- Email: [email protected]
Economic analysis of an aviation bioenergy supply chain
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser
Economic analysis of an aviation bioenergy supply chain ⁎
Jeffrey J. Reimera, , Xiaojuan Zhengb a b
Oregon State University, USA Oregon State University and Dell Corporation, USA
A R T I C L E I N F O
A BS T RAC T
Keywords: Aviation Biofuels Camelina Energy policy General equilibrium Oilseeds
This study develops a regional economic model that can mimic stages of a potential aviation fuel supply chain based on the oilseed camelina. A general equilibrium model is developed that accounts for key sectors of the supply chain, and is parameterized using detailed data for the Pacific Northwest region of the United States. Model scenarios are developed to evaluate the potential effectiveness of policies that would promote the development of the supply chain. While the existing low price of conventional fuel makes camelina-based fuel economically infeasible at present, the supply chain could be viable if consumers, e.g. airline passengers, are willing pay more to use the biofuel, such as if they perceive an environmental benefit from it. Alternatively, if use of the oilseed is consistent with the energy policy priorities of policymakers, one of the following approaches could potentially work: a 17% subsidy on the alternative fuel, a 20% tax on the conventional fuel, or a combination 9% subsidy on the alternative and 9% tax on the conventional fuel. Implications for economic efficiency, regional employment, and economic welfare are quantified.
1. Introduction This study concerns the development of a new bioenergy supply chain for the aviation sector of the Pacific Northwest region of the United States. The focus is on the policies and price signals necessary to induce a new market channel to develop, and on the implications for the efficiency of resource allocation. A general equilibrium economic model with explicit representation of key sectors is developed and applied to aviation fuel made from oilseeds such as camelina. It can be processed into a high grade bio-based jet fuel for military and commercial purposes using approaches such as a Hydroprocessed Esters and Fatty Acids (HEFA) process (Bauen et al. [1], Shonnard et al. [2], IATA [3], and Natelson et al. [4]). Interest in this potential bioenergy resource has grown in recent years. For example, the number of airlines which have used a blend of this alternative fuel for a commercial flight has risen to 21 (IATA [3]). Interest is also strong among military agencies, such as the U.S. Air Force, and operators of commercial airports, including those of Portland and Seattle. These institutions are reportedly interested in using U.S.-sourced biofuels, again for supply diversification, and perhaps for perceived environmental reasons, such as a reduction in net greenhouse gas emissions. Additional support has come from the U.S. Environmental Protection Agency who has declared that aviation biofuel – biojet, in industry parlance – is eligible for Renewable Identification Numbers that can be traded on the open market. This
⁎
has further solidified the long term prospects for this new bioenergy supply chain (IATA [3]). At the other end of the supply chain are oilseed processors and refiners, and ultimately, farmers who would need to plant the appropriate oilseeds. Camelina, for example, is argued to be ideal in the Pacific Northwest, where its cultivation need not displace other crops grown for food when incorporated into a wheat-fallow rotation (Stein [5]). Although currently grown on less than 50,000 acres in the United States, acreage could potentially be expanded by three to four million acres without adversely impacting food prices (EPA [6], Winchester et al. [7]). It also has low input requirements, is suitable for marginal soils, and has natural competitiveness with weeds (Putnam et al. [8], Hulbert et al. [9]). Despite the theoretical potential and technical plausibility of this alternative supply chain, the economics of how the market might work are less clear. There are public good aspects to the problem, meaning that the private sector, by itself, may not make the necessary investments. Interactions up and down the supply chain must be considered, including the incentives of a range of different economic agents who may have competing interests. Farmers, for example, are incentivized by a high price for camelina, while refiners are incentivized by a low price. Whether prices can be found that satisfy all supply chain participants, simultaneously, is an empirical question that can be addressed in part by economic modeling. To shed light on this problem, a general equilibrium economic
Corresponding author. E-mail addresses: jeff[email protected] (J.J. Reimer), [email protected] (X. Zheng).
http://dx.doi.org/10.1016/j.rser.2016.12.036 Received 9 May 2015; Received in revised form 20 April 2016; Accepted 6 December 2016 1364-0321/ © 2016 Published by Elsevier Ltd.
Please cite this article as: Reimer, J., Renewable and Sustainable Energy Reviews (2016), http://dx.doi.org/10.1016/j.rser.2016.12.036
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
similar shocks, following pioneering techniques introduced in Valenzuela et al. [13]. This aspect of the study is novel in the general equilibrium literature and assists the reader gauge the reliability of the results. The study also complements work by Diebel and Ball [14], Walsh [15], and Stein [5]. Building on the first two studies, Stein estimates potential supply curves for camelina in Oregon, Washington, Idaho, and Montana with a break-even price approach. Stein's results suggest that given current market conditions, the supply in the Pacific Northwest will not be enough to meet biofuel targets without an increase in promotion, most likely by government. This finding serves as one motivation for the study at hand. Unlike the approach proposed here, Stein uses a partial equilibrium framework and looks mainly at adoption by farmers, as opposed to the links between different market players. He does not examine policy, the potential for inter-regional trade, labor markets, or any of the macro-economic aspects considered here. The study is also complementary to McCullough et al. [16], who assess the potential for biofuels production in Washington state. As with many other studies, they find that some form of government intervention is likely to be needed to “jumpstart” the system. The study at hand is distinct in that it considers a larger geographical area, recent developments with respect to aviation biojet demand, and a wider variety of policies at different stages of the supply chain. To preview some of the conclusions, this study finds that the new bioenergy supply could be made feasible, or at least price competitive, through a number of particular policy mechanisms. Unless a segment of consumers can be induced to pay extra for use of biojet, one of the following policies would provide the incentive structure to facilitate market development: a 17% subsidy on the alternative fuel, a 20% tax on the conventional fuel, or a combined 9% subsidy on the former and 9% tax on the latter. The latter, so-called tax-cum-subsidy approach provides a “double dividend” in the sense that it combines a targeted tax on conventional fuel, with a targeted subsidy on the alternative fuel, so as to be relatively revenue-neutral and to have smaller distortions on other markets. The remainder of the study is organized as follows. The next section describes the modeling approach taken, addressing the general problem of how to model a market channel for a new product that is not yet established. Since the model shares many features from other established computable general equilibrium models, changes that are unique to this study are emphasized, with separate sections describing the regional adaptation of the modeling framework, the data, calibration of model parameters, and validation of key parameters. Subsequent sections, in turn, examine how the scenarios are developed, and the results. Five different cases are analyzed, each of which shed light on viable options for development of a regional supply chain for biofuels. The final section concludes.
model is developed in this study that simultaneously accounts for a number of sectors key to the analysis. This general type of model is part of a venerable tradition in economics, and has been used to study biofuel energy policy (Cansino et al. [10]). It combines mathematical representation of the incentives and constraints faced by all of the economic actors in the system. To make the model useful for policy, it must be made to replicate, that is, its parameters must be calibrated such that the model can reproduce baseline data for the region in an appropriate recent time period. Parameters of the model are calibrated primarily using highly detailed IMPLAN [11] data for the Pacific Northwest region (Oregon, Washington, and Idaho) of the United States. These data account for trade and transfers between hundreds of economic sectors within the region, including between the aviation, processing, and farming sectors. Once parameterized and validated, the model is used to illustrate the mechanisms that could make this new bioenergy supply chain economically feasible. These include non-policy as well as policy instruments such as subsidies and taxes, including a tax-cum-subsidy approach. A suitably motivated government could conceivably take the estimates of this study and use them to guide policies that would facilitate the emergence of an aviation biofuel supply chain in the Pacific Northwest. Unlike related studies such as Natelson et al. [4], Tabatabaie and Murthy [12], and Winchester et al. [7], the present study does not emphasize the engineering aspects of biofuel production, the change in greenhouse gas emissions for conventional versus alternative fuels, or quantify the potential security benefits of using alternative fuels. For example, Natelson et al. [4] provides a techno-economic analysis of alternative fuel production based upon oilseeds such as camelina. Relative to this study, they provide much more detail on the engineering requirements for alternative refinery operations, but less detail on different aspects of the supply chain, including the welfare consequences of alternative tax and subsidy policies that could make biojet comparable in price to conventional fuel. The study at hand additionally emphasizes labor market impacts and a range of macro-economic outcomes. A special focus is on whether increased demand for oilseeds can be met by local sources, as opposed to sources outside a region. One of the selling points of oilseeds – camelina in particular – is that it can be a “home grown” source of fuel with very low opportunity costs of production. Yet even if the oilseed feedstock is produced by local farms, this may not imply that the processing will occur within the region. Likewise, if the processing is done within the region, the oilseed feedstock could perhaps be procured most efficiently from outside the region, such as from Canadian farmers, thereby diluting the “home grown” nature of the energy resource from the viewpoint of U.S. policymakers. To address these issues, the study explicitly models the separate stages of a vertical supply chain, including sourcing from inside and outside a region of the United States, and from outside the country. In this sense, the study complements Winchester et al.’s [7] analysis of alternative fuels for the aviation industry. Their study uses a global general equilibrium model, and considers other advanced biofuels such as biomass-based diesel and grain-based ethanol cellulosic fuels. The study at hand considers a particular supply chain, distinguishing oilseed production from processing and refining, and emphasizes the Pacific Northwest region, where oilseeds can be integrated into a wheat-fallow operation with little or no impact on wheat supply (Stein [5]). This study also considers different policy instruments, including a tax-cum-subsidy that can minimize the distortion of economic incentives in multiple markets. Another unique aspect of this study is that model parameters are validated using historical price and quantity data for oilseeds. This goes beyond simply calibrating the model to IMPLAN input-output social accounting data. In this study, actual year-to-year price movements are compared to those which arise from model simulations representing
2. Modeling approach and data A diagram of the methodology for the study is in Fig. 1. At the top of the figure are two boxes; the left represents the theoretical economic model, which is a mathematical representation of the behavioral objectives and responses of all decision-makers including government. The theoretical model accounts for numerous economic features including taxes, subsidies, and the other policy instruments to be discussed below. The theoretical model is combined with a social accounting matrix (SAM) to create a calibrated economic model. The SAM itself is created using IMPLAN (2012) data to numerically represent trades and transfers between producers, buyers, households, and government. The SAM also traces the flows of tax revenues, government spending, exports, and imports. Finally, the calibrated model is shocked with a number of policy simulations to generate new equilibrium values of all endogenous variables. These are compared to the original baseline values to evaluate different policy measures. 2
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
Table 1 Sectors.
Fig. 1. Methodological overview.
A more detailed exposition of the elements in Fig. 1 is now provided. We focus first on the theoretical general equilibrium model. It is a modified version of the approach of Reimer, Weerasooriya, and West [17], with multiple markets linked in supply-demand equilibrium through intermediate input use, factors of production (labor and capital), and other linkages. It is a system of highly non-linear equations representing optimizing households and firms, inter-household and government transfers, savings and investment, government, and trade with other regions. A government sector is represented as collecting taxes through various instruments, and receiving transfers from other institutions. It then transfers this revenue back out to other institutions within the model. A representative household receives income from labor, capital, inter-household transfers, federal and state government transfers, and investment income. This household spends money on commodities, inter-household transfers, federal and state government taxes, and investment. Producers of goods choose their level of operation to maximize profits or minimize costs using constant returns to scale production technology. Production factors include labor and capital, paid according to their respective marginal productivities, as well as intermediate inputs. This last feature is important as it captures the vertical supply chain of production. The production function assumes fixed amounts of intermediate input use, along with flexible use of capital and labor. This flexibility is modeled using a constant elasticity of substitution (CES) functional form, similar to other instances in the model in which an agent has substitution possibilities. A full mathematical presentation of the model, as described above, can be found in Löfgren et al. [18]. There are four general modifications from that approach which are made in this study, highlighted below. The remainder of this section can be skipped by readers less interested in the technical aspects of the general equilibrium model.
Number
Name
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Oilseed farming Grain farming Petroleum refining Oilseed processing and refining Animal production including cattle poultry eggs All other crop farming Construction Mining and quarrying Utilities Wholesale and retail trade Processed food Food purchased away Alcohol and cigarettes Manufacturing Housing Transportation by air All other transportation Education Health Other services Miscellaneous
themselves are reported in Table 1. The top row of Fig. 2 depicts the aviation and animal production sector, which purchase either oil (for fuel) and meal (for livestock feed). The third row depicts a composite processed oilseeds sector. Activity in this sector takes place inside or outside the region according to a constant elasticity of substitution function. If oilseeds are processed within the region, the oilseed feedstock itself can come from within the region or outside the region (Fig. 2, rows 4 and 5). Unprocessed oilseeds from outside the region come either from the rest of the United States, or the rest of the world. Regional oilseed farming can be utilized within the region, or sold outside the region. These individual substitution possibilities are each characterized by a CES form.
2.2. Data and parameterization The model described above is a simultaneous system of equations written for GAMS software. The joint equilibrium values of the endogenous variables is calculated using the PATH non-linear programming solver. Model parameters are calibrated with IMPLAN data [11], which in raw form distinguish more than 400 distinct sectors of the economy. To make the analysis practical, these are aggregated to the 21 sectors in Table 1. The first step is to create a Social Accounting Matrix (SAM), which is a highly detailed account of monetary flows between economic agents, commodities, factors, and institutions including government. Model parameters are calibrated to ensure that baseline values of endogenous variables match the SAM regional information. During calibration, all prices are set to unity and the base year factor levels and SAM flows are substituted into the model as equilibrium values of model variables. The process is similar to maximum likelihood estimation with one observation. Model parameters calibrated through this process correspond to the year 2011. The model also contains free parameters set by the user. These are set to values based on econometric estimates or otherwise employed in the literature (Rose and Liao [19], Hertel et al. [13], McCullough et al. [16]). For example, the elasticity of substitution between capital and labor in the production function is set at 2. The rest-of-world export elasticity of demand for oilseeds is set to −1.175, an estimate made in Reimer et al. [20]. Elasticities of substitution between regional and rest of U.S. supplies are from Bilgic et al. [21], and are set at 1.447 and 1.339 for unprocessed and processed oilseeds, respectively. Elasticities
2.1. Regional modification Instead of modeling a country as a whole as in Löfgren et al., a distinct region within the United States is modeled in this study. A list of sectors and related output is listed in Table 1. A representative regional household receives income from labor, capital, and transfers. It spends money on commodities, transfers, taxes, and investment. Regional firms use labor, capital, and intermediate goods to maximize profits. Regional state governments collect taxes and receive transfers from other institutions, with spending constrained to equal revenue. Key aspects of the regional supply chain are highlighted in Fig. 2, which depicts two of the 21 total sectors that are modeled, and which 3
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
Fig. 2. Sectoral flowchart.
between U.S. and the rest of world supplies are from Hertel et al. [13], and are set at 4.9 and 5.2 for unprocessed and processed oilseeds, respectively.
2.3. Model validation An especially unique aspect of this model is the validation exercise undertaken to indicate how well the model replicates historical data for the region, thereby providing a measure of confidence regarding predictions made below. This section can be skipped by readers less interested in the technical aspects of the procedure. The approach is to map out the model's response functions for relevant shocks in observable prices and quantities over time. The approach makes use of 1991–2011 oilseed price and quantity data from the United States of Department Agriculture (USDA), National Agricultural Statistics Service Crop Production, Grain Stocks, and Crop Values and USDA, Foreign Agricultural Service, Global Agricultural Trade System. Since the supply chain is not well established, data on camelina are unavailable. Data on canola are used, since it is very similar in agronomic and use characteristics with respect to camelina, it has a long historical record in the Northwest, and since it comprises much of the production considered as oilseeds within the SAM. The process of calibration can be explained by first examining Figs. 3 and 4, which report actual canola output and prices, respectively, for the U.S. over 1991–2011. There are trends in both output and prices, related to factors such as technology change (Fig. 3) or general inflation in the economy (Fig. 4). Abstracting from these trends, the associated year-to-year volatility can provide information about the extent that prices adjust to supply or demand shocks. To abstract from any less relevant trends in Figs. 3 and 4, two alternative ways of detrending the time series are used: an ordinary least squares (OLS) approach (in which only the year of an observation, and an intercept, are included on the right hand side of the model) and an autoregressive moving average (ARMA) model. A full description of these statistical models are not reported due to space constraints, but is available upon request and also in Valenzuela et al. [13]. Fitted values from the estimated models are presented visually in Figs. 3 and 4. Residuals from the output regression are used to calculate percen-
Fig. 3. Actual versus predicted U.S. canola (million pounds).
Fig. 4. Actual versus predicted U.S. canola price ($ per hundred pounds).
tage changes in output by year. These percentage changes are plugged into the model, which then generates an associated price change. The standard deviation of these price changes is 15. If output was well below trend (such as in 2009 and 2010), then the supply curve for the product would shift to the left, and price would rise. On the other hand, 4
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
gallons of fuel per year. Using the seed to oil conversion factor, the corresponding amount of meal that would be produced is 988.2 million pounds, which is assumed to sell at $0.15 per pound. The average yield in the Pacific Northwest is 1600 pounds per acre [24], so at the 60,000,000 gallon target identified above, approximately 908,250 acres would be required. This is much less than the three to four million acres of fallow land each year that could reportedly support camelina [6,7]. Before implementing the above changes, it is worthwhile to note the baseline values of inter-industry relationships for the year 2011 in the IMPLAN data assembled for the region. These are reported under the heading “input-output relations” in Table 3, which also reports some of the main simulation results to be discussed shortly. In the 2011 baseline, there was no biojet used in the regional aviation industry, but there was $1347.3 million worth of conventional jet fuel use. The regional livestock industry used $34.4 million worth of oilseed meal and $264.3 million worth of other feeds (e.g., grains).
when output was above trend for a year (such as in 2002 and 2003), then the supply curve would shift to the right, and price would fall. The extent to which the price will actually change is an empirical question, depending in part on price elasticities of demand in other regions (e.g., whether extra supply can be absorbed) as well as the ease of trading oilseeds across regions (which is influenced by barriers to trade such as transport costs and tariffs). If the model is calibrated correctly, then the magnitude of the simulated price changes is expected to match the magnitude observed in the historical record. The magnitude of historical price changes can be characterized by the second moment of the residuals from the regressions in Fig. 4. The standard deviation of residuals from the simple trend line regression of actual oilseed prices is 15.4 under the ARMA approach, and 25.1 under the OLS approach. The latter predicts higher volatility since it forces a linear relationship, and thus is less able to mimic year-to-year differences. These two approaches can be viewed as covering the range of the extremes, and the actual level of price volatility is expected to lie between the two values of 15.4 and 25.1. In other words, if the model is calibrated appropriately, it should generate price movements that have a magnitude within this range. As mentioned above, the simulated standard deviation of U.S. canola price changes is 15. This lies at the lower bound of the 15.4 – 25.1 range. If it was lower than this range, simulated price responses would be too moderate. This would arise, for example, if the cross-regional elasticities of substitution were too high. If the response above was higher than 25.1, then it is as if price is too sensitive in the model. In this case, some slack would need to be built into the model, by increasing the magnitude of the regional elasticities of substitution. This does not need to be done, however, as the results obtained above are not inconsistent with the historical experience.
3. Results Design and analysis of five cases is presented below. A complete description of the code for these simulations, as well as that of the model itself, is available from the authors upon request. 3.1. Demand change with no policy This first scenario is an initial exploratory analysis intended to show how increased demand for biojet will change the regional economy unaccompanied by any explicit policy change in this regard. In some ways it is a “business as usual” case by which to reference the policy simulations in later cases. A number of airlines have expressed great interest in alternative fuels, but are not willing or able to pay more than what they pay for conventional fuel (IATA [3]). One approach is to pass along the costs to customers with a program wherein consumers voluntarily pay extra, for example, to purchase carbon offsets to compensate for the greenhouse gas emissions caused by personal air travel. The significance of this is that biojet can be more expensive than conventional fuel, yet still be viable from the standpoint of airline profitability. Results for such a scenario are reported in Tables 3–6. The top of Table 3 shows that by changing the inter-industry demand schedules, biojet use in aviation rises from zero to $185 million. Meanwhile, oilseed meal use in the livestock sector rises by $149 million. There is a corresponding fall in conventional jet fuel use, and a corresponding fall in other feed use in livestock. These changes align with the targets and assumptions in Table 2. The demand changes lead to a cascade of changes throughout the regional economy. Table 3 reports changes in supply and demand, distinguishing between oilseeds as a feedstock, and processed/refined oilseeds (biojet or meal). There is a $21.3 million (4%) rise in the demand for oilseed feedstock, increasing from $527.0 million to $548.3 million. The total local demand for processed oilseeds rise by 9.8%, or $346.3 million. Most of this is met by imports to the region, which rises by 10.7% or $293.3 million. The remainder is met by the local processing/refining industry, which expands by 3.2%, or $78.2 million (any residual is due to exports, which are positive but small and therefore not reported). One implication is that a regional biofuels industry does not necessarily guarantee that local producers/refiners will have increased demand for their product; much of it may be imported from outside the region. The top part of Table 4 shows the change in prices at different stages of the supply chain, and for ancillary markets. The increased demand puts pressure on the price of biojet. With the new demand, the price of biojet is even more expensive than before, rising 0.1% from the base assumption of $3.69/gallon up to $3.71/gallon. The price of conventional fuel falls by 0.2%, from $3.06 to $3.05. This implies that
2.4. Assumptions common to all scenarios To evaluate policies for a regional biofuel supply chain, five counterfactual scenarios are developed. Assumptions common to all cases are presented in Table 2. These are based in part on information in Hodges and Rahmani [22], Schumacher [23], and Stein et al. [24]. It is assumed that the cost of feedstock is 13 cents per pound and the extractable oil is 32%. The seed to oil conversion factor is then 24.22 pounds per gallon. The cost of feedstock for oil is $3.15 per gallon. The cost of pressing seeds to oil is 54 cents per gallon, such that the total cost of oil is then $3.69 per gallon. The $3.69 estimate is consistent with related studies such as Winchester et al. [7], who estimate that HEFA jet fuel from rotation crops in the U.S. such as camelina could be produced at around $3.70 per gallon. This consistency provides a great deal of reassurance that the assumptions below are reasonable. Winchester et al. [7] also report that various rotation crops could potentially be grown on 43 million acres in the U.S. per year, yielding up to 3.2 billion gallons of renewable jet fuel. This study, with a regional focus, considers a relatively modest target of 60,000,000 Table 2 Cost analysis. Variable
Assumed value
U.S. Gulf Coast kerosene-type jet fuel spot price FOB ($/gal) Production cost of oilseed feedstock ($/pound) Oilseed extractable oil (%) Seed-oil conversion factor (lbs/gal at 32% oil content) Meal produced alongside each gallon of fuel (pounds) Cost of feedstock for oil ($/gallon) Cost of pressing seeds to oil ($/gallon) Total cost of oil ($/gallon) Target: Oilseed-based fuel per year (gallons) Corresponding amount of meal (million pounds) Price of oilseed meal ($/pound)
3.06 0.13 32 24.22 16.47 3.15 0.54 3.69 60,000,000 988.176 0.15
5
Regional demand and supply Demand for oilseed feedstock ($ mil) % Output of processed oilseeds ($ mil) % Imports of processed oilseeds ($ mil) % Demand of processed oilseeds ($ mil) %
Input-output relations Biojet use in aviation ($ mil) Conventional jet fuel use in aviation ($ mil) Oilseed meal use in livestock ($ mil) Other feed use in livestock ($ mil)
Variable
6 57.0 10.8 210.1 8.7 351.8 12.9 453.9 12.9
4.0 78.3
3.2 293.3
10.7 346.3
9.8
2420.1
2736.8
3530.3
−147
264.3
21.3
−147
149
34.4
527.0
150
185 −185
186 −175
Simulated change
Case 2: Demand change plus 17.2% biojet subsidy
0.0 1347.3
Initial
Case 1: Demand change only
Table 3 Results concerning oilseeds, farm level and processed.
8.9
10.4 312.9
0.6 283.9
0.8 13.8
4.3
−148
148
174 −254
Case 3: Demand change plus 19.5% conventional fuel tax
8.7
10.3 307.9
0.2 282.6
0.3 4.0
1.7
−148
147
172 −264
Case 4: Demand change plus 22.6% conventional fuel tax
11.0
11.7 389.4
5.0 320.2
6.3 121.9
33.3
−148
149
180 −212
Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax
J.J. Reimer, X. Zheng
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
7
3.71 0.66
3.69 0.63
Gross regional product ($ mil) % Regional govt expenditure ($ mil) % Federal government revenue ($ mil) % Federal govt expenditure ($ mil) % Payments to regional labor ($ mil) % Payments to regional capital ($ mil) % 1.8 8104.9 4.6 2460.6 1.3 4393.1 2.4 5088.1 1.8 1017.5 1.2
−0.1 −849.7 −0.5 −146.5 −0.1 −276.4 −0.1 −280.0 −0.1 −77.4 −0.1
−0.03 −28.8
0.0 −26.0
0.0 −41.1
0.0 −47.0
0.0 −40.3
0.0
176,694
188,141
185,767
278,355.1
81,699.3
11,161.2
−184.2
1.4
2.1 1175.1
2.7 5850.0
1.5 5066.0
5.3 2834.6
2.1 9326.2
12,853.9
−3704.8
−3228.9
248.2 Simulated change −713.1
10.4
Equivalent variation ($)
Case 4: Demand change plus 22.6% conventional fuel tax
2.7 1.5 22.6 3.0 3.2 4.6
3.80 0.05
0.15 3.75
Case 4: Demand change plus 22.6% conventional fuel tax
Case 3: Demand change plus 19.5% conventional fuel tax
2.3 1.1 19.5 2.7 2.7 3.9
Simulated change 0.1 −1.4 −0.2 −17.2 −0.3 −0.5
Case 2: Demand change plus 17.2% biojet subsidy
3.79 0.13
0.15 3.66
Case 3: Demand change plus 19.5% conventional fuel tax
3.06 0.00
Simulated value 0.12 3.05
Case 2: Demand change plus 17.2% biojet subsidy
Initial value 604,414
Case 1: Demand change only
0.1 −1.4 −0.2 0.6 0.0 0.0
0.15 3.05
Initial value 0.15 3.06
Case 1: Demand change only
Variable
Table 5 Macroeconomic results.
Oilseed price (%) Other livestock feed price (%) Conventional jet fuel price (%) Biojet price (%) Livestock price (%) Air transportation price (%)
Price of canola meal ($/lb) Conventional jet fuel price ($/gal) Price of biojet fuel ($/gal) Gap between the above two ($/gal)
Variable
Table 4 Results concerning prices.
0.5
0.8 446.8
1.0 2289.1
0.6 1931.8
1.9 1092.7
0.8 3379.7
4955.3
−1415.4
Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax
1.2 −0.2 8.9 −9.0 1.1 1.6
3.36 0.03
0.14 3.33
Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax
J.J. Reimer, X. Zheng
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
Table 6 Demand for labor by sector. Sector
Oilseed farming Other feeds farming Petroleum refining Oilseed processing and refining Animal production All other crop farming Construction Mining and quarrying Utilities Wholesale and retail trade Processed food Food purchased away Alcohol and cigarettes Manufacturing Housing Transportation by air All other transportation Education Health Other Services Miscellaneous
Case 1: Demand change only
Case 2: Demand change plus 17.2% biojet subsidy
Case 3: Demand change plus 19.5% conventional fuel tax
Case 4: Demand change plus 22.6% conventional fuel tax
Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax
Initial value 0.2 112.1 425.1 77.6
4.8 −10.1 −2.5 8.3
Simulated change (%) 12.8 −9.6 −2.1 22.4
−6.8 −19.0 −35.4 1.7
−8.5 −20.3 −39.5 0.7
3.6 −14.1 −19.3 13.0
650.2 4565.7 9471.4 468.5 2912.8 37,079.9
0.0 0.1 0.0 −0.3 0.0 0.0
1.6 0.3 0.0 0.0 0.2 0.0
−2.8 −4.2 0.2 −7.3 −2.3 −0.4
−3.2 −4.9 0.2 −8.2 −2.6 −0.5
−0.4 −1.9 0.1 −3.6 −1.0 −0.2
3580.6 8233.9 268.9 34,556.4 4798.7 1347.8 4073.6 3079.5 30,235.4 91,857.8 78,765.8
0.1 0.0 0.1 0.0 0.0 0.1 0.1 0.0 0.0 0.0 0.0
0.7 0.1 0.1 0.4 0.0 1.3 0.2 0.0 0.1 0.0 −0.3
−0.2 −0.4 0.0 −1.0 −0.8 −7.9 −3.5 −0.2 −0.3 −0.1 1.9
−0.3 −0.4 −0.1 −1.2 −1.0 −9.0 −4.0 −0.2 −0.3 −0.1 2.2
0.3 −0.1 0.0 −0.3 −0.4 −3.1 −1.6 −0.1 −0.1 −0.1 0.7
equilibrium results are reported in Tables 3–6. Looking first at Table 4, the price of biojet is, for the first time, within one cent per gallon of conventional jet fuel ($3.05 versus $3.06). With such a small a gap between the two fuel prices, there is now little reason for airlines not to choose biojet. Since the demand for conventional jet fuel falls, its price falls by 0.2%, from $3.06 to $3.05 (Table 4). With these new incentives in place, the price of livestock and air transportation falls by 0.3% and 0.5%, respectively, as willingness to supply increases among firms in these two sectors (due to the subsidized inputs). The prices for this scenario are summarized in Fig. 5. Changes in supply, demand, and input-output relations are reported Table 3. As in Case 1, biofuel use rises from zero to $185
the gap between the above two prices is now 66 cents per gallon, up from 63 cents per gallon in the baseline assumption. Without a mechanism to transfer the higher costs of biojet to consumers, this case is unlikely to arise in the real world. On the other hand, this case could be argued to represent the status quo, in the sense that the aviation industry claims to have the preferences approximated by this case, yet the supply chain is undeveloped because of the price disadvantage of biojet. Case 1 macroeconomic effects are summarized in Table 5. There is a relatively insignificant $184.2 million (0.03%) fall in regional gross regional product, due to the reallocation of resources necessary to induce the use of biojet. A measure of peoples’ well-being is also reported, which is equivalent variation. It is approximately the change in gross regional product adjusted for the change in regional prices associated with the demand change. Equivalent variation for Case 1 is $10.4 million, meaning that regional households would (in the aggregate) would pay this amount to avoid the various changes brought about in this scenario. The decline in the overall regional economy is manifested as falling government revenue at the state and federal level; there is slightly less tax revenue (and spending, which is constrained to equal revenue) since resources are allocated away from a traditionally lower-cost supply chain. Changes in labor demand are reported in Table 6. Labor demand falls in other feeds farming (−10.1%) and petroleum refining (2.5%) but rises in oilseed production and refining (8.3%) and in oilseed farming (4.8%). The net effect of these changes is a net $47 million fall in labor demand in the economy (Table 4).
4.00 3.50
U.S. dollars per gallon
3.00 2.50 2.00 1.50 1.00
3.2. Demand change plus biojet subsidy
0.50
If consumers are not willing to directly pay the premium for biojet, yet it is perceived as providing public benefits, then a public policy must be considered. One possibility is a subsidy. To make biojet cost feasible with conventional fuel, the biojet price must be subsidized such that it falls by 17.2% (Table 4). This is carried out in the model by calibrating an ad valorem subsidy, adjusting the appropriate parameter until the prices (as viewed by buyers) are equalized. Through this process, the value 17.2% was converged upon. The associated general
0.00 Conventional jet fuel price
Price of biojet fuel
Case 1: Demand change only (reference case) Case 2: Demand change plus 17.2% biojet subsidy Case 3: Demand change plus 19.5% conventional fuel tax Case 4: Demand change plus 22.6% conventional fuel tax Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax Fig. 5. Price results.
8
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
diverse range of sectors. However, the overarching lesson of this scenario is that trying to achieve parity in the price of the two fuel types by taxing conventional fuels has a unique disadvantage: it forces the price of biojet up as well. This creates a “two steps forward, one step backward” situation and ultimately increases the cost of air transportation. For this reason we turn to an alternative approach.
million. Meanwhile, oilseed meal use in the livestock sector rises by $149 million. There is a corresponding fall in conventional jet fuel use, and a corresponding fall in other feed use in livestock, as shown in the table. Table 3 also shows that local oilseed demand rises by 10.8%, or $57 million. This is used to increase local output of processed oilseeds by $210.1 million, or 8.7%. Overall the new demand for processed oilseeds under the subsidy is $453.9 million, or 12.9% higher than before. Despite the large rise in local production, even more comes from imports, which rise 12.9%. Macroeconomic results for case 2 are reported in Table 5. To maintain budget balance, state governments must reduce expenditures on other goods. This ultimately results in a reduction of the size of the regional economy by $849.7 million. This results in a significantly larger welfare cost than in Case 1; equivalent variation is $248.2 million (Table 5). Changes in labor demand are reported in Table 6. The largest increases in labor demand are for oilseed processing and refining (22.4%) and oilseed farming (12.8%). There is a smaller decrease in labor demand in other feeds (−9.6%) and in petroleum refining (−2.1%).
3.4. Demand change plus biojet subsidy and conventional fuel tax As shown above, a tax increases the price of air transportation by 3.9–4.6%, and is paid for only by those who use conventional fuel. The subsidy has a more immediate effect on the use of biojet and lowers the price of air transportation (−0.5%), but is shouldered by taxpayers as a whole. Given these considerations, the best features of these two policy instruments might be combined, such that there is a tax-cum-subsidy policy. In this new scenario, an approximately 9% tax on the conventional fuel and 9% subsidy on biojet is considered. With these changes, model simulations show that the new price of biojet is $3.33 per gallon, while the price of conventional fuel is $3.36 per gallon (Table 4). The gap is three cents, which may be small enough that it would not prevent aviation operators from choosing biojet when available. The prices implied by this scenario are summarized in Fig. 5. This shows that the tax-cum-subsidy is effective in equilibrating the prices. Table 3 shows that these policies, combined with the preference change in all scenarios, causes biojet use in aviation to rise by $180 million and oilseed meal use for livestock to rise by $149 million. Compared to the pure tax in cases 3 and 4, the tax-cum-subsidy induces larger increases (6.3% and 5.0%) in the local oilseed farming and processing industries. Yet as in previous cases, much of the growth occurs outside the region, as imports of processed oilseeds rise 11.7%. Gross regional product rises by 0.8%, less than under cases 3 and 4, but more than in case 2, in which the regional economy shrunk somewhat (Table 5). Equivalent variation is –$1415.4 million, implying that the aggregate regional household is better off, although less so than in cases 3 and 4 (Table 5). This scenario is relatively revenue neutral compared to the pure subsidy or tax scenario alone, as government revenue and spending rises by 1.9%. Payments to regional labor rise by 0.8% (Table 5). Labor demand falls by 19.3% and 14.1%, respectively, for petroleum refining and other feeds farming, and rises by 13.0% in oilseed processing and refining (Table 6).
3.3. Demand change plus conventional fuel tax The above scenario shows the size of subsidy that would be required to make biojet cost competitive with conventional fuels. It also suggests that this approach is especially costly and likely to lower regional welfare. For this reason, a tax on the conventional jet fuel is now considered. As of 2011, conventional jet fuel (specifically, kerosene for aviation) was taxed by 1 cent, 11 cents, and 6 cents per gallon by the states of Oregon, Washington, and Idaho, respectively (Defense Logistics Agency [25]). Federal taxes were approximately 4.4 cents per gallon (Internal Revenue Service [26]). In this scenario, the additional tax is set such that the gap between the biojet fuel cost, and the conventional fuel cost, is reduced to near zero. In case 3 a tax of 19.5% is imposed, raising the price of conventional fuel to $3.66. In case 4 a tax on conventional fuel of 22.6% is imposed, raising the price of conventional fuel to $3.75. Both taxes bring the price of conventional fuel approximately in line with the price of the alternative fuel (Fig. 5). These two are chosen, however, to illustrate that as the conventional fuel is taxed, it has the potentially unanticipated consequence of pushing up the price of biojet as well. This rises from $3.69 per gallon in the baseline up to $3.79 in case 3, and up to $3.80 in case 4. The gap is therefore $0.13 in case 3, and $0.05 in case 4. The gap narrows as the tax rises, but nonetheless, the very fact that biojet is rising in price is undesirable since jet fuel prices (whether made from kerosene or oilseeds) are rising above what either was initially. This happens for the same reason that it did in Case 1; rising demand for biojet raises its price. As the prices of both fuels rise, air transportation becomes more costly. It rises by 3.9% under case 3% and 4.6% under case 4 (Table 4), in contrast to the slight fall in price in case 2 (the subsidy). Despite this outcome, increasing the tax on conventional fuel by approximately 20% increases the odds that a new supply chain will develop by reducing the price disadvantage of biojet. This also has the advantage of expanding the regional economy, if only slightly. Gross regional product rises by 1.8% in case 3, and by 2.1% in case 4 (Table 5). The tax generates sufficient revenue to slightly boost spending in the regional economy. In contrast to case 2, payments to labor and capital rise in these scenarios by 1.8% and 1.2% in case 3, and by 2.1% and 1.4% in case 4. Changes in labor demand are reported in Table 6. While there is shrinking labor demand in oilseed and other feed farming, along with petroleum refining, these losses are offset by labor demand growth in the rest of the economy, as the new tax revenue provides spending in a
3.5. Discussion Biofuels are sometimes promoted as having the potential to positively impact a given region. For the Pacific Northwest in 2011, the value of the regional oilseed processing sector was $2420.1 million (Table 3). In all cases examined here, activity in this sector rises. However, this local supply is not enough to meet the hypothesized demand in any of the cases. In 2011 there were $2736.8 million imports of processed oilseeds, and in these cases, the imports always increased by more than does activity in this sector in the region itself. This ease of substitution constrains the ability of the biojet industry to have a large effect within the immediate region. The implication is that providing subsidies for processing does not imply that production of the feedstock will occur within the region. The inter-regional substitution possibilities are too great for oilseed processors and refiners to source only from the local area, and to have all operations occur within the Pacific Northwest. 4. Conclusions This study develops a regional multi-market economic policy model to examine supply chain issues as the aviation sector seeks diversify its fuel sources away from conventional fuels. Technological and engineer9
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
constitute a case against camelina as a potential energy feedstock. It is clear that substantial policy interventions will be required to align private incentives with public goals, given current price and market conditions for alternative fuels. For other readers, the results might be encouraging and could serve to narrow the range of policy choices. In either case, it is hoped that this study has shed light on the nature of economic incentives that would be necessary to bring this potential bioenergy resource into fruition, as well as its potential impacts on the broader economy.
ing aspects are not modeled in great detail; rather, the focus is on the transmission and size of price signals from end users to raw feedstock producers such that biojet can be reasonably competitive in price with conventional jet fuel. Five cases are considered, the first of which concerns a consumer demand change only with no new policy changes taking place. This is intended to reflect a scenario in which a subset of airline fliers voluntarily elect to pay more, allowing airlines to use the proceeds to offset the higher cost of biojet, and bring the alternative biofuel supply chain into fruition. This approach leaves a large gap between the prices of the two fuels. Without a segment of consumers willing to pay extra for the use of biojet, this case is unlikely to arise in the real world. In some sense this is therefore a “business as usual” case, for which public policy lies dormant and little progress is made. This would appear to explain the current situation of camelina-based biojet. Given that aviation biofuels appear to be among the energy policy priorities of the United States, a number of public policy mechanisms are next considered through a series of cases. The first is a subsidy on biojet that would bring its cost in line with conventional fuel. The subsidy is determined to be approximately 17%. A problem with this approach is that all taxpayers pay for the subsidy, instead of just users of the fuel. This case would only make sense if biojet can be viewed as providing a public good. This is an expensive approach, generally, and entails the least favorable macro-economic outcomes. Gross regional product declines and other measures of general economic welfare are unfavorable. As another possibility, a tax on the conventional fuel is considered. It would need to be approximately 20% to make end users indifferent between it and alternative fuels. One problem is that this raises the demand for biojet to an extent that it itself becomes ever more expensive. This makes it even harder to narrow the price gap between it and the conventional fuel. This make air transportation approximately 4% more costly, which may be an undesirable outcome from the viewpoint of the aviation industry, who otherwise would be in favor of the supply diversification motivation as envisioned in this study. A more favorable scenario, examined with another case, would involve a combined 9% subsidy on biojet and 9% tax on the conventional fuel. This mitigates the gap between the fuel prices, is approximately revenue neutral for government, and has other favorable macroeconomic outcomes as well. It is less likely to induce distortions in markets beyond those of central interest to the example of this study. Another finding is that providing subsidies for processing does not imply that production of the feedstock will occur within the region. The inter-regional substitution possibilities – in the form of easy transportation across regions – are too great for oilseed processors and refiners to source only from the local area, and to have all operations occur within the region. This limits the ability of this industry to have an effect on regional economic development. This issue matters little with respect to overall economic efficiency in the country as a whole. However, it could make this less appealing to regional policymakers, whose support may be critical to any push for implementation of the policies considered here. One of the novel features of the analysis is the historical validation technique that provides statistical measures of model performance. In particular, the study presents a means of model validation that, upon a careful review of the literature, appears to be a first for a regional computable general equilibrium model. The model is calibrated such that it can replicate the size of past price changes for actual historical oilseed yield changes. Due to this validation, greater confidence can be placed in the measured impacts of new biofuel demand on the regional economy. This provides a baseline with respect to how isolated or connected the Pacific Northwest is to the rest of the United States, and the rest of the world, in the case of oilseeds, as either a raw feedstock or as a processed biofuel, namely biojet. In summary, for some readers, the results presented herein might
Acknowledgements This project is partially supported by Formula Grant no. 2012– 67009-19917 of the United States Department of Agriculture National Institute of Food and Agriculture, as well as Oregon State University. References [1] Bauen A, Howes J, Bertuccioli L, Chudziak C. Review of the potential for biofuels in aviation: final report for CCC. Fuel 2009:6021–5. [2] Shonnard DR, Williams L, Kalnes TN. Camelina-derived jet fuel and diesel: sustainable advanced biofuels. Environ Prog Sustain Energy 2010;29:382–92. http://dx.doi.org/10.1002/ep.10461. [3] IATA. Report on alternative fuels 2014:65. 〈https://www.iata.org/publications/ Pages/alternative-fuels.aspx〉. [4] Natelson RH, Wang W-C, Roberts WL, Zering KD. Technoeconomic analysis of jet fuel production from hydrolysis, decarboxylation, and reforming of camelina oil. Biomass Bioenergy 2015;75:23–34. http://dx.doi.org/10.1016/j.biombioe.2015.02.001. [5] Stein L. Economic analysis of potential camelina oil crop supplies in the northwest U.S. Oregon State University; 2012. [6] EPA. Regulation of Fuels and fuel additives: identification of additional qualifying renewable fuel pathways under the renewable fuel standard program. Washington, DC: 2012. [7] Winchester N, McConnachie D, Wollersheim C, Waitz I a. Economic and emissions impacts of renewable fuel goals for aviation in the US. Transp Res Part A Policy Pract 2013;58:116–28. http://dx.doi.org/10.1016/j.tra.2013.10.001. [8] Putnam DH, Budin JT, Field LA, Breene WM. Camelina: a promising low-input oilseed. In: Janick J, Simon JE, editors. New Crop.. New York: Wiley; 1993. [9] Hulbert S, Guy S, Pan W, Paulitz T, Schillinger W, Sowers K. Camelina production in the dryland Pacific Northwest. Pullman WA; 2012. [10] Cansino JM, Cardenete M a, González-Limón JM, Román R. Economic impacts of biofuels deployment in Andalusia. Renew Sustain Energy Rev 2013;27:274–82. http://dx.doi.org/10.1016/j.rser.2013.06.006. [11] IMPLAN. MIG IMPLAN data and software 2012. [12] Tabatabaie SMH, Murthy GS. Life cycle assessment of biodiesel production from Camelina in the Pacific Northwest U.S. Corvallis OR; 2015. [13] Valenzuela E, Hertel TW, Keeney R, Reimer JJ. Assessing global computable general equilibrium model validity using agricultural price volatility. Am J Agric Econ 2007;89:383–97. http://dx.doi.org/10.1111/j.1467-8276.2007.00977.x. [14] Diebel PL, Ball DA. An economic analysis of alternative wheat rotations for eastern Oregon. Corvallis OR; 1999. [15] Walsh ME. Method to estimate bioenergy crop feedstock supply curves. Biomass Bioenergy 2000;18:283–9. http://dx.doi.org/10.1016/S0961-9534(99)00094-X. [16] McCullough M, Holland D, Painter K, Stodick L, Yoder J. Economic and environmental impacts of Washington state biofuel policy alternatives. J Agric Resour Econ 2011;36:615–29. [17] Reimer J, Weerasooriya SA, West T. How does the Supplemental Nutrition Assistance Program affect the United States economy. Agric Resour Econ Rev 2015;43:233–52. [18] Löfgren H, Robinson S, Harris RL. A standard computable general equilibrium (CGE) model in GAMS. doi:338.1′01'51—dc21; 2002. [19] Rose A, Liao SY. Modeling regional economic resilience to disasters: a computable general equilibrium analysis of water service disruptions. J Reg Sci 2005;45:75–112. http://dx.doi.org/10.1111/j.0022-4146.2005.00365.x. [20] Reimer JJ, Zheng X, Gehlhar MJ. Export demand elasticity estimation for major U.S. crops. J Agric Appl Econ 2012;4:501–15. [21] Bilgic A, King S, Lusby A, Schreiner D. Estimates of U.S. regional commodity trade elasticities of substitution. Reg Anal Policy 2002;32:79–98. [22] Hodges A, Rahmani M. Cost of product of Camelina Oil in North Florida; 2012. [23] Schumacher J. Overview of Oilseed crushing and processing technologies; 2008. [24] Stein L, Seavert C, Macnab S, Petrie S. Oregon State University Enterprise Budget for Camelina Following Winter Wheat. Corvallis OR; 2014. [25] DLA. Compilation of U.S. fuel taxes 2012:1–428. 〈http://www.energy.dla.mil/ DLA_counsel_energy/Documents/Tax Documents/Tax compilation 2012-04.pdf〉. [26] IRS. Internal revenue service report on fuel taxes 2015. 〈http://www.irs.gov/ publications/p510/ch01.html#en_US_2013_publink1000116905〉.
10
Contents lists available at ScienceDirect
Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser
Economic analysis of an aviation bioenergy supply chain ⁎
Jeffrey J. Reimera, , Xiaojuan Zhengb a b
Oregon State University, USA Oregon State University and Dell Corporation, USA
A R T I C L E I N F O
A BS T RAC T
Keywords: Aviation Biofuels Camelina Energy policy General equilibrium Oilseeds
This study develops a regional economic model that can mimic stages of a potential aviation fuel supply chain based on the oilseed camelina. A general equilibrium model is developed that accounts for key sectors of the supply chain, and is parameterized using detailed data for the Pacific Northwest region of the United States. Model scenarios are developed to evaluate the potential effectiveness of policies that would promote the development of the supply chain. While the existing low price of conventional fuel makes camelina-based fuel economically infeasible at present, the supply chain could be viable if consumers, e.g. airline passengers, are willing pay more to use the biofuel, such as if they perceive an environmental benefit from it. Alternatively, if use of the oilseed is consistent with the energy policy priorities of policymakers, one of the following approaches could potentially work: a 17% subsidy on the alternative fuel, a 20% tax on the conventional fuel, or a combination 9% subsidy on the alternative and 9% tax on the conventional fuel. Implications for economic efficiency, regional employment, and economic welfare are quantified.
1. Introduction This study concerns the development of a new bioenergy supply chain for the aviation sector of the Pacific Northwest region of the United States. The focus is on the policies and price signals necessary to induce a new market channel to develop, and on the implications for the efficiency of resource allocation. A general equilibrium economic model with explicit representation of key sectors is developed and applied to aviation fuel made from oilseeds such as camelina. It can be processed into a high grade bio-based jet fuel for military and commercial purposes using approaches such as a Hydroprocessed Esters and Fatty Acids (HEFA) process (Bauen et al. [1], Shonnard et al. [2], IATA [3], and Natelson et al. [4]). Interest in this potential bioenergy resource has grown in recent years. For example, the number of airlines which have used a blend of this alternative fuel for a commercial flight has risen to 21 (IATA [3]). Interest is also strong among military agencies, such as the U.S. Air Force, and operators of commercial airports, including those of Portland and Seattle. These institutions are reportedly interested in using U.S.-sourced biofuels, again for supply diversification, and perhaps for perceived environmental reasons, such as a reduction in net greenhouse gas emissions. Additional support has come from the U.S. Environmental Protection Agency who has declared that aviation biofuel – biojet, in industry parlance – is eligible for Renewable Identification Numbers that can be traded on the open market. This
⁎
has further solidified the long term prospects for this new bioenergy supply chain (IATA [3]). At the other end of the supply chain are oilseed processors and refiners, and ultimately, farmers who would need to plant the appropriate oilseeds. Camelina, for example, is argued to be ideal in the Pacific Northwest, where its cultivation need not displace other crops grown for food when incorporated into a wheat-fallow rotation (Stein [5]). Although currently grown on less than 50,000 acres in the United States, acreage could potentially be expanded by three to four million acres without adversely impacting food prices (EPA [6], Winchester et al. [7]). It also has low input requirements, is suitable for marginal soils, and has natural competitiveness with weeds (Putnam et al. [8], Hulbert et al. [9]). Despite the theoretical potential and technical plausibility of this alternative supply chain, the economics of how the market might work are less clear. There are public good aspects to the problem, meaning that the private sector, by itself, may not make the necessary investments. Interactions up and down the supply chain must be considered, including the incentives of a range of different economic agents who may have competing interests. Farmers, for example, are incentivized by a high price for camelina, while refiners are incentivized by a low price. Whether prices can be found that satisfy all supply chain participants, simultaneously, is an empirical question that can be addressed in part by economic modeling. To shed light on this problem, a general equilibrium economic
Corresponding author. E-mail addresses: jeff[email protected] (J.J. Reimer), [email protected] (X. Zheng).
http://dx.doi.org/10.1016/j.rser.2016.12.036 Received 9 May 2015; Received in revised form 20 April 2016; Accepted 6 December 2016 1364-0321/ © 2016 Published by Elsevier Ltd.
Please cite this article as: Reimer, J., Renewable and Sustainable Energy Reviews (2016), http://dx.doi.org/10.1016/j.rser.2016.12.036
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
similar shocks, following pioneering techniques introduced in Valenzuela et al. [13]. This aspect of the study is novel in the general equilibrium literature and assists the reader gauge the reliability of the results. The study also complements work by Diebel and Ball [14], Walsh [15], and Stein [5]. Building on the first two studies, Stein estimates potential supply curves for camelina in Oregon, Washington, Idaho, and Montana with a break-even price approach. Stein's results suggest that given current market conditions, the supply in the Pacific Northwest will not be enough to meet biofuel targets without an increase in promotion, most likely by government. This finding serves as one motivation for the study at hand. Unlike the approach proposed here, Stein uses a partial equilibrium framework and looks mainly at adoption by farmers, as opposed to the links between different market players. He does not examine policy, the potential for inter-regional trade, labor markets, or any of the macro-economic aspects considered here. The study is also complementary to McCullough et al. [16], who assess the potential for biofuels production in Washington state. As with many other studies, they find that some form of government intervention is likely to be needed to “jumpstart” the system. The study at hand is distinct in that it considers a larger geographical area, recent developments with respect to aviation biojet demand, and a wider variety of policies at different stages of the supply chain. To preview some of the conclusions, this study finds that the new bioenergy supply could be made feasible, or at least price competitive, through a number of particular policy mechanisms. Unless a segment of consumers can be induced to pay extra for use of biojet, one of the following policies would provide the incentive structure to facilitate market development: a 17% subsidy on the alternative fuel, a 20% tax on the conventional fuel, or a combined 9% subsidy on the former and 9% tax on the latter. The latter, so-called tax-cum-subsidy approach provides a “double dividend” in the sense that it combines a targeted tax on conventional fuel, with a targeted subsidy on the alternative fuel, so as to be relatively revenue-neutral and to have smaller distortions on other markets. The remainder of the study is organized as follows. The next section describes the modeling approach taken, addressing the general problem of how to model a market channel for a new product that is not yet established. Since the model shares many features from other established computable general equilibrium models, changes that are unique to this study are emphasized, with separate sections describing the regional adaptation of the modeling framework, the data, calibration of model parameters, and validation of key parameters. Subsequent sections, in turn, examine how the scenarios are developed, and the results. Five different cases are analyzed, each of which shed light on viable options for development of a regional supply chain for biofuels. The final section concludes.
model is developed in this study that simultaneously accounts for a number of sectors key to the analysis. This general type of model is part of a venerable tradition in economics, and has been used to study biofuel energy policy (Cansino et al. [10]). It combines mathematical representation of the incentives and constraints faced by all of the economic actors in the system. To make the model useful for policy, it must be made to replicate, that is, its parameters must be calibrated such that the model can reproduce baseline data for the region in an appropriate recent time period. Parameters of the model are calibrated primarily using highly detailed IMPLAN [11] data for the Pacific Northwest region (Oregon, Washington, and Idaho) of the United States. These data account for trade and transfers between hundreds of economic sectors within the region, including between the aviation, processing, and farming sectors. Once parameterized and validated, the model is used to illustrate the mechanisms that could make this new bioenergy supply chain economically feasible. These include non-policy as well as policy instruments such as subsidies and taxes, including a tax-cum-subsidy approach. A suitably motivated government could conceivably take the estimates of this study and use them to guide policies that would facilitate the emergence of an aviation biofuel supply chain in the Pacific Northwest. Unlike related studies such as Natelson et al. [4], Tabatabaie and Murthy [12], and Winchester et al. [7], the present study does not emphasize the engineering aspects of biofuel production, the change in greenhouse gas emissions for conventional versus alternative fuels, or quantify the potential security benefits of using alternative fuels. For example, Natelson et al. [4] provides a techno-economic analysis of alternative fuel production based upon oilseeds such as camelina. Relative to this study, they provide much more detail on the engineering requirements for alternative refinery operations, but less detail on different aspects of the supply chain, including the welfare consequences of alternative tax and subsidy policies that could make biojet comparable in price to conventional fuel. The study at hand additionally emphasizes labor market impacts and a range of macro-economic outcomes. A special focus is on whether increased demand for oilseeds can be met by local sources, as opposed to sources outside a region. One of the selling points of oilseeds – camelina in particular – is that it can be a “home grown” source of fuel with very low opportunity costs of production. Yet even if the oilseed feedstock is produced by local farms, this may not imply that the processing will occur within the region. Likewise, if the processing is done within the region, the oilseed feedstock could perhaps be procured most efficiently from outside the region, such as from Canadian farmers, thereby diluting the “home grown” nature of the energy resource from the viewpoint of U.S. policymakers. To address these issues, the study explicitly models the separate stages of a vertical supply chain, including sourcing from inside and outside a region of the United States, and from outside the country. In this sense, the study complements Winchester et al.’s [7] analysis of alternative fuels for the aviation industry. Their study uses a global general equilibrium model, and considers other advanced biofuels such as biomass-based diesel and grain-based ethanol cellulosic fuels. The study at hand considers a particular supply chain, distinguishing oilseed production from processing and refining, and emphasizes the Pacific Northwest region, where oilseeds can be integrated into a wheat-fallow operation with little or no impact on wheat supply (Stein [5]). This study also considers different policy instruments, including a tax-cum-subsidy that can minimize the distortion of economic incentives in multiple markets. Another unique aspect of this study is that model parameters are validated using historical price and quantity data for oilseeds. This goes beyond simply calibrating the model to IMPLAN input-output social accounting data. In this study, actual year-to-year price movements are compared to those which arise from model simulations representing
2. Modeling approach and data A diagram of the methodology for the study is in Fig. 1. At the top of the figure are two boxes; the left represents the theoretical economic model, which is a mathematical representation of the behavioral objectives and responses of all decision-makers including government. The theoretical model accounts for numerous economic features including taxes, subsidies, and the other policy instruments to be discussed below. The theoretical model is combined with a social accounting matrix (SAM) to create a calibrated economic model. The SAM itself is created using IMPLAN (2012) data to numerically represent trades and transfers between producers, buyers, households, and government. The SAM also traces the flows of tax revenues, government spending, exports, and imports. Finally, the calibrated model is shocked with a number of policy simulations to generate new equilibrium values of all endogenous variables. These are compared to the original baseline values to evaluate different policy measures. 2
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
Table 1 Sectors.
Fig. 1. Methodological overview.
A more detailed exposition of the elements in Fig. 1 is now provided. We focus first on the theoretical general equilibrium model. It is a modified version of the approach of Reimer, Weerasooriya, and West [17], with multiple markets linked in supply-demand equilibrium through intermediate input use, factors of production (labor and capital), and other linkages. It is a system of highly non-linear equations representing optimizing households and firms, inter-household and government transfers, savings and investment, government, and trade with other regions. A government sector is represented as collecting taxes through various instruments, and receiving transfers from other institutions. It then transfers this revenue back out to other institutions within the model. A representative household receives income from labor, capital, inter-household transfers, federal and state government transfers, and investment income. This household spends money on commodities, inter-household transfers, federal and state government taxes, and investment. Producers of goods choose their level of operation to maximize profits or minimize costs using constant returns to scale production technology. Production factors include labor and capital, paid according to their respective marginal productivities, as well as intermediate inputs. This last feature is important as it captures the vertical supply chain of production. The production function assumes fixed amounts of intermediate input use, along with flexible use of capital and labor. This flexibility is modeled using a constant elasticity of substitution (CES) functional form, similar to other instances in the model in which an agent has substitution possibilities. A full mathematical presentation of the model, as described above, can be found in Löfgren et al. [18]. There are four general modifications from that approach which are made in this study, highlighted below. The remainder of this section can be skipped by readers less interested in the technical aspects of the general equilibrium model.
Number
Name
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Oilseed farming Grain farming Petroleum refining Oilseed processing and refining Animal production including cattle poultry eggs All other crop farming Construction Mining and quarrying Utilities Wholesale and retail trade Processed food Food purchased away Alcohol and cigarettes Manufacturing Housing Transportation by air All other transportation Education Health Other services Miscellaneous
themselves are reported in Table 1. The top row of Fig. 2 depicts the aviation and animal production sector, which purchase either oil (for fuel) and meal (for livestock feed). The third row depicts a composite processed oilseeds sector. Activity in this sector takes place inside or outside the region according to a constant elasticity of substitution function. If oilseeds are processed within the region, the oilseed feedstock itself can come from within the region or outside the region (Fig. 2, rows 4 and 5). Unprocessed oilseeds from outside the region come either from the rest of the United States, or the rest of the world. Regional oilseed farming can be utilized within the region, or sold outside the region. These individual substitution possibilities are each characterized by a CES form.
2.2. Data and parameterization The model described above is a simultaneous system of equations written for GAMS software. The joint equilibrium values of the endogenous variables is calculated using the PATH non-linear programming solver. Model parameters are calibrated with IMPLAN data [11], which in raw form distinguish more than 400 distinct sectors of the economy. To make the analysis practical, these are aggregated to the 21 sectors in Table 1. The first step is to create a Social Accounting Matrix (SAM), which is a highly detailed account of monetary flows between economic agents, commodities, factors, and institutions including government. Model parameters are calibrated to ensure that baseline values of endogenous variables match the SAM regional information. During calibration, all prices are set to unity and the base year factor levels and SAM flows are substituted into the model as equilibrium values of model variables. The process is similar to maximum likelihood estimation with one observation. Model parameters calibrated through this process correspond to the year 2011. The model also contains free parameters set by the user. These are set to values based on econometric estimates or otherwise employed in the literature (Rose and Liao [19], Hertel et al. [13], McCullough et al. [16]). For example, the elasticity of substitution between capital and labor in the production function is set at 2. The rest-of-world export elasticity of demand for oilseeds is set to −1.175, an estimate made in Reimer et al. [20]. Elasticities of substitution between regional and rest of U.S. supplies are from Bilgic et al. [21], and are set at 1.447 and 1.339 for unprocessed and processed oilseeds, respectively. Elasticities
2.1. Regional modification Instead of modeling a country as a whole as in Löfgren et al., a distinct region within the United States is modeled in this study. A list of sectors and related output is listed in Table 1. A representative regional household receives income from labor, capital, and transfers. It spends money on commodities, transfers, taxes, and investment. Regional firms use labor, capital, and intermediate goods to maximize profits. Regional state governments collect taxes and receive transfers from other institutions, with spending constrained to equal revenue. Key aspects of the regional supply chain are highlighted in Fig. 2, which depicts two of the 21 total sectors that are modeled, and which 3
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
Fig. 2. Sectoral flowchart.
between U.S. and the rest of world supplies are from Hertel et al. [13], and are set at 4.9 and 5.2 for unprocessed and processed oilseeds, respectively.
2.3. Model validation An especially unique aspect of this model is the validation exercise undertaken to indicate how well the model replicates historical data for the region, thereby providing a measure of confidence regarding predictions made below. This section can be skipped by readers less interested in the technical aspects of the procedure. The approach is to map out the model's response functions for relevant shocks in observable prices and quantities over time. The approach makes use of 1991–2011 oilseed price and quantity data from the United States of Department Agriculture (USDA), National Agricultural Statistics Service Crop Production, Grain Stocks, and Crop Values and USDA, Foreign Agricultural Service, Global Agricultural Trade System. Since the supply chain is not well established, data on camelina are unavailable. Data on canola are used, since it is very similar in agronomic and use characteristics with respect to camelina, it has a long historical record in the Northwest, and since it comprises much of the production considered as oilseeds within the SAM. The process of calibration can be explained by first examining Figs. 3 and 4, which report actual canola output and prices, respectively, for the U.S. over 1991–2011. There are trends in both output and prices, related to factors such as technology change (Fig. 3) or general inflation in the economy (Fig. 4). Abstracting from these trends, the associated year-to-year volatility can provide information about the extent that prices adjust to supply or demand shocks. To abstract from any less relevant trends in Figs. 3 and 4, two alternative ways of detrending the time series are used: an ordinary least squares (OLS) approach (in which only the year of an observation, and an intercept, are included on the right hand side of the model) and an autoregressive moving average (ARMA) model. A full description of these statistical models are not reported due to space constraints, but is available upon request and also in Valenzuela et al. [13]. Fitted values from the estimated models are presented visually in Figs. 3 and 4. Residuals from the output regression are used to calculate percen-
Fig. 3. Actual versus predicted U.S. canola (million pounds).
Fig. 4. Actual versus predicted U.S. canola price ($ per hundred pounds).
tage changes in output by year. These percentage changes are plugged into the model, which then generates an associated price change. The standard deviation of these price changes is 15. If output was well below trend (such as in 2009 and 2010), then the supply curve for the product would shift to the left, and price would rise. On the other hand, 4
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
gallons of fuel per year. Using the seed to oil conversion factor, the corresponding amount of meal that would be produced is 988.2 million pounds, which is assumed to sell at $0.15 per pound. The average yield in the Pacific Northwest is 1600 pounds per acre [24], so at the 60,000,000 gallon target identified above, approximately 908,250 acres would be required. This is much less than the three to four million acres of fallow land each year that could reportedly support camelina [6,7]. Before implementing the above changes, it is worthwhile to note the baseline values of inter-industry relationships for the year 2011 in the IMPLAN data assembled for the region. These are reported under the heading “input-output relations” in Table 3, which also reports some of the main simulation results to be discussed shortly. In the 2011 baseline, there was no biojet used in the regional aviation industry, but there was $1347.3 million worth of conventional jet fuel use. The regional livestock industry used $34.4 million worth of oilseed meal and $264.3 million worth of other feeds (e.g., grains).
when output was above trend for a year (such as in 2002 and 2003), then the supply curve would shift to the right, and price would fall. The extent to which the price will actually change is an empirical question, depending in part on price elasticities of demand in other regions (e.g., whether extra supply can be absorbed) as well as the ease of trading oilseeds across regions (which is influenced by barriers to trade such as transport costs and tariffs). If the model is calibrated correctly, then the magnitude of the simulated price changes is expected to match the magnitude observed in the historical record. The magnitude of historical price changes can be characterized by the second moment of the residuals from the regressions in Fig. 4. The standard deviation of residuals from the simple trend line regression of actual oilseed prices is 15.4 under the ARMA approach, and 25.1 under the OLS approach. The latter predicts higher volatility since it forces a linear relationship, and thus is less able to mimic year-to-year differences. These two approaches can be viewed as covering the range of the extremes, and the actual level of price volatility is expected to lie between the two values of 15.4 and 25.1. In other words, if the model is calibrated appropriately, it should generate price movements that have a magnitude within this range. As mentioned above, the simulated standard deviation of U.S. canola price changes is 15. This lies at the lower bound of the 15.4 – 25.1 range. If it was lower than this range, simulated price responses would be too moderate. This would arise, for example, if the cross-regional elasticities of substitution were too high. If the response above was higher than 25.1, then it is as if price is too sensitive in the model. In this case, some slack would need to be built into the model, by increasing the magnitude of the regional elasticities of substitution. This does not need to be done, however, as the results obtained above are not inconsistent with the historical experience.
3. Results Design and analysis of five cases is presented below. A complete description of the code for these simulations, as well as that of the model itself, is available from the authors upon request. 3.1. Demand change with no policy This first scenario is an initial exploratory analysis intended to show how increased demand for biojet will change the regional economy unaccompanied by any explicit policy change in this regard. In some ways it is a “business as usual” case by which to reference the policy simulations in later cases. A number of airlines have expressed great interest in alternative fuels, but are not willing or able to pay more than what they pay for conventional fuel (IATA [3]). One approach is to pass along the costs to customers with a program wherein consumers voluntarily pay extra, for example, to purchase carbon offsets to compensate for the greenhouse gas emissions caused by personal air travel. The significance of this is that biojet can be more expensive than conventional fuel, yet still be viable from the standpoint of airline profitability. Results for such a scenario are reported in Tables 3–6. The top of Table 3 shows that by changing the inter-industry demand schedules, biojet use in aviation rises from zero to $185 million. Meanwhile, oilseed meal use in the livestock sector rises by $149 million. There is a corresponding fall in conventional jet fuel use, and a corresponding fall in other feed use in livestock. These changes align with the targets and assumptions in Table 2. The demand changes lead to a cascade of changes throughout the regional economy. Table 3 reports changes in supply and demand, distinguishing between oilseeds as a feedstock, and processed/refined oilseeds (biojet or meal). There is a $21.3 million (4%) rise in the demand for oilseed feedstock, increasing from $527.0 million to $548.3 million. The total local demand for processed oilseeds rise by 9.8%, or $346.3 million. Most of this is met by imports to the region, which rises by 10.7% or $293.3 million. The remainder is met by the local processing/refining industry, which expands by 3.2%, or $78.2 million (any residual is due to exports, which are positive but small and therefore not reported). One implication is that a regional biofuels industry does not necessarily guarantee that local producers/refiners will have increased demand for their product; much of it may be imported from outside the region. The top part of Table 4 shows the change in prices at different stages of the supply chain, and for ancillary markets. The increased demand puts pressure on the price of biojet. With the new demand, the price of biojet is even more expensive than before, rising 0.1% from the base assumption of $3.69/gallon up to $3.71/gallon. The price of conventional fuel falls by 0.2%, from $3.06 to $3.05. This implies that
2.4. Assumptions common to all scenarios To evaluate policies for a regional biofuel supply chain, five counterfactual scenarios are developed. Assumptions common to all cases are presented in Table 2. These are based in part on information in Hodges and Rahmani [22], Schumacher [23], and Stein et al. [24]. It is assumed that the cost of feedstock is 13 cents per pound and the extractable oil is 32%. The seed to oil conversion factor is then 24.22 pounds per gallon. The cost of feedstock for oil is $3.15 per gallon. The cost of pressing seeds to oil is 54 cents per gallon, such that the total cost of oil is then $3.69 per gallon. The $3.69 estimate is consistent with related studies such as Winchester et al. [7], who estimate that HEFA jet fuel from rotation crops in the U.S. such as camelina could be produced at around $3.70 per gallon. This consistency provides a great deal of reassurance that the assumptions below are reasonable. Winchester et al. [7] also report that various rotation crops could potentially be grown on 43 million acres in the U.S. per year, yielding up to 3.2 billion gallons of renewable jet fuel. This study, with a regional focus, considers a relatively modest target of 60,000,000 Table 2 Cost analysis. Variable
Assumed value
U.S. Gulf Coast kerosene-type jet fuel spot price FOB ($/gal) Production cost of oilseed feedstock ($/pound) Oilseed extractable oil (%) Seed-oil conversion factor (lbs/gal at 32% oil content) Meal produced alongside each gallon of fuel (pounds) Cost of feedstock for oil ($/gallon) Cost of pressing seeds to oil ($/gallon) Total cost of oil ($/gallon) Target: Oilseed-based fuel per year (gallons) Corresponding amount of meal (million pounds) Price of oilseed meal ($/pound)
3.06 0.13 32 24.22 16.47 3.15 0.54 3.69 60,000,000 988.176 0.15
5
Regional demand and supply Demand for oilseed feedstock ($ mil) % Output of processed oilseeds ($ mil) % Imports of processed oilseeds ($ mil) % Demand of processed oilseeds ($ mil) %
Input-output relations Biojet use in aviation ($ mil) Conventional jet fuel use in aviation ($ mil) Oilseed meal use in livestock ($ mil) Other feed use in livestock ($ mil)
Variable
6 57.0 10.8 210.1 8.7 351.8 12.9 453.9 12.9
4.0 78.3
3.2 293.3
10.7 346.3
9.8
2420.1
2736.8
3530.3
−147
264.3
21.3
−147
149
34.4
527.0
150
185 −185
186 −175
Simulated change
Case 2: Demand change plus 17.2% biojet subsidy
0.0 1347.3
Initial
Case 1: Demand change only
Table 3 Results concerning oilseeds, farm level and processed.
8.9
10.4 312.9
0.6 283.9
0.8 13.8
4.3
−148
148
174 −254
Case 3: Demand change plus 19.5% conventional fuel tax
8.7
10.3 307.9
0.2 282.6
0.3 4.0
1.7
−148
147
172 −264
Case 4: Demand change plus 22.6% conventional fuel tax
11.0
11.7 389.4
5.0 320.2
6.3 121.9
33.3
−148
149
180 −212
Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax
J.J. Reimer, X. Zheng
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
7
3.71 0.66
3.69 0.63
Gross regional product ($ mil) % Regional govt expenditure ($ mil) % Federal government revenue ($ mil) % Federal govt expenditure ($ mil) % Payments to regional labor ($ mil) % Payments to regional capital ($ mil) % 1.8 8104.9 4.6 2460.6 1.3 4393.1 2.4 5088.1 1.8 1017.5 1.2
−0.1 −849.7 −0.5 −146.5 −0.1 −276.4 −0.1 −280.0 −0.1 −77.4 −0.1
−0.03 −28.8
0.0 −26.0
0.0 −41.1
0.0 −47.0
0.0 −40.3
0.0
176,694
188,141
185,767
278,355.1
81,699.3
11,161.2
−184.2
1.4
2.1 1175.1
2.7 5850.0
1.5 5066.0
5.3 2834.6
2.1 9326.2
12,853.9
−3704.8
−3228.9
248.2 Simulated change −713.1
10.4
Equivalent variation ($)
Case 4: Demand change plus 22.6% conventional fuel tax
2.7 1.5 22.6 3.0 3.2 4.6
3.80 0.05
0.15 3.75
Case 4: Demand change plus 22.6% conventional fuel tax
Case 3: Demand change plus 19.5% conventional fuel tax
2.3 1.1 19.5 2.7 2.7 3.9
Simulated change 0.1 −1.4 −0.2 −17.2 −0.3 −0.5
Case 2: Demand change plus 17.2% biojet subsidy
3.79 0.13
0.15 3.66
Case 3: Demand change plus 19.5% conventional fuel tax
3.06 0.00
Simulated value 0.12 3.05
Case 2: Demand change plus 17.2% biojet subsidy
Initial value 604,414
Case 1: Demand change only
0.1 −1.4 −0.2 0.6 0.0 0.0
0.15 3.05
Initial value 0.15 3.06
Case 1: Demand change only
Variable
Table 5 Macroeconomic results.
Oilseed price (%) Other livestock feed price (%) Conventional jet fuel price (%) Biojet price (%) Livestock price (%) Air transportation price (%)
Price of canola meal ($/lb) Conventional jet fuel price ($/gal) Price of biojet fuel ($/gal) Gap between the above two ($/gal)
Variable
Table 4 Results concerning prices.
0.5
0.8 446.8
1.0 2289.1
0.6 1931.8
1.9 1092.7
0.8 3379.7
4955.3
−1415.4
Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax
1.2 −0.2 8.9 −9.0 1.1 1.6
3.36 0.03
0.14 3.33
Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax
J.J. Reimer, X. Zheng
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
Table 6 Demand for labor by sector. Sector
Oilseed farming Other feeds farming Petroleum refining Oilseed processing and refining Animal production All other crop farming Construction Mining and quarrying Utilities Wholesale and retail trade Processed food Food purchased away Alcohol and cigarettes Manufacturing Housing Transportation by air All other transportation Education Health Other Services Miscellaneous
Case 1: Demand change only
Case 2: Demand change plus 17.2% biojet subsidy
Case 3: Demand change plus 19.5% conventional fuel tax
Case 4: Demand change plus 22.6% conventional fuel tax
Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax
Initial value 0.2 112.1 425.1 77.6
4.8 −10.1 −2.5 8.3
Simulated change (%) 12.8 −9.6 −2.1 22.4
−6.8 −19.0 −35.4 1.7
−8.5 −20.3 −39.5 0.7
3.6 −14.1 −19.3 13.0
650.2 4565.7 9471.4 468.5 2912.8 37,079.9
0.0 0.1 0.0 −0.3 0.0 0.0
1.6 0.3 0.0 0.0 0.2 0.0
−2.8 −4.2 0.2 −7.3 −2.3 −0.4
−3.2 −4.9 0.2 −8.2 −2.6 −0.5
−0.4 −1.9 0.1 −3.6 −1.0 −0.2
3580.6 8233.9 268.9 34,556.4 4798.7 1347.8 4073.6 3079.5 30,235.4 91,857.8 78,765.8
0.1 0.0 0.1 0.0 0.0 0.1 0.1 0.0 0.0 0.0 0.0
0.7 0.1 0.1 0.4 0.0 1.3 0.2 0.0 0.1 0.0 −0.3
−0.2 −0.4 0.0 −1.0 −0.8 −7.9 −3.5 −0.2 −0.3 −0.1 1.9
−0.3 −0.4 −0.1 −1.2 −1.0 −9.0 −4.0 −0.2 −0.3 −0.1 2.2
0.3 −0.1 0.0 −0.3 −0.4 −3.1 −1.6 −0.1 −0.1 −0.1 0.7
equilibrium results are reported in Tables 3–6. Looking first at Table 4, the price of biojet is, for the first time, within one cent per gallon of conventional jet fuel ($3.05 versus $3.06). With such a small a gap between the two fuel prices, there is now little reason for airlines not to choose biojet. Since the demand for conventional jet fuel falls, its price falls by 0.2%, from $3.06 to $3.05 (Table 4). With these new incentives in place, the price of livestock and air transportation falls by 0.3% and 0.5%, respectively, as willingness to supply increases among firms in these two sectors (due to the subsidized inputs). The prices for this scenario are summarized in Fig. 5. Changes in supply, demand, and input-output relations are reported Table 3. As in Case 1, biofuel use rises from zero to $185
the gap between the above two prices is now 66 cents per gallon, up from 63 cents per gallon in the baseline assumption. Without a mechanism to transfer the higher costs of biojet to consumers, this case is unlikely to arise in the real world. On the other hand, this case could be argued to represent the status quo, in the sense that the aviation industry claims to have the preferences approximated by this case, yet the supply chain is undeveloped because of the price disadvantage of biojet. Case 1 macroeconomic effects are summarized in Table 5. There is a relatively insignificant $184.2 million (0.03%) fall in regional gross regional product, due to the reallocation of resources necessary to induce the use of biojet. A measure of peoples’ well-being is also reported, which is equivalent variation. It is approximately the change in gross regional product adjusted for the change in regional prices associated with the demand change. Equivalent variation for Case 1 is $10.4 million, meaning that regional households would (in the aggregate) would pay this amount to avoid the various changes brought about in this scenario. The decline in the overall regional economy is manifested as falling government revenue at the state and federal level; there is slightly less tax revenue (and spending, which is constrained to equal revenue) since resources are allocated away from a traditionally lower-cost supply chain. Changes in labor demand are reported in Table 6. Labor demand falls in other feeds farming (−10.1%) and petroleum refining (2.5%) but rises in oilseed production and refining (8.3%) and in oilseed farming (4.8%). The net effect of these changes is a net $47 million fall in labor demand in the economy (Table 4).
4.00 3.50
U.S. dollars per gallon
3.00 2.50 2.00 1.50 1.00
3.2. Demand change plus biojet subsidy
0.50
If consumers are not willing to directly pay the premium for biojet, yet it is perceived as providing public benefits, then a public policy must be considered. One possibility is a subsidy. To make biojet cost feasible with conventional fuel, the biojet price must be subsidized such that it falls by 17.2% (Table 4). This is carried out in the model by calibrating an ad valorem subsidy, adjusting the appropriate parameter until the prices (as viewed by buyers) are equalized. Through this process, the value 17.2% was converged upon. The associated general
0.00 Conventional jet fuel price
Price of biojet fuel
Case 1: Demand change only (reference case) Case 2: Demand change plus 17.2% biojet subsidy Case 3: Demand change plus 19.5% conventional fuel tax Case 4: Demand change plus 22.6% conventional fuel tax Case 5: Demand change plus 9% biojet subsidy and 9% conventional fuel tax Fig. 5. Price results.
8
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
diverse range of sectors. However, the overarching lesson of this scenario is that trying to achieve parity in the price of the two fuel types by taxing conventional fuels has a unique disadvantage: it forces the price of biojet up as well. This creates a “two steps forward, one step backward” situation and ultimately increases the cost of air transportation. For this reason we turn to an alternative approach.
million. Meanwhile, oilseed meal use in the livestock sector rises by $149 million. There is a corresponding fall in conventional jet fuel use, and a corresponding fall in other feed use in livestock, as shown in the table. Table 3 also shows that local oilseed demand rises by 10.8%, or $57 million. This is used to increase local output of processed oilseeds by $210.1 million, or 8.7%. Overall the new demand for processed oilseeds under the subsidy is $453.9 million, or 12.9% higher than before. Despite the large rise in local production, even more comes from imports, which rise 12.9%. Macroeconomic results for case 2 are reported in Table 5. To maintain budget balance, state governments must reduce expenditures on other goods. This ultimately results in a reduction of the size of the regional economy by $849.7 million. This results in a significantly larger welfare cost than in Case 1; equivalent variation is $248.2 million (Table 5). Changes in labor demand are reported in Table 6. The largest increases in labor demand are for oilseed processing and refining (22.4%) and oilseed farming (12.8%). There is a smaller decrease in labor demand in other feeds (−9.6%) and in petroleum refining (−2.1%).
3.4. Demand change plus biojet subsidy and conventional fuel tax As shown above, a tax increases the price of air transportation by 3.9–4.6%, and is paid for only by those who use conventional fuel. The subsidy has a more immediate effect on the use of biojet and lowers the price of air transportation (−0.5%), but is shouldered by taxpayers as a whole. Given these considerations, the best features of these two policy instruments might be combined, such that there is a tax-cum-subsidy policy. In this new scenario, an approximately 9% tax on the conventional fuel and 9% subsidy on biojet is considered. With these changes, model simulations show that the new price of biojet is $3.33 per gallon, while the price of conventional fuel is $3.36 per gallon (Table 4). The gap is three cents, which may be small enough that it would not prevent aviation operators from choosing biojet when available. The prices implied by this scenario are summarized in Fig. 5. This shows that the tax-cum-subsidy is effective in equilibrating the prices. Table 3 shows that these policies, combined with the preference change in all scenarios, causes biojet use in aviation to rise by $180 million and oilseed meal use for livestock to rise by $149 million. Compared to the pure tax in cases 3 and 4, the tax-cum-subsidy induces larger increases (6.3% and 5.0%) in the local oilseed farming and processing industries. Yet as in previous cases, much of the growth occurs outside the region, as imports of processed oilseeds rise 11.7%. Gross regional product rises by 0.8%, less than under cases 3 and 4, but more than in case 2, in which the regional economy shrunk somewhat (Table 5). Equivalent variation is –$1415.4 million, implying that the aggregate regional household is better off, although less so than in cases 3 and 4 (Table 5). This scenario is relatively revenue neutral compared to the pure subsidy or tax scenario alone, as government revenue and spending rises by 1.9%. Payments to regional labor rise by 0.8% (Table 5). Labor demand falls by 19.3% and 14.1%, respectively, for petroleum refining and other feeds farming, and rises by 13.0% in oilseed processing and refining (Table 6).
3.3. Demand change plus conventional fuel tax The above scenario shows the size of subsidy that would be required to make biojet cost competitive with conventional fuels. It also suggests that this approach is especially costly and likely to lower regional welfare. For this reason, a tax on the conventional jet fuel is now considered. As of 2011, conventional jet fuel (specifically, kerosene for aviation) was taxed by 1 cent, 11 cents, and 6 cents per gallon by the states of Oregon, Washington, and Idaho, respectively (Defense Logistics Agency [25]). Federal taxes were approximately 4.4 cents per gallon (Internal Revenue Service [26]). In this scenario, the additional tax is set such that the gap between the biojet fuel cost, and the conventional fuel cost, is reduced to near zero. In case 3 a tax of 19.5% is imposed, raising the price of conventional fuel to $3.66. In case 4 a tax on conventional fuel of 22.6% is imposed, raising the price of conventional fuel to $3.75. Both taxes bring the price of conventional fuel approximately in line with the price of the alternative fuel (Fig. 5). These two are chosen, however, to illustrate that as the conventional fuel is taxed, it has the potentially unanticipated consequence of pushing up the price of biojet as well. This rises from $3.69 per gallon in the baseline up to $3.79 in case 3, and up to $3.80 in case 4. The gap is therefore $0.13 in case 3, and $0.05 in case 4. The gap narrows as the tax rises, but nonetheless, the very fact that biojet is rising in price is undesirable since jet fuel prices (whether made from kerosene or oilseeds) are rising above what either was initially. This happens for the same reason that it did in Case 1; rising demand for biojet raises its price. As the prices of both fuels rise, air transportation becomes more costly. It rises by 3.9% under case 3% and 4.6% under case 4 (Table 4), in contrast to the slight fall in price in case 2 (the subsidy). Despite this outcome, increasing the tax on conventional fuel by approximately 20% increases the odds that a new supply chain will develop by reducing the price disadvantage of biojet. This also has the advantage of expanding the regional economy, if only slightly. Gross regional product rises by 1.8% in case 3, and by 2.1% in case 4 (Table 5). The tax generates sufficient revenue to slightly boost spending in the regional economy. In contrast to case 2, payments to labor and capital rise in these scenarios by 1.8% and 1.2% in case 3, and by 2.1% and 1.4% in case 4. Changes in labor demand are reported in Table 6. While there is shrinking labor demand in oilseed and other feed farming, along with petroleum refining, these losses are offset by labor demand growth in the rest of the economy, as the new tax revenue provides spending in a
3.5. Discussion Biofuels are sometimes promoted as having the potential to positively impact a given region. For the Pacific Northwest in 2011, the value of the regional oilseed processing sector was $2420.1 million (Table 3). In all cases examined here, activity in this sector rises. However, this local supply is not enough to meet the hypothesized demand in any of the cases. In 2011 there were $2736.8 million imports of processed oilseeds, and in these cases, the imports always increased by more than does activity in this sector in the region itself. This ease of substitution constrains the ability of the biojet industry to have a large effect within the immediate region. The implication is that providing subsidies for processing does not imply that production of the feedstock will occur within the region. The inter-regional substitution possibilities are too great for oilseed processors and refiners to source only from the local area, and to have all operations occur within the Pacific Northwest. 4. Conclusions This study develops a regional multi-market economic policy model to examine supply chain issues as the aviation sector seeks diversify its fuel sources away from conventional fuels. Technological and engineer9
Renewable and Sustainable Energy Reviews (xxxx) xxxx–xxxx
J.J. Reimer, X. Zheng
constitute a case against camelina as a potential energy feedstock. It is clear that substantial policy interventions will be required to align private incentives with public goals, given current price and market conditions for alternative fuels. For other readers, the results might be encouraging and could serve to narrow the range of policy choices. In either case, it is hoped that this study has shed light on the nature of economic incentives that would be necessary to bring this potential bioenergy resource into fruition, as well as its potential impacts on the broader economy.
ing aspects are not modeled in great detail; rather, the focus is on the transmission and size of price signals from end users to raw feedstock producers such that biojet can be reasonably competitive in price with conventional jet fuel. Five cases are considered, the first of which concerns a consumer demand change only with no new policy changes taking place. This is intended to reflect a scenario in which a subset of airline fliers voluntarily elect to pay more, allowing airlines to use the proceeds to offset the higher cost of biojet, and bring the alternative biofuel supply chain into fruition. This approach leaves a large gap between the prices of the two fuels. Without a segment of consumers willing to pay extra for the use of biojet, this case is unlikely to arise in the real world. In some sense this is therefore a “business as usual” case, for which public policy lies dormant and little progress is made. This would appear to explain the current situation of camelina-based biojet. Given that aviation biofuels appear to be among the energy policy priorities of the United States, a number of public policy mechanisms are next considered through a series of cases. The first is a subsidy on biojet that would bring its cost in line with conventional fuel. The subsidy is determined to be approximately 17%. A problem with this approach is that all taxpayers pay for the subsidy, instead of just users of the fuel. This case would only make sense if biojet can be viewed as providing a public good. This is an expensive approach, generally, and entails the least favorable macro-economic outcomes. Gross regional product declines and other measures of general economic welfare are unfavorable. As another possibility, a tax on the conventional fuel is considered. It would need to be approximately 20% to make end users indifferent between it and alternative fuels. One problem is that this raises the demand for biojet to an extent that it itself becomes ever more expensive. This makes it even harder to narrow the price gap between it and the conventional fuel. This make air transportation approximately 4% more costly, which may be an undesirable outcome from the viewpoint of the aviation industry, who otherwise would be in favor of the supply diversification motivation as envisioned in this study. A more favorable scenario, examined with another case, would involve a combined 9% subsidy on biojet and 9% tax on the conventional fuel. This mitigates the gap between the fuel prices, is approximately revenue neutral for government, and has other favorable macroeconomic outcomes as well. It is less likely to induce distortions in markets beyond those of central interest to the example of this study. Another finding is that providing subsidies for processing does not imply that production of the feedstock will occur within the region. The inter-regional substitution possibilities – in the form of easy transportation across regions – are too great for oilseed processors and refiners to source only from the local area, and to have all operations occur within the region. This limits the ability of this industry to have an effect on regional economic development. This issue matters little with respect to overall economic efficiency in the country as a whole. However, it could make this less appealing to regional policymakers, whose support may be critical to any push for implementation of the policies considered here. One of the novel features of the analysis is the historical validation technique that provides statistical measures of model performance. In particular, the study presents a means of model validation that, upon a careful review of the literature, appears to be a first for a regional computable general equilibrium model. The model is calibrated such that it can replicate the size of past price changes for actual historical oilseed yield changes. Due to this validation, greater confidence can be placed in the measured impacts of new biofuel demand on the regional economy. This provides a baseline with respect to how isolated or connected the Pacific Northwest is to the rest of the United States, and the rest of the world, in the case of oilseeds, as either a raw feedstock or as a processed biofuel, namely biojet. In summary, for some readers, the results presented herein might
Acknowledgements This project is partially supported by Formula Grant no. 2012– 67009-19917 of the United States Department of Agriculture National Institute of Food and Agriculture, as well as Oregon State University. References [1] Bauen A, Howes J, Bertuccioli L, Chudziak C. Review of the potential for biofuels in aviation: final report for CCC. Fuel 2009:6021–5. [2] Shonnard DR, Williams L, Kalnes TN. Camelina-derived jet fuel and diesel: sustainable advanced biofuels. Environ Prog Sustain Energy 2010;29:382–92. http://dx.doi.org/10.1002/ep.10461. [3] IATA. Report on alternative fuels 2014:65. 〈https://www.iata.org/publications/ Pages/alternative-fuels.aspx〉. [4] Natelson RH, Wang W-C, Roberts WL, Zering KD. Technoeconomic analysis of jet fuel production from hydrolysis, decarboxylation, and reforming of camelina oil. Biomass Bioenergy 2015;75:23–34. http://dx.doi.org/10.1016/j.biombioe.2015.02.001. [5] Stein L. Economic analysis of potential camelina oil crop supplies in the northwest U.S. Oregon State University; 2012. [6] EPA. Regulation of Fuels and fuel additives: identification of additional qualifying renewable fuel pathways under the renewable fuel standard program. Washington, DC: 2012. [7] Winchester N, McConnachie D, Wollersheim C, Waitz I a. Economic and emissions impacts of renewable fuel goals for aviation in the US. Transp Res Part A Policy Pract 2013;58:116–28. http://dx.doi.org/10.1016/j.tra.2013.10.001. [8] Putnam DH, Budin JT, Field LA, Breene WM. Camelina: a promising low-input oilseed. In: Janick J, Simon JE, editors. New Crop.. New York: Wiley; 1993. [9] Hulbert S, Guy S, Pan W, Paulitz T, Schillinger W, Sowers K. Camelina production in the dryland Pacific Northwest. Pullman WA; 2012. [10] Cansino JM, Cardenete M a, González-Limón JM, Román R. Economic impacts of biofuels deployment in Andalusia. Renew Sustain Energy Rev 2013;27:274–82. http://dx.doi.org/10.1016/j.rser.2013.06.006. [11] IMPLAN. MIG IMPLAN data and software 2012. [12] Tabatabaie SMH, Murthy GS. Life cycle assessment of biodiesel production from Camelina in the Pacific Northwest U.S. Corvallis OR; 2015. [13] Valenzuela E, Hertel TW, Keeney R, Reimer JJ. Assessing global computable general equilibrium model validity using agricultural price volatility. Am J Agric Econ 2007;89:383–97. http://dx.doi.org/10.1111/j.1467-8276.2007.00977.x. [14] Diebel PL, Ball DA. An economic analysis of alternative wheat rotations for eastern Oregon. Corvallis OR; 1999. [15] Walsh ME. Method to estimate bioenergy crop feedstock supply curves. Biomass Bioenergy 2000;18:283–9. http://dx.doi.org/10.1016/S0961-9534(99)00094-X. [16] McCullough M, Holland D, Painter K, Stodick L, Yoder J. Economic and environmental impacts of Washington state biofuel policy alternatives. J Agric Resour Econ 2011;36:615–29. [17] Reimer J, Weerasooriya SA, West T. How does the Supplemental Nutrition Assistance Program affect the United States economy. Agric Resour Econ Rev 2015;43:233–52. [18] Löfgren H, Robinson S, Harris RL. A standard computable general equilibrium (CGE) model in GAMS. doi:338.1′01'51—dc21; 2002. [19] Rose A, Liao SY. Modeling regional economic resilience to disasters: a computable general equilibrium analysis of water service disruptions. J Reg Sci 2005;45:75–112. http://dx.doi.org/10.1111/j.0022-4146.2005.00365.x. [20] Reimer JJ, Zheng X, Gehlhar MJ. Export demand elasticity estimation for major U.S. crops. J Agric Appl Econ 2012;4:501–15. [21] Bilgic A, King S, Lusby A, Schreiner D. Estimates of U.S. regional commodity trade elasticities of substitution. Reg Anal Policy 2002;32:79–98. [22] Hodges A, Rahmani M. Cost of product of Camelina Oil in North Florida; 2012. [23] Schumacher J. Overview of Oilseed crushing and processing technologies; 2008. [24] Stein L, Seavert C, Macnab S, Petrie S. Oregon State University Enterprise Budget for Camelina Following Winter Wheat. Corvallis OR; 2014. [25] DLA. Compilation of U.S. fuel taxes 2012:1–428. 〈http://www.energy.dla.mil/ DLA_counsel_energy/Documents/Tax Documents/Tax compilation 2012-04.pdf〉. [26] IRS. Internal revenue service report on fuel taxes 2015. 〈http://www.irs.gov/ publications/p510/ch01.html#en_US_2013_publink1000116905〉.
10