Abstract
We quantify the emergence of biofuel markets and its impact on world agriculture using the multimarket, multicommodity international FAPRI model. The model incorporates trade-offs between biofuel, feed, and food production and consumption and international feedback effects of the emergence through world prices and trade. We shock the model with exogenous changes in ethanol demand, first in the United States, then in Brazil and other countries, and compute shock multipliers for land allocation decisions for important crops and countries. The Brazilian ethanol expansion using sugarcane has fewer consequences on existing arable land allocation than the U.S. ethanol expansion does using corn feedstock. (JEL Q17, Q42)
I. Introduction
The global emergence of biofuel production is bringing forth new trade-offs between food, feed, energy, and the environment (Tilman et al. 2009; Searchinger et al. 2008). These trade-offs are multidimensional, with both local and global implications (Elobeid et al. 2007; Farrell et al. 2006; Hill et al. 2006; Searchinger et al. 2008; Secchi, and Babcock 2007; and Walsh et al. 2003). Our paper sheds light on several dimensions of these trade-offs, with a focus on land allocation. We explore how significant the trade-offs are and where they occur in terms of geographical and market location (land use, commodity, and processed food markets). Our investigation of the land allocation effects of the biofuel emergence pays particular attention to ethanol production expansion and its location, its effects on land devoted to feedstock, and competing crops.
In particular, we focus on ethanol expansion in the United States and, alternatively, in Brazil, China, the European Union (EU), and India.1 These five countries constitute the bulk of the world ethanol market. Brazil and the United States are the world’s largest ethanol producers and markets. Brazil is the only significant exporter of ethanol. These two countries use two different feedstocks, which have contrasting implications for global land use. The other three (China, the EU, and India) also have expanding ethanol sectors, but they are relatively small compared to those in the United States and Brazil. Biodiesel is the other major existing biofuel, but biodiesel is currently unprofitable, and its expansion is much less likely, particularly in the United States.2 The latter motivates our focus on ethanol. An expansion of ethanol production and/or consumption in these five countries has significant local land allocation effects that propagate globally through world trade and price effects. The prices induce land reallocation away from crops for which relative prices fall and toward crops for which relative prices rise. As competition for land intensifies, aggregate land use in crop production is expected to increase. We examine the projected evolution of land allocation under this biofuel emergence by type of crop and by pasture use for key countries growing feedstock for ethanol (corn, sorghum, wheat, sugarcane, and other grains) and major crops competing with feedstock for land resources, such as oilseeds. The biofuel expansion outside of the United States tends to have fewer consequences for land allocation beyond sugarcane production, in contrast to the expansion of U.S. ethanol using corn, which affects many temperate crops and food production. As a consequence, the food-biofuel trade-off could be either exacerbated or decreased depending on the location of the biofuel production underlying the expansion (in the United States versus in Brazil).
Our analysis relies extensively on the international FAPRI (Food and Agricultural Policy Research Institute) model, a multimarket, partial-equilibrium model of world agriculture, food, fiber, and bioenergy markets. We use the international FAPRI model to quantify a sequence of two ethanol shocks: first, an exogenous increase in U.S. ethanol demand, and second, an exogenous increase in world demand for ethanol (specifically in Brazil, China, the EU, and India). We compute the effects of these two shocks in deviation from the 2007 U.S. and world FAPRI baseline for the years 2007/08 to 2016/17 (FAPRI 2007). To compare the shocks, we compute proportional impact multipliers on key variables (dln(variable)/dln(shock)) and report their values summarized in a 10-year average. The variables of prime interest are land, prices, trade, production, and consumption. The land multipliers show, at the margin, how sensitive (or not) land allocation is to the growing demand for ethanol, not only in countries with sizeable ethanol markets, but also in other countries growing feedstock crops and crops competing for land with these feedstocks. We highlight the movement of land away from major crops competing for land with feedstock crops. Because of the high U.S. tariff on ethanol, the U.S. and world ethanol markets are nearly segmented.
Higher U.S. demand for ethanol only translates into a U.S. ethanol production expansion, with little ethanol expansion elsewhere. However, this U.S. expansion has strong global effects on land allocation, as coarse grain prices transmit significant shocks worldwide. Changes in U.S. coarse grain prices also affect the prices of U.S. wheat and oilseeds through land shifting away from these crops. These agricultural price increases also transmit to world markets. We also compare our multiplier estimates to those of Searchinger et al. (2008), which were based on different assumptions and tend to be larger than ours.
In contrast to the U.S. expansion, an expansion in Brazilian ethanol use and production chiefly affects the world ethanol market and land used for sugarcane production in Brazil and, to a lesser extent, in other sugar-producing countries. However, the Brazilian expansion has a small impact on other land uses in most other countries, as sugar crops, especially sugarcane, tend to compete less for land than do corn and coarse grains.
II. Recent Literature on Biofuels
There are a growing number of studies on ethanol and other biofuel markets, as the industry has experienced a boom in recent years. The contributions of De La Torre Ugarte et al. (2003); Elobeid and Tokgoz (2008); Elobeid et al. (2007); English, Menard, and De La Torre Ugarte (2001); English et al. (2004, 2006a, 2006b); Gallagher, Otto, and Dikeman (2000); Gallagher et al. (2003, 2006); House et al. (1993); Searchinger et al. (2008); Tokgoz and Elobeid (2006); Tokgoz et al. (2007); USDA-OCE (2000, 2002); and Walsh et al. (2003) stand out. Gallagher et al. (2006) look at the competitive position ofBrazilian ethanol produced from sugar processing vis-a-vis the U.S. ethanol produced from corn, under the assumption of no tariffs in the ethanol market. Koizumi and Yanagishima (2005), among the first to establish an international ethanol model, examine the implications of a change in the compulsory ethanol-gasoline blend ratio in Brazil on world ethanol and sugar markets.
Few papers fully endogenize the prices of major feedstock crops used in ethanol production (sugarcane and corn). Most studies hold these constant (Gallagher et al. 2006; Koizumi and Yanagishima 2005), with the exception of that of Ferris and Joshi (2005). Recently, Elobeid and Tokgoz (2008) and Tokgoz and Elobeid (2006) have endogenized these crop prices using a large set of models, including the international crop markets and explicit market equilibrium mechanisms. These recent studies of Elobeid and Tokgoz incorporate linkages between an international ethanol model, an international sugar model, and a U.S. crops model. FAPRI (2007) and Tokgoz et al. (2007) incorporate model developments proposed by Elobeid and Tokgoz and represent the first attempt to fully endogenize all major prices through explicit modeling of world markets for ethanol, feedstock crops, and other agricultural commodities. Searchinger et al. (2008) address the indirect land use change including deforestation and savanna use brought about by a likely U.S. ethanol expansion. The indirect land use changes are inferred from Tokgoz et al. (2007) and information from the Woods Hole Research Center to assign land use to different ecosystems. Their net carbon consequences are then assessed using the GREET model.
Elobeid and Tokgoz (2008) also analyze the impact of ethanol trade liberalization and removal of the federal tax credit in the United States on U.S. and Brazilian ethanol markets.3 The U.S. tariff insulates U.S. producers from most of the world market discipline. The removal of ethanol trade distortions induces an increase in the price of world ethanol and jolts the U.S. ethanol market. The U.S. domestic ethanol price decreases, which results in a significant decline in production and an increase in consumption through larger imports. Brazil responds to the higher world ethanol price by increasing its production. Total ethanol consumption in Brazil decreases, and net exports increase.
Tokgoz and Elobeid (2006) analyze the impact of price shocks in three input and output markets critical to ethanol: gasoline, corn, and sugar. They investigate the impact of these shocks on ethanol and related agricultural markets in the United States and Brazil. They find that the composition of a country’s vehicle fleet determines the direction of the response of ethanol consumption to changes in the gasoline price. They also find that a change in feedstock costs affects the profitability of ethanol producers and the domestic ethanol price. In Brazil, where ethanol and sugar compete for the supply of sugarcane, changes in the sugar market affect the competing ethanol market and price.
English et al. (2006a) look at the impact of achieving the goal of “25 × 25” (25% of the projected energy needed in 2025 coming from renewable energy sources) on the U.S. agricultural sector. They use a computer simulation model ofU.S. agriculture, which provides annual estimates of changes in U.S. land use resulting from the demand generated by bioenergy industries. This comprehensive study of the U.S. agricultural sector abstracts from impacts of the changes in the U.S. agricultural sector on international agricultural markets and the resulting adjustments on world prices. English et al. (2006b) analyze similar issues, using the same model for a shorter time horizon (2014) and focus on the southern region of the United States.
The various analyses by English, De la Torre, and associates (De La Torre Ugarte et al. 2003; English et al. 2006a, 2006b; Walsh et al. 2003) have been outstanding and have supported the competitiveness of cellulosic feedstock in ethanol production. However, most recent estimates (e.g., Popp and Hogan 2007) provide a unit cost of cellulosic feedstock nearly twice as large as the original ones (e.g., Hallam, Anderson, and Buxton 2001). The new estimates clearly suggest that cellulosic ethanol is not competitive at current market conditions. In our analysis, we abstract away from cellulosic-ethanol expansion to reflect its lack of competitiveness based on its current cost structure and technology.4
III. The International Fapri Model
Model Structure
The international FAPRI model is a set of multimarket (multicommodity, multicountry), nonspatial, partial-equilibrium models developed at Iowa State University.5 The international FAPRI model includes econometric and simulation submodels covering all major temperate crops, sugar, ethanol, dairy, and livestock and meat products for all major producing and consuming countries and calibrated on most recently available data (see Table 1 for commodity and country coverage). The international FAPRI model is used extensively for market outlook and policy analysis (FAPRI 2007). Extensive market linkages exist in the model, reflecting derived demand for feed in livestock and dairy sectors, competition for land in production, and consumer substitution possibilities for sets of close substitutes such as vegetable oils and meat types. The international FAPRI model and associated numerical analyses have been validated through numerous academic publications, external reviews, and internal annual updates.
The International Fapri Model
The modeling system captures the biological, technical, and economic relations among key variables within a particular commodity and across commodities. The model is based on historical data analysis, current academic research, and a reliance on accepted economic, agronomic, and biological relationships in agricultural production and markets. In general, for each commodity sector, the economic relationship that supply equals demand is maintained by determining a market-clearing price for the commodity. In countries where domestic prices are not solved endogenously, these prices are modeled as a function of the world price using a price transmission equation. Since the submodel for each sector/commodity is linked to the other submodels, changes in one commodity sector impacts other sectors.
Agricultural supply comes from land harvested multiplied by yields. Land responds to relative agricultural returns, reflecting the competition for land among crops within defined geographical areas. Oilseeds and grains compete for land in many countries. Within grains, corn and other coarse grains also compete for land. Sugarcane production is often on land unsuitable for other crops. However, it does compete with soybeans in Brazil and with rice in some Asian countries.
More specifically, acreage functions in the international FAPRI crops model are expressed as a function of crop prices (own price and the price of competing crops), and lagged acreage and prices. Prices are real, that is, deflated by a GDP deflator specific to the country (no money illusion). In all crop area equations, the own price effects dominate the cross-price effects of competing crops. Symmetry is not imposed. Prices enter acreage functions either as part of real gross returns per unit of land, or simply as prices depending on fit and analyst opinion. The two corresponding alternative equations for harvested area for each crop are
[1a]or
[1b]where A is area harvested in thousand hectares, GR is gross returns in real terms, P is real price, index j stands for competing cross commodities j = 1,. . .n, subscript i is the crop of interest, and r denotes country or region, and the time index t represents the year. GR and p are defined in local currency per hectare. For each crop, GR is determined by
[2]where Y is the yield of each crop in metric tons per hectare. As an example, planted area for corn in Brazil is expressed as a linear function of lagged corn area, corn, soybean, and wheat gross returns with corresponding elasticities provided in the Appendix in Table A2. Table A2 specifies when returns or prices are used in different equations and shows when lagged acreage is not present in some equations. Table A2 provides acreage elasticities for key commodities and countries. Because extensive data are available for U.S. agriculture, the U.S. model is divided into nine regions with equations for each crop grown within each region. Expected net returns for the crop and its competing crops determine the planted area under each crop and in each region. The planted area for each crop is adjusted for area under the Conservation Reserve Program. The equation for planted area is given as follows:
[3]where Ai is area planted in million acres for crop i, DP is real decoupled payment per acre, index j stands for competing commodities j = 1,. . .n, i denotes the crop of interest, and R denotes region R in the United States. ENR represents the real expected net returns of crop i and competing crops in dollars per acre in each region R. For each crop, this is determined by the following equation:
[4]where EP is the expected real price of commodity i in dollars per unit (e.g., bushel), EYD is expected yield in units per acre, VC is the real variable production expenses in dollars per acre, ELDP is the real expected loan deficiency payments in dollars per acre, and ECCP is expected real countercyclical payments in dollars per acre. The time index t denotes the year. The equation for corn planted area in region R, for example, is given a linear function depending on net expected returns for corn, barley, cotton, oats, rice, soybeans, sorghum, wheat, sunflower, and peanuts and decoupled payments. The elasticities corresponding to the price responses in [4] are presented in Table A2. For regions where certain crops are not grown, the price response and elasticities that correspond to that particular crop are equal to zero.
Land allocation in the FAPRI model does not include an explicit land market with a clearing rent because of data constraints on such markets in many countries. Instead, the sum of acreage allocations over all crops and pasture use is constrained not to exceed the arable land endowment of the country. All analyses conducted always include an ex post check to make sure that land allocation satisfies the constraint. An example of this check is the case of Argentina. With pasture land accounting for the largest land use category, we derive a time path for dairy and cattle stocking rate improvement to make sure that the sum of all agricultural uses of land covered in our models as a proportion of total agricultural area is within historical rates and does not exceed one. An annual stocking rate improvement of 1.7%, which is well below the historical rate of 2.5%, is necessary to meet the use land constraint.6
We now turn to the structure of the ethanol submodel. Like the other FAPRI submodels, the ethanol submodel is a nonspatial, multimarket world model. The submodel specifies ethanol production, use, and trade between countries/regions. Country coverage consists of the United States, Brazil, the EU, China, India, Japan, South Korea, and a Rest-of-World aggregate. The model incorporates linkages to the agriculture and energy markets, namely, feedstock crops, world sugar, and gasoline markets.
The general structure of the country ethanol model is made up of behavioral equations for production, consumption, stocks, and net trade. Complete country models are established for the United States, Brazil, China, the EU, and India, while only net trade equations are set up for Japan, South Korea, and the Rest-of-World because of limited data availability. The model solves for a representative world ethanol price (Brazilian anhydrous ethanol price) by equating excess supply and excess demand across countries. Using price transmission equations, the domestic price of ethanol for each country is linked with the representative world price through exchange rates and other price policy parameters. The exception is the domestic ethanol price in the United States, which is solved endogenously as long as the U.S. price remains below the Brazilian price adjusted for transportation. All prices in the model are expressed in real terms. U.S. ethanol is produced primarily from corn predominantly from a dry-milling process with distillers dried grains (DDGs) as a major by-product. The U.S. ethanol market is nearly insulated from the world ethanol market because of a high U.S. tariff imposed on nonpreferential imports of ethanol. Some limited imports occur through the Caribbean Islands, but these do not threaten the protection of U.S. producers.7 Brazilian ethanol is produced from sugarcane. Indian ethanol is produced from molasses. EU ethanol production uses a combination of grains as feedstock.
The FAPRI model is capable ofhandling most, but not all, important structural and regime changes. Ethanol production capacity is explicitly modeled as a function of expected net profit for both dry and wet mills separately. Utilization rates for each type of mill are modeled as a function of net profit margin. Output of ethanol is the sum of ethanol production from dry and wet mills and co-evolves with profits.8 Demand for feedstock follows the same pattern. In the livestock industry and feed demand, we also model the dynamics of livestock and their influence on feed use and demand. However, the FAPRI model does not impose zero return in farming via exit or entry. Hence, profit margins have an important role at the margin but do not serve as a signal for exit when margins substantially deteriorate, except in the ethanol capacity equation where capacity shrinks with negative profit. The structure of the U.S. component of the ethanol submodel is more elaborate, especially on the ethanol demand side. Because these demand characteristics have limited implication for land use, we refer interested readers to Tokgoz et al. (2007) for further information. Further description of the FAPRI model is available on the Internet at www.fapri.iastate.edu/models/.
Data and Calibration
The model is calibrated on 2006/07 marketing-year data for crops and 2006 calendar-year data for livestock and biofuels, and 10-year projections are generated for the period between 2007 and 2016. The model is recursively solved for 10 successive annual equilibria. The submodels also adjust for marketing-year differences by including a residual that is equal to world exports minus world imports, which ensures that world demand equals world supply.9 Elasticity values for supply and demand responses are based on econometric analysis and on consensus estimates.
Agricultural and trade policies for each commodity in a country are included in the submodels to the extent that they affect the supply and demand decisions of the economic agents. These include taxes on exports and imports, tariffs, tariff rate quotas, export subsidies, intervention prices, other domestic support instruments, and set-aside rates. The models assume that existing agricultural and trade policy variables will remain unchanged in the outlook period. Elasticity parameter estimates and policy variables used in the model are available at www.fapri.iastate.edu/tools/ for the international component of the model. Elasticities for the ethanol submodel (including its U.S. component) are shown in the Appendix (Table A1).
Data for commodity supply and utilization are obtained from the F. O. Lichts online database,10 the Food and Agriculture Organization (FAO) of the United Nations (FAOSTAT Online),11 the Production, Supply, and Distribution View (PS&D) of the U.S. Department of Agriculture (USDA),12 the European Commission Directorate General for Energy and Transport,13 and UNICA,14 among others. Macroeconomic data such as GDP, GDP deflator, population, and exchange rate are exogenous variables that drive the projections of the model. They were gathered from the International Monetary Fund and Global Insight.
These data sets provide historical data that are used to calibrate the models, and the models provide projections for supply and utilization of commodities and prices. Supply and utilization data include land use, yields, production, consumption, net trade, and stocks.
IV. Simulation Scenarios
Baseline
The baseline is the FAPRI2007 U.S. and World Outlook (FAPRI 2007), which establishes a credible reference trajectory for all variables of interest (land use by country and by crop, production, consumption, commodity prices, uses of crops including ethanol feedstock, feed, and food use, and trade flows). The FAPRI baseline assumes continuity in current policies for the coming decade (2007/08 to 2016/17). The baseline includes a significant increase in ethanol production and use, with a sustained increase in land allocation to corn in the United States away from soybeans, and toward sugarcane and soybeans in Brazil. As explained in the following sections, three important U.S. policies condition the analysis: the U.S. tariff on ethanol, the U.S. mandated use of ethanol via renewable fuel mandates, and the U.S. blender’s tax credit. Detailed information on the FAPRI baseline assumptions and results is posted at www.fapri.iastate.edu/outlook2007/.
Scenario 1
In scenario 1, we shock U.S. demand for ethanol with a permanent 10% exogenous expansion (a permanent horizontal shift of demand with respect to its baseline level), leading to an equilibrium increase in U.S. ethanol use of roughly 3%. As the U.S. demand for ethanol is very elastic at the margin, a large permanent shift is necessary to induce a net increase in the equilibrium of the U.S. ethanol market. The mandatory component of ethanol demand is price inelastic but inframarginal. At the margin, ethanol-based gasoline competes in price with regular gasoline; a small increase in ethanol price decreases the competitiveness of ethanol-based gasoline, and the use of ethanol by refiners falls rapidly at the margin. What matters for land allocation is the change in the equilibrium value of ethanol quantity/use, which affects the derived demand for feedstock crops. The shift in demand is really the way to get to the shift in equilibrium use in a sensible way.
The adjustment mechanism to this U.S. shock is first through a direct impact on U.S. ethanol production, then U.S. use and production of coarse grains, chiefly corn, and then of other crops. Ethanol imports are little affected, as the U.S. price remains lower than the world price inclusive of the high U.S. tariff and the transportation cost. In the United States, there is a trickle-down of price effects from crude oil to gasoline, to ethanol, to corn, to other coarse grains, to other crops, to feed prices, to livestock, and so on, which in turn affects these markets and the associated land allocation. The crop price changes freely transmit to the world markets. Since the United States is a major producer and exporter of corn, wheat, sorghum, and soybeans, any changes in the U.S. price of these commodities affect the world markets to a great extent. The higher U.S. ethanol price does not transmit to the world ethanol market because the two ethanol markets are effectively segmented by the large U.S. ethanol tariff.
We compute annual proportional impact multipliers on key variables in proportional deviation from the baseline (dln(variable)/ dln(U.S. ethanol use)) and report their values summarized in the 10-year average. The denominator of the multiplier is based on the change in ethanol use, because it is the relevant change that matters for land allocation because it affects the derived demand for feedstock crops. The horizontal shift in demand is the sensible way to induce the shift in equilibrium use.
These multipliers are shown for key variables in the first row of Table 2 under each country label. The multipliers can be interpreted as indicating the average annual percent change in a variable induced by a 1% increase in U.S. ethanol use, or multiplied by 100, they provide an estimate of the impact of a doubling ofU.S. ethanol use on the variable of interest. The variables shown are ethanol output and trade, land area, and world price of corn, sugar beet, sugarcane, feed grains, wheat, soybean, rapeseed/canola, and sunflower. The countries reported in Table 2 are Brazil, China, the EU, India, and the United States for ethanol-producing countries; Argentina, Australia, Canada, Mexico, and South Africa for non-ethanol-producing countries; and a world aggregate. Impact multipliers for world and U.S. ethanol and commodity prices are reported in Table 3.
Impact Multipliers (10-Year Average) From Scenariosfor Major Variables
Impact Multipliers (10-Year Average) From Scenarios for World and U.S. Prices
Scenario 2
Scenario 2 considers a permanent 5% exogenous increase of ethanol demand (a permanent horizontal shift of ethanol demand in deviation from the baseline level) in Brazil, China, the EU, and India, leading to an equilibrium increase in aggregate ethanol use in these countries of roughly 3%. We compute a second set of impact multipliers for 2007/08 to 20016/17 in deviation from the baseline values (dln(variable)/dln(aggregate ethanol use in Brazil, China, the EU, and India)) and again we report their values as summarized in the 10year average. The interpretation of these multipliers is similar to the former one (percent change in variable induced by a 1% change in aggregate ethanol demand in these four countries, or if scaled by 100, an estimate of the impact of a doubling of ethanol use in these four countries). They appear in the second row of Table 2 for each country, below the country results for Scenario 1. Detailed results for all variables and all countries are available from the authors.
V. Results
Scenario 1 Results
The increase in U.S. ethanol use directly affects U.S. feedstock markets, especially corn, and to a lesser extent, sorghum. U.S. exports of coarse grains decrease (trade has the highest multipliers); stocks are reduced (second-largest multipliers) as they respond negatively to prices; then land devoted to coarse grains expands. Long-term land responses would be higher, as full adjustment takes a long time.15 Substitution possibilities in feed demand for corn contribute to the propagation of higher prices to all feed products (sorghum, barley, oats, and wheat).
In the U.S. corn market, the derived demand for feedstock in ethanol production increases by more than one (multiplier value of 1.04) as corn displaces other grains.16 Corn feed use falls (multiplier value of -0.19), seed use increases (0.18) with the expansion of land devoted to corn, and food corn use falls slightly. The most significant drop in food use is in highfructose corn syrup production (-0.06); the multiplier for other-food use is -0.02. In aggregate, total corn use (domestic use) increases but by much less than the initial ethanol increase (multiplier value of 0.29). Corn exports decrease dramatically (-0.6), and stocks fall substantially (-0.45).
U.S. land area allocated to corn increases (0.14). This could potentially increase by higher rates in the long run when inventories bottom out at their minimum required levels for markets to function. The impacts on U.S. sorghum and barley are qualitatively similar but smaller. U.S. land area increases slightly for sorghum (0.035) and more substantially for barley (0.103).
In U.S. oilseed markets, there is a sharp reduction in land devoted to soybeans (-0.10) and, to a lesser extent, to sunflowers (-0.046). Changes in land allocations to other oilseeds exhibit smaller magnitudes in absolute value. These reductions lead to higher oilseed prices, which lead to a fall in biodiesel production (multiplier value of -0.15). In livestock and meat markets, the ethanol shock translates into higher feed grain prices, lower DDG prices, and a small increase in soybean meal prices.
The lower DDG price has to be qualified. The expansion of ethanol production induces a similar expansion in its by-products such as DDGs. Higher DDG supply translates into a lower equilibrium price. Although the historical pattern suggests that corn and DDG prices move in close tandem, with a much larger ethanol market projected in our study, our results show a possible departure in price movement from their historical pattern in the short run, but returning back to their strong correlation in the long run. Short-run corn and DDG price movements can diverge when the corn price is pulled up by strong ethanol demand while the supply of DDG expands substantially to depress the price in the short run. Ethanol plants usually sell a third of DDG as wet with 65% to 70% moisture content. This product is very perishable, having a shelf life of less than a week. In contrast, it takes time to build animal numbers to expand DDG demand to absorb substantial expansion in supply. Both factors can lead to a decline in the DDG price in the short run when the corn price is rising.
The shock leads to a small reduction in aggregate meat production. Substitution in consumption induces net gains to some sectors. U.S. beef production increases slightly and wholesale meat prices increase moderately. Retail prices increase by even less. The trickle-down of price effects is summarized in Figure 1.
U.S. Ethanol Shock: Price Multipliers (average over 2007-2017; DDGs, distillers dried grains)
The world impact of the U.S. ethanol shock occurs first via preferential ethanol trade expansion. Preferential ethanol imports by the United States increase, but from a very small base (multiplier of 0.6). This expansion has a negligible impact on the world ethanol markets except for net trade adjustments, which are large in percentage terms because they are computed from a small base. In terms of feedstock, there are negligible effects on world sugarcane land allocation (0.002), as world sugarcane-based ethanol production does not change much.
By contrast, the higher U.S. feed grain and oilseed prices transmit to other countries, but not fully, because tariffs and other trade costs prevent full transmission of world price effects into local markets. World land area devoted to corn increases moderately in aggregate (0.06), but more substantially in Argentina (0.106). Argentina is the second-largest exporter of corn, behind the United States. Reduction in U.S. corn exports driven by increasing corn use for ethanol gives Argentina an opportunity to expand its world market share and, hence, is a strong response in the allocation for corn. Land used for corn in Egypt also expands more than the world average (0.075). Growth in land devoted to corn in Brazil and India follows the aggregate corn supply (multipliers of 0.05 and 0.059, respectively). Additional land devoted to corn in other important corn production countries, such as China, Mexico, South Africa, and Canada, is also observed (multipliers range between 0.020 and 0.048).
Higher world prices for other feed grains also occur. They translate into smaller net imports and larger net exports, but these effects are more moderate than for corn. Land devoted to feed grains other than corn tends to expand moderately. For example, world sorghum area expands proportionally by about a fourth of the world corn area expansion (multiplier of 0.016 versus 0.056). Sorghum area in Mexico expands above the world average level (multiplier of 0.046). World soybean land allocation falls slightly (−0.026), but it expands in Brazil, the most competitive soybean producer in the world (0.033). However, soybean land area falls in Argentina in favor of land devoted to corn (soybean multiplier of −0.04). Rapeseed land allocation falls globally (multiplier of −0.02 or less in absolute value), except for in the EU. Sunflower land area falls in Argentina (−0.068) but increases in the EU (0.015). As the competition for land increases, there are moderate decreases in pastureland in Brazil and Argentina.
Recent U.S. ethanol policy changes provide an opportunity to illustrate the usefulness of our multipliers. The Energy Independence and Security Act (EISA) of 2007 increases the implied starch-based ethanol Renewable Fuel Standard from 7.9 billion gallons in the 2005 Energy Act to 15 billion gallons, an increase of nearly 90%. The U.S.-shock multipliers in Table 1, multiplied by 90, provide an estimate of the impact of the starch-based ethanol mandated expansion under the EISA relative to the 2005 mandate, assuming it is binding. For example, with such a policy shock, corn area in the United States, Argentina, and Brazil would increase by nearly 13%, 9.5%, and 4.5%, respectively. The world corn price would increase by nearly 26%. If the 2005 mandate is not binding, the ethanol shock in percentage changes will be smaller than 90% moving from the current use to the new mandated use. Similarly, one could look at the removal of the U.S. blender’s tax credit. The impact of the tax credit depends on the binding nature of the renewable fuel mandate. If the mandate is binding, then the tax credit is redundant and ethanol use is invariant to the credit. If the mandate is not binding, then the tax credit stimulates ethanol demand. Mapping the removal of the tax credit into a decrease in U.S. ethanol use, our multipliers can be used to predict the land allocation implications of removing the credit.
Table 4 presents multipliers for total crop area for the world and major countries for the last year of the projection, namely, 2016. In response to a U.S. ethanol expansion, world crop area increases, with a multiplier of 0.011. Most of the increase in world crop area is through a world corn area increase. Brazil and South Africa respond the most, with multipliers of 0.034 and 0.046, respectively. Mexico follows with a multiplier of 0.025, followed by the United States, Thailand, and Egypt.
Impact Multipliers (The Last Year of the Projection) From Scenarios for Total Crop Area
Scenario 2 Results
The shock on the world ethanol market has a direct impact on the world ethanol price, as well as on the local ethanol markets in which the shock is initiated. The average impact on the world ethanol price is very high (multiplier of 3.11). In sharp contrast, the U.S. ethanol price (Omaha price) is left nearly unaffected (multiplier of 0.01). This lack of impact is motivated by the segmentation of the U.S. and world markets as previously explained. World ethanol trade is impacted and exhibits large multipliers because trade is thin: U.S. imports fall (multiplier of -2.27), Brazilian exports rise despite the exogenous increase in their demand (1.31), and the EU’s imports respond sharply given the shock imposed on the EU ethanol market and the net-importing status of the region (multiplier of 8.92). Brazilian ethanol production increases substantially (multiplier of 0.94). By contrast again, U.S. ethanol production and feedstock use are barely affected (multiplier of 0.04). Given the increase in the world ethanol price, there is a small decrease in U.S. use via its imported consumption (multiplier of −0.02). Land effects in the United States are even smaller.
In feedstock markets, the largest price effects are registered for sugar, given the importance of sugarcane and sugar byproducts as feedstock in Brazil and India (sugar price multiplier of 0.13). The effect on world corn prices is a tenth of that on the sugar price (0.013) because of the limited size of grain-based ethanol production outside of the United States, namely, in China and the EU. Similarly, the price of other feed grains increases slightly (multiplier of 0.009 or less). The world ethanol shock also has some impact on grain stocks and grain trade flows, but land area devoted to grains and grain production remains nearly unchanged in most countries. Feedstock use increases in China (0.75), the EU (0.24), India (molasses), and Brazil (sugarcane multiplier of 0.94).
In this scenario, the impact on sugarcane and sugar is the only significant change in feedstock markets. Brazilian sugarcane area increases substantially (multiplier of 0.44), sugar production falls as it competes with ethanol for the sugarcane feedstock (−0.11), and sugar exports fall (−0.168). Other competitive sugar exporters expand their land area devoted to sugar crops, production, and exports (multipliers of0.01 to 0.04). A small scale ofland expansion for sugarcane is found in other major sugar-exporting countries, for example, Australia (0.018), India (0.017), and Thailand (0.004). Worldwide, sugarcane land area increases with a multiplier of 0.138, but world sugar output falls as expected, albeit rather slightly (multiplier of −0.012). Although ethanol demand in China and the EU is also exogenously raised in this scenario, the impact on the feedstock market is obviously lower, as ethanol industries there are still in their infancy stage. Hence, changes in the area allocation of corn (primary feedstock for ethanol production in China) and wheat (major feedstock used for ethanol production in the EU) are very modest in the world. The impacts on most of the other crops and sectors are negligible.
Table 4 also presents total crop area multipliers in response to a global (nonU.S.) ethanol expansion in 2016. The world crop area multiplier is 0.001 in this scenario, which is very small relative to the multiplier for the U.S. ethanol expansion. This is also the case for most of the other countries, with the notable exception of Brazil, which has a total crop area multiplier of 0.064. South Africa, Thailand, the EU, and Egypt follow, with total crop area multipliers between 0.003 and 0.005. Since the United States is a major exporter of grains and oilseeds, any change in U.S. corn demand impacts world markets considerably. This leads to larger land use changes in the world in response to a U.S. ethanol market expansion relative to a global (non-U.S.) shock.
Table 4 includes total crop area multipliers from Searchinger et al.’s study (2008), one that used the same modeling system but assumed that there was no bottleneck in the E-85 gasoline market, and that long-run equilibrium conditions prevailed in the U.S. ethanol industry and livestock sectors in 2016, the end year of their analysis. They introduced a $10-per-barrel crude oil price shock. Searchinger et al. used indirect land use generated from the model to infer changes in associated ecosystems and then used the GREET model to estimate greenhouse gas emissions and savings from ethanol. The crop area multipliers from the Searchinger et al. study are generally larger than the crop area multipliers in this study, as shown in the table. For example, Searchinger et al.’s world crop area multiplier is 0.017, whereas it is 0.011 in this study for the U.S. demand shock. This difference in results exists chiefly because Searchinger et al. impose a long-run equilibrium assumption for the U.S. ethanol market (the U.S. ethanol industry grows until the net profit margin for the cornbased ethanol sector is exhausted to zero) and the demand for E-85 fuel does not exhibit any bottleneck.
VI. Conclusions
This study analyzes and quantifies the foreseeable emergence of biofuel markets in U.S. and world agriculture using the international FAPRI modeling system and two ethanol demand scenarios that contrast the implications of using corn feedstock in the U.S. expansion, versus sugarcane in the expansion in Brazil. We examine the projected evolution ofland allocation under these biofuel emergence scenarios for major feedstocks and crops, as well as for key countries. We report the movement of land away from major crops competing for land with feedstock crops for the two scenarios.
The major effects ofU.S. ethanol expansion on U.S. agriculture occurs through corn prices trickling down to other feed grains and other crop prices, inducing significant U.S. land allocation changes. These price effects also transmit worldwide, especially for feed grains and soybeans, and thus affect the land allocation in some countries such as Argentina and Brazil. Land allocation effects may be understated in our results because of large stock adjustments occurring in the short term, and because we do not impose long-run equilibrium conditions, as do Searchinger et al.
U.S. and world ethanol markets are segmented but could be integrated by removing the U.S. ethanol tariff (free trade). This in turn would reduce the significant corn land area effect of the current U.S. ethanol expansion. Brazilian ethanol producers, rather than U.S. producers, would perceive the increased profit opportunities on the U.S. market and increase their ethanol production to meet U.S. ethanol demand.
When ethanol production expands outside of the U.S. (global scenario), the land effect of this ethanol expansion is mostly felt in sugarcane and sugar markets because Brazil is the largest and most competitive ethanol producer and relies on sugarcane feedstock. Sugarcane competes with fewer crops for land than corn does. The resulting impact of this foreign ethanol expansion is negligible on U.S. agriculture, even including the U.S. sugar market, because of the insulation of the U.S. sugar and ethanol industries through trade protection.
In sum, the global effects of the biofuel expansion on land are smaller than its local effects on land and on other local industries competing for the feedstock. Effects on land reallocation within countries are sometimes quite large, such as in the case of the U.S. corn and soybean reallocation or the grain/ oilseed allocation in Latin America brought about by the U.S. ethanol expansion. The small global land effect of the Brazilian ethanol expansion is a striking result.
Given the emerging nature of the world ethanol markets, our study comes with some caveats. Limited data availability for ethanol markets makes econometric estimations of elasticities used in the biofuel model difficult. The scenario results provided here are dependent upon several assumptions, such as the lack of cellulosic ethanol production, the ability of the livestock sector to adapt to the use of biofuel coproducts in feed rations, and the ability of the world ethanol market to move through supply and demand bottlenecks. As entrepreneurs around the world push for new breakthroughs in biofuel and coproduct production and usage, it is possible that some of the assumptions used for this analysis may no longer be relevant.
Appendix
International Ethanol Model Elasticities
International Crop Model Elasticities
Footnotes
The authors are, respectively, scientist, Department of Economics, Iowa State University; Marlin Cole professor of international agricultural economics, Department of Economics, Iowa State University, and also affiliated with INRA Grignon, France; associate scientist, Department of Economics, Iowa State University; associate scientist, Department of Economics, Iowa State University; research fellow, International Food Policy Research Institute; and assistant professor, Department of Agricultural Economics, University of Tennessee. Earlier versions of this work were presented at the Farm Foundation workshop Biofuels, Food, and Feed Tradeoffs in April 2007 in Saint Louis, Missouri, and the OECD conference Modelling Global Land Use and Social Implications in the Sustainability Assessment of Biofuels in June 2007 in Copenhagen, Denmark. The authors thank the editor, Daniel Bromley, and two anonymous referees for useful suggestions and comments, and Bruce Babcock, Chad Hart, Dermot Hayes, Patrick Westhoff, and conference participants for discussions and comments on the analysis.
↵1 The European Union trading block is hereafter referred to as a “country” for ease of exposition, and “EU” refers to the EU-25.
↵2 The EU, on the other hand, has an established biodiesel market.
↵3 Specifically, the federal tax credit is the volumetric ethanol excise tax credit, which is given to refiners blending ethanol with gasoline. At the time of the study, the credit was 510 per gallon of ethanol. It is currently 450 per gallon.
↵4 Ethanol from cellulosic feedstock can become economically viable if it receives a significant enough subsidy to compensate for its higher cost structure relative to corn-based ethanol or if technological advancements result in lowering the cost of production.
↵5 The U.S. crops, livestock, and dairy models, which are part of the FAPRI modeling system, have been developed and maintained by the University ofMissouri.
↵6 If a binding constraint occurred, the procedure to reconcile the land allocation and constraint would be as follows: A logistic function would be used where the land limit can be directly imposed. As the total land allocation gets closer to the constraint, the aggregate land response falls to zero and a second equation, a land allocation equation, allocates the given aggregate land allocation from the first equation into the various agricultural sectors in the country according to relative returns.
↵7 The U.S. ethanol trade policy includes a 2.5% ad valorem tariff and a per unit tariff of 540 per gallon. However, under the Caribbean Basin Economic Recovery Act (CBERA), if ethanol is produced from at least 50% agricultural feedstock grown in a CBERA country, it is admitted into the U.S. free of duty.
↵8 In each process, ethanol production is obtained by Multiplying production capacity by the utilization rate.
↵9 We attempt to harmonize our world supply and utilization data to the Northern Hemisphere marketing-year. For crops, a residual term is needed to balance world supply and world demand in history. This residual is equal to world exports minus world imports for each crop. This residual term is then held fixed at the level of the last period of data observation so that it does not influence equilibrium solutions in the projection period.
↵10 F. O. Lichts, Online Database, Lichts Interactive Data. Available at www.agra-net.com/portal2/home.jsp?template=productpage&pubid=ag083.
↵11 FAOSTAT Online Database. Available at http://faostat.fao.org/faostat/collections?subset=agriculture (accessed July 2006).
↵12 U.S. Department of Agriculture, Foreign Agricultural Service, Production, Supply, and Distribution Online. Available at www.fas.usda.gov/psdonline/psdHome.aspx (accessed January 2007).
↵13 European Commission Directorate General for Energy and Transport, various publications. Available at http://ec.europa.eu/dgs/energy_transport/figures/pocketbook/2004_en.htm (accessed July 2006).
↵14 UNICA (Sao Paulo Sugarcane Agroindustry Union), lectures and presentations. Available at http://english.unica.com.br/dadosCotacao/estatistica/ (accessed July 2006).
↵15 This land adjustment is partial because lagged acreage and prices are present in acreage functions. The adjustment to early years’ shocks in later years would be long run, especially by year 10. However, in later years there is still a displacement of ethanol use with respect to its baseline value in the corresponding years. Hence, to that last shock (say, in year 10), adjustment is still a shortterm partial adjustment. Formally derived long-term multipliers would be based on acreage specification characterized by long-term Nerlovian price responses with no lagged variable in them.
↵16 Because of space limitations, multipliers for corn utilizations including stocks are not reported. Information can be requested from the authors.
