Documentation for GCAM
The Global Change Analysis Model
View the Project on GitHub JGCRI/gcam-doc
Table 1: Inputs required by the supply module 1
| Name | Resolution | Unit | Source |
|---|---|---|---|
| Land use and land cover | By GLU, land type, and year | thousand \(km^2\) | Land Allocator |
| Historical harvested area (used to calculate historical yield) | By GLU, crop, management practice, and year | thousand \(km^2\) | Exogenous |
| Historical production (used to calculate historical yield) | By GLU, crop, management practice, and year | thousand \(km^2\) | Exogenous |
| Agriculture productivity growth (used to calculate future yield) | By GLU, crop, management practice, year | % per year | Exogenous |
| Non-land variable cost (also no water/fertilizer/labor/capital) of production | By GLU, crop, management practice, and year | 1975$/kg | Exogenous |
| Labor coefficients | By GLU, crop, management practice, and year where applicable | physical input per unit output | Exogenous |
| Capital coefficients | By GLU, crop, management practice, and year where applicable | capital input per unit output | Exogenous |
| Labor prices | By GCAM region and year | 1975$/person or related calibrated unit | Economy module |
| Capital prices | By GCAM region and year | unitless gross rate of return | Economy module |
| Fertilizer coefficients | By GLU, crop, management practice, and year | 1975$/kg | Exogenous |
| Water coefficients | By GLU, crop, management practice, and year | 1975$/kg | Exogenous |
| Commodity prices | By region, commodity, and year | 1975$/kg | Marketplace |
| Fertilizer prices | By region, commodity, and year | 1975$/kg | Marketplace |
| Water prices | By basin and year | 1975$/m3 | Marketplace |
1: Note that this table differs from the one provided on the Supply Inputs Page in that it lists all inputs to the land supply module, including information passed from other modules. Additionally, the units listed are the units GCAM requires, rather than the units the raw input data uses.
Variable costs are the non-land shutdown costs of producing a crop, expressed per unit of output. They create price floors in GCAM: if the crop price falls to or below variable cost, production can fall to zero.
GCAM represents several input costs explicitly, including fertilizer, water, agricultural labor and capital where applicable. These explicit inputs should not be double counted in the residual varCost category. Variable costs should therefore be interpreted narrowly as minimum production costs that are not otherwise represented.
Variable costs should not include land costs, returns to capital, profits, depreciation, or owner wages. These value-added components are represented through land allocation, profit rates, or explicit labor and capital inputs rather than through shutdown costs. If labor-cost data are used in residual variable costs, they should be limited to hired labor rather than owner labor or expected farm income.
Region-specific variable costs should be used carefully because they create region-specific price floors. If technical change lowers global prices, high variable-cost regions may stop producing a commodity for reasons driven by the imposed floor rather than by broader land-allocation dynamics.
In summary:
Labor and capital inputs are explicit inputs to primary agricultural production. These inputs are downscaled to agricultural technologies and used to trace labor costs, wages, capital returns, and agricultural value-added by region, sector, and technology. Labor-intensive and capital-intensive technologies can be differentiated through input coefficients, so changes in wage rates or capital rental prices can affect technology profitability and production choices.
Within primary agriculture, a common regional agricultural wage rate is assumed across agricultural sectors, reflecting labor mobility within agriculture. Total regional labor supply is linked to socioeconomic employment assumptions and allocated between agriculture and the Materials sector in the broader GCAM-Macro framework. Agricultural capital is also connected to investment tracking and the regional savings-investment closure through GCAM-Macro.
Profit rate for all agricultural production technologies is calculated as:
\[profitRate = 1e9*( price + subsidy - varCost - inputCosts + secondaryValue ) * yield + impliedSubsidy\]where \(price\) is the commodity price, \(subsidy\) is any exogenously-specified subsidy, \(varCost\) is the non-land variable cost, \(inputCosts\) are the costs of inputs (e.g., fertilizer, water), \(yield\) is the yield for the technology, and \(impliedSubsidy\) is an implicit subsidy calculated in the calibration periods to ensure profits are above a specified threshold. Note that the subsidy is multiplied by \(1e9\), as the land allocator expects profit rates in 1975$/billion m2.
Note, \(inputCosts\) may include explicit labor and capital inputs in addition to inputs such as fertilizer and water.
See calcProfitRate in ag_production_technology.cpp.
Agricultural supply is calculated as:
\[supply = yield * land\]where \(yield\) is the yield for the technology and \(land\) is the land allocation (retrieved from the land allocation module).
See calcSupply in ag_production_technology.cpp.
For technologies that have production in the historical period, yield for the historical period is calculated as:
\[yield = supply / land\]For all technologies, future yield is calculated as:
\[yield_{t} = yield_{t-1} * (1 + APG)^{timestep}\]where \(yield_{t}\) is the yield in time \(t\), \(APG\) is the agricultural productivity growth rate and \(timestep\) is the number of years between \(t\) and \(t-1\).
See initCalc in ag_production_technology.cpp.
GCAM uses one of two different logit formulations to calculate the shares for each technology or subsector. For livestock, subsectors represent different production systems, where technologies represent different feed sources.
The first option, also known as the relative-cost-logit, is:
where \(s_i\) is the share of technology or subsector \(i\), \(alpha_i\) is the share weight, \(c_i\) is the cost of technology or subsector \(i\), and \(beta\) is the logit exponent.
The second option, also known as the absolute-cost-logit, is:
where \(s_i\) is the share of technology or subsector \(i\), \(alpha_i\) is the share weight, \(c_i\) is the cost of technology or subsector \(i\), and \(beta\) is the logit exponent.
See relative cost logit and absolute cost logit.
There are a number of ways that policies can be applied directly to influence the land sector in GCAM. These include the following.
Protected Lands: With this policy, GCAM users can set aside a fraction of natural land, removing it from economic competition. This land cannot be converted to crops, pasture, or any other land type. This is similar to real-world policies such as reducing emissions from deforestation and forest degradation (REDD). See more on how these assumptions can be adjusted here.
Valuing carbon in land: When applying a price on carbon through any of the emissions-related policy approaches, GCAM users can choose whether that price extends to land use change CO2 emissions. This policy is modeled as a subsidy to land-owners for the holding carbon stocks as opposed to a price on the emissions themselves.
Bioenergy constraints: GCAM users can impose constraints on bioenergy within GCAM. Under such a policy, GCAM will calculate the tax or subsidy required to ensure that the constraint is met. By default two bioenergy related constraints are enabled in GCAM and described below. Alternative approaches could be used for more direct constraints as show in the examples.
Negative emissions budget: This constraint limits the total gross value of all negative emissions, be it bioenergy or otherwise, to a certain fraction of GDP as defined by energy.NEG_EMISS_GDP_BUDGET_PCT in constants.R. To enforce the constraint the model will scale back the value of the subsidy given to bioenergy to stay within the budget. Thus limiting the value but not necessarily the quantity of negative emissions. Note this budget is also applied to the valuation of carbon in land when running such a policy, however the actual value of those emissions are not included in this budget at the moment.
Biomass externality cost: We include an additional constraint which is meant to represent the costs paid for various externalities resulting from large scale production of purpose grown bioenergy crops. This constraint is represented with an increasing cost with higher levels of production as defined in A27.GrdRenewRsrcCurves.csv.
Land constraints: GCAM users can constrain the amount of land of a particular type in a given region. Under such a policy, GCAM will calculate the tax or subsidy required to ensure that the constraint is met. See example.
Land-use modification strategies reduce the cost of limiting atmospheric CO2 concentrations, but can make crop prices rise and transform human diets, for example, when people consume less beef and other carbon-intensive protein sources. The rate at which crop productivity is improved has a strong influence on emissions from land-use change. Thus, the technology used for growing crops is potentially as important for limiting atmospheric CO2 as are approaches like CO2 capture and storage (Wise et al. 2009).
Agricultural commodities