Agricultural soils are a major source of atmospheric nitrous oxide (N2O), a potent greenhouse gas (GHG). Because N2O emissions strongly depend on soil type, climate, and crop management, their inventory requires the combination of biophysical and economic modeling, to simulate farmers' behavior. Here, we coupled a biophysical soil-crop model, CERES-EGS, with an economic farm type supply model, AROPAj, at the regional scale in northern France. Response curves of N2O emissions to fertilizer nitrogen (Nf) inputs were generated with CERES-EGC, and linearized to obtain emission factors. The latter ranged from 0.001 to 0.0225 kg N2O-N kg-1 Nf, depending on soil and crop type, compared to the fixed 0.0125 value of the IPCC guidelines. The modeled emission factors were fed into the economic model AROPAj which relates farm-level GHG emissions to production factors. This resulted in a N2O efflux 20% lower than with the default IPCC method. The costs of abating GHG emissions from agriculture were calculated using a first-best tax on GHG emissions, and a second-best tax on their presumed factors (livestock size and fertilizer inputs). The first-best taxation was relatively efficient, achieving an 8\% reduction with a tax of 11 euro/t-CO2-equivalent, compared to 68 euro/t-CO2eq for the same target with the second-best scheme.