Voluntary emissions offset programs between developing and industrialized countries suffer from adverse selection, because participants will self-select into the program. In contrast, pure subsidies for mitigation lead to full participation and hence efficiency, but require large financial transfers which make them unattractive to industrialized countries. We present a simple model to demonstrate the impact of three policy options on the performance of offset programs: (1) baseline scale increases, (2) offset discounting and (3) setting stringent baselines. With baseline scale increases, entire political jurisdictions such as regions or nations are assigned a single, aggregate baseline and must choose whether to participate as one entity. We find that increasing scale both improves efficiency and reduces transfers from offset buyers to sellers. Offset discounting means paying less than the value of abatement and can be paired with trading ratios between offsets and allowances in a cap-and-trade system. We show that discounting is inefficient, but can make offsets more attractive to industrialized countries. Setting stringent baselines also involves a tradeoff between efficiency and transfers. We finally show that Pareto efficient policies that are individually rational for buyers and sellers entail some combination of discounting and/or stringent baselines: offset policies are never first-best, but can be efficiency improving, especially with increased scale.