the impacts of biofuels targets on land-use change and food supply: a global cge assessment

18
AGRICULTURAL ECONOMICS Agricultural Economics 43 (2012) 315–332 The impacts of biofuels targets on land-use change and food supply: A global CGE assessment Govinda R. Timilsina a,, John C. Beghin b , Dominique van der Mensbrugghe c , Simon Mevel d a Development Research Group, The World Bank, 1818 H Street, NW, Washington, DC 20433, USA b Iowa State University, Ames, IA 50011-1070, USA c Food and Agriculture Organization of the United Nations (FAO), Viale delle Terme di Caracalla, Rome I-00153, Italy d United Nations Economic Commission for Africa, Addis Ababa, Ethiopia Received 14 January 2011; received in revised form 14 July 2011; accepted 27 October 2011 Abstract We analyze the long-term impacts of large-scale expansion of biofuels on land-use change, food supply and prices, and the overall economy in various countries or regions using a multi-country, multi-sector global computable general equilibrium model augmented with an explicit land-use module and detailed biofuel sectors. We find that an expansion of biofuel production to meet the existing or even higher targets in various countries would slightly reduce GDP at the global level but with mixed effects across countries or regions. Significant land re-allocation would take place with notable decreases in forest and pasture lands in a few countries. The expansion of biofuels would cause a moderate decrease in world food supply and more significant decreases in developing countries like India and Sub-Saharan Africa. Feedstock commodities (sugar, corn and oil seeds) would experience significant increases in their prices in 2020, but other price changes are small. JEL classifications: D58, Q17, Q42, Q48, Q54 Keywords: Biofuels; Ethanol; Biodiesel; Land use; Food supply; Food prices; CGE model; Impacts 1. Introduction The last five years have been characterized by strong price movements in commodity markets. From 2006 to summer 2008, commodity prices increased to historically record-high levels in nominal terms. Since, this commodity bubble subsided but a new price upswing started in spring 2011. Agricultural com- modity prices have remained high, well above the historical trend. These recent sharp rises in agricultural and primary food prices have made real the issue of food-biofuels trade-offs. A key question to investigate is: Is the global biofuel expansion creating further food scarcity and hunger, especially among re- gions and population at risk? In addition, a second concern has Corresponding author. Tel.: 1 202 473 2767; fax: 1 202 522 1151. E-mail address: [email protected] (G. R. Timilsina). Data Appendix Available Online A data appendix to replicate main results is available in the online version of this article. Please note: Wiley-Blackwell, inc. is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article. emerged regarding the environmental impact of biofuels, and in particular, their local and global impact on land use and car- bon release. Many stakeholders now question the urgency of further developing biofuels because of this triple global inter- face between food, bioenergy, and the environment (Runge and Senauer, 2007; Searchinger et al., 2008). We investigate some of these concerns and provide some answers to these complex questions, with a focus on the impact of biofuel expansion on land allocation and the resulting impact on food supply and affordability. In the midst of the price surge of spring 2008, analysts have suggested that the production of biofuels was one of the key reasons for the spike in global food prices (Mitchell, 2008; Rosegrant, 2008, among others). However, there is no established consensus on the causes of the recent food cri- sis and its potential determinants, which include the surge in biofuels, cost-push factors such as the increase in energy and fertilizer prices, high food consumption growth in large emerg- ing economies, the role of market restrictions, investment funds’ hoarding, and the devaluation of the U.S. dollars on dollar- denominated commodity prices (Banse et al., 2008b; Timmer, 2008). c 2012 International Association of Agricultural Economists DOI: 10.1111/j.1574-0862.2012.00585.x

Upload: govinda-r-timilsina

Post on 27-Sep-2016

215 views

Category:

Documents


2 download

TRANSCRIPT

AGRICULTURALECONOMICS

Agricultural Economics 43 (2012) 315–332

The impacts of biofuels targets on land-use change and food supply:A global CGE assessment

Govinda R. Timilsinaa,∗, John C. Beghinb, Dominique van der Mensbrugghec, Simon Meveld

aDevelopment Research Group, The World Bank, 1818 H Street, NW, Washington, DC 20433, USAbIowa State University, Ames, IA 50011-1070, USA

cFood and Agriculture Organization of the United Nations (FAO), Viale delle Terme di Caracalla, Rome I-00153, ItalydUnited Nations Economic Commission for Africa, Addis Ababa, Ethiopia

Received 14 January 2011; received in revised form 14 July 2011; accepted 27 October 2011

Abstract

We analyze the long-term impacts of large-scale expansion of biofuels on land-use change, food supply and prices, and the overall economy invarious countries or regions using a multi-country, multi-sector global computable general equilibrium model augmented with an explicit land-usemodule and detailed biofuel sectors. We find that an expansion of biofuel production to meet the existing or even higher targets in various countrieswould slightly reduce GDP at the global level but with mixed effects across countries or regions. Significant land re-allocation would take placewith notable decreases in forest and pasture lands in a few countries. The expansion of biofuels would cause a moderate decrease in world foodsupply and more significant decreases in developing countries like India and Sub-Saharan Africa. Feedstock commodities (sugar, corn and oilseeds) would experience significant increases in their prices in 2020, but other price changes are small.

JEL classifications: D58, Q17, Q42, Q48, Q54

Keywords: Biofuels; Ethanol; Biodiesel; Land use; Food supply; Food prices; CGE model; Impacts

1. Introduction

The last five years have been characterized by strong pricemovements in commodity markets. From 2006 to summer 2008,commodity prices increased to historically record-high levelsin nominal terms. Since, this commodity bubble subsided buta new price upswing started in spring 2011. Agricultural com-modity prices have remained high, well above the historicaltrend. These recent sharp rises in agricultural and primary foodprices have made real the issue of food-biofuels trade-offs. Akey question to investigate is: Is the global biofuel expansioncreating further food scarcity and hunger, especially among re-gions and population at risk? In addition, a second concern has

∗Corresponding author. Tel.: 1 202 473 2767; fax: 1 202 522 1151.E-mail address: [email protected] (G. R. Timilsina).

Data Appendix Available Online

A data appendix to replicate main results is available in the online version ofthis article. Please note: Wiley-Blackwell, inc. is not responsible for the contentor functionality of any supporting information supplied by the authors. Anyqueries (other than missing material) should be directed to the correspondingauthor for the article.

emerged regarding the environmental impact of biofuels, andin particular, their local and global impact on land use and car-bon release. Many stakeholders now question the urgency offurther developing biofuels because of this triple global inter-face between food, bioenergy, and the environment (Runge andSenauer, 2007; Searchinger et al., 2008). We investigate someof these concerns and provide some answers to these complexquestions, with a focus on the impact of biofuel expansion onland allocation and the resulting impact on food supply andaffordability.

In the midst of the price surge of spring 2008, analystshave suggested that the production of biofuels was one ofthe key reasons for the spike in global food prices (Mitchell,2008; Rosegrant, 2008, among others). However, there is noestablished consensus on the causes of the recent food cri-sis and its potential determinants, which include the surge inbiofuels, cost-push factors such as the increase in energy andfertilizer prices, high food consumption growth in large emerg-ing economies, the role of market restrictions, investment funds’hoarding, and the devaluation of the U.S. dollars on dollar-denominated commodity prices (Banse et al., 2008b; Timmer,2008).

c© 2012 International Association of Agricultural Economists DOI: 10.1111/j.1574-0862.2012.00585.x

316 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

Similarly, there are many questions relating to the long-termimpacts of increased biofuels production: How does the expan-sion affect food availability and prices in the long run? Whowould gain and lose from potentially higher commodity andfood prices, where would agricultural production change? Willbiofuel production generate income and reduce poverty despitehigher food prices? What are the land supply responses? Ourarticle examines some of these long-term issues using a globaldynamic computable general equilibrium (GDCGE) model aug-mented with an explicit land allocation module and detailedbiofuel production sectors.

The GDCGE model accounts for the competition betweenbiofuel and food industries for agricultural commodities. Themajor biofuel feedstock is composed of corn, sugarcane, soy-bean oil, palm oil, and other vegetable oils and their backwardlinkages to oilseeds. The approach pays particular attention toproductivity gains through increases in yield and to changesin land allocation between various uses between forest (aggre-gate commercial and noncommercial forest) land and agricul-ture and within agricultural uses. Yield assumptions have beencontentious in the biofuel literature because of their implica-tions for land expansion (Keeney and Hertel, 2009; Searchingeret al., 2008). An expansion of biofuels would result in diversionof land used for other agricultural commodities toward pro-duction of biofuel feedstock. Grassland and forestland couldbe converted to agricultural land to produce biofuel feedstock.Yield responses to higher prices mitigate these land diversionsand reallocations although the exact magnitude of these re-sponses remains uncertain. This uncertainty is an importantcaveat qualifying our results.

Our analysis aims first, to estimate land-use mix at national,regional, and global levels in the baseline scenario. This isfollowed by the estimation of land-use mix under two mainscenarios of biofuels penetration in the global energy supplymix for road transportation. All types of land such as cropland,grassland, and forestland are included. The study assesses, atnational, regional, and global levels, the impacts of the increasedproduction of biofuels on land allocation, agricultural and foodproduction, and their prices.

Finally, this study also investigates the role of increaseddemand for biofuel feedstock as a promoter of monocultureof some feedstock. How does the biofuel expansion impactforest and other resources? Existing studies (e.g., Sielhorstet al., 2008) argue that the expansion of biofuels seriously im-pacts wetlands. In Indonesia, deforestation is linked with an in-creased production of palm oil, a feedstock for biodiesel. Thus,this study examines impacts of biofuels on various types ofland use and natural resources at national, regional, and globallevels.

We find that the impact of a considerable global expansionof biofuel production on agricultural prices is moderate, ex-cept for sugar crops, which increase by about 9.2%–11.6%.Food prices and supplies are moderately affected by these in-creases in biofuels unlike predicted by earlier estimates (DeSanti, 2008; Mitchell, 2008). However, the land-use impact is

more significant with large expansion of cropland offset by re-duction in forest and pasture. Within the land used for crops,a sizeable reshuffling of land takes place to accommodate theexpansion of feedstock crops for biofuels. Decreases in forestareas are notable in Brazil and Canada. These results do notexclude the occurrence of a perfect storm in the short term asthe one that occurred in 2008 with a spike in prices and adverseconsequences for food security and poverty. Our results shouldbe carefully considered as estimated long-term impacts and donot shed light on these potential short-term food price crises.

2. Literature background

Following the price hikes of spring 2008, Mitchell (2008) andTimmer (2008) provided ad hoc insightful background analysisof key factors affecting food prices including the role of theemergence of biofuels and the resulting increase in agriculturalfeedstock demand. These articles did not formally address theglobal multicrop land allocation problem. Recent investigationsbased on the FAPRI model (Elobeid and Tokgoz, 2008; Fabiosaet al., 2010; Searchinger et al., 2008; Tokgoz et al., 2007) es-timate the impact of various scenarios of biofuel expansion onprices and land use. In these partial equilibrium investigations,yields are assumed constant and most of the production changesoccur through land allocation changes, overstating the land andprice effects. Land allocation is country-specific and dependson local crop mix, suitability of land, and relative crop prices.There is no explicit modeling of land markets. Despite theserigidities, global land expansion and agricultural price effectsare surprisingly moderate, partly because supply adjustmentsare substantial via stock adjustment, flexibility in existing landuse, and world market response to higher prices.

Rosegrant et al. (2008) uses the IMPACT model to analyzeseveral scenarios departing from the recent (2007) biofuel de-mand. The IMPACT model incorporates land area and yieldresponses to prices. Rosegrant suggests that biofuel demandcontributed about 30% of the price increases during the periodcovering the early 2000s until 2007. The latter figure refers tothe effect on the weighted average grain prices, with the largesteffect on corn prices (39%) but lesser effects on rice and wheatprices (21% and 22%, respectively). Roberts and Schlenker(2010) use a calorie-equivalent approach to aggregate wheat,corn, rice, and soybean production. They estimate aggregatesupply and demand for calories and quantify the impact of abiofuel shock equivalent to the current U.S. biofuel mandate onthe price of calories from these four crops. They find that thecalorie price increases by about 20% when accounting for calo-ries saved through dried distillers grains with solubles (DDGS),and 30% when abstracting from these by-product caloriessavings.

General equilibrium analyses provide a more encompassingassessment of the biofuel emergence on the whole economy andfood prices because these multiple linkages are explicit in themodel. The disadvantages of these analyses are the aggregation

G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332 317

of crops in a few sectors, the lack of realistic biofuel policyparameterization, and trade elasticities, which tend to dampentrade responses to changing relative agricultural prices.

Taheripour et al. (2007) look at the impact of biofuel expan-sion on land through reallocation of land into corn, sugarcane,and oilseeds sectors. Biodiesel is treated as crushing oilseedsand producing meal as a byproduct (as DDGS are in the case ofethanol), a drastic assumption. Land supply follows the standardone-level CET assumption incorporated in the GTAP modelstructure. Land adjusts sluggishly according to relative rentsobtained in respective activities. All countries have a commoncost structure. The trade matrix is rather limited. It analyzesseveral policy shocks such as the increase in crude oil prices,the phasing out of MTBE, and biofuel subsidies. Keeney andHertel (2009) propose a model incorporating land, yield, andtrade responses to the biofuel expansion. Aggregate land sup-ply is fixed but land can move across uses according to relativereturns. The yield response to prices is a much-needed addition.Keeney and Hertel make the yield response explicit. It dependson the substitution between land, labor, capital, and other fac-tors. Keeney and Hertel assume that U.S. imports of ethanolfrom Brazil increase proportionally to the total U.S. ethanol de-mand, irrespective of price levels, a questionable assumption asexplained later in the modeling section and which has importantconsequences for land allocation.

Birur et al. (2008) use the GTAP database developed by Leeet al. (2005, 2008) to analyze the impact of biofuel productionon global agricultural markets. The latter authors decomposeland into 18 agro-ecological zones (AEZs). Birur et al. con-sider three biofuels, ethanol from grains, ethanol from sugarcrops, and biodiesel from vegetable oils. They treat the twoethanol productions as imperfect substitutes. The crops consid-ered are coarse grains, oilseeds, sugarcane, other grains, andother agriculture. Crops, grazing, and forestry use the total landendowment. Composite land supply is made of land in the 18AEZs, which are treated as highly substitutable (CES with anelasticity of substitution of 20). Within any AEZ, land shifts be-tween forest, pasture, and crops with some CET value of –0.11.Within crops, land shifts moving with a CET elasticity of –0.5.Landowners maximize returns on land by choosing an optimumallocation across uses according to relative returns. The nestedCET tree structure of land supply essentially parallels Keeneyand Hertel within each AEZ.

Banse et al. (2008a) use a modified GTAP-E model whichincorporates a refined three-level nested approach to land allo-cation first proposed by Huang et al. (2004). Banse et al. analyzethe impact of the EU biofuel directive on agricultural markets.Gohin (2008) and Bouet et al. (2008) analyze the impact of theEU Biofuel directive. Gohin’s analysis accounts for set-a-sideland coming back into agricultural production and for the im-pact on the livestock sector. Gohin’s land supply follows that ofKeeney and Hertel (2009). Bouet et al. use a land decompositionbased on AEZ data.

Our study differs from these existing CGE analyses on sev-eral fronts. First, our study simulates biofuels mandates and

targets for both developed and developing countries, whereasthe existing studies focus on mandates of developed countries’only. Second, our model presents more detailed representationof land-use change than the existing studies. Third, our studyrepresents all major types of biofuels differentiating betweenethanol and biodiesel and further classifying ethanol into threecategories by type of feedstock (i.e., corn ethanol, sugar ethanol,and other grains ethanol). Fourth, this article presents very de-tailed impacts on economic activity, trade, land use, and foodsupply for 25 countries/regions around the world.

3. Land module for biofuel modeling

Ethanol and biodiesel are the two major commercial biofuelconsidered in our model. The major feedstocks used for ethanolare grains, sugar beet, sugarcane, and molasses. For biodiesel,the major feedstocks are vegetable oil from rapeseed, soybean,and palm oil. The major feedstock for ethanol in warmer coun-tries is sugarcane, whereas grains are used in more temperateclimates (corn, but also wheat, and barley). Other feedstocktypes are not yet used in commercial operations and are notexpected to be for the foreseeable future. All countries grow-ing grains, oilseeds, or sugar crops are potentially affected bythe biofuel expansion through the transmission of higher worldprices for these commodities as well as the changes in theirrelative magnitude, even though they may not produce biofuel.

3.1. Modeling approach

The overall structure of the CGE model is briefly presented inAppendix A available in our longer working article (Timilsinaet al., 2010). We direct our discussions on the land-use com-ponent of the model as the article mainly deals with impactson land-use change and food supply. We refer the interestedreader to van der Mensbrugghe (2008) for the detailed descrip-tion of the model and to Appendix B of our working article fora description of the modifications associated with the biofuelmodule. The following sections describe the approach to landallocation adopted in the analysis.

Our analysis uses an acreage response model parallelingBirur et al. (2008), Keeney and Hertel (2009), and Lubowskiet al. (2006), but with significant departures regarding the CETassumptions impacting land allocation decisions. For the latterwe follow a nested approach to land allocation reminiscent ofHuang et al. (2004) and as applied by Banse et al. (2008a).

Huang et al. (2004) and Banse et al. (2008a) provide a morerealistic land allocation tree and nesting than Keeney and Herteldo. The nested structure allows for different CET elasticities forsubsets of agricultural activities and crops. For each country,a specific CET tree is designed to capture various and distinctsubsets of crops competing for the same agricultural land with ahigher elasticity of transformation within nests and lower onesbetween crop nests. As noted by Huang et al., land devotedto rice in most countries does not directly compete with other

318 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

Fig. 1. Land allocation.

crops. Similarly, we note that sugarcane production competeswith a limited number of crops.1 By contrast, oilseeds and grainproductions are in direct and strong competition with each otherin most countries where they are produced. Hence, consistentwith this fact, we lump them within the same nest and to allowfor greater land mobility between them.

As in Birur et al., we assume that on the top nest, forest(commercial and noncommercial), pasture, and crops competefor the same land. Then, we depart from the simple nesting ofcrops of Birur et al. We assume that crops are divided into landgoing to four crop categories (rice, sugar crops, an aggregategrains and oilseeds, and an aggregate other crops). These fourcrop categories form a second nest. Finally, grains and oilseedsconstitute a last nest with wheat, corn, other coarse grains,and oilseeds as competing for the use of the same land. Thisstructure accounts for the relative isolation of rice and sugarcaneareas, and the geographical proximity and competition betweencoarse grains and food grains, and oilseeds for the same land.Fig. 1 illustrates the nesting.

Next, we derive the price responses of land to unit revenuefor the nested land allocation explained previously. We firstintroduce a simple land allocation within a single nest, and thenextend the intuitive structure to the 3-nest actual structure. Thesimple structure mimics what happens within each nest holdingconstant the land allocated to the nest. The simple structurealso describes what happens in the top nest, for which the

1Sugar beet is included in the sugar nest, which probably results in under-stating the substitution between beet area and grains and oilseeds in temperatezones.

land constraint is fixed. For the two subsequent nests, the totalland allocated to the respective nest is price-dependent and willrespond to price changes. This additional price response is thenexplained for the two nests.

3.2. Derivation of the land supply specification

We start by assuming first a single nest with a total landconstraint A. Under such setup, equation (1) shows the firstoptimality condition for land allocation to activity i, Li, thatmaximizes land revenues under an aggregate land constraint A.The land constraint is shown in equation (2). The land constraintaccounts for land quality differences. The CET transformationelasticity, σ CET, is negative. The composite price PP reflectsthe composite revenue per unit of land in aggregate land A.

Li = θσi

CETA

(PP

Ri

)σ CET

, with

PP =[∑

i

θσi

CETR1−σ CET

i

] 11−σCET

(1)

and

A =[∑

i

θiLσCET −1σCET

i

] σCETσCET −1

. (2)

G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332 319

The responses of land activities with respect to changesin unit revenues, εij= dlnLl/dlnRj, are derived from the log-differentiation of equation (1) holding the land constraint Aconstant. The responses are shown in equation (3)

εij = −σ CET(δij − αj ), (3)

where δij indicates the kronecker delta, and αj indicates theshare of revenue from activity j in total land revenues.

Next, we derive the land supply price elasticities for the three-nest structure. The optimization problem assumes a sequence ofseparable decisions given prices for the respective choice withinthe nest and following a set of constraints for each respectivenest (A, An2, and An3). We start from the top nest allocatingland A between forest, pasture, and aggregate crops (An2) givenreturns Ri (i = f, p, n2). The second nest allocates land An2

between rice, sugar crops, an aggregate other crops, and grainsand oilseeds (An3) given prices Ri (i = r, s, oc, n3). Last in thethird nest, the aggregate land devoted to grains and oilseedsAn3 gets allocated between wheat, corn, other coarse grains,and oilseeds according to returns Ri (i = w, c, ocg, os). Foreach nest a structure similar to equations (1) and (2) can bederived by substituting the respective crop choice and returnsand aggregate land constraint for the corresponding nest.

In the top nest, land responses to any price j are given by

εij = −σ CET1(δij − αj ), i = f, p, n2, and

j = f, p, r, s, oc,w, c, ocg, os,(4)

where σ CET1 is the CET elasticity for the top nest and δij andαj are defined as previously. The land allocation responses in thelower nests have two components. A first component derivedas in equation (3), or as in the top nest, shows the price effectwith the land allocated to the nest assumed fixed, and a secondcomponent shows the price response on the land allocated tothe nest. It is best to separate the price response by nest. Wefirst derive the price response for land allocated in the secondnest to prices in the middle level, lower level, and then in thetop level, respectively. This is shown in equation (5)

εij = −σ CET1 (1 − αn2)

(αj

αn2

), + −σ CET2

(δij − αj

αn2

),

i = r, s, oc, n3 and j = r, s, oc, n3,

εij = −σ CET1(1 − αn2)

(αj

αn2

), + −σ CET2

(δij − αj

αn2

)i = r, s, oc, n3 and j = w, c, ocg, os

with δij = 1if i = n3 and j = w, c, ocg, os, else δij = 0, and

εij = −σ CET1(δij − αj ), i = r, s, oc, n3 and j = f, p,

(5)

where σ CET2is the CET elasticity for the second nest.Finally, the response for land use in the bottom nest is derived

in equation (6) following a similar decomposition of the priceresponse and noting that the land adjustment goes through thethree nests for returns of activities in the bottom nest, two nests

for prices of activities in the middle nest, and one nest for pricesof activities located in the top nest

εij = −σ CET1(1 − αn2)αj

αn2− σ CET2 αj

αn3

(1 − αn3

αn2

)

−σ CET3

(δij − αj

αn3

),

i = w, c, ocg, os, and j = w, c, ocg, os,

εij = −σ CET1(1 − αn2)αj

αn2− σ CET2

(δij − αj

αn2

),

i = w, c, ocg, os, andj = r, s, oc, n3,

and with δij = 1 if i = w, c, ocg, os and

j = n3, else δij = 0, and

εij = σ CET1αj , i = w, c, ocg, os, and j = f, p, n2,

(6)

where σ CET3is the CET elasticity for the third nest. We use thefollowing values –0.2, –0.5, and –1 for the three nested CETelasticities.

3.3. Yield elasticity specification

We follow the derivation of Mullen et al. (1988) as recentlyapplied by Keeney and Hertel to estimate yield responses toprices. The approach derives the comparative statics of changesin input prices on market equilibrium in a competitive agricul-tural output sector characterized by perfect competition. Vari-able Xi is input i (i =1, . . . , k) used in a given crop productionQ characterized by constant return to scale. In log-differentialform, the supply of the input is expressed as

d ln Xsi = μid ln Wi, (8)

Where μi denotes the price elasticity of supply and Wi denotethe input price. Under cost minimization in the crop sector, thederived demand for Xi is given by (9) in log-differential form

d ln XDi =

∑j

σij sj d ln Wj + d ln Q, (9)

where σ ij denotes the Allen–Uzawa elasticity of substitutionbetween inputs i and j in the production of the agriculturaloutput Q; cost shares are denoted by sj.

The zero profit equilibrium condition in the output marketequates the cost per unit and the price of the agricultural goodP. This condition is expressed in log-differential form in (10)

d ln P =∑

j

sj d ln Wj. (10)

Market equilibrium in input markets implies that XSi = XD

i .

Combining equations (8) and (9) and the input market equilib-rium conditions leads to (11)

d ln Q =∑

j

(δij (μj/sj ) − σij

)sjd ln Wj, i = 1, . . . k, (11)

320 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

with δij defined again as the kronecker delta. The system of kequations (11) can be solved for the share-weighted changes ininput prices (sj d ln Wj) as a function of output change d ln Q.The latter result can then be substituted back in equation (10) toeliminate variable W. This transformation yields equation (12),which expresses the total response of output Q to changes inoutput price P induced by changes in input markets

d ln P = ι′[N − �]−1ιd ln Q, or

d ln Q/d ln P = [ι′[N − �]−1ι

]−1,

(12)

where N indicates a diagonal matrix with diagonal elements(μj/sj), � is the matrix of Allen Uzawa elasticities σ ij, and ι in-dicates a unitary column vector. Assuming that the land supplyresponse is zero (μland = 0), then the response d ln Q/d ln Pprovides the yield response to the change in P. Different timehorizons can be considered for the yield response by varyingthe supply response of all inputs. Large supply responses μj

indicate a longer run. At the limit an infinite supply of theseinputs other than land indicate a long run and provide an upperbound on the yield response.

3.4. Trade assumptions impacting land allocation decisions

The Armington structure of differentiating products is pre-served but it assumes higher substitution elasticities than thosechosen in the Keeney and Hertel approach, because corn, sugar-cane, and soybean products are largely homogenous commodi-ties. Although the grain and oilseed aggregates in GTAP arecomposite commodities, the shocks emanating from the biofuelexpansion really involves a small subset of the commodities inthese aggregates. Trade response in the real world will be large.Corn from Argentina and Brazil will easily substitute for U.S.corn, and Brazilian soybean products will substitute perfectlyfor U.S. soybeans. Similarly, ethanol trade substitution shouldreflect nearly perfect substitution between Brazilian ethanol,and domestic ethanol in the demand structure of any importingcountry, including the United States.

A second pivotal trade policy assumption is to correctlymodel U.S. biofuel trade policy. In particular, the U.S. bio-fuel trade policy includes an ethanol ad valorem tariff of 2.5%and a specific tariff or 54 cents/gallon. These trade impedimentsare often missing in CGE analyzes. If U.S. demand expansionpushes the U.S. price to the Brazilian price inclusive of the tar-iffs, then Brazilian imports take place and the U.S. price movesin tandem with the Brazilian ethanol price and trade is positive.Historically, trade has happened at some point in most yearsavoiding the zero trade share issue with the Armington struc-ture. Note that small ethanol imports also occur under a tariffrate quota (TRQ) allowed by the Caribbean Initiative. The TRQmostly goes under filled. In sum, the U.S. tariff regime blockingBrazilian ethanol imports over a large price band is a key policyelement incorporated in our analysis.

4. Definitions of baseline and biofuel scenarios

Baseline refers to a scenario where business as usual willcontinue for economic development, population growth, andbiofuel development. Existing biofuel policies (e.g., alreadyimplemented mandates, subsidies, and import duties) are partof the baseline.2 Future mandates, which are incremental to ex-isting mandates, are part of targets and included in the scenarios.Like in most dynamic CGE models, three-variables: populationgrowth, savings and investment, and productivity are the keydrivers to generate a baseline. Population growth, depreciationrate of capital stock, and productivity growth are exogenous tothe model. Growth in aggregate labor supply tracks the workingage population (defined as those between 15 and 65 years). Sav-ings and investment determines the overall level of the capitalstock along with the rate of depreciation. Sectoral productivitygrowths are consistent with recent trends (World Bank, 2009).Another exogenous variable in the model baseline is the growthof energy prices, which are calibrated with projections madeby the Energy Information Administration (2006) of the U.S.Department of Energy.3

We define two scenarios. The first scenario considers theimplementation of announced targets (AT) of biofuel usage bycountries (hereafter AT scenario) with all targets implementedlinearly by 2020. Some countries meet their AT before 2020 dueto existing policies (e.g., mandates and subsidies) and marketconditions (e.g., U.S., Brazil, Malaysia, and South Africa). Forthese countries we assumed that the AT scenario follows thebaseline starting from the year they first become binding. Forexample, Brazil meets the AT before 2009 due to the existingmandate and market conditions, therefore, the study does notimpose any additional policy instrument until 2020 in Brazil. Inthe United States, the baseline would be equal to AT scenario by2015 and we assume this is maintained until 2020. Moreover,we abstract from second generation or cellulosic biofuels dueto limitations in data and their unknown cost.

The second scenario generally considers a doubling of theAT keeping the timing of the implementation of the targets un-changed. In India no increase is added to the AT target level,because the latter is already extremely high (20% in 2017).We refer to this scenario as the enhanced scenario (hereafterthe “ET” scenario). Table 1 presents biofuel targets in terms ofenergy contents in both scenarios. Each country has a differ-ent initial share and therefore a specific path to the target; forexample, Brazil already meets its target by 2010. Once percent-age targets are reached, these targets are constant as shares butthey still evolve in terms of physical volume as transportationenergy consumption changes over time. The study assumes gov-ernment revenue neutrality, in other words, governments need

2Impacts of removal of subsidies and import duties have been analyzed indetailed in a separate paper.

3A module that can represent both conventional and unconventional oil andgas reserves and production would be ideal; here, however, the model used heredoes not have that capacity. Hence, we used energy price forecasts from othersources instead of generating them endogenously in the baseline.

G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332 321

Table 1Scenarios: Share of biofuels in the total liquid fuel demand for transportation

Country/regions Target (%)∗ Subsidy rated (%) Tax rate (%)

AT ET AT ET AT ET

Australia and New Zealand 1.23 2.46 36.71 57.92 0.16 0.34Japan 0.60 1.20 14.85 45.76 0.05 0.19Canada 4.10 8.20 50.07 68.86 0.56 1.08United States∗∗ 4.07 8.14 0.96 29.78 0.04 0.84France 10.00 20.00 58.36 74.57 1.11 2.65Germany 10.00 20.00 43.35 65.08 1.03 2.78Italy 10.00 20.00 65.32 78.40 1.21 2.45Spain 10.00 20.00 61.01 75.13 1.12 2.21United Kingdom 10.00 20.00 74.00 83.08 1.00 1.75Rest of EU and EFTA 10.00 20.00 75.62 84.83 1.09 1.99China 3.65 7.30 22.57 50.01 0.45 1.40Indonesia 5.00 10.00 22.10 50.91 3.39 3.47Malaysia 1.81 3.62 1.79 39.41 0.02 0.60Thailand 5.20 10.40 51.92 73.95 0.92 1.92Rest of East Asia and Pacific 1.49 2.98 42.45 58.10 0.22 0.39India 16.70 16.70 53.26 53.26 3.35 3.35Rest of South Asia − − − − − −Argentina 5.00 10.00 52.27 70.43 0.87 1.61Brazil∗∗∗ 9.50 19.00 − 2.48 − 0.83Rest of LAC 1.48 2.96 16.74 47.97 0.10 0.41Russia − − − − − −Rest of ECA − − − − − −MENA − − − − − −South Africa 2.00 4.00 0.88 10.73 0.03 0.68Rest of Sub-Saharan Africa − − − − − −EFTA = European Free Trade Association; LAC = Latin America and Caribbean; EAC = Eastern Europe and Central Asia; MENA = Middle East and North Africa.∗Refers to ratio of consumption of ethanol and biodiesel to consumption of ethanol, biodiesel, gasoline, and diesel in road transportation (the consumption areexpressed in energy unit).∗∗Mandates for cellulosic ethanol are not included.∗∗∗Because the biofuels penetration in the baseline is same as that in AT scenario, no additional subsidy would be required to meet this scenario.∗∗∗∗Subsidies equivalent to the targets and corresponding taxes on gasoline and diesel required to finance the subsidies in year 2020.

to find additional revenue to finance the subsidies required torealize the targets. Taxation of gasoline and diesel would be thebest source as it not only provides needed revenue but also helpslower the level of subsidies on biofuels. The revenue neutral taxrates turn out to be small due to the large tax base.

5. Simulation results

We present key impacts on production and prices of biofuelsand agricultural commodities, land-use mix, and food supply.Whenever possible we relate our results to the existing litera-ture.

Table 1 presents the subsidy level or shadow price requiredif the mandates or targets are to be met through an economicinstrument for the two scenarios.4 The subsidy rate varies acrosscountries depending upon the biofuels penetration gaps betweenthe scenarios and the baseline. Most European countries exhibita higher subsidy level because their 10% target by 2020 is muchhigher than their biofuel penetration levels in the baseline. On

4There could be other economic instruments. However, we consider subsidy,as it would be the most direct economic instrument to meet the targets.

the other hand, the level of subsidy in the United States is smallas its biofuels target is small and would nearly be met by 2020due to the existing subsidy.

We present results in deviation from the baseline trajectoryboth in percentage deviation from the model baseline, and inbillion of dollars in 2004 prices when relevant.5 We focus thediscussion on the ET scenario which represents a larger shockon the world economy than the AT scenario does. The ET sce-nario leads to a 13.1% share of biofuels in the total consumptionof liquid fuels in the transport sector. We begin by explainingwhat each shock represents in biofuel expansion in deviationfrom the baseline followed by discussions on land use and foodsupply impacts as well as food price consequences. We presentkey results in this section; detailed results from the main andsensitivity analysis are available from the authors upon request.

5.1. Impacts on biofuels production and trade

The AT scenario induces significant changes in biofuelproduction although some of these are moderate in several

5The change in billion dollars is relevant in cases where the percentage changenumbers, appear high due to small base.

322 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

Table 2Increase in biofuel production in 2020 in AT and ET scenarios relative to thebaseline (dollar figures in 2004 prices)

Country/regions AT ET

US$ billion % US$ billion %

World total 38.8 64.5 92.1 153.2High-income 25.4 89.6 67.7 238.8Australia and New Zealand 0.0 20.9 0.2 145.2Japan 0.0 3.8 0.1 28.3Canada 0.2 45.0 0.8 167.3United States 0.2 1.1 0.8 4.2France 7.7 268.0 19.8 691.2Germany 3.3 112.0 11.1 371.2Italy 2.7 321.0 7.0 842.2Spain 2.8 367.5 7.3 959.3United Kingdom 2.7 500.1 6.1 1,135.9Rest of EU and EFTA 5.7 538.8 14.5 1,355.9Middle and low-income 13.4 42.1 24.4 76.8China 2.2 55.4 8.7 217.9Indonesia 0.3 49.2 1.4 210.1Malaysia 0.0 0.5 0.0 1.3Thailand 0.4 88.1 1.3 294.7Rest of East Asia and Pacific 0.1 53.7 0.4 155.1India 3.9 205.8 3.9 207.4Rest of South Asia 0.0 58.1 0.2 213.5Argentina 0.1 45.9 0.6 189.9Brazil 6.3 31.1 7.9 39.2Rest of LAC 0.0 4.2 0.2 51.4Russia 0.0 −1.3 −0.1 −3.0Rest of ECA 0.0 −0.9 0.0 −2.5MENA 0.0 −1.2 0.0 −2.8South Africa 0.0 0.5 0.0 0.7Rest of Sub-Saharan Africa 0.0 0.1 0.0 −1.8

countries with sizeable biofuel programs. The latter countrieshave been implementing their biofuel targets for several years.Global biofuel production increases by 64.5% in 2020 com-pared to what it would have been in the baseline with the largestchanges in the EU, China, Thailand, and India. By contrast, theET scenario has much larger effects in most countries, than theAT scenario does, except India. This is by design because thepenetration targets are doubled or more. The targets in the ETscenarios translate into a vast increase in world biofuel produc-tion and use (+153.2% or $92.1 billion at 2004 prices) with thelargest increases in the EU and European Free Trade Associa-tion (EFTA) countries ($65.8 billion), Brazil ($7.9 billion), andChina ($8.7 billion) as shown in Table 2.

Countries like Brazil, China, France, and India would realizerelatively higher production of biofuels. Whereas the produc-tion increase in Brazil is mainly driven by international trade,the increases in other countries are driven by domestic targets.Some countries, which do not have biofuel targets, could ex-perience decrease in production (of biofuels) due to increasedexport demand for their biofuel feedstock, but the reductions arenegligible. Table 3 presents scenario impacts on internationaltrade of biofuels. The expansion of biofuels would induce anexpansion of biofuel trade of 259% under the AT scenario.The expansion of biofuel trade volume under the ET scenario

Table 3Change in biofuel trade (%) relative to the baseline in 2020—-by region

Country/regions Imports Exports

AT ET AT ET

World total 258.7 520.7 258.7 520.7High-income 310.9 794.2 370.6 934.7Australia and New Zealand 0.0 0.0 0.0 0.0Japan −1.6 19.1 0.0 0.0Canada 65.5 249.1 0.3 0.5United States 0.6 2.3 38.3 163.4France 153.8 564.8 486.1 1,204.7Germany 78.9 303.2 873.5 2,220.7Italy 319.7 820.0 0.0 0.0Spain 362.1 909.6 375.9 1,017.4United Kingdom 1,042.4 2,556.6 472.7 1,096.3Rest of EU and EFTA 637.2 1,527.9 82.7 332.3Middle and low-income 203.6 232.6 181.1 234.1China 0.0 0.0 25.9 79.5Indonesia 0.0 0.0 1.3 3.6Malaysia 0.0 0.0 1.7 4.9Thailand 0.0 0.0 −39.0 −77.4Rest of East Asia and Pacific 54.5 170.1 0.0 0.0India 420.3 425.8 0.0 0.0Rest of South Asia 0.0 0.0 306.1 1,134.2Argentina 0.0 0.0 1.5 26.5Brazil 0.0 0.0 198.5 250.6Rest of LAC 0.0 46.2 0.0 0.0Russia 0.0 0.0 0.0 0.0Rest of ECA −5.8 −8.1 0.0 0.0MENA −2.2 −4.9 0.0 0.0South Africa 0.0 0.0 0.0 0.0Rest of Sub-Saharan Africa −3.6 −6.0 0.0 0.0

Note: The large percentage changes in many countries are due to the small basein the baseline.

is twice that of the AT scenario. Countries that would expe-rience the highest level of imports of biofuels include Indiaand some EU countries. Countries that would experience thehighest level of exports of biofuel include Brazil and some EUcountries, namely France and Germany. These countries canexpand their biofuel feedstock production.

5.2. Impacts on agricultural production and trade

The effects on agricultural output are shown in Table 4. Everycountry or region considered experience an agricultural expan-sion. Even the countries, which experience decrease in biofuelproduction, would see an increase in agricultural production dueto increased export demand for biofuel feedstock. EU countriesexhibit the highest percentage increase in agricultural outputs.

Agricultural output reaches a global expansion of 1.1%($38.3 billions) in 2020 in the ET scenario. All crops, exceptrice,6 increase in the ET scenario, and notably so for sugarcrops (13.5%), other coarse grains (11.2%), oilseed (7.6%),corn (4.1%), and wheat (3.7%). Rice production falls notably

6Rice production decreases as land is reallocated toward oilseeds, sugar crops,and grains as discussed in detail later.

G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332 323

Tabl

e4

Perc

entc

hang

ein

agri

cultu

ralo

utpu

t(va

lue

atco

nsta

ntpr

ices

)re

lativ

eto

the

base

line

in20

20

Cou

ntry

/reg

ions

Agr

icul

ture

Padd

yri

ceSu

gar

crop

sO

ther

crop

sW

heat

Cor

nO

ther

grai

nsO

ilsee

dsL

ives

tock

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

Wor

ldto

tal

0.4

1.1

−0.2

−0.3

8.1

13.5

−0.1

−0.1

1.3

3.7

1.0

4.1

4.5

11.2

2.4

7.6

0.0

−0.1

Hig

h-in

com

e0.

82.

00.

10.

210

.124

.00.

0–0

.12.

66.

90.

72.

511

.127

.36.

318

.30.

00.

1A

ustr

alia

and

New

Zea

land

0.3

0.8

0.1

0.3

0.4

1.8

0.3

0.6

1.0

3.5

1.2

4.8

0.9

2.7

1.9

5.3

0.2

0.5

Japa

n0.

30.

90.

10.

20.

53.

20.

41.

10.

93.

10.

42.

21.

13.

20.

92.

70.

30.

8C

anad

a0.

30.

80.

00.

00.

20.

60.

0−0

.40.

72.

34.

015

.11.

02.

71.

13.

00.

00.

1U

nite

dSt

ates

0.2

0.7

0.1

0.3

0.1

0.1

0.1

0.1

1.0

3.4

0.7

2.7

0.7

1.8

0.9

2.6

0.1

0.3

Fran

ce2.

66.

3−2

.0−6

.068

.815

7.5

−1.4

−4.3

5.2

12.7

0.2

0.5

8.2

20.0

28.3

83.0

−0.5

−1.6

Ger

man

y1.

33.

60.

00.

00.

72.

90.

0−0

.57.

118

.11.

86.

87.

719

.717

.556

.90.

10.

2It

aly

1.1

2.9

0.6

1.4

0.4

1.0

0.4

1.1

2.2

6.4

0.8

2.4

2.1

5.9

14.1

36.4

0.4

1.0

Spai

n1.

53.

9−0

.1−0

.30.

51.

80.

20.

52.

67.

61.

34.

836

.894

.83.

28.

70.

20.

4U

nite

dK

ingd

om1.

12.

30.

00.

019

.949

.6−0

.7−1

.8−1

.4−2

.40.

00.

062

.113

4.9

11.2

30.4

−0.5

−1.3

Res

tof

EU

and

EFT

A0.

71.

70.

41.

13.

07.

70.

20.

42.

15.

7−0

.5−0

.89.

323

.57.

921

.50.

00.

1M

iddl

ean

dlo

w-i

ncom

e0.

20.

5−0

.2−0

.47.

410

.0−0

.1−0

.20.

41.

41.

35.

40.

30.

90.

31.

8−0

.1−0

.3C

hina

0.2

0.5

−0.1

−0.2

0.2

0.3

−0.1

− 0.4

1.1

3.2

4.7

17.9

−0.2

−1.9

−0.2

−1.0

−0.1

−0.3

Indo

nesi

a0.

10.

1−0

.1−0

.68.

129

.7−0

.2−0

.80.

00.

00.

0−0

.20.

00.

00.

10.

5−0

.1−0

.5M

alay

sia

0.1

0.2

0.2

0.4

0.2

0.4

0.1

0.2

0.0

0.0

0.0

0.0

1.4

3.4

0.4

1.2

0.0

−0.1

Tha

iland

0.4

0.7

−0.9

−3.3

28.0

92.3

−0.7

−3.0

0.0

0.0

0.4

1.7

−0.3

−2.1

−0.5

−2.8

−0.5

−2.2

Res

tof

Eas

tAsi

aan

dPa

cific

0.1

0.2

0.0

−0.1

3.2

8.3

0.1

0.2

0.4

1.5

0.9

3.7

0.6

1.6

0.3

1.0

0.0

−0.1

Indi

a0.

20.

3−0

.7−0

.613

.313

.5−1

.0−0

.9−0

.4−0

.3−0

.4−0

.3−0

.4−0

.4−0

.6−0

.5−0

.6−0

.5R

esto

fSo

uth

Asi

a0.

10.

20.

0−0

.10.

93.

20.

00.

10.

30.

80.

20.

80.

30.

51.

02.

90.

0−0

.1A

rgen

tina

0.4

0.9

0.6

0.7

−0.1

−0.7

0.1

0.0

0.5

1.1

1.5

5.6

0.3

0.7

0.5

1.2

−0.2

−0.7

Bra

zil

0.3

1.1

−1.2

−2.0

17.9

22.3

−1.4

−2.1

−1.2

−0.2

−1.4

−1.3

−0.9

−0.6

1.0

5.4

−1.0

−1.6

Res

tof

LA

C0.

10.

40.

0−0

.20.

21.

10.

20.

30.

51.

60.

10.

60.

10.

21.

75.

20.

0−0

.2R

ussi

a0.

20.

50.

10.

30.

30.

70.

41.

10.

31.

00.

41.

10.

61.

41.

74.

60.

0−0

.1R

esto

fE

CA

0.3

0.8

0.1

0.4

0.5

1.3

0.3

0.7

0.7

2.0

0.7

2.2

0.8

2.2

1.9

4.6

0.1

0.2

ME

NA

0.3

0.8

−0.1

−0.2

0.6

1.5

0.3

0.6

1.1

3.3

1.4

4.9

0.9

2.5

1.4

3.6

−0.1

−0.3

Sout

hA

fric

a0.

41.

10.

00.

00.

61.

40.

51.

41.

85.

30.

41.

31.

43.

40.

72.

10.

10.

3R

esto

fSu

b-Sa

hara

nA

fric

a0.

10.

20.

00.

10.

30.

60.

20.

41.

03.

20.

00.

20.

0−0

.10.

20.

7−0

.2−0

.4

Not

e:T

hebo

ldva

lues

repr

esen

tagg

rega

tion

ofth

ere

mai

ning

.

324 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

in Brazil and Thailand because competing crops such as sugarare more profitable. The strong expansion of biofuel in theET scenario explains this agricultural expansion of feedstockcrops. The sugar expansion takes place principally in the EU,Thailand, Brazil, and India. The global increase in grains takesplace in the EU with wheat in France, and other coarse grains inSpain; corn in the United States expands as well. Limited expan-sion takes place in grains in Middle and low-income countriesexcept for corn production in China. This limited expansion ingrains occurs because other crops such as sugarcane are moreprofitable in many of these countries. The Armington struc-ture is also partially responsible for these localized expansions.Oilseed expansion is strong in EU countries (France, Germany,and Italy) and Brazil. Oilseed production expands because veg-etable oil is the feedstock for biodiesel. Malaysia would haveexpansion of palm oil production as reflected in the increasein oilseed production and food processing related to oilseeds.Finally, livestock production contracts in most countries as thecost of feed goes up. Protein meal is slightly cheaper than in thebaseline but it does not offset the more expensive feed grains.

Our results on agricultural impact are close to those of Bouetet al. (2008). They obtain large increases in EU sugar cropsand relatively moderate increases in grains. They also showa vast expansion of oilseeds as we do, but with a strongerexpansion in Brazil, reflecting the high global integration ofoilseeds markets across countries and oil types. Gohin (2008)also obtains sizeable expansion of oilseed production in the EUinduced by EU biodiesel targets and an expansion of livestockproduction from the cheaper protein in feed. Our results onagriculture output are also close to those of Kretschmer et al.(2009) and Al-Riffai et al. (2010) and consistent with those ofFischer et al. (2009).

Changes in agricultural trade are notable and illustrate theimportant role of global trade to mitigate domestic shocks inbiofuel and feedstock markets. These are shown in Tables 5 and6. All countries with biofuel targets would experience increasesin total trade of agricultural commodities, mostly commoditiesused for biofuel feedstock and which experience the largestprice changes. The increase in agriculture imports would berelatively higher in EU countries and India. The percentage in-crease in sugar and oilseeds imports would be relatively higheras compared to imports of other agriculture commodities. Theexpansion of biofuels would cause substantial decrease in ex-ports of agriculture commodities in countries such as India,Thailand, and the United Kingdom. On the other hand, exportof agriculture commodities, mainly biofuel feedstock, wouldincrease in countries like Brazil, Argentina, Russia, and manyregions, such as Sub-Saharan Africa, Middle East and NorthAfrica (MENA), and South Asia (except India). These signifi-cant two-way trade effects can be attributed to various factors.

The heterogeneity of the basket of commodities, such asoilseeds and grains of different quality, distance, and locationof production and net surpluses and deficits areas prevailingin many countries contribute to the trade patterns. Within alarge country, surplus and deficits areas can export and import

more cheaply than they can access the domestic market. Inthe European region, trade in oilseed is quite large with EFTAcountries and Germany importing from other EU countries likeFrance. Brazil and the United States extend their exports ofoilseeds as they are the world’s largest soybean producers andexporters.

5.3. Impacts on prices of agriculture and other commodities

Changes in world prices are detailed in Table 7. The changesvary from –1.7% (crude oil) to 9.7% (sugarcane) under the ETscenario. These large increases for sugar prices are followed byhigher prices for ethanol (4.1%), corn (3.6%), biodiesel (3.5%),oilseeds (2.9%), wheat (2.3%), and other coarse grains (2.2%).Processed food prices change moderately (0.5%). The prices ofgasoline and diesel products fall by 1.4%. These lower energyprices mitigate the higher cost of farm ingredients in processedfood. We also compute price multipliers (percentage change inprice/percentage change in world biofuel production), to gaugethe price impact of the two scenarios. The sugar price multi-pliers are the largest at 22.3% (AT scenario) and 12.3% (ETscenario). The value of 22.3% could be interpreted as the per-centage increase in price if world biofuel doubled. The nextmultiplier is for the ethanol price with a multiplier of 6.9%for the AT scenario. The multipliers for the aggregate agricul-tural price index are 5.5% and 4%, respectively for the AT andET scenarios. The price multipliers for other commodities aremoderate. For example, the processed food price multiplier is0.5%, again clearly showing the moderate impact biofuels haveon processed food prices. These small changes do not precludehigher localized price changes.

5.4. Impacts on land-use mix

Changes in aggregate land allocations between forest, pas-ture, and crops under both scenarios are shown in Table 8 andFig. 2. Results for the AT and ET scenarios show similar patternsbut muted by the smaller biofuel targets. In most countries (theEU, Thailand, South Africa, India, and Brazil) land moves awayfrom pasture and forest and into crops. The key driver of the landallocation away from forest and pasture into crops is the higherrelative price of agricultural output relative to the price of activ-ities using pasture and forest (see Table 7). Accordingly, landis reallocated to crops, especially feedstock to biofuels produc-tion, which experience the highest price increases. The shadowprice of forestland in most countries falls relative to the shadowprice of land in other activities with associated contraction oftrade and output in forest products. Large biofuel targets am-plify this mechanism. Russia, in contrast, has no biofuel targetand is a natural exporter of forest products. Its forest productionand trade expand along with a modest relative increase in landallocated to forest (0.014%) offsetting the contraction in forestproduction in other countries. As a large agricultural producer,

G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332 325

Tabl

e5

Perc

entc

hang

ein

agri

cultu

rali

mpo

rts

(val

ueat

cons

tant

pric

es))

rela

tive

toth

eba

selin

ein

2020

–by

regi

on

Cou

ntry

/reg

ions

Agr

icul

ture

Padd

yri

ceSu

gar

Oth

ercr

ops

Whe

atC

orn

Oth

ergr

ains

Oils

eeds

Liv

esto

ck

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

Wor

ldto

tal

1.0

3.0

0.3

0.7

1.6

4.2

0.7

1.8

0.9

3.0

1.8

6.9

4.3

12.1

4.2

13.2

0.4

1.1

Hig

h-in

com

e2.

37.

00.

41.

214

.062

.30.

82.

13.

07.

71.

64.

914

.037

.713

.343

.50.

51.

3A

ustr

alia

and

New

Zea

land

0.8

2.3

0.8

1.9

−0.6

1.6

0.8

2.2

2.8

16.3

1.1

3.8

−0.3

1.0

1.5

4.4

0.5

1.2

Japa

n0.

51.

50.

20.

6−0

.91.

4−0

.2−0

.30.

51.

40.

82.

50.

41.

10.

82.

4−0

.1−0

.2C

anad

a1.

34.

10.

51.

40.

51.

80.

82.

09.

336

.39.

134

.81.

76.

31.

44.

10.

61.

8U

nite

dSt

ates

0.6

1.6

0.3

1.3

−3.1

−5.2

0.7

1.8

0.6

2.1

0.2

0.8

0.0

−0.2

1.0

3.0

0.1

0.3

Fran

ce4.

513

.71.

13.

121

1.2

983.

42.

16.

234

.297

.11.

44.

3−1

.30.

184

.627

4.4

1.2

3.6

Ger

man

y5.

019

.10.

51.

50.

02.

80.

72.

114

.641

.81.

44.

87.

923

.232

.312

6.4

0.4

1.3

Ital

y1.

74.

40.

10.

3−2

.0−3

.90.

61.

60.

20.

10.

30.

00.

61.

218

.450

.60.

81.

9Sp

ain

2.8

7.5

0.9

2.4

−0.2

0.1

0.5

1.2

−0.7

−2.1

0.6

1.4

43.5

120.

61.

22.

50.

20.

4U

nite

dK

ingd

om2.

87.

50.

41.

119

.164

.51.

23.

04.

510

.71.

23.

513

3.6

377.

820

.552

.81.

23.

0R

esto

fE

Uan

dE

FTA

2.7

6.9

0.3

0.8

4.5

14.0

0.6

1.6

4.5

10.9

2.3

5.9

15.5

39.5

13.2

34.6

0.4

1.1

Mid

dle

and

low

−inc

ome

0.7

1.8

0.3

0.6

1.4

3.3

0.7

1.7

0.5

2.0

1.8

7.9

−0.3

−0.2

1.2

3.2

0.4

1.1

Chi

na0.

62.

10.

21.

1−0

.10.

20.

51.

71.

77.

014

.265

.41.

76.

31.

13.

40.

41.

1In

done

sia

1.1

3.7

1.5

4.0

67.5

511.

00.

93.

00.

82.

00.

61.

40.

72.

12.

07.

40.

82.

5M

alay

sia

0.9

2.3

1.7

2.7

1.1

1.5

0.6

1.2

1.1

3.1

1.0

3.1

0.9

2.6

1.4

3.5

0.3

0.4

Tha

iland

1.0

3.3

1.9

4.7

159.

31,

324.

41.

13.

40.

2−0

.12.

59.

5−0

.5−1

.01.

02.

40.

61.

4R

esto

fE

astA

sia

and

Paci

fic0.

71.

80.

40.

615

.445

.80.

71.

60.

82.

30.

92.

90.

20.

50.

92.

60.

30.

7In

dia

3.9

4.6

6.8

7.8

173.

617

1.1

4.3

5.0

6.2

2.3

6.4

2.8

5.1

0.7

6.1

4.8

2.1

3.2

Res

tof

Sout

hA

sia

0.4

0.7

0.3

−0.5

4.4

16.7

0.5

0.9

0.0

−0.6

0.4

0.8

0.0

−0.4

0.6

1.1

0.2

−0.1

Arg

entin

a0.

72.

3−1

.8−3

.50.

30.

70.

41.

60.

00.

89.

640

.90.

11.

50.

82.

10.

50.

6B

razi

l1.

22.

71.

93.

50.

00.

01.

83.

40.

31.

61.

93.

51.

32.

13.

810

.01.

21.

9R

esto

fL

AC

0.6

1.6

0.5

1.2

0.5

3.4

0.6

1.7

0.4

1.0

0.3

0.4

0.3

0.8

1.3

3.8

0.3

0.8

Rus

sia

−0.6

−1.4

−0.5

−1.5

−2.1

−5.7

−0.1

−0.3

−1.4

−3.4

−0.9

−2.6

−3.6

−9.4

−1.5

−4.4

−1.1

−2.6

Res

tof

EC

A0.

0−0

.20.

10.

1−1

.9−5

.20.

40.

8−1

.5−4

.0−0

.5−1

.6−3

.8−8

.70.

61.

00.

20.

4M

EN

A−0

.3−0

.90.

20.

3−1

.7−2

.90.

10.

3−1

.3−3

.80.

10.

6−0

.9−2

.40.

0−0

.1− 0

.2−0

.5So

uth

Afr

ica

0.0

−0.1

0.7

1.7

−0.7

−2.6

0.3

0.9

−0.5

−1.5

−0.8

−3.6

−0.9

−3.2

0.4

1.2

0.2

0.5

Res

tof

Sub−

Saha

ran

Afr

ica

−0.1

−0.4

−0.2

−0.7

−0.5

−1.6

−0.2

−0.6

0.6

1.7

−0.7

−2.2

−2.8

−6.3

−1.2

−3.4

−0.5

−1.1

Not

e:T

hebo

ldva

lues

repr

esen

tagg

rega

tion

ofth

ere

mai

ning

.

326 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

Tabl

e6

Perc

entc

hang

ein

agri

cultu

rale

xpor

ts(v

alue

atco

nsta

ntpr

ice)

rela

tive

toth

eba

selin

ein

2020

—by

regi

on

Cou

ntry

/reg

ions

Agr

icul

ture

Padd

yri

ceSu

gar

Oth

ercr

ops

Whe

atC

orn

Oth

ergr

ains

Oils

eeds

Liv

esto

ck

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

Wor

ldto

tal

1.0

3.0

0.3

0.7

1.6

4.2

0.7

1.8

0.9

3.0

1.8

6.9

4.3

12.1

4.2

13.2

0.4

1.1

Hig

h-in

com

e0.

92.

50.

71.

7−2

.6−5

.60.

61.

40.

72.

22.

07.

24.

211

.74.

312

.80.

41.

0A

ustr

alia

and

New

Zea

land

1.0

2.6

0.6

1.4

2.9

4.5

1.0

2.5

1.6

4.5

7.1

30.4

2.0

6.1

3.2

8.9

0.6

1.6

Japa

n2.

05.

52.

57.

14.

79.

32.

46.

43.

410

.81.

04.

12.

95.

85.

616

.31.

44.

0C

anad

a1.

02.

60.

00.

02.

04.

30.

81.

11.

44.

34.

017

.52.

15.

92.

57.

10.

30.

8U

nite

dSt

ates

1.1

3.2

0.6

1.7

6.9

20.8

0.7

1.8

2.0

6.7

2.1

7.6

1.4

3.8

2.5

7.5

0.8

2.2

Fran

ce0.

50.

8−1

.6−4

.9−4

4.1

−67.

4−1

.9−6

.7−1

.3−4

.51.

44.

213

.935

.513

.040

.1−0

.9−3

.1G

erm

any

1.1

1.9

0.0

0.0

6.1

13.3

0.5

0.4

0.3

−1.1

4.3

16.4

10.9

27.2

4.1

1.8

0.5

1.0

Ital

y1.

02.

61.

12.

74.

612

.61.

22.

94.

112

.32.

710

.05.

614

.79.

728

.50.

20.

3Sp

ain

1.2

3.2

0.3

0.8

39.7

184.

90.

92.

46.

018

.75.

823

.97.

218

.213

.843

.50.

81.

7U

nite

dK

ingd

om−1

.1−2

.00.

00.

0−2

8.8

−58.

6−0

.9−2

.6−1

.3−0

.30.

00.

0−2

5.1

−44.

011

.641

.5−1

.0−2

.6R

esto

fE

Uan

dE

FTA

0.6

1.8

1.0

2.7

−1.3

−3.7

0.8

2.1

−1.4

−2.4

−1.5

−1.1

1.3

4.9

6.2

21.0

0.2

0.7

Mid

dle

and

low

-inc

ome

1.4

4.2

−0.2

−1.1

4.4

10.7

0.9

2.5

2.4

8.4

1.2

6.1

5.5

15.5

4.1

13.7

0.7

2.0

Chi

na1.

01.

00.

60.

58.

217

.91.

31.

80.

6−1

.3−1

2.0

−40.

50.

0−4

.83.

15.

70.

60.

6In

done

sia

0.3

−0.9

0.3

−0.7

−52.

0−9

2.9

0.4

−0.9

0.0

0.0

9.2

41.6

0.0

0.0

2.7

0.8

−0.3

−1.5

Mal

aysi

a1.

03.

0−0

.6−1

.11.

25.

31.

23.

30.

00.

00.

00.

02.

15.

43.

08.

20.

62.

0T

haila

nd−0

.1−1

.4−1

.9−6

.4−6

9.8

−96.

7−0

.1−1

.60.

00.

03.

011

.8−0

.1−1

.60.

5−0

.1−0

.2−1

.5R

esto

fE

astA

sia

and

Paci

fic1.

43.

81.

13.

3−1

6.4

−37.

71.

33.

12.

06.

69.

944

.14.

212

.23.

19.

10.

72.

3In

dia

−6.0

−1.7

−5.7

−4.7

−79.

7−7

8.4

−6.3

−4.6

−5.9

3.6

−6.9

11.1

−8.8

−0.8

−2.4

3.4

−2.8

−2.6

Res

tof

Sout

hA

sia

2.7

5.1

1.5

5.0

11.9

−0.9

2.7

5.0

3.8

10.6

6.3

28.0

8.9

14.0

3.6

7.0

1.4

3.6

Arg

entin

a1.

54.

04.

28.

00.

82.

91.

02.

51.

43.

9−0

.1−1

.73.

910

.02.

57.

30.

10.

9B

razi

l0.

54.

8−2

.0−1

.6−5

0.0

−57.

6−1

.9−2

.01.

16.

5−1

.11.

52.

813

.24.

115

.1−2

.0−1

.8R

esto

fL

AC

1.4

3.8

0.8

1.9

2.8

3.7

1.1

2.7

3.4

10.5

5.3

21.7

5.6

16.4

5.9

18.4

0.6

1.8

Rus

sia

3.7

10.1

2.3

6.0

0.0

0.0

2.9

7.6

7.5

22.3

8.5

31.3

8.4

23.7

13.0

35.5

2.9

7.6

Res

tof

EC

A1.

74.

51.

23.

35.

414

.51.

33.

62.

78.

32.

810

.55.

616

.07.

417

.30.

82.

4M

EN

A2.

25.

41.

23.

34.

712

.41.

94.

35.

216

.77.

530

.87.

320

.49.

024

.51.

74.

0So

uth

Afr

ica

1.4

3.9

0.0

0.0

4.6

13.2

1.4

3.9

3.1

9.7

0.9

3.4

3.1

6.7

4.2

13.1

0.7

2.0

Res

tof

Sub−

Saha

ran

Afr

ica

2.2

5.4

1.7

4.9

6.1

16.8

2.1

4.8

3.5

10.9

5.8

25.1

7.3

20.2

4.9

13.5

1.7

4.5

Not

e:T

hebo

ldva

lues

repr

esen

tagg

rega

tion

ofth

ere

mai

ning

.

G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332 327

Table 7Change in world prices and multipliers

AT % ET % AT ETprice price multiplier Multiplierchange change

Sugar 9.2 11.6 22.3% 12.3%Ethanol 2.8 4.6 6.9% 4.9%Biofuels 2.7 4.5 6.5% 4.8%Agriculture 2.2 3.8 5.5% 4.0%Corn 1.1 3.7 2.7% 3.9%Biodiesel 1.1 3.3 2.7% 3.5%Oilseeds 1.5 3.1 3.7% 3.3%Wheat 1.1 2.4 2.8% 2.6%Other cereal grains 1.0 2.3 2.5% 2.5%Paddy rice 0.8 1.6 2.0% 1.7%Vegetables, fruits 0.7 1.5 1.6% 1.6%Livestock 0.4 1.1 1.0% 1.1%Forestry 0.3 0.9 0.8% 0.9%All goods and services 0.5 0.8 1.3% 0.8%Processed food 0.2 0.5 0.5% 0.5%Other services 0.0 0.0 0.0% 0.0%Other manufacturing 0.0 0.0 0.0% 0.0%Construction 0.0 0.0 0.0% 0.0%Other mining 0.0 −0.1 0.0% −0.1%Coal 0.0 −0.1 −0.1% −0.1%Gas distribution −0.1 −0.1 −0.2% −0.1%Chemicals −0.1 −0.1 −0.1% −0.1%Electricity −0.1 −0.2 −0.2% −0.2%Total industry and services −0.2 −0.4 −0.4% −0.4%Natural gas −0.2 −0.4 −0.5% −0.4%Transport services −0.2 −0.6 −0.6% −0.6%Other industrial sectors and

services−0.3 −0.6 −0.7% −0.7%

Diesel −0.6 −1.4 −1.5% −1.4%Gasoline −0.6 −1.4 −1.5% −1.5%Refined oil −0.6 −1.4 −1.5% −1.5%Crude oil −0.8 −1.7 −1.9% −1.8%

Notes: World prices are trade-weighted indices of bilateral export prices. Mul-tipier is 100 × price change/world biofuel output change.

Russia also experiences land moving away from pasture intocrops.

Our land variable does not distinguish commercial and non-commercial forest. Hence, we potentially have a positive bias inour forest effects as noncommercial forest is harder to convertthan commercial forestland (more difficult access and clearingcost).

Aggregate effects in the three major land uses are small (inpercentage change from the baseline), less than 1% in absolutevalue in most countries, except in Brazil, Thailand, and the EU(Germany, France, Spain, and the United Kingdom).

For the latter, effects are relatively larger, between 1.1% and5.1% in absolute value for changes in forest and pasture. Franceexhibits 5.1% and 4.1% decreases in forest and pasture lands.7

Within the crop allocations, large changes take place withinmany countries, with massive land reallocation away from rice

7Some countries have legal constraints on land conversion, which we do notmodel. Rather, we consider a low elasticity of transformation (–0.2) for landallocated between forest, crops, and pasture. This is a limitation of our analysis.

and other crops to sugar crops, coarse grains, and corn. Theimplementation of the ET to 2020 increases land use in cropsglobally. In the longer run, technical progress would take placeand productivity gains would reduce the ramping up of land use.Under the ET scenario, worldwide land devoted to sugar andother coarse grains increases by 9.4% and 6.9%, respectively.The impacts on land-use in Al-Riffai et al. (2010) and Fischeret al. (2009) are close to ours with significant expansion of landfor sugar crops and rapeseed, and declines in land for rice andvegetable and fruit. Fischer et al. (2009) find that expansionof biofuel targets could increase total arable land by 1%–3%,depending upon the scenarios, in 2020.

Fig. 2 shows the change in the distribution of forest land bycountry for both scenarios. The ET scenario leads to nearly 18.4million hectares of forest loss as compared to the baseline in2020 with the largest losses taking place in Brazil and Canada.In Brazil, the loss of forest relative to the baseline amountedto 6.1 million hectares in 2020, which is the largest changegenerated in physical terms; it represents 1.2% of Brazil’s totalforest area in 2020.8

5.5. Impacts on food supply

Food supply includes direct consumption of crops, fruits andvegetables, and livestock, and processed food. The composi-tion of the food supply changes however, because the shares oflivestock products, sugar, and some grains decrease the most.This global effect does not exclude stronger localized effectswhen feedstock and food use directly compete locally and withcostly transportation. Nevertheless regional trade within a coun-try should help dampen these potential effects. If biofuels areproduced locally, trade costs must have been reduced for allcommodities and arbitrage in food markets would take place iflocal commodity prices rise significantly.

In 2020, under the ET scenario, world food supply decreasesby $14.1 billion or 0.2% from the corresponding food supply inthe baseline. Under the AT scenario, the reduction in food sup-ply is about half as large. The negligible percentage reductionis caused by a large base as we include the entire food sector,including cereals, processed food, and livestock. Consideringonly cereals, Fischer et al. (2009) show that expansion of bio-fuels to meet the existing targets would cause a reduction in arange of 8–29 million tons of food supply in 2020 relative toits reference case, that is, a reduction of food supply between0.29% and 1.05%. The most important reason for the differentresults is that Fischer et al. fixed the penetration biofuels inyear 2020 at the level of 2008 in the reference case, whereasour model allows the penetration of biofuels to increase due toexisting tax incentives.

Global effects seem small especially in percentage terms;importantly low- and middle-income countries are more neg-atively affected than high-income countries. About two-third

8Fischer et al. (2009) find that expansion of biofuels could cause up to 20million hectares of deforestation by 2020.

328 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

Tabl

e8

Cha

nges

inla

ndsu

pply

(%)

rela

tive

toth

eba

selin

ein

2020

Cou

ntry

/reg

ions

Tota

lcro

pFo

rest

Past

ure

Ric

eSu

gar

crop

sO

ther

crop

sW

heat

Cor

nO

ther

grai

nsO

ilsee

ds

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

AT

ET

Wor

ldto

tal

0.1

0.2

−0.2

−0.6

−0.2

−0.5

−0.3

−0.6

6.4

9.4

−0.5

−1.2

0.0

0.4

0.3

1.7

3.1

6.9

0.6

2.2

Hig

h-in

com

e0.

10.

4−0

.4−1

.1−0

.3−0

.8−0

.2−0

.66.

312

.8−0

.7−2

.10.

41.

0−0

.2−0

.17.

817

.03.

08.

0A

ustr

alia

and

New

Zea

land

0.1

0.2

−0.1

−0.2

0.0

−0.1

−0.1

−0.4

0.0

0.6

0.0

−0.2

0.5

1.7

0.6

2.7

0.4

1.1

1.2

3.2

Japa

n0.

00.

0−0

.1−0

.30.

0−0

.1−0

.2−0

.60.

11.

30.

00.

00.

41.

30.

00.

70.

51.

50.

31.

1C

anad

a0.

10.

4−0

.2−0

.5−0

.2−0

.60.

00.

0−0

.2−0

.6−0

.3−1

.30.

00.

32.

710

.20.

20.

50.

40.

9U

nite

dSt

ates

0.0

0.1

−0.1

−0.3

−0.1

−0.2

−0.1

−0.4

−0.1

−0.4

−0.1

− 0.4

0.5

1.7

0.3

1.4

0.3

0.5

0.5

1.3

Fran

ce0.

30.

7−1

.8−5

.1−1

.4−4

.1−3

.2−8

.937

.073

.0−2

.8−7

.90.

80.

7−2

.8−7

.53.

15.

817

.346

.0G

erm

any

0.3

0.8

−0.7

−2.2

−0.5

−1.5

0.0

0.0

−0.4

−0.8

−0.9

−3.0

3.3

7.3

−0.8

−1.0

3.9

8.7

11.0

33.6

Ital

y0.

20.

4−0

.7−1

.9−0

.5−1

.3−0

.7−2

.0−0

.8−2

.0−0

.8−2

.1−0

.4−0

.6−1

.4−3

.3−0

.4−0

.88.

620

.7Sp

ain

0.1

0.4

−0.7

−1.8

−0.5

−1.3

−1.0

−2.7

−0.6

−1.3

−0.8

−2.2

−0.1

0.2

−1.1

−1.9

25.4

58.9

0.5

1.3

Uni

ted

Kin

gdom

0.4

1.0

−1.2

−3.1

−1.2

−3.0

0.0

0.0

10.8

24.6

−2.1

−5.3

−4.3

−9.3

0.0

0.0

40.0

76.8

4.9

13.1

Res

tof

EU

and

EFT

A0.

20.

5−0

.3−0

.8−0

.3−0

.8−0

.3−0

.81.

43.

5−0

.5−1

.30.

71.

8−1

.4−3

.16.

214

.95.

013

.1M

iddl

ean

dlo

w-i

ncom

e0.

10.

2−0

.1−0

.2−0

.2−0

.4−0

.3−0

.66.

58.

3−0

.3−0

.6−0

.3−0

.10.

73.

0−0

.1−0

.1−0

.3−0

.1C

hina

0.1

0.3

−0.1

−0.2

−0.1

−0.3

−0.1

−0.5

0.0

−0.1

−0.2

−0.8

0.4

0.7

3.3

12.1

−0.7

−3.6

−0.6

−2.6

Indo

nesi

a0.

00.

1−0

.1− 0

.4−0

.1−0

.4−0

.1−0

.76.

322

.8−0

.2−0

.90.

00.

0−0

.1−0

.40.

00.

00.

00.

1M

alay

sia

0.1

0.2

0.0

0.0

0.0

−0.1

0.1

0.2

0.1

0.2

0.0

0.0

0.0

0.0

0.0

0.0

1.1

2.8

0.2

0.8

Tha

iland

0.0

0.1

−0.3

−1.1

−0.3

−1.2

−0.7

−2.6

15.2

44.8

−0.6

−2.5

0.0

0.0

0.2

0.8

−0.3

−2.1

−0.6

−2.8

Res

tof

Eas

tAsi

aan

dPa

cific

0.0

0.0

0.0

−0.1

−0.1

−0.1

−0.1

−0.2

2.2

5.9

0.0

0.0

0.2

0.9

0.7

2.7

0.3

0.9

0.2

0.5

Indi

a0.

20.

2−0

.5−0

.6−0

.5−0

.5−0

.9−0

.910

.010

.0−1

.1−1

.1−0

.7−0

.8−0

.7−0

.6−0

.6−0

.7−0

.8−0

.8R

esto

fSo

uth

Asi

a0.

00.

10.

00.

00.

0−0

.10.

0−0

.20.

62.

30.

0−0

.10.

10.

40.

10.

40.

10.

10.

82.

2A

rgen

tina

0.1

0.2

−0.2

−0.5

−0.2

−0.6

0.0

−0.6

−0.4

−1.2

−0.2

−0.6

0.0

−0.1

0.8

3.4

−0.2

−0.5

0.0

−0.1

Bra

zil

0.2

0.3

−0.7

−1.2

−0.7

−1.4

−1.1

−2.2

10.7

12.8

−1.3

−2.3

−1.3

−1.6

−1.5

−2.4

−1.1

−1.9

0.4

2.7

Res

tof

LA

C0.

00.

1−0

.1−0

.2−0

.1−0

.3−0

.1−0

.40.

00.

50.

0−0

.10.

20.

80.

00.

0−0

.1−0

.31.

23.

6R

ussi

a0.

10.

30.

00.

0−0

.1−0

.3−0

.1−0

.30.

0−0

.10.

10.

20.

00.

00.

10.

20.

20.

51.

13.

0R

esto

fE

CA

0.1

0.2

−0.1

−0.3

−0.1

−0.2

−0.1

−0.3

0.1

0.2

0.0

−0.1

0.2

0.6

0.2

0.9

0.3

0.8

1.1

2.7

ME

NA

0.1

0.1

−0.2

−0.4

−0.1

−0.4

−0.2

−0.6

0.2

0.4

0.0

−0.2

0.5

1.5

0.7

2.6

0.3

0.8

0.7

1.7

Sout

hA

fric

a0.

10.

3−0

.1−0

.2−0

.1−0

.20.

00.

00.

20.

30.

10.

21.

02.

90.

00.

10.

71.

70.

20.

7R

esto

fSu

b-Sa

hara

nA

fric

a0.

00.

10.

00.

0−0

.1−0

.20.

00.

00.

10.

20.

10.

10.

72.

3−0

.10.

0−0

.1−0

.20.

10.

4

Not

e:T

hebo

ldva

lues

repr

esen

tagg

rega

tion

ofth

ere

mai

ning

.

G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332 329

Fig. 2. Change in forest land relative to baseline, million ha.

of the world food supply decrease is imputable to developingcountries. Lower income economies generally rely largely onagriculture and therefore are significantly dependent on foodproducts. China, Rest of Sub-Saharan Africa, MENA, and In-dia would suffer the most if biofuel targets are actually im-plemented. Food availability would be reduced by $1.4 billionin India alone under AT scenario. The United States wouldcarry most of the burden among developed economies (seeTable 9).

5.6. Aggregate economic impacts

Beyond, effects on land, food production, and prices, it isalso interesting to gauge the overall effect of a large biofuel ex-pansion on the whole economy of various countries/regions.Fig. 3 presents the impacts of biofuel expansion on coun-tries’ GDP for both scenarios. The expansion of biofuels wouldslightly decrease global GDP (0.02% and 0.06% reductionsunder AT and ET scenarios, in year 2020 as compared to thebaseline). At country/regional level, the results are mixed.

Brazil, Argentina, Thailand, and Indonesia experience in-creases in GDP, whereas countries like the United States, China,and India experience reductions in their GDP. Countries, which

have small penetration in the baseline but have set ambitioustargets, would face GDP decreases because the targets signifi-cant alter the allocation of resources in production and changetrade patterns, resulting in a worsening of the terms of trade.India is a good example, where the government has set a targetof 20% (or 17% in terms of energy content) for biofuels by2020, but its current level of biofuels use is small. Achievingthis target would cause India’s imports of agricultural com-modities, to increase, in aggregate, by 3.9%–4.6% dependingon the scenarios (see Table 5). Its export would drop by 1.7%–6%. The terms of trade deteriorates thereby causing significantGDP loss (0.13%–0.23%).9 Countries like Brazil, Argentina,and Indonesia would experience increases in their GDP due toincreased exports of biofuels or feedstock required for biofu-els. These countries experience improvements of their terms oftrade. Energy exporting countries/regions, such as MENA andRussia, will face reductions in GDP from reduced oil exportsand negative terms of trade effects.

9India has different impacts for different scenarios despite similar targets butwith different terms of trade effects.

330 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

Fig. 3. Change in real GDP at market price relative to baseline, percentage.

5. Conclusions

In this study, we analyzed the impacts of a large-scale globalexpansion of biofuels on agriculture, land use, and food supply.The first core finding is the moderate long-term impact on com-modity prices induced by a large expansion of biofuels underboth scenarios. This finding is conditioned on yield gains takingplace as a response to increases in commodity and land prices.In addition, and for both scenarios we also compute food priceeffects as found in recent CGE and PE analyses which fullymodeled biofuel and feedstock markets. PE analyses do not ac-count for all sources of productivity gains and generate higherland and food price effects. Our results suggest that plannedbiofuel targets would not cause large impacts on food supplyat the global level although the impacts would be significant indeveloping countries like India and Sub-Saharan Africa. Agri-cultural commodities such as sugar, corn, and oil seeds (asoil), that serve as the main biofuel feedstock, would experience1%–9 % price increases in 2020 as compared to that in thebaseline due to the expansion of biofuels to meet the existingtargets.

We show that land-use allocation between forest, pasture, andcrops would be significantly altered, and leads to considerableforest and pasture destruction in several countries, especially

under enhanced targets. Vast expansion of biofuel does lead toglobal forest losses of about 18.4 million hectares in 2020 underenhanced targets. Technical progress eventually would reducethe reliance on land to expand agricultural output and forest-land could be regained in the longer run. Within the expandedcropland uses, we obtain large effects in several countries im-plementing large biofuel targets. The general tendency is to ex-pand land devoted to feedstock crops (sugar crops, grains; andoilseeds for vegetable oil for biodiesel). These effects are largein 2020, because they correspond with the expansion phase ofthe targets. In the longer run, the land expansion would recedeand productivity gains would reduce land use for feedstockand the tradeoff between food and biofuel. We have conductedextensive sensitivity analysis. Various configurations of key pa-rameters have been tried (elasticities of transformation in land,elasticities of substitution in production and fuel use, productiv-ity assumptions, and trade elasticities). The qualitative resultsfor the two scenarios are invariant to major changes in theseparameters. The robustness of the results has to do first with therelative small share of farm values in food processing, hence themoderate food price effects. Second, the increase in demand forfeedstock can be accommodated given the various mechanismsat play via trade in agricultural commodities, land allocation,and biofuels production, even under limited land substitution

G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332 331

Table 9Change in food supply relative to the baseline in 2020—by region

Country/regions AT ET

US$ billion % US$ billion %

World total −6.5 −0.1 −14.1 −0.2High-income −2.2 −0.1 −5.8 −0.2Australia and New Zealand 0.0 0.0 −0.1 −0.1Japan −0.1 0.0 −0.2 −0.1Canada 0.0 −0.1 −0.1 −0.2United States −0.9 −0.1 −2.5 −0.3France −0.1 0.0 −0.3 −0.1Germany −0.2 −0.1 −0.6 −0.2Italy −0.1 −0.1 −0.3 −0.2Spain −0.1 0.0 −0.2 −0.1United Kingdom −0.2 −0.1 −0.4 −0.1Rest of EU and EFTA −0.5 −0.1 −1.1 −0.2Middle and low-income −4.3 −0.2 −8.3 −0.3China −0.5 −0.1 −1.3 −0.2Indonesia −0.1 −0.1 −0.1 −0.1Malaysia 0.0 −0.1 0.0 −0.3Thailand 0.0 0.1 0.0 0.1Rest of East Asia and Pacific 0.0 0.0 −0.2 −0.1India −1.4 −0.4 −1.2 −0.3Rest of South Asia 0.0 0.0 −0.1 −0.1Argentina 0.0 0.0 0.0 −0.1Brazil 0.0 0.0 0.0 0.0Rest of LAC −0.2 −0.1 −0.6 −0.3Russia −0.3 −0.2 −0.7 −0.6Rest of ECA 0.0 0.0 0.0 0.0MENA −1.3 −0.4 −3.0 −1.0South Africa 0.0 0.0 0.0 0.0Rest of Sub-Saharan Africa −0.5 −0.2 −1.2 −0.5

across uses. Small relative land variations can accommodate ad-ditional feedstock demand. Putting land into production takesresources and time, hence these results assume that adjustmentscan take place over time. The long-tern nature of our analysiscannot be over stated in the present context of high commod-ity prices. In addition, we have considered dry distiller grains(DDGS) as a feed crop substitute. DDGS are a byproduct ofethanol and provide revenues and reduce feed grain demand.Accordingly, the implied indirect land use and price effectslinked to feed grains are dampened once DDGS are accountedfor.

Considering the size and complexity of the model, we couldnot exactly incorporate all policy parameters which might affectthe quantitative results, particularly in the ET scenario. More-over, we have not accounted for existing distortions in sugarand energy markets. TRQs in sugar markets tend to increaseprice effects by dampening supply response. Sugar price ef-fects could be understated. The impact of distortions in energymarkets is less clear a priori.

Acknowledgment

The views and findings presented here should not be at-tributed to our respective institutions. We thank Masami Ko-

jima, Peter Berck, Mark Rosegrant, Stephen Mink, and areferee for their valuable comments and David Zilberman,Jean Christophe Bureau, Harry de Gorter, Amani Elobeid, JayFabiosa, Alexandre Gohin, Will Martin and John Nash for dis-cussions. We acknowledge the Knowledge for Change Program(KCP) Trust Fund for the financial support. Beghin also ac-knowledges support from the Marlin Cole Professorship.

References

Al-Riffai, P., Dimaranan, B., Laborde, D., 2010. “Global Trade and Environ-mental Impact Study of the EU Biofuels Mandate”, Report for the EuropeanCommission, DG TRADE, ATLASS Consortium.

Banse, M., van Meijl, H., Tabeau, A., Woltjer, G., 2008a. Will EU biofuelpolicies affect global agricultural markets? Eur. Rev. Agric. Econ. 35(2),117–141.

Banse, M., Nowicki, P., van Meijl, H., 2008b. Why are current food prices sohigh? In: Zuurbier, P., van de Vooren, J. (Eds.), Sugarcane Ethanol. Con-tributions to Climate Change Mitigation and the Environment. WageningenAcademic Publishers, The Netherlands, Chapter 10, pp. 227–247.

Birur, D., Hertel, T.W., Tyner, W., 2008. Impact of biofuel production on worldagricultural markets: A computable general equilibrium analysis, GTAPWorking Paper No 53, Center for Global Trade Analysis, Purdue University,Indiana, USA.

Bouet, A., Curran, L., Dimaranan, B., Ramos, M.-P., Valin, H., 2008. Biofuels:Global Trade and Environmental Impact Study. CEPII draft report, Paris,17th December.

De Santi, G., 2008. Biofuels in the European Context: Facts and uncertainties.Joint Research Centre (JRC), Petten, the Netherlands.

Elobeid, A., Tokgoz, S., 2008. Removal of U.S. ethanol domestic and tradedistortions: Impact on U.S. and Brazilian ethanol markets. Am. J. Agric.Econ. 90(4), 918–932.

Energy Information Administration (EIA), 2006. Annual Energy Outlook2006 with Projections to 2030. Report#: DOE/EIA-0383. Available at:http://www.eia.doe.gov/oiaf/aeo/, accessed July 2008.

Fabiosa, J., Beghin, J., Dong, J.F., Elobeid, A., Tokgoz, S., Yu, T.-H., 2010.Land allocation effects of the global ethanol surge. Predictions from theinternational FAPRI model. Land Econ. 86(4), 687–706.

Fischer, G., Hizsnyik, E., Prieler, S., Shah, M., van Velthuizen, H., 2009. “Bio-fuels and Food Security,” The OPEC Fund for International Developmentand International Institute for Applied Systems Analysis.

Gohin, A., 2008. Impacts of the European Biofuel Policy on the farm sector: Ageneral equilibrium assessment. Rev. Agric. Econ. 30(4), 623–641.

Huang, H., van Tongeren, F., Dewbre, J., van Meijl, H., 2004. “A New Repre-sentation of Agricultural Production Technology in GTAP” Paper Presentedat the 7th Annual Conference on Global Economic Analysis, WashingtonDC, USA.

Keeney, R., Hertel, T.W., 2009. The indirect land use impacts of U.S. biofuelpolicies: The importance of acreage, yield, and bilateral trade responses.Am. J. Agric. Econ. 91(4), 895–909.

Kretschmer, B., Narita, D., Peterson, S., 2009. The economic effects of the EUbiofuel target. Energy Econ. 31, S285–S294.

Lee, H.-L., Hertel, T.W., Sohngen, B., Ramankutty, N., and U.S. EnvironmentalProtection Agency. 2005. GTAP greenhouse gases emissions data base. Cen-ter for Global Trade Analysis, Purdue University, West Lafayette, IN47907,USA.

Lee, H.-L., Hertel, T.W., Rose, S., Avetisyan, M., 2008. An integrated globalland use data base for CGE analysis of climate policy options. Center forGlobal Trade Analysis, Purdue University GTAP, Working Paper No. 42,Purdue University.

Lubowski, R., Plantinga, A., Stavins, R., 2006. Land-use change and carbonsinks: Econometric estimation of the carbon sequestration supply function.J. Environ. Econ. Manage. 51, 135–152.

332 G. R. Timilsina et al. / Agricultural Economics 43 (2012) 315–332

Mitchell, D., 2008. A note on rising food prices. Policy Research WorkingPaper WPS 4682, World Bank, Washington, DC.

Mullen, J., Wohlgenant, M.K., Farris, D.E., 1988. Input substitution and thedistribution of surplus gains from lower US beef processing costs. Am. J.Agric. Econ. 70 (1988), 245–254.

Roberts, M.J., Schlenker, W., 2010. “The U.S. Biofuel Mandate and World FoodPrices: An Econometric Analysis of the Demand and Supply of Calories,”Presented at the NBER Agricultural Economics Conference March 4-5,Cambridge, MA.

Rosegrant, M.W., 2008. Biofuels and grain prices: Impacts and policy re-sponses. Testimony for the U.S. Senate Committee on Homeland Securityand Governmental Affairs May 7, 2008.

Rosegrant, M.W., Zhu, T., Msangi, S., Sulser, T., 2008. Global scenarios forbiofuels: Impacts and implications. Rev. Agric. Econ. 30(3), 495–505.

Runge, C.F., Senauer, B., 2007. How Biofuels Could Starve the Poor. ForeignAffairs 86(3). Available at: http://www.foreignaffairs.com/articles/62609/c-ford-runge-and-benjamin-senauer/how-biofuels-could-starve-the-poor, ac-cessed July 1, 2009.

Searchinger, T.D., Heimlich, R., Houghton, R.A., Dong, F., Elobeid, A.,Fabiosa, J., Tokgoz, S., Hayes, D., Yu, T., 2008. Use of croplands for bio-fuels increases greenhouse gases through emissions from land use change.Science 319(5867), 1238–1240.

Sielhorst, S., Molenaar, J.W., Offermans, D., 2008. Biofuels in Africa. Anassessment of risks and benefits for African wetlands, Commissioned paperby Wetland International. AIDEnvironment.

Taheripour, F., Birur, D.K., Hertel, T.W., Tyner, W.E., 2007. Introducing LiquidBiofuels into the GTAP Data Base, GTAP Research Memorandum N0 11,2007, revised May 1, 2008.

Timilsina, G.R., Beghin, J.C., van der Mensbrugghe, D., Mevel, S., 2010. Theimpacts of biofuel targets on land-use change and food supply: A globalCGE assessment, Policy Research Working Paper 5513. The World Bank,Washington, DC.

Timmer, C.P., 2008. Causes of High Food Prices. ADB Economics WorkingPaper No. 128, October 2008.

Tokgoz, S., Elobeid, A., Fabiosa, J., Hayes, D., Babcock, B., Yu, T., Dong, F.,Hart, C., Beghin, J., 2007. Emerging biofuels: Outlook of effects on U.S.grain, oilseed, and livestock markets. Center for Agricultural and Rural De-velopment, Iowa State University, Staff Report 07-SR 101, July (Revised).

van der Mensbrugghe, D., 2008. The environmental impact and sustainabilityapplied general equilibrium (ENVISAGE) model. The World Bank, January2008. Available at: http://go.worldbank.org/9ALYVKJHD0, accessed July2009.

World Bank, 2009. Global Economic Prospects 2009: Commodities at theCrossroads. The World Bank, Washington, DC.