Modeling Carbon Emissions Charges and Carbon Sequestration Payments
in U.S. Agriculture: A Description of the USMP Model 1
Robert House, Jan Lewandrowski,
Mark Peters, and Carol Jones 2
September 28, 2001
- This paper was prepared for posting on the web page of the, “Forestry and Agriculture Greenhouse Gas Modeling Forum.” The Forum was held October 1-3, 2001 at the U.S. Fish and Wildlife’s National Conservation Training Center in Shepherdstown WV. The material presented is meant to serve only as background material for the USMP based presentations at the conference. The material has not been peer reviewed and should not be cited in that context without the authors’ permission.
2. The authors are economists with USDA’s Economic Research Service.
Modeling Carbon Emissions Charges and Carbon Sequestration Payments
in U.S. Agriculture: A Description of the USMP Model
Model Overview
To estimate the farm sector impacts of a national program to reduce GHG emissions using both carbon charges and sequestration payments, we employ the U.S. Regional Agricultural Sector Model (USMP). USMP is a spatial and market equilibrium model designed for assessing a wide range of economic, environmental, and policy issues of interest to the U.S. farm sector. The model is linked with regularly updated USDA production practices surveys (USDA, 1992 and 1996), the USDA multi-year baseline ( and geographic information system (GIS) databases such as the National Resources Inventory (USDA, 1994). USMP simulates how changes in commodity market conditions, agricultural technologies, and farm policies related to commodity production, resource use, the environment, and trade would impact the farm sector. These impacts are captured through changes in regional commodity supplies, commodity prices, commodity demands, use of production inputs, farm income, farm welfare measures, government expenditures, participation in government commodity programs, and various indicators of environmental quality.[1]
USMP depicts the U.S. farm sector in considerable geographic, commodity, and production enterprise detail. Figure 1 shows USMP’s geographic coverage. The model parses the contiguous 48 states into 45 regions defined by the intersection of the 10 USDA farm production regions and 26 land resource regions. Table 1 details USMP’s commodity coverage. Crop production includes corn, sorghum, oats, barley, wheat, rice, cotton, soybeans, hay and silage. Collectively, these 10 crops account for about 75 percent of the value of U.S. agricultural production (USDA, 1999).[2] USMP’s commodity coverage also includes 16 primary livestock products (the most important being dairy, swine, beef cattle, and poultry) and over 2 dozen processed and retail products (including dairy, pork, fed beef, nonfed beef, poultry, soy meal, soy oil, and livestock feed). With respect to enterprise management, USMP includes nearly 1,000 production activities that reflect different choices farmers can make regarding crop rotations, input mixes, and production technologies (particularly the choice of tillage system). Another 70 production activities process farm commodities into intermediate and final demand products.
Structurally, U.S. agriculture is assumed to operate in a competitive market framework – subject to government program interventions - with producers maximizing profits and consumers maximizing utility. Operationally, USMP’s objective function is to maximize the sum of consumer and producer surplus across all commodities. With respect to inputs, the markets for cropland, pasture land, family labor, hired labor, and irrigation water are modeled at the regional level with upward sloping supply curves (i.e., the supply of these inputs increase with price). All other farm input markets – some 23 in all including fuels, fertilizers, pesticide, seed, machinery, and custom operations - are modeled at the national level. In national input markets supply functions are perfectly elastic - implying supply can be increased without increasing price. [3]
Production activities are generally represented by fixed coefficient production functions. In the case of crop production enterprises, fertilizer inputs per acre (and corresponding yields) are variable coefficient. In the case of livestock feed rations, relative feed inputs are permitted to vary according to relative feed grain and livestock market prices (subject to physical feed ration restrictions pertaining to specific types of livestock.). For crop commodities, production activities are differentiated by tillage practice, multi-year crop rotation, dryland or irrigated system, participation in government farm programs, and other characteristics. USMP production units reflect representative production units over specified geographic areas (e.g. a state or region). Hence, each production activity in USMP is an average of production techniques in the geographic area it represents.
Final product markets are modeled at the national level. On the demand side USMP distinguishes between the demands for domestic consumption, export, commercial stocks, and government stocks. Government farm programs in USMP include production flexibility contract payments, target prices, acreage reduction, acreage flexibility, acreage diversion, conservation reserve, and CCC commodity loan programs.[4] Participation in farm programs is voluntary and is determined endogenously in response to market forces affecting the costs and returns associated with commodity production, participation costs, and program benefits.
Baseline and Model Simulations
USMP provides comparative static analysis from any base year between 1988 and 2010 inclusive. Simulations begin by calibrating the model’s acreage, tillage shares, production, consumption, trade, and price variables exactly to a specified historical or USDA baseline year and corresponding geographic information. [5] A change in farm policy or market conditions is then introduced and the model endogenously determines new levels of acreage, tillage shares, production, consumption, trade, and prices for all crop and livestock commodities and most processed products that would result after all output and input markets have fully adjusted. Because adjustment paths are not modeled, USMP simulations are properly viewed as a comparison between an initial and a new medium- run equilibrium state.[6] Within USMP farm sector responses to price changes are determined by all producers adjusting their input use, output choices, and production levels such that the net returns to the last units of each commodity produced equilibrate (i.e., marginal revenue = marginal cost), and likewise, the marginal values of the last unit of different inputs used equilibrate (i.e., input marginal value product = input marginal cost).
To facilitate comparisons of our results with those presented in other studies, we refer to the literature in selecting both the timing of our GHG mitigation program and the range of carbon prices we analyze. Due largely to the ongoing international negotiations related to the Kyoto Protocol, the recent scientific literature related to climate change mitigation has focused on achieving GHG emissions reductions between 2008 and 2012 - the timeframe mandated in the Protocol. Hence , we assume our carbon charges and sequestration payments are implemented in 2010. Viewing carbon charges and sequestration payments generically as a price for carbon emissions, we assess six alternative carbon prices - specifically $10, $25, $50, $75, $100, and $125 per metric ton carbon. These prices intersect the ranges considered in several recent studies (Francl, 1997; McCarl et al, 1999; Stavins, 1999; Plantinga et al., 1999; and Peters et al., 2001).
To obtain our reference – or business as usual (BAU) - scenario we calibrate USMP to the supply, demand, production, acreage, regional crop tillage, government program, input cost, and other conditions projected in the USDA baseline for 2010. Hence the scenarios examined in this analysis are properly interpreted as reflecting differences in economic conditions relative to a 2010 world in which no actions to reduce GHG emissions are taken.
With USMP calibrated to produce the commodity market, input market, and agricultural trade conditions in the USDA Baseline Projection for 2010, we assess the farm sector impacts of a national GHG emissions mitigation strategy that includes charges on the carbon emissions of energy intensive inputs and payments to farmers for adopting various land use and production practice changes that sequester additional carbon on agricultural lands.
Charges on the carbon emissions of energy intensive inputs, in one form or another, are a standard feature of proposals to reduce GHG emissions because fossil fuel combustion accounts for 98 percent of U.S. CO2 emissions and 81 percent of all U.S. GHG emissions (EPA, 2001). While agriculture is heavily dependent on energy, its share of national energy consumption is relatively small. In this analysis we assume that the charge on carbon emissions is determined outside of the farm sector and that input suppliers pass all of any related increases in the prices on energy intensive inputs on to farmers. In USMP then, the agricultural sector responds to carbon charges by adjusting production levels and shifting to less energy intensive mixes of inputs and outputs.
The carbon sequestration component of our GHG mitigation strategy pays farmers to afforest marginal cropland and pasture, shift cropland into grasses, expand the use of conservation tillage systems (particularly no-till systems), and to switch to less energy intensive crop rotations. To prevent farmers from “churning” lands among uses and/or practices to take advantage of program payments while not increasing the net amount of carbon stored on agricultural lands, our sequestration program penalizes farmers by the amount of the carbon charge for making changes in land uses and production practices that emit more carbon than what they are presently doing.[7] From a policy perspective the penalty has two important implications. First, it ensures that farmers are paid only for the net sequestration that results from the incentives. Second, it limits sequestration payments to farmers who are not already practicing the desired land uses and production practices.
Following USDA’s historical tendency to limit the duration of conservation programs – particularly ones that emphasize land retirements – we assume our carbon sequestration program lasts for 15 years. The finite time horizon of the sequestration program has an important implication for the structure of sequestration payments. Specifically, with respect to GHG mitigation, the value of a unit of carbon sequestered in agricultural sinks would only equal the value of an equivalent unit of carbon emissions reduction if the sequestered carbon remained in the sink permanently. In the framework developed here, this outcome is unlikely. If farmers stop receiving sequestration payments after 15 years then it is possible that many – if not most – of the affected lands would soon be shifted back to uses and practices that emit more GHGs. If this occurs much of the carbon sequestered over the 15-year program period could be quickly released back into the atmosphere.
To the extent that lands managed for carbon sequestration are returned to their present use and/or management system, society would receive much less GHG mitigation than from an equivalent amount of carbon that had not been emitted into the atmosphere (i.e., emissions reduction). The emission reduction would count as a permanent reduction in net emissions while the carbon sequestered in agricultural sinks would only be locked up for a period of 1 to 15 years. Specifically, the carbon sequestered in year one would be stored for 15 years, the carbon sequestered in year two would be stored for 14 years, and so forth until year 15 when the incremental carbon sequestration would only be stored for 1 year. To reflect this time path of temporary carbon accumulation we discount the sequestration payments assuming that all carbon sequestered as a result of the program is released back into the atmosphere at the end of the program.
Finally, one management option on lands that are afforested as a result of offering farmers carbon sequestration payments would be to sell the associated timber after the program ends. Since landowners have a reasonably secure expectation that this asset will exist as a result of their participation in the afforestation part of the carbon sequestration program, the expected value of this timber should also be reflected in farmer’s afforestation decisions. Our framework adjusts the payments farmer receive to afforest agricultural lands to reflect the annualized value of a 15 year old stand of timber.
To help readers interpret our empirical results, Box 1 summarizes the key assumptions underlying our USMP simulations.
Implementing carbon charges for energy emissions in USMP
To implement a carbon charge on energy emissions in USMP we first estimate how the charge would impact the prices of gasoline, diesel, LP gas, natural gas, three fertilizers (nitrogen, phosphorous, and potash), a composite chemical called pesticide, and electricity. Price increases for these inputs are based on the level of the carbon charge and the carbon embodied in a unit of the input. For gasoline, diesel, LP gas and natural gas “embodied” means the carbon emitted in the on farm use of the inputs. For fertilizer and pesticide embodied means the carbon emitted in manufacture and distribution of the inputs to the farm. For electricity embodied means the carbon emitted in the generation of electricity used on the farm. Estimates of the per unit carbon content of these inputs are detailed in table 2 and were obtained or derived from the energy literature.
The inputs listed in table 2 are key components of several more aggregate energy intensive inputs in USMP – including fuels and oils, electricity, fertilizer and lime, pesticides, marketing and transportation, and custom services. Expenditures for fuel, lubrication, and electricity in crop and livestock budgets are referred to as “direct energy costs.” Expenditures for energy products embodied in other inputs such as chemicals and custom operations are referred to as “indirect energy costs.” For each crop and livestock enterprise in USMP, the direct and indirect energy expenditure categories were decomposed into their component parts using ERS cost-of-production (COP) data. The resulting expenditures for diesel, gasoline, lubrication, natural gas, LP gas, and electricity were then divided by their respective prices to calculate the quantities of each input used per unit of output for each production system. Finally, the estimated price increases were applied to the costs of diesel, gasoline, lubrication, natural gas, LP gas, and electricity in USMP’s energy intensive inputs for each of the model’s crop and livestock production systems assuming that input suppliers pass the entire carbon charge on to producers. This reflects our assumption of perfectly elastic supply curves in national input markets and results in input prices increasing in direct proportion with the level of the charge and the per unit carbon content of each input.[8]
Implementing payments for afforestation
To implement a program of payments to shift marginal cropland and pasture into forest in USMP we need to specify regional values for the quantity of carbon sequestered annually in growing forests and the cost of establishing new forests. Estimates of the quantity of carbon sequestered by afforesting agricultural lands in each USMP region are developed from data in Birdsey (1996). Birdsey’s work appears in a larger study of forests and global change (Sampson and Hair, 1996) in which the 48 contiguous states are grouped into 8 forest regions. For each of these regions, Birdsey details per acre carbon accumulation in forests of selected tree species. Data are given in 5-year intervals starting in year 0 (i.e., conversion from pasture or cropland to forest) out to 120 years and reflect fully stocked timberland under average management conditions. Carbon values are presented for trees, soils, under story, litter and total ecosystem.
For this analysis we assigned to each USMP region the tree species associated with the most geographically similar region in the Birdsey study. For regions where Birdsey described more than one tree species, we selected the species the highest carbon accumulation value that most closely mix in Birdsey. We then took the total quantity of carbon accumulated in the ecosystem over the first 15 years of forest growth and divided it by 15 to obtain an annual rate of carbon sequestration. Our values for carbon sequestered on lands shifting from cropland and pasture to forest then reflect average per acre annual sequestration in trees, soils, under story, and litter over first 15 years of growth. The actual carbon sequestration values used for afforesting cropland and pasture are shown in table 3 for each USMP region.
Periodic estimates of forest establishment costs are not generally available for most regions of the country but are published biannually for the South. Data for 1998 (DuBois et al., 1999) indicate that the total cost of seedlings, prescribed burning, and hand planting in the Southeast in 1998 averaged about $93.29 per acre. To estimate regional forest establishment costs state level data on the value and quantity of timber harvested from National Forests in 1998 (USDA Forest Service, 1999) were used to derive a share of harvest weighted timber price for each Farm Production Region and for the Southeast as defined by the Forest Service. The Farm Production Region prices were then divided by the Forest Service Southeast region price and these ratios were multiplied by $93.29 to obtain an estimated forest establishment cost in each Farm Production Region. [9]