Innovation and productivity: the recent empirical literature and the state of the art

Which types of innovation reliably increase productivity, and under what conditions?

Jacques Mairesse, Pierre Mohnen, and Ad Notten examine a decade of empirical work on innovation and productivity in their paper "Innovation and productivity: the recent empirical literature and the state of the art" (2025).

They review more than a thousand peer-reviewed papers published between 2013 and 2023, most of which linking R&D, innovation output, and firm productivity across advanced and emerging economies.

Their main conclusions include:

• Productivity is consistently related to innovation, though the strength of the relationship varies with firm size, industry intensity, and the formal or informal nature of the business.
• Process and organisational innovations tend to have a more reliable positive effect on productivity than product innovations, except in advanced economies and high-tech sectors.
• In services, non-technological innovations such as changes to business processes or marketing often deliver larger productivity improvements than technological ones.
• Studies found adopting modern management practices raised productivity by ~17% in Indian textile firms, and Chinese innovating firms outperform matched non-innovators by ~11%.
• German firms using AI show higher productivity independent of conventional research and development, and patent data suggest AI intensity is positively associated with labour productivity worldwide.
• Evidence for complementarity between product, process, organisational, and marketing innovations remains mixed.
• There is stronger support for complementarity between innovation types and external conditions such as human capital and competitive intensity.

This study makes the case for broader intangible disclosure beyond R&D spend, like management practices, artificial intelligence adoption, and organisational capabilities to assess productivity potential.

Industry context is decisive: for manufacturing firms, process and patent-based signals dominate, while for services portfolios, organisational and marketing innovation often explain more productivity dispersion.

Many underlying studies rely on cross-sectional or short-panel innovation surveys, which the authors acknowledge tend to overstate the innovation-productivity relationship compared to panel analyses with firm fixed effects.

Practitioners should therefore discount point estimates taken from single country cross-sectional studies and weight more heavily results from panel designs or quasi-experiments to sharpen causal claims.