A key feature of the 2008 revision of the System of National Accounts was the treatment of R&D expenditure as investment. The question arises whether the standard approach towards accounting for growth contribution of assets is justified given the special nature of R&D that provides capital services by affecting the working of other inputs as a whole – akin to technical change and often requires up-front investment with sunk costs. We model R&D inputs with a restricted cost function and compare econometric estimates with those derived under a standard index number approach but find no significant differences. However, we cannot reject the hypothesis of increasing returns to scale. The standard MFP measure is then broken down into a scale effect and a residual productivity effect, each of which explains about half of overall MFP change. The scale effect points to the importance of the demand side and market size for productivity growth. We also compute mark-up rates of prices over marginal cost and find widespread evidence of rising mark-ups for the period 1985-2016.