AI Summary & Key Takeaways
AI InsightReserve estimates computed on a four-branch classification are roughly a third of those computed on ten branches, so coarse sectoral data systematically hide the potential they are used to measure.
📌 3 Key Takeaways
Abstract
Structural change raises aggregate labour productivity when workers move from branches in which they are less productive to branches in which they are more productive. How much of that gain remains available to a region at a given moment has never been measured directly. This paper defines and computes the reallocation reserve: the proportional productivity gain a region would obtain by adopting the employment composition of the European frontier while holding its own branch productivity levels fixed. Using Eurostat regional accounts for 231 NUTS 2 regions in 26 countries over 2000–2022, we show that the reserve is large, highly unequal and only partly spent. Its unweighted mean falls from 35.4 to 19.0 per cent, but the median region ends the period with a reserve of 10.4 per cent, and the median ratio of realised reallocation to the reserve available in 2000 is 0.18. The reserve measured in one period predicts realised reallocation in the next, with a coefficient of 0.049 (standard error 0.008) on five-year windows conditional on country by window fixed effects, and it survives an out-of-sample test on the final window. A corollary follows for convergence: within countries, reallocation is the only component of productivity growth that converges, and once the reserve is controlled for, initial income carries no further information about it. Reserve estimates computed on a four-branch classification are roughly a third of those computed on ten branches, so coarse sectoral data systematically hide the potential they are used to measure.