analysis were based on experimental or field data and the USDA's Erosion Productivity Impact Calculator (EPIC) (Williams and Renard, 1985).
As with WRI's case studies, the USDA's EPIC model was used to estimate soil erosion rates, short- and long-term crop yields, nutrient runoff, potential groundwater contamination, and soil carbon sequestration for each production system in each region. These representative estimates were based on the principal land resource regions (LRRs) for each of the 10 U.S. production regions and the predominant soils in those LRRs (Figure 1). The analysis was disaggregated into 48 LRRs for agronomic and environmental evaluation.4
The U.S. Math Programming (USMP) model (House, 1987), developed by the ERS over the past decade for national economic policy analysis, was adapted for use in this study. In collaboration with ERS, WRI produced an NRA version of the USMP model by extending it to include alternative commodity production systems, soil depreciation allowances, soil carbon sequestration, energy budgets, and regional natural resource damages (Figure 2). Prior to this effort, the model covered predominant production practices only and did not include any environmental impacts.
Using the completed NRA version of the USMP model, WRI tested a variety of agricultural policy scenarios for the 1995-1996 farm bill discussions. The analysis estimated several variables for each policy scenario for each region, including commodity production, commodity prices, farm income, net economic value of agricultural production, fiscal cost of income support, value and level of agricultural trade, land use, gross soil erosion, value of soil depreciation, value of soil carbon sequestered, and value of off-farm surface water impacts.
The analysis was done from the standpoint of maximizing farmers' incomes over the long term, with postsolution calculations of public welfare (Chandler et al., 1981), because farm production decisions are made by farmers, not policy makers. To estimate the value of production to the farmer, we calculated net farm income, incorporating gross operating margin, a soil depreciation allowance for changes in soil productivity, and commodity program payments. The value of off-site resource damages was excluded because farmers do not pay these.
Public welfare was estimated by calculating the net economic value of production, using gross operating margin, a soil depreciation allowance, the value of off-site surface water damages (because society absorbs the costs of these damages), and the value to society of soil carbon sequestration (because mitigating global warming benefits society as a whole). Income support was excluded because these transfer payments do not alter the net economic value.
Through the analysis described above, the revised USMP model can identify the optimal technologies for each policy scenario and estimate their potential extent of use as determined by relative profitability. In this way, estimates can be generated of the physical extent and economic value of natural resource impacts for a given policy and technology mix.