Modelling imperfectly appropriable R&D via spillovers

We provide an extensive comparison of the well-known R&D models of d'Aspremont-Jacquemin and Kamien-Muller-Zang, exposing full or partial conflict in key conclusions and policy prescriptions. Using a mix of formal and/or intuitive criteria, we argue that the first model is of questionable validity for large spillover values, the latter being characterized by jointly increasing returns to scale in the R&D process. An argument leading to equivalence between the two R&D processes, but not to agreement on some key conclusions, is exposited. In conclusion, the models emerge as fundamentally different, although a unilateral shift in the cost function of the first leads to agreement in all equilibrium outcomes. (C) 2000 Elsevier Science B.V. All rights reserved.