The Great British productivity puzzle and regional industries: a dynamic multilevel spatial frontier analysis

Using two approaches, we contribute to sub-national economic performance measurement by netting out the in uence of a region s performance on its industries. The rst approach measures performance using technical e¢ ciency and involves introducing the rst multilevel spatial stochastic frontier model. The second involves using the results from the rst approach to introduce a dynamic multilevel spatial TFP growth decomposition. Using panel data, we apply the approaches to a production frontier for the eight industries nested in Great Britain s NUTS3 regions (2004 20). The motivation is the well-known UK productivity puzzle, which concerns a lack of de nitive reasons why its post-2008 productivity slowdown has been noticeably bigger than comparator countries. One suggested contributing factor are the regional economic performance di¤erences within the country, where both our approaches provide clearer evidence on this by reporting estimates of net regional industry economic performance. For three of the eight industries (Construction; Distribution, hotels and restaurants; Other services), we observe negative mean own and spillover TFP growth, which is consistent with the country s productivity slowdown and its economy being service oriented.