Biofuels Role in Mexico s Rural Development

Biofuels Role in Mexico s Rural Development George A. Dyer* Te James Hutton Institute Craigiebuckler, Aberdeen AB15 8QH United Kingdom george.dyer@utton.ac.uk * corresponding autor: georgie.dyer@gmail.com; Tel: +44(0)1224 395383; Fax: +44(0)844 928 5429; Craigiebuckler, Aberdeen, AB15 8QH UK 1

Abstract. Biofuels are expected to promote growt, create jobs, and reduce poverty and inequality in developing areas wile contributing to environmental sustainability. Acieving tese goals will require accounting for teir potential synergies and tradeoffs wile addressing causal interactions across scales. An agent-based, general-equilibrium model of rural Mexico is used to analyze te callenges of tis country s recent biofuels initiative. Results reveal tat modest increases in crop prices could generate substantial feedstock surpluses for biofuels based on eiter corn or alternative crops, representing US$236 - $273 million in annual sales. Significant profits in feedstock production would raise land rents at te forest margin. A corn-based industry would increase forests opportunity costs up to 2.0%, particularly in Souteast Mexico. One based on alternative crops could raise tem substantially more, particularly in Central Mexico, entailing significant land-cover canges. Te feedstock sector would promote commercial agriculture but generate scant economic growt in rural areas. A igly mobile migrant labor force could keep wage increases small and temporary. Stagnant wages and rising food prices ultimately could reduce rural purcasing power wit mixed implications on food security. Rural food deficits could decrease, but on-farm consumption would also contract markedly weter feedstocks are based on staples or cas crops. Wile absentee producers and landlords will be te major beneficiaries of te Mexican biofuels industry, no mecanisms ave been contemplated to mitigate adverse impacts on rural wellbeing and land use. Assessments of biofuels potential in oter developing countries, particularly Sub-Saaran Africa, sould be reconsidered using a similar metodology. Keywords: corn; cas crops; feedstocks; food security; land use cange; agriculture. 2

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 1. Introduction Only a few years ago many observers predicted tat bioenergy would elp satisfy growing demands for energy wile protecting te environment and simultaneously create jobs, promote equity and raise te living standard of te world s poor [1-4]. But oter analysts seemed equally confident tat a rapidly expanding biofuels industry would entail iger crop prices, price volatility, and food insecurity witout clear environmental benefits [5, 6]. Arguments on bot sides of te debate remained largely teoretical in late 2006, wen crop prices began to surge, ultimately leading to a global food crisis in 2008. Te factors driving tis surge in prices included te low value of te dollar and world-wide canges in crop production/utilization ratios tat decimated inventories [7-10]. In te case of corn, drivers also included te growt of te biofuels industry, particularly etanol, wic drove corn prices in te United States up sarply. Energy and food prices are expected to continue affecting developing economies for te foreseeable future, influencing teir balance of payments and prospects of growt [11]. Te effects will not be uniform across te developing world. Analysts distinguis countries like Haiti, wose weak finances will likely deteriorate wit canges in te terms of trade, and oters like Mexico, wose large energy sector sould provide wider scope for adjustment [7, 8]. At te micro level, neverteless, expected price trends are bound to urt poor consumers everywere tat is, unless governments intervene [7, 11]. Notwitstanding tese expectations, many analysts remain confident tat biofuels will provide ample opportunities for growt in developing areas [7, 12-15]. Agricultural growt an expected outcome of te industry s demand for feedstocks will drive economic development and reduce poverty, wic will elp safeguard food security. Potentially adverse effects on food supplies and natural resources will constitute at most a sort-term trade-off. Adequate policies and good timing sould allow governments to arrive at a desirable outcome in te long run, wen tecnological advances will relieve te pressure on food prices and te environment [12, 13]. Accordingly, governments around te world are promoting te development of a domestic bioenergy industry in te name of national security, as a buffer against oil-market socks and a way of conserving foreign reserves (tat migt ten be used to finance food imports) [5]. 1

27 28 29 30 31 32 33 34 35 36 37 38 39 Government initiatives under way, suc as te recent Mexican Biofuels Law, ave wide and ambitious goals. Acieving all teir targets neverteless will require integrating widely disparate food, energy and environmental policies wile accounting for teir potential synergies and tradeoffs [5, 7]. Te success of tese initiatives could depend on anticipating and addressing effectively te multiple effects of policies over social, economic and environmental variables. Tis will require in turn te development and implementation of an integrated analytical framework capable of addressing te interactions between causal factors and outcomes at different scales i.e., field, farm, regional, national and international levels wile accounting for te specificity of market and policy conditions witin and across individual localtions. Alas, integrating micro and macroeconomic processes witin a variable context can be an elusive goal [16, 17]. Here we use an agent-based, general equilibrium model of rural Mexico a simple analytical framework tat integrates across organizational and spatial scales in order to analyze te callenges of te Mexican Biofuels initiative and its implications on agricultural markets, rural development, food security and land-use cange. 40 41 42 43 44 45 46 47 48 49 50 51 52 1.1. Te Mexican context As an early participant in te recent corn-price saga, Mexico became te poster cild of te food crisis [18-20]. Corn is a food staple in Mexico, were per-capita consumption is te world s second igest. Imports represent 30% of domestic consumption, and te country is te second largest importer of American corn. Domestic corn prices are tigtly linked to US prices [21, 22], and since te peso also is tigtly linked to te dollar, a weak dollar could not buffer price increases as in oter countries. Not surprisingly, Mexico was expected to bear te most adverse impacts of crop-price increases after Sub- Saaran Africa [23]. Actual impacts of te corn-price surge migt ave been more complex [20]. In late 2006, consumer prices of corn, corn-flour and tortillas soared across Mexico, creating a major crisis for te new federal administration, wic immediately called te public s attention to te spike in international prices and specifically blamed biofuels. It migt seem surprising tus tat te government s first response was to raise corn-import quotas, but it ultimately resorted to controlling 2

53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 tortilla prices directly in order to sort out a persisting crisis. Paradoxically, federal autorities began drafting at te same time a plan to promote a domestic biofuels industry, wic eventually was signed into law in early 2008. Te Mexican government s response to te crisis was consistent neverteless wit remarks made by te autonomous Bank of Mexico, wic attributed te spike in prices to speculation in te domestic grain market. Indeed, for te last fifteen years, consumer prices for corn, corn-flour and tortillas ave moved largely independently of corn import prices [20]. Te Mexican Biofuels Law s professed objective is to diversify te nation s supply of energy, reduce green-ouse emissions, promote rural development, create jobs and improve te rural poor s standard of living wile safeguarding food security [24]. Te Mexican Biofuels Program was launced a year later [25]. Its immediate goal was to guarantee a sufficient supply of feedstocks to generate 1.76x10 8 liters of etanol in its first year (8.1x10 8 liters in te second). In order to produce te 2.3 x10 6 tons of biomass required, te Program anticipated incorporating 3x10 5 ectares of land into te feedstock sector. Officials ave been vague regarding te type of incentives tat could be used to reac tis goal, but te focus seems to be on enancing producer profitability [25, 26]. Deficiency payments and price supports are tus a prime candidate. Te Law explicitly states tat price trends sould be considered wen determining te type and amount of incentives needed by te budding industry [24, 27]. Concerns wit food security ave been addressed by restricting te use of staples, namely corn, as feedstock. Te Program will tus focus initially on five cas crops: sugar cane, oil palm, sweet sorgum, jatropa and castor beans. Altoug imported corn can be used as feedstock unrestrictedly, use of domestic grain requires te government s express autorization. Te Ministry of Agriculture simultaneously as 73 74 75 76 promoted a reform tat would deregulate use of yellow corn. 1 are tigtly linked. Wite and yellow corn prices neverteless 1 Domestic corn grain can be used as feedstock only wen surpluses satisfy te country s consumption demand [24, 27]. Te Ministry of Agriculture argues tat only wite corn sould be regulated, since yellow corn is used mostly in industrial uses. 3

77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 2. Metods Our analysis of te Mexican Biofuels Law focuses on te feedstock sector s implications on land use and te rural economy. Biofuel production per se would employ skilled and unskilled labor, but tis employment could be insignificant compared to te feedstock sector use of rural unskilled labor [14, 15]. Te various fuel crops considered in te Mexican Biofuels Program constitute an alternative for most farmers in Mexico [25, 26]. Wile some of te impacts of an expanding feedstock sector could depend on wic of tese crops is used [14, 15], te present analysis focuses on te differences between cas crops and corn, te staple crop. Two scenarios are used to analyze tese two cases separately: te first simulates a 5% increase in producer prices for corn; te second an analogous increase in cas-crop prices. Given te vagueness of prevailing policy te scale of te socks migt seem arbitrary, but preliminary results sowed tat tese prices would acieve te Program s target, generating similar feedstock surpluses for te biofuel industry. Our analytical framework is an agent-based model of te Mexican rural economy, used previously in analyses of various policies [20, 28]. Te complete model nests individual models of multiple rural ouseolds and oter (non-rural) agricultural producers into a single economy (i.e., into a general equilibrium context), wic in turn is linked to te world economy troug trade and migration (see Appendix). Te basic ouseold model is similar to te one described by Sing et al. [29] wen all goods and variable inputs are tradable; it resembles te model of de Janvry et al. [30] wen one or more markets are missing. As in tose models, ouseolds are assumed to maximize teir utility troug consumption subject to several constraints, including a self-sufficiency constraint for subsistence ouseolds tat sets on-farm consumption equal to production. Nesting proceeds in tree stages. First, rural ouseolds in eac of five regions are grouped into four types tat describe well te socioeconomic landscape of rural Mexico [31]: (i) landless ouseolds; (ii) small-olders (<2 a); (iii) medium-olders (2-5 a), and (iv) large-olders (>5 a). Houseold models are calibrated at tis stage using data from te 2003 Mexico National Rural Houseold Survey (Encuesta Nacional a Hogares Rurales de México or ENHRUM). 4

103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 In a second stage, te four ouseold groups are integrated wit absentee (non-rural) producers in teir region (i.e., individual landolders and corporations not based in rural areas but tat noneteless participate in rural markets) into five different regional models based on aggregate statistics for te agricultural sector (SIAP). Rental and wage rates are determined at tis stage reflecting observed wage and rents disparities across regions due to transaction costs. Given te importance of factor markets to te present analysis, te model was modified to incorporate an explicit supply of land and migrant labor. In order to reflect te sluggisness of land-use cange due to conversion costs and managerial inertia, te allocation of land across uses was restricted troug a nested constant elasticity of transformation (CET) supply function [32-34]. Te function was calibrated using ENHRUM data and econometric estimates available in te literature [34-36]. In modeling migrant labor, supply elasticities were estimated econometrically using survey data [31]. Elasticities vary across ouseold groups and regions, reflecting actual differences in access to foreign and domestic labor markets. In a final stage, te five regional models are integrated into a model of rural Mexico were oter market balances are determined. Tis tree-stage, agent-based modeling approac provides muc greater detail tan aggregate general equilibrium (CGE) models [14, 15], incorporating differences in prices, 118 production tecnologies and market participation across regions and individual agents. 2 It also provides 119 120 121 122 123 124 125 126 greater flexibility, as it allows eterogeneous producers and consumers to respond idiosyncratically and interact wit eac oter in local markets. For instance, a ouseold tat does not participate in food markets may still employ wage labor in subsistence crop production using a tecnology tat is distinct from tat of commercial ouseolds. It may also sell its labor to oter farmers. Eac farmer s valuation of te subsistence crop is represented by a ouseold-specific sadow price, and eac ouseold s demand for market and non-market goods is distinct from oters. Simulation models often ave been criticized, described as black boxes supported by poor data. Tis migt apply to some aggregate models used in te past. Our disaggregated, agent-based approac 2 Some aggregate models ave separate demand equations for ouseold groups, but market access, prices and production accounts (and tus tecnologies and participation in all production activities) typically are sared and performed by a single representative agent [14, 15]. 5

127 128 129 130 131 overcomes te first misgiving by explicitly modeling in minute detail te processes tat generate a particular outcome [20]. It addresses te second criticism troug use of farm-level survey data. Out-ofsample prediction and sensitivity analysis are te two main tools used to validate simulation models. Accurate ex-ante prediction of land-use canges in 2008, following te corn price surge, establises our model s validity in land-use analysis [20, 28]. 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 2.1. Data Data on non-rural production (i.e., te activities of absentee or non-rural producers) were derived from te Mexican government s Agricultural Information System (SIAP) after adjusting for te sare of production pertaining to rural producers [28]. Data on non-rural tecnologies were derived from publised input-output matrixes for different regions [36]. Te Mexico National Rural Houseold Survey (ENHRUM) was te source of all data for rural areas. Designed in collaboration wit Mexico s Census Bureau, te survey s sampling frame provides a statistically reliable caracterization of Mexico s population living in rural areas, i.e., in communities wit fewer tan 2,500 inabitants. Te survey provides detailed data on assets, endowments, productive activities and market participation for 1765 ouseolds. Net incomes for sample ouseolds were estimated based on detailed data on agricultural and non-agricultural activities and oter income sources, including, on- and off-farm labor, migration and public transfers. Te sum of income from tese sources equals ouseold total net income. Summary statistics for eac of te four ouseold groups can be found in Ref. 31. Rural incomes are igly diversified, and income sources vary sarply across te four groups. Altoug most rural ouseolds grow crops, te sare of net income from tis activity in total income ranges from less tan one per cent (for landless ouseolds) to 15.3% (for te largest landolding group). Livestock accounts for anoter 0.9% to 6.9% of total rural income. Rental income constitutes between 3.4 and 4.4% of total income for te landed groups but logically was negligible for te landless group. Te lack of correlation between landoldings and average per-capita income is due to te weigt of off-farm income from wages, migrant remittances, public transfers, and non-agricultural production. By far te largest rural income 6

153 154 155 156 sare is from wages (43.1 to 64.1%), but only 13% of wages come from farm work, reflecting rural ouseolds integration wit non-agricultural labor markets. Predictably, wile te sare of total income from crop production activities is igest for large-older ouseolds, te small-older and landless ouseolds receive most of teir income from wage work, mostly off-farm. 157 158 159 160 161 162 163 164 165 166 2.1.1. Corn production and output Of te 21.7 million ectares of cropland in 2002, te year of te survey, 40% were sown to corn. Rural ouseolds cultivated 37% of tis area but produced only 10.5% of total corn output. Rural ouseolds were responsible for 60% of te area and 23% of te output in Souteast Mexico, but only 11% and 2%, respectively, in te Norteast. Average corn yields were 3.2 tons per ectare (ton/a): 2.3 ton/a in rainfed areas and 8.5 ton/a under irrigation. Average yields ranged from 1.5 ton/a in te Souteast to 7.2 ton/a in te Nortwest. Yields obtained by non-rural producers across Mexico were 5.6 times iger tan tose of rural producers (0.8 ton/a). Te difference in yields between non-rural and rural producers was 720% in West-central Mexico but only 25% in te Nortwest. 167 168 169 170 171 172 173 174 175 2.1.2. Land endowments and markets Te average rural landolding is 4.8 ectares but te standard deviation is large (25.1), reflecting a ig degree of landlessness. Except for te landless, every ouseold group in rural Mexico rents land bot out and in. In Norteast Mexico, rural ouseolds rent more land out tan in, wic means tat 65% of rented land is rented to absentee producers. In oter regions, rural ouseolds rent up to 8 times more land in tan out. At least 25% of land rented by rural ouseolds in tese regions belongs to absentee landowners. Accordingly, in te model, excess demand is satisfied by absentee producers in every region but te Norteast, were absentee producers absorb te excess supply of rural land. 176 177 178 7

179 180 181 182 183 184 185 186 187 2.1.3. Labor markets Rural ouseolds are net suppliers of local wage labor and migrant labor. Unsurprisingly, te supply of local wage labor is greatest among landless ouseolds, but even large-olders are net suppliers in every region but te Nortwest. In te latter region, around 25% of te rural supply of wage labor commutes into urban areas on a daily basis. In oter regions, less tan 15% of te local wage labor commutes. Up to 20% of te remaining labor is employed by neigbors; te rest is employed locally by outside agents or institutions, including absentee producers and te tree levels of government. In Nort and Central Mexico, labor demand by absentee producers is large enoug to absorb te excess supply of local labor and employ migrants from oter regions. 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 3. Results Corn output grows 1.5% following te increase in tis crop s price in te first scenario. Similarly, te output of cas crops grows 1.5% after teir own prices increase in te second scenario. Corn sales neverteless grow considerably more in te first scenario tan cas-crop sales in te second: sales receipts are 16% greater for corn tan for cas crops. In percentage terms, cas-crop sales grow by 1.5%, wile corn sales grow by 10.6% (Tables 1, 2). Clearly, te market supply of corn is surprisingly elastic. Te reason is tat corn market surpluses increase largely at te expense of on-farm consumption, wic falls in response to prices. In contrast, given tat on-farm consumption of cas crops is nil, cas-crop sales grow only to te extent tat output does. Tis result migt seem surprising in ligt of te classical farmouseold model [29], were consumption of staples grows wit prices due to iger incomes, reducing market surpluses. In fact, te same income effect is observed ere in bot scenarios; but price canges ave additional, indirect effects on bot farmers and ouseolds tat ultimately result in greater corn sales. Indirect impacts suc as tese, known as general equilibrium effects, are observed only wen te interactions among agents are analyzed in an economy-wide context, as done ere. In tis case, tese effects are felt first witin te agricultural sector, but tey ripple gradually across te entire economy wit unexpected albeit important implications for rural development, land use and food security. 8

205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 Eac sector s contribution to agricultural growt clearly depends not only on te extent of its own expansion, wic is similar for cas crops and corn, but on te sector s relative size. Te gross value of te cas-crop sector is over four times greater tan tat of corn, so te former s expansion as a greater weigt in agricultural gross product. Its contribution to growt also depends on weter te focus is on gross product or in te generation of value added. A major determinant of canges in tese two variables is te way in wic scarce productive factors, particularly land, are reallocated across economic activities after te price sock. In tis case, land in corn expands 2.5% after an increase in corn prices, but total cropland grows only 0.1%, wile demand for labor does not register a significant increase (Table 1). By comparison, te area under cas crops expands only 0.9% after an increase in cas-crop prices, but total cropland and demand for rural labor increase 0.5 and 0.7%, respectively. Overall, te expansion of cas crops is associated wit a 2.8% increase in te agricultural gross product and a 3.6% increase in its net product or value added (Table 1); te expansion of corn (after an increase in its own price) results in 0.4 and 0.8% growt in tese variables, respectively (Table 2). Canges in demand for land and labor ave distinct effects on wages and rents across Mexico depending on te relative importance of cas crops and corn in eac region. After corn prices rise, landowners rents increase between 0.2% in Norteast (NE) Mexico and 3.0% in te Souteast (SE) (Table 4). Farm wages increase up to 0.8% in te Nortwest (NW) region but not significantly in te NE. Increases are considerably iger after cas-crop prices rise: cropland rents increase between 4.0% in te NW region and 5.9% in Central (C) Mexico, wile wages increase between 0.2% in te NE and 3.4% in C Mexico (Table 3). Higer demand for land and labor combined wit iger rents and wages explains wy agricultural value added increases substantially more wit cas-crop prices tan corn s (Tables 1, 2). Canges in land rents, wages and profits determine te opportunity cost of uncultivated land, wic ultimately drives land-cover cange and deforestation. Crop prices considered ere generate profits equivalent to at least 5% of cropland rents in every region (Tables 3, 4); but te availability of pastures mitigates te encroacment of forests tat oterwise would ensue, since pastures are more easily converted into cropland tan forests. Expected canges in te opportunity cost of land at te forest edge 9

231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 reflect te relative availability of pastures across regions: te opportunity cost of uncultivated land rises between 0.1% in NE Mexico and 2.0% in te SE after an increase in corn prices (Table 4). Tis opportunity cost increases considerably more after cas-crop prices rise, i.e., between 2.0 and 4.9%; but in tis case, increases are largest in C and West-central (WC) Mexico, were cas crops ave te greatest weigt (Table 3). Since te expansion of te agricultural frontier is not immediate, existing agricultural land is reallocated in te sort-term according to relative canges in cropland rents and profits. Tus, corn land and output contract 1.7% and 1.3%, respectively, after an increase in cas-crop prices (Table 1). Te land-intensive livestock sector experiences a similar contraction. An increase in corn prices and an expanding corn sector ave a smaller toll on oter land-based activities: cas crops and livestock contract only 0.2% after corn prices rise (Table 2). Again, tese responses vary widely across regions reflecting canges in local rents, wages and profits (Tables 5 14 in Supplementary data). Crop expansions entail te reallocation of production factors not only across activities but also across farms, e.g., between rural farms and farms owned by absentee (i.e., non-rural) producers. Tis reallocation is greatest after an increase in corn prices: absentee producers employ 0.4% more cropland and labor after te sock, wile demand for tese factors among rural producers decreases 0.6 and 0.3%, respectively (Table 2). Accordingly, corn output increases 3.7% in non-rural farms but decreases 0.6% in rural farms. Te reason is tat te expansion pursued by rural and non-rural large-olders (and te ensuing increase in rents and wages) forces small and subsistence farmers to down-scale teir own activities. Tis as implications on te demand and supply of corn. Sales of corn increase sarply as land is reallocated from subsistence to commercial farms, but on-farm consumption necessarily declines. Since corn purcases also diminis after prices increase, te deficit of corn in rural areas contracts nearly 30% even as te output of rural farms drops. In contrast, non-rural farms surpluses increase 3.7%, contributing an additional 10.6% to corn volumes available for urban and industrial uses, including feedstocks. Significant differences are observed across regions. In SE Mexico, te consumption of corn of all rural-ouseold groups falls more tan teir output, reducing te region s large deficit by almost 16% (Table 10). In C Mexico te current deficit of corn is transformed into a small surplus despite a 10

257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 2.5% contraction of rural farms output (Table 11). In contrast, every group in te NW except te landless expands its output and sales of corn, expanding te region s large surplus by 4.0% (Table 11). Te demand and supply of corn is affected also after an expansion of cas crops. In tis case, nonrural corn surpluses contract 1.3% as producers turn to cas crops; and rural deficits grow 16% as ouseolds substitute purcased corn for on-farm production. Corn volumes available for non-rural uses decline 4.9% as a result. Agricultural price socks ave implications beyond agriculture. As crop production expands, te increase in rural wages is felt across te rural economy. Non-agricultural activities contract 0.4 and 1.9% after te expansion of corn and cas crops, respectively (Tables 1, 2). Tus, te entire rural economy grows at considerable lower rates tan te agricultural sector. Te gross product of rural areas still grows 1.7% after cas-crop prices rise but only 0.2% after corn prices. Net growt is sligtly iger: rural value added increases 2.6 and 0.6% after cas crops and corn, respectively (Tables 1, 2). Overall, te expansion of cas crops contributes more to rural economic growt in every region but te SE, were te economy grows as muc as wit cas crops as wit corn. But growt is not distributed omogeneously witin tese regions. Aggregate-growt statistics reported above conceal important differences between rural and non-rural farms. For instance, after accounting for non-agricultural activities, non-rural producers generate 3.6% more value added tan before cas crops expand; rural ouseolds only 1.6% (Table 1). Differences between tese groups are more pronounced after corn expands. In tis case, non-rural producers generate 0.9% more value after te expansion; rural ouseolds generate only 0.2% more (Table 2). Tese differences ave important repercussions on te distribution of income, but identifying winners and losers requires accounting for te redistribution of value added among landowners, workers and capital owners. Te contraction of subsistence output and non-agricultural income observed in bot scenarios represents a considerable loss for most rural ouseolds; but teir wage earnings rise as ouseold members ire out teir labor. Simultaneous increases in on-farm employment and wage rates result in a 4.9% increase in wage income after cas-crop prices rise; but wage income rises only 0.7% after corn 11

283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 prices because demand for labor is tempered by te eterogeneity of supply responses witin tis sector. Incomes do not rise uniformly across te population; canges are loosely associated wit landolding sizes. Wages constitute an outlay for rural large-olders tat employ labor, e.g., in NE Mexico. Rural large-olders in general earn 1.5% less in wages after corn prices increase, wile landless ouseolds earn 1.1% more (Table 2). In contrast, every ouseold group experiences an increase in wage income after cas-crop prices rise. Gains remain associated wit landolding sizes, ranging from 1.9% for large rural landolders to 5.6% for te landless. At te same time, wage gains entail losses in oter income sources. For instance, employment in local farms entices migrants to return to teir communities and would-be migrants not to leave. Accordingly, remittances decrease 1.3% after an expansion of cas crops but only 0.1% after an expansion of corn (Tables 1, 2). As opposed to wages, rents constitute a net outlay for rural ouseolds but a net gain for absentee landowners, wo experience a 3.2% gain, after all income sources are considered, wen biofuels are based on cas crops. Income grows only 1.0% for rural ouseolds. Te same pattern is observed in every region, but it is sligtly more equitable in te NW, were rural and non-rural incomes increase 2.6 and 4.2%, respectively (Tables 5-9). It is least equitable in te SE, were rural incomes barely increase. In contrast to wage-income gains, full-income canges are positively associated wit landolding sizes. However, differences across groups and regions depend not only on factor endowments, as in aggregate models [14, 15], but also on eac group s activities. After an increase in te price of cas crops, income gains range from 0.4% for te landless to 2.3 and 3.2% for rural and non-rural large-olders, respectively. Gains do not differ significantly in nominal and real terms due to te small weigt of cas crops in ouseold consumption. Te outcome is qualitatively different wen corn prices rise. In tis case, rural ouseolds income grows 0.2% nominally, but tese gains become 0.4% losses in real terms due to corn s weigt in rural diets (Table 2). Absentee producers income still grows 0.9% in real terms. Nominal-income canges remain tied to landoldings sizes: absentee producers and rural large-olders experience 0.9 and 0.7% gains, respectively; ouseolds wit less tan 5 ectares or no land do not experience any nominal gains. All rural ouseold groups experience real-term losses, but differences 12

309 310 311 312 across groups depend on consumption patterns more tan landolding sizes. Outcomes are similar in every region but te NE, were incomes register few canges because of corn s low weigt in te local economy. In SE and C Mexico, were corn is most important, real incomes decrease up to 1% after corn prices rise. 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 4. Discussion Simulation results suggest tat modest increases in crop prices could generate considerable surpluses for an emerging biofuels industry weter tis is based on cas crops or te staple corn. Sales of feedstocks could represent a significant flow of revenue into te agricultural sector. A 5% increase in price could raise sales of corn and cas crops by US$273 and $236 million, respectively. However, te implications on agricultural growt and te rural economy depend on factors oter tan revenue. Some of tese implications will be similar watever fuel crop is used. In bot cases examined, for instance, surpluses will not derive from productivity-enancing innovation but from iger profits associated wit sales. Altoug some growt will be tied to iger productivity, namely in corn, most will be associated wit an area expansion. Tus, feedstock supplies will come at te expense of oter land-based activities, particularly if te agricultural frontier were tigtly constrained. To tis extent, te ability of feedstocks to foster agricultural growt could be limited. Oter repercussions will depend on te particular fuel crop used. Cas crops, given teir iger revenue per unit area, could ave a greater aggregate impact on te rural economy tan corn. As opposed, production of corn, particularly for on-farm consumption, often constitutes a drain of cas flows into rural Mexico [28]. Te specific costs and benefits experienced by various sectors of te population will depend neverteless on te microeconomics of te supply response of eac type of crop. Aggregate supply responses conceal te reallocation of land and labor among individual farms [20]. As demand for feedstocks grows, prices will determine te optimal allocation of tese resources. In te case of corn, only te most efficient farms migt be capable of adding value and expand in te face of iger land and labor costs. Tus, te output of large farms migt expand significantly, but small farms 13

335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 could ave few reasons to expand. Subsistence corn farmers are relatively inefficient and do not always value corn at market prices [37, 38]. Tus, tey migt not benefit from price increases but would still experience iger wages and rents. Weter tese constitute actual outlays or only te opportunity costs of family land and labor, many of tese farmers could forsake subsistence activities in favor of wage work in order to secure an income. A corn-based biofuels industry terefore could entail an expansion of commercial corn output at te expense of subsistence agriculture. A corn sector restructured in suc a way could generate substantial feedstock surpluses witout radically altering land and labor markets. Supply responses are muc more omogeneous in te cas-crop sector, so demand for land and labor could rise considerably more if biofuels are based on alternative crops. Differences between a feedstock sector based on cas crops and one based on staples could ave important implications on local livelioods. Feedstocks in general could be a large source of added value, but tis value would not necessarily be appropriated by farmers. According to our results, rural farms migt generate 1.6% more value after a 5% increase in cas-crop prices, but rural incomes would increase only 1.0%. Absentee producers could face an analogous situation since profits tend to dissipate as supply adjusts. But producers ave oter income sources. Value generated in te feedstock sector ultimately will be distributed between producers and te owners of land, capital and labor. Some payments to factor owners will elp distribute biofuels benefits more equitably. Large-olders relying on ired labor migt need to sare teir increased revenue wit farm workers troug wage increases. Small farms also would fail to fully appropriate te value of teir output, aving to pay rents to landolders and te owners of agricultural macinery [28]. Altoug te size of tese transfers is tigtly linked to eac stakeolder s endowments, factors beyond stakeolders control will play a significant role. Aggregate demand and supply determine relative rent and wage canges, as do restrictions on te excange of land, labor and capital troug local markets. Te cost of commuting or migrating restricts te mobility of labor, creating temporary disparities in regional wages. Te immobility of land and restrictions on land use can create even greater disparities in rental rates across regions. Wage increases would benefit te poor, mostly working rural ouseolds, but 14

361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 canges in rents and profits can ave regressive effects. Rents and profits also tend to flow out of rural areas. Oter market imperfections and limited availability of arable land could allow rural farmers to maintain a profit; but over alf of all corn produced in rural areas generates no profits; it is consumed onfarm eiter as food or feed: 87% of rural ouseolds do not sell corn [28]. In sum, few ouseolds in rural areas would benefit directly from te feedstock sector, particularly one based on corn. Alternatively, analysts believe tat te biofuels industry could ave a multiplier effect on te rural economy tat would elp spread its benefits more widely across te population. However, most revenue generated by te feedstock sector could elude rural Mexico entirely: 60% of te area in corn and nearly 70% of tat in cas crops is managed by absentee producers [28]. Moreover, given teir iger productivity, non-rural farms currently account for an overwelming sare of agricultural output: nearly 90% in te case of corn [28]. Since some rural farms could move out of corn production after a price sock, a corn-based feedstock sector migt not generate any growt in te rural agricultural sector. Tis would not be te case for cas crops. Some of te revenue generated by cas crops migt leak out of rural areas troug capital and land rents or input purcases, but wages would flow in te opposite direction. In some places, tis cas flow could increase demand for locally-produced goods and promote rural economic activity, but several factors could preclude suc outcome in Mexico. Paid employment presumably could ave a similar effect on te rural economy to tat of migration, wic elps fund productive investment in rural areas but reduces te availability family labor [39]. In tis case, owever, employment in te feedstock sector would be mostly a substitute for increasingly 380 unaffordable subsistence activities. 3 Moreover, te amount of income available for investment would 381 382 383 384 depend on te scale of wage increases, wic would be small and temporary as long as rural ouseolds remain a igly mobile labor force via migration. In te long run, rural wages can only rise to te extent tat labor productivity increases; but tere is no indication tat feedstocks would induce suc cange. Temporary wage increases would still benefit ouseolds, but tey also constitute a cost for te rural 3 Tis migt not be unique to te feedstock sector. In fact, wage income as increased substantially in rural Mexico since te mid nineties, largely replacing on- and off-farm activities as a source of income [40]. 15

385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 economy. Since wage increases would not be transmitted to urban markets, tey would also diminis te relative competitiveness of local goods vis-à-vis industrial substitutes. Commercial integration wit urban areas makes suc substitutes widely available. As a result, commerce could sipon cas flows out of rural areas as soon as labor markets brougt tem in. More generally, analyses tat overlook price effects witin rural economies overestimate te indirect benefits and underestimate te costs of te feedstock sector and biofuels industry. Crop expansions based on prices impose costs on oter sectors, so demand for feedstocks alone will not likely promote widespread growt of te rural economy. Absentee producers wo are not part of te rural economy do not experience tese costs, so teir earnings could rise as muc or more tan agricultural revenue (Tables 1, 2). Since rural ouseolds depend more closely on te local economy, teir incomes could grow considerably less or not at all. Absence of real income gains in fact could be te foremost factor precluding a rural multiplier effect, most clearly in te case of corn. As discussed above, feedstock surpluses would derive mostly from an increase in area. An additional 325 x10 3 ectares could be sown to cas crops in Mexico after a 5% increase in prices sligtly more tan te Biofuels Program s initial goal. If tis ad no effect on te agricultural frontier, as government officials expect, some 66 x10 3 ectares could move out of corn and into alternative fuel crops. Tis would exacerbate te rural corn deficit, ultimately reducing domestic corn volumes available for urban and industrial uses. Since US and Mexican corn markets are integrated, tere sould be no sortfall in supply. Tat is, alternative fuel crops migt not increase te price of food in urban Mexico, but tey would exacerbate te country s dependence on food imports, wic poses its own risks to national food security [8]. Mexico would not need to import more corn if producer prices for tis crop rose witin te scope of te Biofuels Program. Some 245 x10 3 ectares could be added to tis crop after a 5% price increase. Surpluses available as biofuel feedstock would increase significantly, but tese would come largely at te expense of subsistence production. Overall, because of feedstocks inevitable impact on te opportunity cost of land (i.e., rents), on-farm consumption will likely drop markedly weter corn or cas crops are 16

411 412 413 414 415 416 417 used for fuel. However, te consequences on food security at te ouseold level remain uncertain, particularly for smallolders and te landless. Tese groups migt devote more time to wage labor (probably in te feedstock sector) but spend teir earnings substituting for declines in subsistence production. Tis would replace one type of risk for anoter: ouseolds would be less vulnerable to weater socks but more exposed to canges in te terms of trade between food and labor [41]. On te oter and, reducing te biofuel industry s repercussions on food safety would require expanding te agricultural frontier. 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 5. Conclusions Te Mexican Biofuel Strategy is based on te assumption tat cas flows from feedstocks will reactivate agriculture, delivering considerable benefits to te rural population [26, 27]. It also assumes, peraps because of wide access to corn imports, tat food security will be safeguarded by te use of cas crops for fuel [26, 27]. Simultaneously, it expects tat te program will ave no impact on te agricultural frontier [26]. Our results confirm tat te demand for feedstocks foreseen for te Program s first year could be satisfied troug domestic supplies of eiter cas crops or corn given modest price increases. Supply responses would promote growt, particularly if cas crops were used. Neverteless, te repercussions on rural incomes, food security and land use would not be as expected in te Strategy. Te Mexican Biofuels Commission coordinates efforts to acieve te Biofuels Law s goals: diversifying te nation s supply of energy wile promoting rural development, improving te poor s living standard and reducing carbon emissions [24]. However, te Strategy on wic tis coordination is based does not fully recognize te trade-offs among tese goals. For instance, te study tat underlies te Strategy considers only risks to aggregate food availability [25, 26, 42]. It cautions against te use of corn given te country s dependence on imports but concludes tat incorporating 800 x10 3 ectares to alternative fuel crops would not compromise food security. As discussed above, avoiding te use of staples as feedstocks does not guarantee tat te rural poor will ave te same access to food. Moreover, te Strategy literally confuses agricultural growt for rural development [26]. A significant expansion of 17

437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 agricultural output neverteless could entail scant growt of te rural economy or, in te case of corn, a contraction of its gross product. Even if te value added of rural farms grew, tis need not imply income gains for rural ouseolds. Unless labor productivity rises, a combination of stagnant wages and rising food prices could redistribute te wealt generated by te biofuel industry regressively, increasing inequality. As for te implications on land use, it is a truism tat as long as total cropland remains constant, food and fuel crops will compete for land. Te Mexican Biofuels Program aims to sow 300 x10 3 ectares to fuel crops in its first year witout impinging on te forest margin [25]. However, te Biofuels Strategy does not contemplate raising te productivity of eiter labor or land [26]. Tus, it is unlikely tat feedstock production will not raise te opportunity costs of forest lands. Conversion of land from oter uses migt reduce forest-cover losses in NE Mexico, were pasture is considerably more abundant tan cropland. But te opportunity costs of land in oter regions could rise up to 5% wit a biofuel industry based on cas crops and 2% wit one based on corn (Tables 3, 4). Tis would not necessarily translate into deforestation in every region, particularly wit corn. Water availability could constrain te agricultural frontier in NW Mexico, were corn is grown on irrigated land. But a corn-based industry would raise te opportunity costs of land most sarply in te SE, were forests are commonly cleared for corn and deforestation rates already are Mexico s igest [43]. An industry based on cas crops would ave a muc greater impact in te SE but most of all in oter regions. Since tese crops can grow on rainfed land, a biofuels industry could entail significant land-cover canges. Tese impacts will be greater if te Biofuel Program expands five-fold in its second year as planned. Given te ineffectiveness of current conservation programs, particularly in SE and C Mexico, tis could translate into deforestation [28]. Around te world, farmers are responding to demand for feedstocks from an emerging biofuels industry. Initiatives promoting biofuels proclaim ambitious social, development and environmental goals. A realistic assessment of biofuels potential requires considering all aspects of tis complex system simultaneously. An elastic supply of feedstocks, for instance, is essential to maintain biofuels 18

463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 competitiveness; yet a rise in crop prices is necessary to draw some of te benefits into te agricultural sector. Likewise, measures tat could distribute biofuels benefits more widely (e.g., troug smallolder out-grower scemes) tend to reduce te industry s profitability [14, 15]. Accurate assessments also require detailed data. Current assessments seem to suggest tat biofuels could enance growt and reduce poverty in various Sub-Saaran countries witout endangering food security; but te studies on wic tey are based assume ideal conditions [14, 15]. Stylized, aggregate models cannot be te basis for detailed assessments or accurate forecasts. A common assumption in tese models is tat te industry s benefits would be distributed widely across te population irrespectively of wo undertakes production, i.e., tat returns to land and labor would increase omogenously. Our analysis sows wy tis migt not be te case in a real economy. Our assessment of te Mexico Biofuels initiative reveals significant trade-offs among its goals. Biofuels migt elp diversify Mexico s energy supplies but could contribute little to rural development or reducing poverty and equality. Te implications on food security also are uncertain. Poverty reduction and food security depend crucially on returns to productive assets [41]. An emerging feedstock sector could cange te relative value of land vis-à-vis labor, wic is te main asset of landless rural ouseolds. It could also increase land rents and profits; but tis could bring canges in control over te land and te distribution of rents and profits, wic usually urts te most vulnerable [44, 45]. Witout additional policies, tese canges could compromise food security and emissions reductions. In order for government to play its role, it must acknowledge tese trade-offs. A more in-dept analysis of te Mexican Biofuels initiative will require disclosure of iterto unpublised data on te type and amount of public resources flowing into te feedstock sector. Recent administrative reforms in te Ministry of Agriculture ave made tis information unavailable. 485 486 487 488 6. References 1. FAO (2005) Bioenergy and te millennium development goals. Forestry Department, FAO, Rome. 2. Koonin SE (2006) Getting serious about biofuels. Science 311: 435. 19

489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 3. Hazell P, Pacauri RK (2006) Bioenergy and agriculture: promises and callenges. IFPRI Focus 14. 4. United Nations-Energy (2007) Sustainable bioenergy: a framework for decision makers. UN. 5. FAO (2007) Assessment of te world food security situation. Committee on World Food Security, Tirty-tird session. FAO, Rome. 6. Pimentel D, Patzek T (2006) Green plants, fossil fuels, and now biofuels. Bioscience 56: 875. 7. World Bank (2007) World Development Report 2008: Agriculture for Development. World Bank, Wasington. 8. FAO (2009) Te state of food insecurity in te world. Economic crises impacts and lessons learned. FAO, Rome. 9. Abbott, P.C., C. Hurt, & Tyner, W.E. (2009). Wat s driving food prices? Farm Foundation Issue Report. Marc 2009 Update. Oak Brook, IL: Farm Foundation. 10. Hocman, G., Rajagopal, D., & Zilberman, D. (2010). Are biofuels te culprit? OPEC, food, and fuel. American economic review: Papers & proceedings 100, 183-187. 11. Sapouri S, Rosen S (2006) Food security assessment. GFA-18. Economic Researc Service, USDA. 12. Rajagopal D, Sexton S, Hocman G, Zilberman D (2009) Recent developments in renewable tecnologies: R&D Investment in Advanced Biofuels. Annu Rev Resour Econ 1:621-644. 13. Byerlee D, De Janvry A, Sadoulet E (2009) Agriculture for rural development: toward a new paradigm. Annu Rev Resour Econ 1:15-31. 14. Arndt C, et al. (2009) Biofuels, poverty, and growt: a computable general equilibrium analysis of Mozambique. Environment and Development Economics 15: 81 105. 15. Turlow J (2010) Economy wide effects of bioenergy developments. In Bioenergy and Food Security: te BEFS Analysis for Tanzania. Maltsoglou I, Kwaja Y (Eds) Te Bioenergy and Food Security Project Food and Agriculture Organization of te United Nations. FAO, Rome 16. Van Ittersum et al., (2008) Integrated assessment of agricultural systems A component-based framework for te European Union (SEAMLESS). Agricultural Systems 96: 150 165. 20