Removing Distortions in the U.S. Ethanol Market: What Does It Imply for the United States and Brazil?

U.S. Ethanol Demand

Total U.S. ethanol demand is divided into fuel-ethanol demand and a residual demand that consists of nonfuel alcohol use (industrial and beverage). Fuel-ethanol demand is a derived demand from the cost function for refiners blending gasoline with additives, including ethanol. Given that only aggregate data are available on U.S. motor gasoline consumption, we are constrained to model an aggregate composite gasoline production representing all types of gasoline available on the U.S. market. Let C denote the cost function for the refiners supplying all types of gasoline blended with additives, including gasoline blended with ethanol. We abstract from the time dimension when not necessary.

The cost function is written as C = C(P^US_E, P^US_O, Policy, Q^US_GS), where Q^US_GS is the refiners' output, which is the gasoline supply, P^US_E is the domestic price of ethanol, P^US_O is the U.S. price of crude oil, and Policy is federal legislation that impacts refiners' ethanol demand. Under the constant-returns-to-scale assumption, the cost function can be written as . The marginal cost (MC_G) of gasoline is constant as long as input prices are constant. Gasoline output Q^US_GS is eventually determined by the intersection of gasoline demand and MC_G at the equilibrium in the gasoline market. By Shephard's lemma, the intermediate demand for fuel ethanol, (∂C/∂P^US_E), is derived as

(1)

where E^US_F is the fuel ethanol demand in million gallons and is the derived demand for ethanol per unit of gasoline. Accounting for the specific policy interventions affecting refiners, we obtain the following equation:

(2)

where TR^US stands for the tax rebate of 51¢ per gallon that refiners get when they blend ethanol with gasoline, Mandate is the requirement of ethanol blend in percentage in certain states, and RFS denotes the renewable fuels standard created by the Energy Bill of 2005 in million gallons. In equation (2), and β₁ < 0.

The term Q^US_GD denotes the Marshallian demand for gasoline in the U.S. market, that is, the amount of gasoline consumption used in transportation in million gallons.

(3)

where P^US_G is the price of unleaded gasoline in dollars per gallon and is a function of P^US_O. The price P^US_G is included in equation (3) as final consumers see the unleaded gasoline price.³ Real gross domestic product (GDP) in 1995 U.S. dollars is denoted by GDP^US, and Pop^US is population. Consumers respond positively to a decrease in the price of the composite fuel, which is a function of the prices of gasoline and ethanol. The ethanol component of the composite aggregate fuel consumption increases as the ethanol price falls relative to the price of gasoline to capture the substitution between the types of gasoline at the gas-station pump. In equation (3), β₂, δ₂ < 0 and α₂, φ₂, ϕ₂ > 0.

In equilibrium in the gasoline market, quantity of gasoline supplied by refiners is equal to the quantity of gasoline demanded by final consumers (Q^US_GD), that is, Q^US_GS = Q^US_GD = Q^US_G. Substituting equations (2) and (3) into equation (1) yields the derived demand of ethanol evaluated at the equilibrium of the gasoline market, E^US_F*:

(4)

where Q^US_GD is defined in equation (3). At the equilibrium of the gasoline market, can be interpreted as the share of fuel ethanol in total gasoline consumption (E^US_F*/Q^US_GD).

In U.S. gasoline production, fuel ethanol is mainly used as an additive to gasoline. Thus, ethanol acts as a complementary good to pure gasoline. However, in demand, ethanol is a substitute to gasoline, through E85 cars, which run on gasoline blended with up to 85% ethanol, as well as a fuel enhancer. In this analysis, through the parameterization of equation (4), it is assumed that the complementary relationship is more dominant than the substitute relationship because currently ethanol is blended mostly at 10% and is not available in all states. Furthermore, E85 cars represent a negligible portion of the U.S. vehicle fleet. Substitution effects are currently limited but may get larger in the future if E85 cars become popular. Due to complementarity, an increase in the price of gasoline translates into a net decrease in demand for ethanol E^US_F*.

The magnitude of the complementary and substitute relationships also depends on the assumptions made about the composition of the U.S. vehicle fleet in the future. As long as the number of FFVs (E85) in the United States remains relatively small, there is only limited substitution for regular cars in terms of substituting gasoline for ethanol. Finally, to complete the specification of total ethanol demand, the residual ethanol demand is simply set up as a function of the U.S. domestic ethanol price.

U.S. Ethanol Supply

To model the domestic ethanol production in the United States, we use a restricted profit function for the ethanol plants. Profit maximization under capacity constraint yields a profit function, which can be expressed as a function of a return per bushel of corn net of energy cost. To account for the different processes of ethanol production, the relative marginal revenues from the by-products from each process are weighted by the share of production by each mill type; s_DM is the share of dry-mill production in total ethanol production, and s_WM is the share of wet-mill production. Thus, the net return per bushel of corn for ethanol plants in the U.S., π^NET, is expressed as

(5)

In equation (5), P^US_GF is the price of gluten feed in dollars per ton, P^US_GM is the price of gluten meal in dollars per ton, P^US_CO is the price of corn oil in dollars per gallon, P^US_DDG is the price of DDG in dollars per ton, and P^US_C is the price of corn in dollars per bushel. P^US_NGt is an index of the price of natural gas, which is multiplied by m = 0.0038 to scale the index to dollars per bushel of corn. The conversion rates (γ_i) are used to convert each price to dollars per bushel of corn.⁴ The ethanol production function (Y^US) is given as:

(6)

where PC denotes the production capacity in million gallons.⁵ In equation (6), α₃, β ₃, δ₃ > 0. The equation for the production capacity is PC_t = PC_t−1 · (1 + g_t), where g_t is the growth rate of this capacity and t denotes the time period. We model g_t as

(7)

where E(D_E) is defined as the expected future demand that investors project for ethanol and 35¢ per bushel is the trigger fixed cost of building a new ethanol plant expressed per bushel of corn.⁶ As a proxy for E(D_E), we used a five-year average of ethanol demand projected five years into the future provided by the EIA's Annual Energy Outlook 2006. In equation (7), α₄, β₄, δ₄ > 0. U.S. production capacity has been increasing at an unprecedented pace, which prompted us to set up the above capacity equation.

U.S. Ethanol Stocks

Next, the ending stock (ES^US_t) equation is expressed as follows:

(8)

where α₅, β₅ > 0 and δ₅ < 0.

U.S. Ethanol Trade

The trade equations consist of export and import equations. Because U.S. ethanol exports are small, they are kept constant. The U.S. ethanol imports are the sum of imports from CBI countries (M_CBI) and imports from other countries (M_Other). For the CBI countries, there is a tariff rate quota (TRQ) rule. The in-quota tariff rate is τⁱ, which is zero. The out-of-quota tariff rate is τ^o, which is 2.5% plus 54¢ per gallon. The TRQ is set at 60 million gallons or 7% of U.S. consumption, whichever is greater. The CBI import equation is based on the relative world ethanol price to the U.S. price as follows:

(9)

where Capacity is the CBI countries' maximum capacity of their dehydration plants, tc is the transportation cost, and α₆, β₆ > 0. Here θ₆ and φ₆ are transmission coefficients, where θ₆ = 0.85 and φ₆ = 0.70. They are included to account for the transaction costs between firms, the time lag between contracts and delivery, and the daily volatility in ethanol prices that are not captured in the annual price data. The coefficients are based on historical relationships between domestic and world ethanol prices. Transportation cost (tc) is 11¢ per gallon.⁷ For CBI, tc also includes the transformation (dehydration) costs. In the above equations, τ^A = τⁱ if M_CBI ≤ TRQ, and τ^A = τ^o if M_CBI > TRQ.

Imports from other countries are subject to the out-of-quota tariff rate. The import equations for other countries are as follows:

(10)

where Supply is the sum of production, beginning stocks, and imports from CBI countries, Demand is the sum of consumption, ending stocks, and exports, and φ₇ = 0.70.

Through equations (9) and (10), we see that when the tariff is not prohibitive, import demand is positive making the domestic U.S. price dictated by the world ethanol price through a price transmission equation. However, since the United States is a large country importer, the world ethanol price is also affected by U.S. import demand. When the tariff is prohibitive and there are no imports from other countries, the domestic U.S. price is solved endogenously within the model, equating supply to demand. To account for this, we construct a price-switching regime. The domestic price of ethanol can be solved either endogenously (P^Endogenous_E) or it can be a price transmission from the world price of ethanol. If P^Endogenous_E >P^W_E · (1 + τ^o) + tc, the domestic ethanol price equals P^W_E · (1 + τ^o) + tc. If P^Endogenous_E < P^W_E · (1 + τ^o) + tc, the domestic ethanol price is P^Endogenous_E.

Brazilian Ethanol Demand

In Brazil, the ethanol demand is divided into anhydrous and hydrous ethanol demand, as they respond to different economic incentives depending on the three types of vehicles (alcohol, flex-fuel, and gasohol cars). The alcohol vehicles use only hydrous ethanol, the gasohol vehicles use only anhydrous ethanol (blended with gasoline), while the FFVs can use both hydrous ethanol and anhydrous ethanol (blended with gasoline). Therefore, we model anhydrous ethanol demand (E^B_A) and hydrous ethanol demand separately (E^B_H), where total ethanol demand in Brazil E^B_Total equals (E^B_H + E^B_A).

The behavioral equations for anhydrous and hydrous ethanol consumption are:

(11)

(12)

where P^W_E represents the price of Brazilian anhydrous ethanol in reals per gallon, which is also the world ethanol price. Although there is a price for hydrous ethanol, only one price for ethanol, namely anhydrous, is used in both demand equations. The two prices are highly correlated as, in general, the price of anhydrous ethanol is the price of hydrous ethanol plus the cost of dehydration, which is assumed constant. The price of gasoline in reals per gallon is given by P^B_G, and I is an interaction term that is equal to P^B_G times the ratio of FFVs in the total vehicle fleet. The term F^B denotes the number of FFVs in the vehicle fleet in units. The terms GDP^B and Pop^B are the GDP in 1995 reals and population for Brazil, respectively. Blend is the mandate of 20–25% of ethanol blended with gasoline. In equation (11), β₇, δ₇, φ₇ < 0 and α₇, ϕ₇, γ₇, η₇ > 0. In equation (12), β₈ < 0 and α₈, δ₈, φ₈, ϕ₈, θ₈, σ₈ > 0.

The interaction term I is used to capture the higher demand responsiveness of FFVs to changes in the price of gasoline. As the number of FFVs increase in the projection period, demand for both anhydrous and hydrous ethanol becomes increasingly responsive to the change in the price of gasoline. In the case of anhydrous demand, as the price of gasoline rises, the demand for ethanol declines as FFVs substitute hydrous ethanol for gasoline blended with anhydrous ethanol. Conversely, for the demand for hydrous ethanol, if the price of gasoline increases, the demand increases as FFVs increase their use of hydrous ethanol relative to anhydrous ethanol blended with gasoline.

Brazilian Ethanol Supply

In modeling the supply of ethanol in Brazil, the link between sugar and ethanol markets is critical as both ethanol and sugar are produced from sugarcane in Brazil. Therefore, the derived demand for sugarcane that goes into ethanol production comes from the profit-maximization problem of sugarcane producers.

In the Brazilian sugar model, we obtain the area harvested for sugarcane in Brazil (AHA^C_t) from the cane producers' profit maximization, which is given as

(13)

where P^B_S,t is the price of sugar in reals per ton (the Caribbean f.o.b. raw sugar price × the exchange rate), and P^B_AC,t is the price of competing crops (namely, soybeans) in reals per ton.⁸ In equation (13), α₉, β₉, δ₉, φ₉, γ₉ > 0 and ϕ₉ < 0. Sugarcane production is area harvested for sugarcane multiplied by the yield. In the ethanol model, the behavioral equation for the share of sugarcane in ethanol production (S^C_E) is given by

(14)

where α₁₀, β₁₀ > 0. Sugarcane used in ethanol production equals S^C_E multiplied by total sugarcane production. Ethanol production equals sugarcane used in ethanol production × the conversion rate of 22.98 gallons per metric ton of sugarcane.

Brazilian Ethanol Stocks

The ethanol ending stock (ES^B_t) equation is constructed as

(15)

where α₁₁, β₁₁ > 0 and δ₁₁ < 0.

Brazilian Ethanol Trade

Net exports are derived as a residual, that is, equal to production plus beginning stocks minus consumption minus ending stocks. Although there is an ethanol import tariff in Brazil, it is not incorporated into the model, as Brazil is a net exporter of ethanol.

U.S. Crops Model

The U.S. crops model is a partial-equilibrium econometric model, which includes behavioral equations that determine crop acreage planted, domestic feed, food and industrial uses, trade, and ending stocks. The model solves for the set of prices that brings supply and demand into balance in all crop markets.⁹ The U.S. crops model is divided into regions with equations for each crop grown within each region.¹⁰ For crops with by-products, behavioral equations for the by-products are also included.

Estimated regional net returns for modeled crops determine the area planted under each crop in each region. The submodels for all the crops and by-products in the U.S. crops model follow the same general framework. In the case of corn, the expected net return in a given region is determined by the expected corn price, the expected yield, variable production expenses, and the expected loan deficiency and counter-cyclical payments. Corn area planted is a function of the expected net returns for corn and competing crops, decoupled payments, and the area allocated to the Conservation Reserve Program. Corn production is given by corn area harvested multiplied by corn yield.

Total corn use is the sum of feed use, seed use, food use, high fructose corn syrup (HFCS) use, ethanol use, exports, and stocks. Feed use is based on feed use per grain-consuming animal unit and an estimate of grain-consuming animal unit for the U.S. livestock sector. Corn exports are estimated through a reduced form equation that reflects the response of international grains and oilseeds markets to changes in U.S. grain and oilseed prices.

International Sugar Model

The international sugar model is a nonspatial, partial-equilibrium econometric model including 29 countries and ROW. The model specifies sugar production, use, and trade between countries expressed in raw sugar equivalence. The general structure of the country submodel includes behavioral equations for area harvested, yield, production for sugar beet and sugarcane on the supply side, and per-capita consumption and ending stocks on the demand side. The world sugar price (Caribbean f.o.b.) is determined by equating excess supply and excess demand across countries. Through price transmission equations, the domestic sugar price of each country or region is linked to the world price through exchange rates and other price policy wedges.

The general framework for each country submodel includes an equation for area harvested in each country similar to equation (13). Area harvested is then multiplied by yield to obtain sugarcane or beet crop production. Total sugar production is obtained by converting beet production and raw cane production into raw sugar equivalent. On the demand side, sugar consumption in each country depends on the price of raw sugar and per-capita income. Ending stocks depend on beginning stock, consumption, and the price of raw sugar.

Scenario 1: Impact of Trade Liberalization

With the removal of the tariffs, the U.S. domestic ethanol price decreases by 13.6%, which results in a 7.2% decline in ethanol production and a 3.8% increase in consumption. The lower domestic price leads to a rise in the share of fuel ethanol in gasoline consumption by 3.7%. Given the lower domestic ethanol price, consumers are substituting gasoline blended with ethanol for gasoline blended with other additives. The removal of trade distortions in the United States and the corresponding higher U.S. ethanol demand increases the world ethanol price by 23.9% on average over the simulation period (table 1). With the open border, the world ethanol price is $1.57 per gallon and the U.S. domestic ethanol price is $1.68 per gallon. Thus, the U.S. domestic ethanol market now sees the world ethanol price plus the transportation cost. Because of their significantly smaller domestic ethanol markets, the impact of demand by countries such as Japan and the EU-15 on the world ethanol price is marginal relative to the United States and Brazil.

U.S. net imports of ethanol increase by 199% (figure 2). Given that net imports make up only 5.3% of domestic consumption in the baseline, the large increase in net imports in the first scenario translates to a 15.1% share of imports in total U.S. domestic consumption. Since the tariffs are removed, Brazil can now export ethanol to the United States directly without having to go through the CBI countries to avoid the tariff as is the case under the closed border. Therefore, trade diversion occurs, and we assume that ethanol imports from CBI countries decline to zero and that Brazil makes up for the decline with higher exports to the United States. However, it is possible that some ethanol imports may continue to come from CBI countries. If ethanol prices are competitive, CBI countries may use domestic feedstock to produce ethanol for exporting to the United States.

The lower domestic production of ethanol translates into a reduced demand for corn in the United States. Thus, the corn price declines by 1.5% on average relative to the baseline. Given the decline in corn used in ethanol production, the production of by-products decreases, by 7.1% for DDG, and by 1.7% each for gluten feed, gluten meal, and corn oil. The reduction in the production of DDG increases the price of DDG by 0.7%. The price of gluten meal increases by 0.9% because of its production decline. However, the prices of gluten feed and corn oil fall by 0.5% and 0.2%, respectively, as the impact from the lower corn price, which decreases the cost of production, exceeds the impact from lower production.

Brazil responds to the higher world ethanol price by increasing its production by 9.1% on average relative to the baseline. Total ethanol consumption decreases by 3.3% and net exports increase by 64%. The higher ethanol price leads to an increase in the share of sugarcane used in ethanol production by 4.9%. This results in less sugarcane used in sugar production, which decreases sugar production in Brazil. The lower supply of Brazilian sugar leads to an increase in the world raw sugar price by 1.8% on average.

Since trade barriers are usually set in place to protect the domestic industry and keep competing foreign products out, the results show that the import tariff in the United States is achieving the desired goal by keeping Brazilian ethanol out of the U.S. domestic market. However, trade barriers contribute to price volatility in the U.S. ethanol market, which can be dampened by opening the borders. The world ethanol market deepens with the removal of the tariff. With the increase in imports in the U.S., this provides less volatility in ethanol prices in the domestic ethanol market. Hence, trade is a way to mitigate price surges from shocks within the United States from ethanol expansion.

Scenario 2: Impact of Trade Liberalization and Tax Credit Removal

In this scenario, in addition to the removal of the tariffs in the United States, the 51¢-per-gallon federal tax credit is removed. The simulation results are presented in table 3. U.S. ethanol consumption decreases by 2.1% as the tax credit for refiners is removed. The U.S. domestic ethanol price decreases by 18.4%, which is a larger drop compared to the first scenario, since U.S. ethanol consumption is lower as consumers now see a higher effective price. In response to the lower domestic price, production decreases by 9.9% compared to the 7.2% decline in the first scenario. The world ethanol price increases by 16.5% on average compared to the baseline, which is also lower than in scenario 1. The impact on the corn by-products market is similar to that in the first scenario, particularly in direction.

Welfare Analysis Results

To calculate welfare changes, each scenario is evaluated relative to the baseline using change in producer surplus, consumer surplus, and tariff revenue. The welfare impacts are computed for the U.S. and Brazilian ethanol industries, the U.S. corn market, and the Brazilian sugarcane market.

Table 4 shows the welfare impacts of U.S. trade liberalization on U.S. and Brazilian consumers and producers. With trade liberalization, U.S. ethanol and corn producers lose because of the decline in the U.S. domestic ethanol and corn prices. Alternatively, U.S. ethanol consumers experience a gain in consumer surplus. The loss of tariff revenue totals $74.6 million. Brazilian ethanol producers benefit from the increase in the world ethanol price while Brazilian consumers lose a part of their consumer surplus. The higher demand for ethanol increases the sugarcane price in Brazil, which in turn translates into an increase in the producer surplus of sugarcane producers.

Table 5 shows the welfare impacts of the removal of the tariff and tax credit. In this scenario, the decrease in U.S. domestic ethanol price results in a decline in the producer surplus of U.S. ethanol producers. The producer surplus for corn producers also declines as the U.S. corn price falls. The change in U.S. consumer surplus is affected by the decline in the U.S. domestic ethanol price and the increase in the effective ethanol price (domestic price minus tax credit) that the ethanol consumers see. The second effect comes from the removal of the tax credit, which is no longer passed on to the consumers. When computing the change in consumer surplus, both effects are taken into consideration. Thus, ethanol consumers see a net decline in their consumer surplus. Given the higher world ethanol price, Brazilian ethanol and sugarcane producers see a net gain while ethanol consumers experience a loss in consumer surplus.

Sensitivity Analysis

We conduct sensitivity analysis and rerun the two scenarios in a series of alternative assumptions on prices responses of supply and demand.¹⁶ We first double the original price response elasticities in supply, demand, and inventory equations for all countries in the ethanol model. We then decrease them by half with respect to their original values.

In the first scenario, when the elasticities are doubled and U.S. trade distortions are removed, the model becomes more price responsive. The trade policy reforms induce larger marginal changes in supply, demand, trade, and the world ethanol price. The increase in ethanol demand in the United States is larger. This results in a larger increase in net imports of ethanol and a larger increase in the world ethanol price compared to the original scenario. In response, Brazil expands its ethanol production beyond the levels in the original scenario, with a larger decline in Brazilian total ethanol consumption. Consequently, Brazilian net exports of ethanol increase by a more significant amount. We find that the same holds true for the second scenario when elasticities are doubled.

With the halving of elasticities the tendencies are reversed with the responses being smaller than they were in the original scenarios. As in the doubling of the elasticities, the qualitative results of the original analysis are maintained. The direction of changes is unaffected by changes in elasticities with the exception of the world sugar price when elasticities are halved. The world sugar price declines instead of increasing as it did in the original scenarios. In the first scenario, the removal of the trade barriers induces a smaller increase in the world ethanol price and, with elasticities halved, the increase in Brazilian ethanol production is smaller than in the original scenario. Brazilian ethanol production increases by 5.5% instead of 9.1% so that the increase in the share of sugarcane in ethanol production is 2.4% instead of almost 5% in the original scenario 1. Since the elasticities are unchanged in sugarcane production, sugarcane production responds to the higher ethanol price and more sugarcane goes into sugar production. The higher supply in Brazilian sugar induces the reduction in the world sugar price. In the original scenario with higher elasticities, the increase in ethanol production is larger, therefore more sugarcane went into the production of ethanol than sugar such that the supply of sugar fell and the world sugar price increased.