Measuring What Converts: Identification Limits of the European Regional Innovation Data

Résumé fourni par la source

The Regional Innovation Scoreboard is the principal evidence base for European regional innovation policy, and a substantial applied literature estimates econometric models on it. This paper asks a prior question: what can that source identify? Using the commercialisation of novelty as a demonstration case — the sales of new-to-market and new-to-firm innovations across 218 European regions over four Community Innovation Survey waves — we establish three boundaries. First, the published annual file is not an annual dataset: biennial survey values are replicated across the reference years each wave covers, so the nominal eight-year panel contains four independent observations per region. Collapsing to survey waves changes standard errors on the annually varying registry indicators by up to forty per cent and reverses one coefficient sign. Second, the cross-sectional and longitudinal dimensions of the source do not agree. A Mundlak decomposition reveals a complete sign reversal for the leading regressor, negative between regions and strongly positive within them, and leave-one-country-out validation returns negative out-of-sample fit for every learner, locating much of the between variation in national measurement regime rather than economic behaviour. Third, flexible learners outperform the linear specification under conventional cross-validation but cease to do so under leakage-free panel designs. Within these boundaries the substantive picture is narrow: of seven innovation inputs only SME product-innovation incidence and trademark activity are associated with innovative revenue, and a taxonomy of innovation modes derived net of performance level proves nearly orthogonal to the official classification. The Scoreboard supports longitudinal inference about a region's own trajectory, not the cross-sectional benchmarking it was designed to invite.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.