We use public goods games to study farmers’ willingness to cooperate in the context of agri-environmental schemes in Germany. In a workshop with stakeholders, four treatments of a linear public goods game were co-designed, addressing the role of (1) heterogeneous endowments, (2) leading-by-example, (3) social norms, and (4) pinpointing the socially optimal solution. Initial results from a sample of approximately 350 farmers indicate that contributions in the game generally exceed those typically found in laboratory studies. Treatment effects also deviate substantially from laboratory studies. In addition to the experiment, we elicited incentivized predictions on experimental outcomes from approximately 200 experts. Expert beliefs appear to be calibrated on laboratory studies, highlighting the need to conduct experimental studies outside the laboratory.