This paper presents the first empirical test with German firm level data of a hypothesis derived by Bustos (AER 2011) in a model that explains the decision of heterogeneous firms to export and to engage in R&D. Using a non-parametric test for first order stochastic dominance it is shown that, in line with this hypothesis, the productivity distribution of firms with exports and R&D dominates that of exporters without R&D, which in turn dominates that of firms that neither export nor engage in R&D. These results are in line with findings for Argentina. The model, therefore, seems to be useful to guide empirical work on the relation between exports, R&D and productivity.