What drives endogenous growth of lagging regions? DUI innovations, knowledge bases, and regional context

This paper aims to investigate how lagging regions can achieve knowledge-driven, endogenous economic growth through doing-using-interacting (DUI)-driven innovation activities. While traditional theories of endogenous growth emphasize the importance of codified, R&D-based knowledge (STI mode of innovation), they largely neglect the innovative capacities of SMEs in lagging regions that are primarily reliant on tacit knowledge. Recent research has highlighted the importance of synthetic knowledge and informal innovation processes in such settings, but solid empirical evidence of their regional embeddedness remains scarce. Our study addresses this gap by qualitatively examining 63 SMEs across ten lagging German regions, focusing on how regional contextual factors shape three types of DUI learning: learning-by-doing, learning-by-using, and learning-by-interacting. The empirical findings reveal three central insights. First, internal learning processes are present across all ten study regions, yet their form and effectiveness are highly dependent on local vocational training structures, business culture, and the availability of skilled labor. Second, external learning via cooperation with regional partners only has lasting innovation effects when local institutions and trust-based networks support it. Third, learning-by-using is contingent on regional infrastructure and customer proximity, highlighting the significant spatial disparities in innovation capacity across lagging regions. Our results show that DUI innovations are not only viable in lagging regions; they can also serve as key drivers of regional economic growth, so long as regional characteristics are adequately considered. Ultimately, the study proposes a typology of lagging regions based on their DUI learning environments and calls for a shift in regional innovation policy towards support for informal, decentralized innovation practices. Future research may test this typology quantitatively and explore the findings’ transferability across different national contexts.