This article investigates the role of greenhouse gas (GHG) offset payment design on abatement efficiency in agriculture. We develop a regionally disaggregated positive mathematical programming model of California agriculture calibrated to economic and agronomic information. Regional yield and GHG emission responses to production practices are derived from a biophysical process model. The economic optimization model allows for simultaneous and continuous changes in water, nitrogen fertilizer, and tillage intensities, and captures crop substitution effects. Empirical results show that second-best policies relying on regionally aggregated emission factors lead to small abatement efficiency losses relative to the first-best policy with finer-scale emission factors. Because the costs of such second-best policies are substantially lower, this finding suggests that they could be cost-effective in California. In contrast, second-best policies targeting a single GHG or a single input entail significant abatement efficiency losses, which nonetheless can be reduced by combining policy instruments.