Since we canno have definie informaion abou he fuure demand changes, his range of carbon sequesraion supply curves can provide a leas inferences on he carbon sequesraion poenial range given he scope of demand change assumpions in Nepal. Table 10. Annual equivalen of carbon gains for 60 years (000 ons CO 2 e/year) under 3 differen modeled demand scenarios $1/ on CO 2 e $3/ on CO 2 e $5/ on CO 2 e $11/ on CO 2 e $27/ on CO 2 e $41/ on CO 2 e BASE High Crop Demand Page 25 of 66

26 Low Crop Demand Figure 6. Marginal cos curves under 3 differen demand scenarios Page 26 of 66

27 3.2 Compuable General Equilibrium model 10 A fairly simple Compuable General Equilibrium (CGE) model has been consruced specifically for his projec. Like oher CGE models 11 he model consiss of equaions describing: Demands by indusries and final demanders for commodiies are influenced by prices. Demands by indusries for capial, land and labour are again influenced by prices. Commodiy axes affecing user prices are explicily modeled. Marke clearing: prices adjus o equae supply and demand for each commodiy. How facor income is disribued and he sysem of ransfer paymens beween households, governmen and he res-of-he-world (ROW) is clearly embodied in he model. Various macro indicaors, such as GDP, GNP, CPI, ec. are generaed in he model. Producion echnology is srucured by a series of "ness", shown in Figure 7 below. In more deail: 62 indusries are disinguished, each producing one commodiy. These are lised underneah Table 11. They include 13 agriculural indusries: Paddy, Maize, Whea, OhGrain, VegeFrui, Oilseed, Sugarcane, PlanFiber, OhCrops, Cale, OhAnmlPrd, RawMilk, Wool; and 4 fores indusries: Firewood, Timbers, GrassFoddr, OForesPrd. Each indusry and non-expor final demander use 62 composie commodiies -- each of which is a CES combinaion of a domesic good and is impored equivalen. Commodiy prices and impor/domesic shares are he same for all local users. For inermediae (indusry) demand he 62 composie commodiies are combined (via CES) ino an aggregae "Inermediae goods" inpu, which is demanded in proporion o oupu. Indusries also require, in proporion o oupu, a composie primary facor which is a CES aggregae of capial, land and labour used by ha indusry. Only agriculural and fores indusries have a land inpu. Some local oupu goes direcly o expor; he res is combined wih impors o form he composie good. 10 Marke clearing mechanism is he building block of he CGE model and hence quie ofen quesions are raised on he suiabiliy of such a model in an economy wih less well developed marke sysem. However, for he presen naure of exercise, similar model is recommended and used widely and hence such a model was proposed for he presen exercise. The model is reliable and applied in oher big REDD+ counries such as Brazil, Indonesia. 11 The model is solved using GEMPACK. I is very similar o he PhilGem model described in Corong and Horridge (2012) and o he sandard IFPRI GAMS CGE model (Lofgren e al, 2002), excep ha rade and ranspor margins are reaed as direc demands. To compue and compare 40-year base and policy scenarios akes abou 5 minues. Archive Iem TPMH0151 a hp:// conains maerials sufficien o replicae he simulaion repored here. Page 27 of 66

28 Each of he 7 household ypes demand he 62 composie commodiies following he Linear Expendiure Sysem (LES). The commodiy mixes of invesmen and of governmen demands are exogenous. Exporers face foreign demand curves of consan elasiciy -- ypical values 4-8. Mos of he CES, expendiure, and expor demand elasiciies are aken from he GTAP daabase 12. The figure shows some ypical values. The supply of agriculural land is disribued beween agriculural indusries using a CET funcion wih ransformaion elasiciy The oal supply of agriculural land o all agriculural indusries is exogenous, as is he oal supply of land o all fores indusries. Land inpus o fores indusries move ogeher (hey are sharing he same land). Alhough no used in his projec, he model allows for several ypes of labour, and for indusries o produce a range of commodiies. 12 GTAP daa base is relaively mos reliable and used exensively for global, regional and counry specific models. Page 28 of 66

29 Paddy Oupu Expor Inermediae goods Leonief Primary facors CES=0.1 CES=0.24 Com 1 o Com C Labor Land Capial All users CES=2 CES=0.4 Type 1 o Type O Domesic Impored Figure 7. Producion Technology The model is a muliperiod; of he recursive-dynamic ype. The only dynamic mechanism is ha which ses nex-period secor-specific indusry capial socks o las-period socks less depreciaion plus las-period invesmen. Invesmen for each indusry is posiively relaed o ha indusry's rae of reurn on capial. Invesors are myopic, bu invesmen in each secor changes so ha, afer some ime, capial earns an exogenous indusry-specific rae of reurn. An open capial marke is assumed, so ha invesmen is no limied by Nepali saving Model Daabase The iniial model daabase consiss mainly of a Social Accouning Marix (SAM) wih 139 rows and columns (Table 11). The daa sources and mehods used o consruc SAM are presened in he Appendix 4. As exhausively discussed here, apar from using supply and use able of CBS, oher numerous sources had o be used for fulfilling he daa gaps. For furher breaking down of foresry and oher few secors, new esimaes had o be made based on he daa available from he naional accouns. As well known, for SAM which is a building block of CGE huge daa sources ha oo Page 29 of 66

30 providing informaion on supply and use of inpus, facor disribuion and final demand vecors as per disinguished secors is needed. Such a daa se is hardly produced in Nepal despie regular annual and periodic surveys covering differen secors of he economy. From ha perspecive, daa gap in foresry secor is equally high and hence for furher exercise as per REDD+ requiremen by avoiding or minimizing bold assumpions, here is a need of fulfilling he daa gaps in a sandard inpu and oupu framework. A SAM shows all monies received or spen by each secor and final demander [a cell in row i, column j shows funds flowing from j o i]. A summary appears in Table 11. Apar from he SAM, he daabase conains: Various elasiciies (like hose shown in Figure 7). For each of Terai/Churia agriculure and foresry: land use in hecares, and share in value of naional oupu. Invesmen in, and capial sock of, each of 62 secors defined in he SAM. The proporion of consumpion of each good by each household ha is regarded as "subsisence" in he Linear Expendiure Sysem (LES). Table 11. Summary of he Social Accouning Marix (SAM), Nepal, 2007, aggregaed million rupees. Yellow-highlighed cells show flows of goods and primary facors (as seen in an inpu-oupu able), while oher cells show ransfer paymens or suboals. (ROW = Res of World) 62 Indusries 62 Commod. Indusries Commodiies Labour Capial Land Households ComTax Tariff Governmen ROW Invesmen Toal Labour Capial Households Land ComTax Tariff Governmen ROW Invesmen Toal Noe: A 2007 SAM of Nepal was prepared by Dr D.R. Khanal, as described in Appendix 4. His SAM has 139 rows and columns; in he summary above rows and column for he 62 secors and 7 household ypes have each been combined. The original 62 secors are: Paddy, Maize, Whea, OhGrain, VegeFrui, Oilseed, SugarCane, PlanFiber, OhCrops, Cale, OhAnmlPrd, RawMilk, Wool, Firewood, Timbers, GrassFoddr, OForesPrd, Fishing, Coal, Oil, GasMining, OhMining, Mea, MeaPrd, VegeblOil, DairyPrd, GrainMill, Sugar, OhFoodPrd, DrinkTobac, Texile, Clohing, LeaherPrd, Lumber, Paper, Peroleum, ChemRubber, MineralPrd, IronSeel, NonFerrMl, FabricMel, MoorVehcl, OhTrnsEqp, ElcrncEqp, OhMechEqp, Furniure, OhManuf, Elecricy, Gas, Waer, Consruc, Trade, OhTrnspr, WrTrnspr, AirTrnspr, Communicn, Finance, Insurance, OhBusSvc, RecOhSvc, GovSvc and Dwelling. The 7 household ypes have differen paerns of income and expendiure: he able below shows source shares in disposable income. Page 30 of 66

31 The main componens of Nepal s GDP and GNP are summarized in Tables 11 and 12. A feaure of Nepal's economy is he imporance of non-expor income from overseas. The sum of firs five componens of GNP labour, capial, land and commodiy axes) in Table 11 equaes o he oal GDP shown in Table 12. The las 2 componens of GNP in Table 12 represen income from overseas: 'remiances' are mainly he earnings of Nepali migran workers, and 'aid' shows ransfers o he governmen from he res of he world. Togeher hese accoun for 13% of oal GNP, relaively a high share. Table 13 shows he main componens of expendiure-side GDP. Impors are riple expors; he rade defici is funded by income from he res of he world. Also noable is he low share of governmen spending in GDP (jus 7%). Table 12. Componens of GNP; from Nepal SAM, 2007, million rupees GNP Labour Capial Land Commodiy GNP Tariffs Remiances Aid Taxes Toal Values % Shares Table 13. Componens of GDP; from Nepal SAM, 2007, million rupees GDP Households Invesmen Governmen Expors - Impors GDP Toal Values % Shares Income is derived from various sources, depending on wheher households are locaed in urban or rural areas (Table 14). In rural areas, households wih no or lile land derive mos of heir annual income from labor (i.e., from working). Households wih greaer landholdings derive more and more income from boh land and capial sources. In all cases, rural households derive a significan proporion of heir income (8% o 18%) from he res of world, eiher hrough remiances or oher inernaional ransfers. Less educaed urban households derive much income from labor. Wih higher income levels a comparaively higher level of income is derived from capial. Page 31 of 66

32 Table 14. Income sources by household ype, percenages of afer-ax income Labour Capial Land Governmen ROW LessTax Toal RurLndNone RurLndSmall RurLndMedium RurLnLarge UrbLowEduc UrbMedEduc UrbHighEduc RurLndNone = rural households wih no land; RurLndSmall= rural households wih small landholdings; RurLndMedium= rural households wih medium landholdings; RurLnLarge = rural households wih large landholdings; UrbLowEduc= urban households wih low educaion levels; UrbMedEduc=urban households wih medium educaion levels; UrbHighEduc=urban households wih high educaion levels; The Closure Mos CGE model conains more variables han equaions. Hence some variables have o be exogenous -- se by he model user. There is some flexibiliy in he choice of exogenous variables: he chosen se is called he closure. Here he same closure is used in boh base and policy simulaions. The following variables are exogenous: Tax raes, and raes of echnological changes Impor prices, and he posiions of expor demand curves ROW ransfers o Nepal households and governmen (including REDD paymens) Employmen; and land available for foresry and agriculure Targe raes of reurn o capial: invesmen in each indusry goes up or down according o wheher he curren rae of reurn is above or below he arge rae. The exchange rae (acing as numeraire). Exernal balance (= expors + ROW ransfers o Nepal Households and governmen - impors). Household spending follows household income, and governmen spending on goods follows GNP. Bu in each case he propensiy o consume is adjused by a single scale facor ha allows he exernal balance condiions o be me. The effec of his is ha REDD paymens allow boh governmen and household spending o increase (implicily he REDD paymen is shared beween households and governmen). Obviously he quesion of how he REDD paymens are disribued is imporan; we simulae a neural scheme. Page 32 of 66