Abstract

What difference does government support of business R\&D make to the rate of innovation? Addressing this important question has deep theoretical roots and broadening practical applications in OECD countries. Analysis of output additionality has been hampered by incomplete data combined with inherently difficult methodological challenges. In this light, we contribute to the formative literature in three main ways: we analyze comprehensive panel data of Norwegian enterprises across the past 25 years; we include trademarks as well as patents to improve measures of innovation output; and we apply machine learning methods to estimate treatment effect functions to address the problem of confounding factors. Our findings support and elaborate earlier work, such as Nilsen et al (2020), that fiscal stimulus tends to have greatest impact on previously non-innovative firms. The impact of support measures, alone or in combination, is on the extensive rather than intensive margin.