Impacts of the European Biofuel Policy on the Farm Sector: A General Equilibrium Assessment
Abstract
We assess the impacts of the European indicative biofuel policy on the farm sector with a farm-detailed computable general equilibrium model. Our simulations suggest that most of the biodiesel demand will be satisfied by imports while the bioethanol demand will be satisfied mostly by domestic production. We also show that the downstream livestock sectors are never negatively affected. Finally, the positive farm income effect is rather robust to our modeling assumptions: our central estimate is a 3.2 billion farm-income increase. However, the transfer efficiency of this policy is invariably limited.
Like many other countries, European Union (EU) Member States are currently contemplating stringent policy instruments to boost their production and use of biofuels to the detriment of imported fossil fuels. The recent significant rise in the real world price of oil contributes to legitimizing this policy for oil net-importing countries (Rajogopal et al., 2007). The current EU biofuel policy is mostly motivated as a way to address the adverse environmental effect of fossil oil consumption (greenhouse gas emissions). However, the environmental efficiency of the biofuel solution is highly disputed, with opponents claiming that biofuel production requires nearly as much nonrenewable energy as it saves in transport activities (Doornbosch and Steenblik). There is no disagreement that this policy will allow the farm sector to be supported by offering new outlets for farm products and thus stimulating farm prices and incomes. For instance, the European Commission (EC) estimates that nearly 200,000 farm jobs will be created with this biofuel solution, which in turn will generate positive social induced effects in poor and remote rural areas.
In the complex debate on the relevance of the EU biofuel policy, the main purpose of this article is to offer a new numerical evaluation of the effects of the EU biofuel policy on EU markets of agricultural and food products as well as on farm incomes. The extent to which this policy will benefit EU farmers may be lower than expected for two reasons. On the one hand, this new demand for agricultural products may not be completely fulfilled by domestic agricultural production. Biofuel refineries may rely on imports from the world market, particularly in the case of biodiesel production that is not so highly protected (by tariffs) as bioethanol production. Indeed, the EC evaluation assumes that nearly one-third of EU consumption of biofuel will be imported. On the other hand, the EU biofuel policy will likely strengthen prices of arable crops and consequently downstream agricultural (livestock) sectors may suffer through an increase in production costs. In the case of the U.S. biofuel policy, Elobeid et al. indeed find such negative effects for the U.S. livestock sectors due to a significant increase of corn prices. More generally, induced effects on the livestock sectors (and ultimately, on the entire farm sector) are numerous and pull in opposite directions. For instance, the expected increase in arable crop production may intensify competition on land usages with a possible decrease in pastureland. However, this effect may be counterbalanced by the possibility of growing energy crops on set-aside land. Moreover, the expected increase in arable crop profitability may lower the opportunity costs of some shared farm-specific primary factors of production (such as machinery or labor) in mixed farming systems.
In order to capture all these conflicting effects, we use an original Computable General Equilibrium (CGE) model of the EU-15 economy where the farm sector is finely detailed (in terms of product coverage, behavioral specification, and policy instrument modeling). This model is first used as a projection tool in order to determine the evolution of the EU farm sectors and markets without the biofuel policy. We then simulate the EU biofuel policy as an exogenous increase of (public-financed) demand of both bioethanol and vegetable oils. Relative to the EC evaluation, our main contributions are first to estimate rather than assume the share of imported biofuel, second to estimate induced effects on livestock sectors, and third to estimate production and employment effects consistently within a unique modeling framework.
As expected, the simulation results show significant positive effects on the arable crop sectors with significant price and production increases. Despite the possibility of growing energy crops on set-aside land, the new demand for vegetable oils will be partly satisfied by increased imports from the world market. In contrast, the new demand for bioethanol will be fully met domestically. This will induce a large drop in EU cereal exports. Quite surprisingly, we observe that livestock domestic prices decrease slightly and production expands marginally. It appears that three marginal effects push in the same direction and together they reverse the small negative feed cost effect on livestock profitability. These effects are first, a small increase in animal fat prices following substitution between all fats at the human demand; second, a significant increase of the price of organic fertilizers following the expansion of arable crop production; and third a lower opportunity price of farm specific factors used in these livestock sectors due to their imperfect mobility. Finally, our results show a positive effect on farm jobs. Our figure (the creation of 43,000 farm job) is much less optimistic than the EC evaluation because the biofuel inflow is partly diluted in higher variable production costs, partly transmitted to world markets, and finally partly capitalized in increased land values.
It is no surprise these impacts are sensitive to modeling assumptions about use of set-aside land for energy crops, reactions of foreign producers-consumers on biofuel world markets, and responses of agricultural production capacity in the EU (captured by our price elasticities of farm supply). However, our estimates on the benefits of the EU biofuel policy for the whole farm sector are rather robust thanks to compensating cross-market effects. On the methodological front, our results logically underline the usefulness of modeling all linkages operating both through commodities and through primary factor markets, as well as of modeling the structural determinants of agricultural supply functions.
Modeling Framework
The CGE Approach
There are two major benefits of using a CGE modeling framework rather than a standard partial equilibrium analysis for the assessment of the EU biofuel policy on the farm sector. The first benefit relates to the modeling of (farm) supply functions with both the specification of globally regular functional form for production technologies and the imposition of accounting identities (namely the zero profit condition, which imposes that farm receipts are equal to all farm expenditures including the rewards to primary factors of production). This type of modeling ensures that, in the case of large price or quantity changes, specified behavioral equations are still consistent with the underlying profit maximization assumption. The second benefit is the joint and consistent estimation of impacts on product and primary factor markets. In contrast, the EC evaluation, for instance, relies on two modeling frameworks with differing assumptions. The first is the use of a partial equilibrium model to simulate the impacts on agricultural markets and prices with some given macroeconomic assumptions (on the gross domestic product [GDP], GDP price index, labor costs, etc.). The second is the use of an input-output model to simulate these macroeconomic effects, including employment effects, where agricultural prices are assumed to be fixed. The CGE model used in this article consistently includes all these effects.
CGE models are often criticized on the ground that they are too aggregated and thus unable to represent the internal complexity of sectors/policies. This argument does not apply to our CGE model as it has product coverage similar to that in agricultural partial equilibrium models. More precisely our CGE model distinguishes thirty-two agricultural products, thirty farm-processed products, ten farm input goods, and finally only two goods/services for the rest of the economy (see table 1). We built the underlying Social Accounting Matrix relying mostly on Eurostat databases and World Trade Organization (WTO) notifications as far as the agricultural sectors are concerned and with the Global Trade Analysis Project (GTAP) database otherwise.
| Sectors | Commodities | |
|---|---|---|
| Agriculture | ||
| Sector | Agricultural subsectors | |
| Agriculture | Soft wheat | Soft wheat |
| Barley | Barley | |
| Maize | Maize | |
| Rape | Rape | |
| Sunflower | Sunflower | |
| Soya | Soybean | |
| Protein crops | Protein crops | |
| Sugar beet | A&B sugar beet, C sugar beet | |
| Fodder | Fodder on arable land | |
| Grass | Grass | |
| Poultry | Poultry, organic nitrogen, organic phosphate, organic potassium | |
| Pigs | Pigs, organic nitrogen, organic phosphate, organic potassium | |
| Laying hen | Eggs, poultry, organic nitrogen, organic phosphate, organic potassium | |
| Dairy cows | Bovine cattle, raw milk, calves, dairy cows, organic nitrogen, organic phosphate, organic potassium | |
| Suckling cows | Bovine cattle, calves, suckling cows, organic nitrogen, organic phosphate, organic potassium | |
| Beef calf | Beef calf, organic nitrogen, organic phosphate, organic potassium | |
| Calf rearing | Bovine cattle, heifers, bulls and steers, organic nitrogen, organic phosphate, organic potassium | |
| Heifers | Bovine cattle, dairy cows, suckling cows, organic nitrogen, organic phosphate, organic potassium | |
| Bulls and steers | Bovine cattle, organic nitrogen, organic phosphate, organic potassium | |
| Sheep and goats | Sheep and goat milk, sheep and goat animals | |
| Fruits vegetables | Fruits, potatoes, and vegetables | |
| Other agricultural activities | Other agricultural products | |
| Food processing | ||
| Meat industry | Bovine meat, pig meat, poultry meat, veal, sheep and goat meats, carcass meals, animal fats | |
| Dairy industry | Butter, skimmed milk powder, cheese from cows, cheeses from sheep and goat, whole milk powder, fluid milk, other dairy products | |
| Compound feed industry | Compound feed | |
| Cereal processing industry | Grains bran, corn gluten feed, isoglucose, other cereal processed products, bioethanol | |
| Oilseed crushing industry | Rape oil, sunflower oil, soybean oil, rape cake, sunflower cake, soybean cake, palm oil | |
| Sugar industry | A&B sugar, C sugar, sugar beet pulp, molasses | |
| Agricultural stationaries | ||
| Mono product sectors supplying | Mineral nitrogen, Mineral phosphate, Mineral potassium, pesticides, veterinary products, fish meals, other energy rich feed, other protein rich feed, other feed ingredients, seeds | |
| Other sectors | ||
| Food retail trade | Food retail trade services | |
| Other sectors | Other products and services | |
The Supply Module for EU Farm Products
The impacts of the EU biofuel policy on the whole farm sector depend first on its capacity to increase domestic production. We model this supply in an original way thanks to the detailed coverage of farm and food products. For instance, the feed module, which is at the heart of the interaction between arable crop and livestock activities, captures the technological relationship between energy rich and protein rich ingredients. Rather than the standard Constant Elasticity of Substitution (CES) function, we use a latent separability approach to represent the substitution patterns between these ingredients (Gohin). The allocation of primary factors of production also plays a significant role in this interaction between arable crop and livestock activities. As far as land is concerned, we adopt an imperfect mobility structure specified with a two level Constant Elasticity of Transformation (CET) function with a first level between pastureland and arable land and a second level between all different arable land uses. The set-aside land usable for energy crop production is modeled as a fixed percentage (the set-aside rate) of the aggregate arable land. We also assume an imperfect mobility of labor and capital across agricultural and nonagricultural sectors. In order to represent the multiproduct nature of agricultural firms and simultaneously avoid the allocation of labor and capital (including overhead expenses) to each farm enterprise, we specify a revenue function in the same vein as Peterson, Hertel, and Preckel. This modeling implies, for instance, that the reward to labor and capital is not the same over all farm activities. Finally, the availability of domestic production to satisfy the biofuel demand will also depend on the possibility of eventually increasing yields per hectare (if land supply is not sufficient). In CGE models, these yields are endogenously determined and depend on substitution patterns between land and other inputs, including chemicals and fertilizers (mineral and organic). Again, a latent separability structure determines these patterns.
The Demand Module for EU Farm Products
By increasing demand of agricultural products, the biofuel policy is likely to affect other demands, notably food consumption. Our model assumes that agricultural products are mostly processed domestically by food industries according to fixed proportions technologies (cereal exports are obviously a notable exception). For instance, oil seeds are processed by the oil and fat industry into vegetable oils and meals with linear technological relationships. The food products are then traded or sold domestically to food retailers and from there to consumers. At the final demand level, we assume high substitution between vegetable oils, intermediate substitution between them and other animal fats (butter), and finally limited substitution between all fats and other food products.
The Trade Module for EU Farm Products
The expected positive impacts of the EU biofuel policy on the whole farm sector also depend on the evolution of trade flows. In our model, we start with the traditional (CES) Armington specification for differentiated products like poultry meat or cheeses. However, most products are defined at a detailed level so that we assume that imports and exports (for instance, of corn, sugar, oilmeals, etc.) are of the same quality (this specification is detailed in Gohin, Guyomard, and Le Mouël). The trade flows that will take place also depend on the “net” capacity of foreign producers to supply the European markets. In that respect, we assume that the EU is a major player on world food markets and thus the net import supplies of farm/food products are positively sloped functions with respect to the corresponding world prices. In most single-country CGE models, the net import supply functions are (log linear) reduced forms that depend only on the price of the concerned product (see, for instance, de Melo and Tarr). This specification assumes a constant price elasticity, which is the default assumption in non-CGE models. However, this specification does not allow the country to pass from a net importer to a net exporter position. Moreover, it captures the substitution relationships at work in foreign countries between products (for instance, between vegetable oils) imperfectly while these are acknowledged in multicountry models. Ideally, one should develop a multicountry model with detailed coverage of all products but such a project may be confronted with data availability issues. The solution adopted in this article is to improve the specification of CGE-type net import supply functions by assuming the existence of leading products. For instance, we assume that soybean oil is the reference vegetable oil on world markets. Accordingly, the world prices of vegetable oils are linked to the world price of soybean oil through price transmission elasticities, which are calibrated, based on the simulation results of Dronne and Gohin. Finally, the world price of soybean oil is determined by the net trade flows of all vegetable oils. This new specification of import supply functions allows the EU to eventually become a net importer/exporter of some oils and simultaneously better capture the substitutions operating in foreign countries. We implement the same trade specification for the group of oil meals (with soybean meal as the leading product and including the Corn Gluten Feed) and the group made of sugar and bioethanol.
The No Biofuel Benchmark
Main Assumptions
The quantitative impacts of any policy critically depend on the situation that would prevail in its absence. In that respect, the recent observed years are not adequate benchmarks for the following reasons. First, the production/consumption/trade of biofuels has already started in some EU countries (significantly in Germany and Sweden). Second, the EU biofuel directive provides an indicative target of a 5.75% market share for biofuels in the overall transport fuel supply for the year 2010 and not for today. Moreover, the EC analysis (EC) of current market development concludes that achieving this target is unlikely without new measures at that time. Third, the last wave of Common Agricultural Policy (CAP) reforms (during 2003–5) is likely to alter the equilibrium of agricultural markets and incomes.
Accordingly, we define a benchmark situation aimed to project the EU economy to 2015 with the following main assumptions. We assume that EU demand and production of biofuels are zero. We make this assumption so that we can subsequently measure the effects of the full EU directive and not simply part of it.
On the supply side, we assume that farm input productivities increase annually by 1.25% in the cereal sectors, by 1% in other arable crop sectors and by 0.75% in livestock sectors. Primary factor productivities in other sectors, including food industries, also increase by 1.2% annually. On the consumption side, we assume that butter consumption decreases each year by 0.5%, compensated for by an equal increase in vegetal oils. Regarding macroeconomic conditions, we assume that the euro stabilizes at a rate of 1.2 to the US dollar. Labor and capital endowments increase by respectively 0.25% and 1.2% annually. Farmland decreases by 0.25% per annum.
In our large country, open economy CGE model, we need to formulate assumptions on the evolution of the world market conditions that domestic agents are facing. That depends, for instance, on the evolution of the climate as well as on the future productivity gains that some countries may reach (like new EU Member States). Our strategy is to use available projections of world prices with some care so that they are consistent with our other assumptions. In particular, most recent projections made by the FAPRI show EU biofuel production and consumption lower than the EU 5.75% indicative target. To avoid this inconsistency in the construction of our baseline, we use the FAPRI projections of world prices made in 2005 where the EU biofuel policy was not taken into account (FAPRI). However, these 2005 projections do not recognize the current development of biofuels in the United States. There is wide recognition that this has had some impact on world market prices but as usual, the extent varies considerably between studies. For instance, the projections of the U.S. corn price for 2015 ranges from 2.56$/bushel (Walsh et al.) to 3.35$/bushel (USDA). One critical assumption behind these results lies in the availability of second-generation biofuels. In this article, we assume that the world price of corn in 2005 is 20% higher than the level projected in 2005. We also increase the projected world prices of soybean products by 10% and of wheat by 5% to reflect the increased competition for land in the United States. The percentages are in the average of available estimates of U.S. biofuel policy.
The last assumptions concern the evolution of agricultural policy instruments. We introduce the CAP reforms of 2003 and 2005 with the (partial) decoupling of direct payments, the reduction of the intervention prices of dairy products and sugar. We assume a fixed 10% mandatory set-aside rate. Finally, we assume that the new single farm payment has a limited production effect (due to wealth, dynamic, and capital market imperfections effects) calibrated to a 7.5% market price support in case of arable crops.1 Even if the EU does offer to lower import tariffs and stop export subsidies in the context of WTO negotiations, we do not introduce these changes in the baseline because this offer is only conditional. On the other hand, we assume that C sugar exports are no longer feasible following the WTO panel ruling on European sugar.
Main Characteristics of the Benchmark
We project that the EU cereal markets in 2015 will be characterized by significant unsubsidized exports of soft wheat to the world market (18.6 MT). However, the corn market is protected from out-of-quota imports by an import tariff (table 2). Yields per hectare (ha) increase moderately to reach 7.0 T/ha for soft wheat and 9.4 T/ha for corn because the CAP reforms dampen the effects of exogenous technical changes. The soft wheat area stabilizes at 13.6 Mha while we project a nonmarginal reduction of the sugar beet area due to the 2005 sugar reform and the WTO ruling on the C sugar production. The EU still uses some export subsidies in that sector.
| Soft Wheat | Corn | Rapeseed | Sunflower | Sugar Beet/Sugar | |
|---|---|---|---|---|---|
| Area (millions ha) | 13,642 | 3,897 | 2,177 | 1,532 | 1,548 |
| 14,332 | 3,779 | 3,744 | 2,001 | 1,758 | |
| 5.1% | −3.0% | 71.9% | 30.6% | 13.5% | |
| Yields per ha (T/ha) | 7.0 | 9.4 | 3.4 | 1.5 | 60.0 |
| 7.0 | 9.4 | 3.4 | 1.5 | 59.7 | |
| 0.6% | 0.2% | 0.0% | −0.6% | −0.5% | |
| Production (millions T) | 94,918 | 36,582 | 7,414 | 2,336 | 13,877 |
| 100,342 | 35,530 | 12,738 | 3,032 | 15,682 | |
| 5.7% | −2.9% | 71.8% | 29.8% | 13.0% | |
| Total demand (millions T) | 76,143 | 38,886 | 8,617 | 4,869 | 13,099 |
| 91,330 | 37,847 | 14,197 | 5,535 | 14,938 | |
| 19.9% | −2.7% | 64.8% | 13.7% | 14.0% | |
| Net exports (millions T) | 18,577 | −2,500 | −469 | −2,115 | 426 |
| 8,630 | −2,500 | −469 | −2,115 | 392 | |
| −53.5% | 0.0% | 0.0% | 0.0% | −8.0% | |
| Domestic prices (€;/T) | 110 | 134 | 211 | 225 | 404 |
| 122 | 142 | 294 | 301 | 404 | |
| 10.8% | 5.7% | 39.0% | 33.7% | 0.0% | |
| World prices ($/T) | 132 | 116 | 253 | 270 | 281 |
| 146 | 116 | 352 | 361 | 281 | |
| 10.8% | 0.0% | 39.0% | 33.7% | 0.2% |
As far as the vegetable oil complex is concerned, the rapeseed area is projected at 2.2 Mha. We project the EU to be a small net importer of rapeseed and a significant importer of sunflower seed and soy. The net trade situation is similar for oil meals with the EU importing 17.1 MT of soybean meal (table 3). On the other hand, the EU is projected to be a net exporter of rape oil as well soybean oil. However, imports of palm oil outweigh these exports so that the EU is a net importer of all vegetable oils.
| Rape Oil | Rape Meal | Soya Oil | Soya Meal | Palm Oil | |
|---|---|---|---|---|---|
| Production (millions T) | 3,454 | 4,057 | 2,183 | 9,658 | |
| 5,679 | 6,678 | 2,182 | 9,653 | ||
| 64.5% | 64.5% | 0.0% | 0.0% | ||
| Total demand (millions T) | 2,447 | 4,620 | 1,852 | 25,798 | 3,819 |
| 6,175 | 7,245 | 4,020 | 24,258 | 5,609 | |
| 152% | 56.8% | 117.0% | −6.0% | 46.9% | |
| Net exports (millions T) | 1,011 | −104 | 272 | −17,100 | −3,819 |
| −363 | −67 | −1,579 | −15,544 | −5,609 | |
| −136% | −35.4% | −681% | −9.1% | 46.9% | |
| Domestic prices (€;/T) | 499 | 111 | 495 | 183 | 476 |
| 737 | 82 | 662 | 179 | 622 | |
| 47.6% | −26.0% | 33.6% | −2.5% | 39.1% | |
| World prices ($/T) | 590 | 132 | 586 | 216 | 563 |
| 871 | 97 | 782 | 211 | 783 | |
| 47.6% | −26.0% | 39.6% | −2.5% | 39.1% |
As far as animal products are concerned, we emphasize the fact that the huge reduction of the butter intervention price has a negative effect on domestic production and positive effect on the domestic consumption (table 4). However, these two effects are not sufficient to resolve the EU butter market imbalance and the EU price is still higher than the world price. In the same table, one can observe that the EU is projected to be a net importer of beef (309 MT) with some out-of-quota imports. In addition, our model suggests that the EU will be a significant net exporter of pork and a small net exporter of poultry meat.
| Pork | Poultry Meat | Beef | Butter | Compound Feed | |
|---|---|---|---|---|---|
| Production | 19,139 | 8,888 | 6,505 | 1,846 | 121,032 |
| (000 T) | 19,250 | 8,947 | 6,573 | 1,846 | 122,122 |
| 0.7% | 0.6% | 1.0% | 0.0% | 0.9% | |
| Total demand | 18,269 | 8,728 | 6,822 | 1,507 | 121,032 |
| (000 T) | 18,311 | 8,731 | 6,830 | 1,536 | 122,122 |
| 0.2% | 0.0% | 0.1% | 1.9% | 0.9% | |
| Net exports | 835 | 128 | −309 | 340 | |
| (000 T) | 890 | 201 | −266 | 310 | |
| 6.9% | 57.0% | −12.5% | −6.6% | ||
| Domestic prices | 2,652 | 2,823 | 3,812 | 2,462 | 260 |
| (€;/T) | 2,613 | 2,786 | 3,794 | 2,462 | 260 |
| −1.2% | −1.2% | −0.5% | 0.0% | −0.2% | |
| World prices | 3,182 | 3,864 | 1,928 | 1,705 | |
| ($/T) | 3,135 | 3,837 | 1,909 | 1,753 | |
| −1.2% | −0.7% | −1.0% | 2.8% |
The EU Biofuel Policy Experiment
Assumptions
Many policy instruments can be used to promote biofuel and they have their own market and welfare effects (de Gorter and Just). We assume that EU governments ensure a domestic demand of biofuel with tax reductions and/or consumption mandates.2 Using the EC figures translated to the EU-15 level, we assume that this demand amounts to 13.8 Mote of biofuels in 2015. The ratio between diesel and petrol consumption is 55:45 and we assume the same ratio for biodiesel and bioethanol. Adjusting for the energy content of biofuels, we assume that the new demand of biodiesel is equal to eight MT and the new bioethanol demand equals 7.3 MT.
On the supply side, we first assume, for methodological convenience, that only first-generation biofuels will be commercially available in 2015. We also assume that the changes induced on the producer prices by these biofuels will not alter available technologies for farming (like the adoption of new genetically modified organisms). Accordingly, our price impacts should be considered as maximum ones. The new demand for biodiesel may be fulfilled by the esterification of domestic/imported rape/soy/palm oil. This process generates glycerine as a coproduct, which is not included in our model. However, Smeets, Junginger, and Faaij argue that the long run value is low. Accordingly, we focus the remaining discussion on the vegetable oils. The substitution possibilities between these different vegetable oils in biodiesel production cannot be derived from past evolutions. Moreover, technical regulations differ with the usage of biodiesel. If it is blended with diesel, then the current EU regulation requires the use of rapeseed and sunflower oils. If it is used in its pure form, then soy or palm oils are also possible. The choice between blended or pure consumption heavily depends on the taxation policy implemented by Member States. For instance, Germany fully exempted from taxation the B100 (pure form) until August 2006 but is presently gradually reintroducing taxes. In the analysis conducted by the EC, the assumptions are that the share of rape oil in biodiesel production will not be able to fall below 50% and the remainder is accounted for by soy and palm oils. In this article, we adopt these figures and assume that 50% (4 MT) will be made from (domestic/imported) rape oil, 25% (2 MT) from imported soybean oil and 25% (2 MT) from imported palm oil. Since there are significant substitution elasticities for EU food consumption as well as for foreign net supply functions between all vegetable oils, it appears that these share assumptions have a limited impact.3
In the same vein, the new demand for bioethanol may be satisfied by imported and/or through the processing of different domestic raw materials (cereals and sugar beet). Bioethanol is mainly made from corn in the United States and sugar cane in Brazil but soft wheat is presently the main feedstock used in EU-15 bioethanol production. The EC projects that EU bioethanol production will be mainly from soft wheat, followed by corn, and finally from sugar beet. We assume that in 2015 the relative production costs will determine the feedstock used in EU bioethanol productions. In that respect, table 5 provides the input-output coefficients of bioethanol production associated with the main EU feedstocks. These figures show that there are no major technological differences between soft wheat and corn processing into bioethanol. We assume that these coefficients will remain in 2015 following Smeets, Junginger, and Faaij. Given the EU corn price is higher than the EU soft wheat price, corn will be used only if the processing costs of bioethanol from corn are lower than costs of processing soft wheat. This is not the case because one plant can produce both kinds of bioethanol. So, we exclude EU production of bioethanol from corn. The last column of table 6 reports the input-output coefficient of bioethanol production from sugar beet. The processing costs of this industry are hard to establish because the sugar sector has been regulated for decades. The complex EU sugar policy results in the cross-subsidization of out-of quota sugar beet and processing costs (Gohin and Bureau). We assume that the sugar sector will use the bioethanol outlet to pursue this cross-subsidization strategy. Practically we assume that one MT of bioethanol will always be made from domestic out-of-quota (previously C) sugar beet.4 This assumption is intended to reflect the technological constraints for the processing of perishable sugar beets. The remaining 6.3 MT of bioethanol are either imported or produced from the processing of soft wheat depending on their domestic prices.
| Bioethanol from Soft Wheat | Bioethanol from Corn | Bioethanol from Sugar Beet | |
|---|---|---|---|
| Raw materials (T) | 3.5 | 3.2 | 12.9 |
| Coproducts | |||
| Distillers dries grains with solubles (T) | 1.5 | 0.9 | 0.75 |
| Wheat bran (T) | 0.1 | ||
| Beet pulp and molasses (T) | |||
| Processing costs | 407€; |
- Source: Smeets, Junginger, and Faaij
| Wheat Sector | Rapeseed Sector | Pig Sector | Milk Cows Sector | All Farm Sectors | |
|---|---|---|---|---|---|
| Labor/capital unit | 1 | 1 | 1 | 1 | 1 |
| Opportunity price | 1.146 | 1.708 | 0.941 | 0.994 | 1.012 |
| (Index) | 14.6% | 70.8% | −5.9% | −0.6% | 1.2% |
| Labor/capital use | 6,642 | 848 | 5,531 | 17,486 | 107,432 |
| (Millions €; equivalent) | 6,994 | 1,325 | 5,564 | 17,486 | 108,614 |
| 5.3% | 56.3% | 0.6% | 0.0% | 1.1% |
The domestic production of bioethanol from soft wheat is a new activity introduced in our model. We calibrate its input-output coefficients with figures reported in the table 6. We make the assumption of constant returns to scale for bioethanol processing technologies. In the initial stage of development, firms may experience increasing returns to scale. However, we assume that the projected volume of production allows firms to exhaust these. Moreover, this assumption is consistent with the assumption made in other food processing industries. Finally, the unit production costs will change with the prices of soft wheat, wheat bran, and Dried Distillers Grains with Solubles (DDGS) byproducts. The latter is assumed to have the same role as Corn Gluten Feed (CGF) in animal feeds.
Bioethanol can also be imported and will compete with domestic production if the domestic price, net of the specific tariff of the 243€/T, is lower or equal to the price of domestic bioethanol. The future world price of bioethanol is also highly uncertain. In recent years, it has fluctuated between 270$/T and 700$/T for several reasons. In particular, it partly depends on the fossil oil price because the two commodities are substitutes in some foreign countries (Brazil, for instance). The evolution of the Brazilian Real also has had some effects on trade flows and world prices. In medium-term projections, it is important to abstract from short-term effects and to concentrate on structural determinants. According to OECD production cost estimates (OECD 2006) Brazil is able to produce bioethanol from sugar cane at 276$/T. According to Elobeid et al., a 60$ per barrel crude oil price translates into a 462$/T bioethanol price which is roughly the level in the last FAPRI projections. We use the latter as our starting point. We furthermore assume that the world price of bioethanol will evolve with EU imports, if any, by specifying a joint net import supply function with sugar (see the modeling section). This specification implies that greater imports of bioethanol by the EU contribute to increases in the world prices of both sugar and bioethanol because there is some substitutability between these two productions in foreign countries (notably Brazil).
Our last assumption relates to the possibility of growing energy crops on set-aside land. We assume that only half of mandatory set aside is available for arable crop cultivation due to agronomic conditions. We also assume that the Blair House limit on oilseeds production cultivated on set aside has no market effects because farmers are allowed to grow cereals/sugar beet on set-aside land. Finally, we check ex post if the biofuel domestic production requires more than the “recultivated” set aside land and the energy area limit benefiting from the 45€;/ha specific payment. If it is, then we assume that this energy crop payment is a pure transfer to farmers without any market effects. If not, we specify it as an input subsidy to soft wheat use in bioethanol production.
Results
On the biodiesel sector
The new demand for rape oil for biodiesel production (4 MT) is considerable with respect to the initial domestic demand (2.4 MT). Our model suggests that this new demand will be mostly satisfied by larger domestic production (2.2 MT or 64.5%) and by the EU moving from being a net exporter (1 MT) to a net importer (0.4 MT) (table 3). In addition, food use of rape oil decreases by 0.3 MT because the domestic and world prices increase considerably (by 47.6%). Indeed, the final world price amounts to 871$/T, a level never reached in the last decade. However, this figure makes sense when compared to the even higher price projections last made by FAPRI. The decrease of rape oil food consumption (10%) is significant but finally limited by the fact that the prices of other vegetable oils increase quite similarly. For instance, we estimate that the world price of soybean oil increases by 33.6%, while that of palm oil increases by 39.1%. If the food consumption of most vegetable oils decreases, that of soybean oil increases by 9% thanks to improved relative prices. However, total vegetable oil consumption decreases by 2.8%.
On the supply side, the increased production of rape oil induces a proportional increase of rape meal production. These additional byproducts are sold on the domestic market due to a significant price decrease (26.0%). The imports of rape meal from the world markets become marginal. The domestic consumption of rape meal increases considerably to the detriment of that of soybean meal. As expected, the EU becomes less dependent on the imports of soybean meal (a reduction of 9.1%).
These oil and meal price effects lead to a significant increase in the prices of oilseeds, by as much as 39% for rapeseed, so that the final price reaches 294€;/T (table 2). It appears that production only increases through an expansion of areas (1.6 Mha). On the other hand, yields are unchanged because three opposite effects interact. On the positive side, the output price increase favors the intermediate consumption of variable inputs (like fertilizers and pesticides) and hence raises yields per hectare. On the negative side, our model assumes that the additional land devoted to rape production is less productive due to decreasing marginal productivities. Moreover, the increased rapeseed production requires more labor and capital. According to our modeling assumptions, this is only possible with higher returns to these factors because they are imperfectly mobile. Indeed, our simulation suggests that the unit opportunity price of the labor/capital bundle for rape production (70.8%) increases more than the land unit price in that sector (21.6%) and consequently the labor/capital mix becomes a “limiting” factor for yield growths (table 6). Obviously, more land is also required for the production of rapeseed but the possibility of cultivating set-aside land softens the competition for this factor. Hence, land is relatively less scarce with the consequence that yields per ha tend to decrease on that ground (more on this point in the sensitivity analysis).
Finally, we point out that the EU sunflower sector also expands (by 29.8%) despite the huge increase in the rapeseed sector. In fact, the production price of sunflower seed follows the rapeseed price since their oils/meals are good substitutes. This positive own price effect dominates the negative cross-price effect of rapeseed.
On the bioethanol sectors
In our experiment, we assume that one MT of bioethanol will be made with EU sugar beet (or equivalently 1.8 MT of out-of-quota sugar). In this respect, the main contribution of our model is to compute the production costs of this volume. In our benchmark, the unitary production cost of sugar equals 376€;/T, hence a level slightly lower than the new intervention price. In order to secure their production quota, we assume that the sugar sector (both the sugar beet farmers and sugar processing firms) cross subsidizes the out-of-quota production as in the past (Gohin and Bureau), and that the bioethanol demand allows them to sell this additional production. However, the cross-subsidy level is now lower because it is assumed a linear function of the gap between the (lower) intervention price and the unit production cost. Moreover, additional production of sugar beet and sugar implies higher production costs (again because of the decreasing marginal productivity of land) as well as higher production of byproducts (sugar beet pulp). Quite surprisingly, the sugar beet pulp domestic price is stable. This is an energy rich feed ingredient, which is modeled as a net substitute to cereals and a net complement to protein rich ingredients. The price decrease of the latter (see above) and the price increase of the former (discussed later) both contribute to increase the domestic use of sugar beet pulp in feed rations and hence to stabilize its price. All these effects combine to a final out-of-quota sugar production cost of 336€;/T and a bioethanol (made from sugar) production cost of 611€;/T (table 7). This production cost is slightly lower than the net of tariff price of imported bioethanol.
| Quantity (000 T) | Domestic Price (€;/T) | |
|---|---|---|
| Bioethanol from domestic wheat | 6,300 | 620 |
| Bioethanol from domestic sugar beet | 1,000 | 611 |
| Imported bioethanol | 0 | 628 |
| Domestic demand | 7,300 | 619 |
The remaining demand for bioethanol is fully supplied by domestic production. Production costs of bioethanol made from soft wheat (620€;/T) are again projected to be lower than imported ones despite a considerable increase in the soft wheat domestic price. The demand for soft wheat represents 21 MT or equivalently 27.3% of initial total domestic demand. It appears that this demand “only” increases by 19.9% (15 MT) because there are simultaneously reductions in the feed (11.6% or 4 MT) and food (4.6% or 2 MT) demands. Domestic production of soft wheat significantly increases (by 5.7% or 5 MT). However, the market balance of soft wheat is mainly obtained by a huge reduction in soft wheat exports to the world markets (by 53.5% or 10 MT). The shock on the soft wheat market is proportionally lower than that on the rape market and, accordingly, price effects are lower. The processing of soft wheat to bioethanol induces an increase in the production of wheat bran and DDGS. In our modeling framework, we assume that the former is an energy rich feed ingredient while the latter has a composition similar to CGF and hence is a net substitute to other protein rich feed ingredients. Accordingly, the price of wheat bran increases slightly while the price of CGF (DDGS) decreases by 27%.
The positive effect on the soft wheat market obviously has some impacts in other cereal markets. For instance, the increase in the area devoted to soft wheat production is partly compensated for by a decrease in corn area. In this market, the main effect is a corresponding decrease in domestic production, which leads to an increase in the domestic price.
On the downstream livestock sectors
As expected, the experiment leads to an increase in cereal prices and a decrease in the price of protein-rich byproducts. The final effect on the price of livestock feed rations depends on the initial shares of these feed ingredients as well as on the possibility of changing the feed ration composition. Our model results show that the second effect dominates. For instance, the production cost of compound feed decreases slightly (by 0.2%). This compound feed cost effect is counterbalanced in the cattle sectors by the higher production costs of fodders (2%); the pasture area is nearly unchanged (−0.6%). Quite surprisingly, our model suggests that all animal/meat productions will expand marginally (from 0.6% for pork to 1% for beef). This is in sharp contrast with Elobeid et al. where they evaluate a much larger shock on the U.S. biofuel. Three marginal effects contribute to our result. First, the domestic price of animal fats increases following those of vegetable oils. Consequently, for a given animal price and prices of the other processing inputs, the meat industry may sell meats at a lower price and hence stimulate domestic consumption. Second, as we explained earlier, arable crop production requires more pesticides and fertilizers. There are substitution possibilities between organic and mineral fertilizers and hence there is additional demand for organic fertilizers. For instance, we evaluate an increase in the organic nitrogen domestic price of 6.3%. This effect also supports the livestock sectors and productions. Finally, our labor/capital mobility modeling allows for cross-subsidization between farm activities. For instance, we evaluate that the labor/capital unit opportunity price in the pig sector decreases by 5.9% (table 6). Intuitively, this means that (mixed) farmers get lower rewards for pig production but, nevertheless, earn more from their farms thanks to their arable crop activities (we discuss this further in the sensitivity analysis section).
This additional production of meats leads to marginal increases in domestic consumption (at most, 0.2% for pork). In fact, the additional production is mainly exported (for pork and poultry) or displaces out-of-quota imports (for beef). Finally, the impacts of this experiment on the dairy sector are, as expected, limited because milk production is constrained by milk quotas. We observe a slight increase in butter consumption (of 1.9%) due to a substitution effect with the vegetable oil consumption.
On the whole farm sector
Our experiment leads to an increase in almost all types of farm production. This is feasible since yields per hectare slightly increase and above all because energy crops are allowed on previously set-aside land. We estimate that 2.8 Mha return to farm production and that overall 5.5 Mha are devoted to biofuel production. This is much larger than the maximum area eligible for the energy crop direct payments.
As expected, the biofuel policy supports farm incomes. The agricultural value added increases by 3.8% or €;3.3 billion (table 8). This supplementary income transmits in higher land values (17.5% or €;0.8 billion since the land unit price increases as well the volume of productive land), higher production quota rents (70% or €;0.3 billion), higher capital rewards (3.3% or €;0.7 billion), and finally higher farm labor remuneration (2.5% or €;1.5 billion). The latter explains the increase in farm employment: farm job creation amounts to 43,000 units (1.3% of initial level). This figure is significantly lower than the EC estimate of nearly 200,000 jobs created. Our tentative explanation is as follows. In our modeling framework, we assume that the EU governments inject €;10.5 billion in purchasing biofuel. The benefit to the EU farm sector (in terms of value added) is “only” €;3.3 billion. Around €;4.8 billion are captured by foreign agents (with lower cereal exports and higher vegetable imports) and €;2.5 billion cover processing costs (of soft wheat to bio ethanol). Accordingly, “only” 30% of the public inflow really does support farm incomes and only half of this supports farm labor. Indeed, nearly 45% of the public inflow goes externally. By comparison, the EC evaluation study assumes that 27% of biofuel demand will be imported. Probably here lies the main source of the difference.
| Land Unit Price (€;/ha) | Agricultural Value Added (Millions Euros) | Farm Labor (000 Persons) | |
|---|---|---|---|
| Baseline | 43 | 85,966 | 3,182 |
| Policy scenario | 51 | 89,263 | 3,225 |
| Percentage changes | 17.5% | 3.8% | 1.3% |
| Sensitivity analysis | |||
| To land availability | 3.6% | 3.6% | 1.5% |
| To labor/capital mobility | 68.5% | 4.9% | 1.7% |
| To fossil oil price | 11.1% | 2.9% | 1.1% |
Sensitivity Analysis
The figures reported so far depend on our modeling assumptions. In this section, we test their robustness to three critical assumptions. The first is on the land available in the EU for energy crop productions. The second is the capacity of the EU farm sector to attract labor and capital from the nonagricultural sectors and hence to respond to the surge in domestic demand. The last is about the fossil oil world price and related world bioethanol price. We focus the analysis on the farm income impacts.
On the Availability of Land
In our central experiment, we assume that for agronomic reasons energy crop production is possible on only half of set-aside land. In this sensitivity analysis, we assume that it is possible to cultivate all set-aside land. In a nutshell, the main results are equal to our central estimates. The additional cultivated areas are much less productive such that average yields decrease. These results are not unexpected. For instance, Elobeid et al. also get limited effects of the addition of Conservation Reserve Program (CRP) area.
On the Mobility of Labor and Capital to Farm Sectors
In our central results, it appears that agricultural production is partially constrained by the availability to attract labor and capital. Let us now assume a long-term horizon where these two primary factors of production are perfectly mobile among all activities. This is an extreme assumption not usually adopted in other CGE simulations. However, it is interesting to give an upper boundary to the farm impacts. Before analyzing these results, this modeling alternative implies that the domestic supply of farm products is more sensitive to price shocks and the main limiting factor is then land.
As expected, the supply response by EU farmers is now much larger. For instance, domestic rape oil production increases by 154% (71.8% in our central estimate). Interestingly, the impacts on livestock sectors are unchanged because all the indirect effects we previously identified move together. For instance, the price of feedstocks increases less or decreases more. However, the unit cost of labor and capital used in these sectors no longer declines.
Are farm incomes better supported in this alternative scenario? The response is unambiguously yes but by a rather limited amount. In this alternative, the leakages of the biofuel benefit to the world markets are clearly lesser. But the increased EU farm production requires more inputs (for instance, the expenditure on chemical inputs increases by €;0.8 billion compared to the central estimates) and the support is more highly capitalized in land values. Finally, we find that farm job creation now amounts to 55,000 (compared to 43,000 in the central case).
On the Fossil Oil and Bioethanol World Prices
Our central estimates are based on the assumption that the world price of bioethanol is equal to 462$/T when the world price of oil is equal to 60$ per barrel. All economic studies recognize that there are great uncertainties about the evolution of these two prices, even at the 2015-year horizon. In the EC evaluation study, the oil price is assumed to vary between 48$ and 70$ per barrel. In this sensitivity analysis, we assume this oil price to stabilize at 45$ per barrel and by consequence that the world price of bioethanol (without any EU imports) is 346$/T. We finally assume that EU governments will fully arbitrate between all types of bioethanol due to increased competition represented by potential imports.
As expected, this alternative assumption has a major impact on the bioethanol sectors: domestic production made from sugar beet disappears, that from soft wheat amounts to only 3.3 MT and now imports total four MT. The specific tariff of 243€;/T on bioethanol obviously prevents more massive imports. As a result, the domestic production of soft wheat expands less than in the central case. At the same time, exports of soft wheat decrease by a smaller amount. Basically, the EU imports of bioethanol are partly compensated for by the higher net trade positions in cereals. Impacts on the livestock sectors are very similar to the central estimates. Finally, the effective agricultural support of the EU biofuel policy is logically more limited with the creation of only 34,000 farm jobs.
Concluding Comments
At present, the efficiency of the European indicative biofuel policy is highly disputed as a means of tackling both EU energy security and climate change issues. However, there is no doubt that, if fully enforced, it will support the EU farm sector as a provider of biofuel raw materials. The purpose of this article is to offer a numerical evaluation of this potential benefit with a farm-detailed CGE model. Two side effects may indeed lower the expected positive direct effect on the EU arable crop sector. First, this new (policy supported) demand may be satisfied by imports. Second, downstream (livestock) sectors may suffer from an increase of their production costs. In our standard case, the simulation results suggest that a great part of the biodiesel demand will be satisfied by (marginally taxed) imports while the demand for bioethanol will mostly be satisfied by domestic production (thanks to significant import taxes). This share between imported/domestic biofuel is logically quite sensitive to the assumptions on world prices and on the EU farm supply responses. On the other hand, all simulations reveal that downstream livestock sectors are not negatively affected. Finally, given our assumptions the positive farm income effect is also robust due to compensating cross-market effects. The transfer efficiency of this policy is, nevertheless, invariably limited; as a consequence, the EU biofuel policy cannot be justified only on those grounds.
There are obviously several extensions to this article that may better address this vast biofuel issue. For instance, with the same model, it would be interesting to investigate trade policy shocks on the new biofuel markets compared to “old” agricultural markets. In terms of model improvements, we believe that the priorities are first to develop an energy module in order to allow more insights into the global welfare effects and second to enlarge the model in both country coverage (extension to new EU Member States and main players on the world markets) and product coverage (introduction of second-generation biofuel products).
Acknowledgments
The author acknowledges financial support by the Agricultural Trade Agreements (TRADEAG) project, funded by the European Commission, DG Research (Specific Targeted Research Project, Contract n°513666). The author thanks Y. Dronne, H. Guyomard as well as the two anonymous referees and the editor for the constructive comments. The author is solely responsible for the contents of this article.
