Firm price-cost markups may reflect (a) bigger step sizes from quality innovations that confer significant knowledge spillovers onto other firms, and/or (b) higher process efficiency than competing firms. We write down an endogenous growth model in which, compared with the laissez-faire equilibrium, the social planner would generally like to reallocate research resources towards high markup firms in case (a) so as to capitalize on knowledge spillovers but not in case (b). We then exploit unit price variation across firms in French manufacturing to assess the relative strength of these two forces. Viewed through the lens of our model, the French data are consistent with significant variation in innovation step sizes, and hence gains from mitigating R&D misallocation. The policy implication is that, to reach the social optimum, French research subsidies should favor only those high markup firms with “good” rents.