Research by Prof Diane Coyle and Dr Jen-Chung Mei identified how the primary sectors contribute to the UK's productivity slowdown despite them generally being considered strengths of the UK market. Lucy Hampton explains their findings and the need for further investigation.

This ‘puzzle’ of the UK’s productivity growth slowdown has been extensively explored by economists and policymakers alike. Potential explanations are numerous and include factors on both the supply and demand side, as well as the difficulty of accurately measuring productivity in the modern economy.
One important question about the slowdown is to what extent it reflects slowdowns within sectors or shifts in economic activity between sectors. Labour moving from sectors with high to low growth – indicative of structural change – will reduce productivity for the whole economy, for example.
Another important question is about the sectors responsible: is the slowdown concentrated in particular sectors, or evenly distributed across the economy?
Prof Diane Coyle and Dr Jen-Chung Mei address both these questions in their paper – Diagnosing the UK productivity slowdown: which sectors matter and why? – published by Economica.
Using recently released data on double-deflated real value-added[1] from the Office for National Statistics (ONS), they found that the slowdown occurred within sectors and, surprisingly, that it was concentrated in the high-value industries usually considered strengths of the UK economy.
Reallocation between or slowdown within sectors?

Coyle and Mei distinguish between the effects of productivity growth within industries and reallocation between sectors[2]. They find that the whole-economy slowdown was largely a result of slowdown within sectors and that reallocation played a small role. Figure 1 shows that the contribution of within-sector growth to whole-economy productivity growth (in blue) slowed substantially post-2008, and took on large negative values during the recession, while the contribution of labour reallocation (in red) was generally small both pre and post-2008.
Their paper also documents the slowdowns within particular sectors. The largest slowdowns occurred in agriculture, information and communication, manufacturing, finance, electricity and transportation. The within-sector slowdowns are mainly attributable to transport equipment and pharmaceuticals in manufacturing, and to computer software and telecommunications in information and communication.
Strikingly, with the exception of agriculture, these are among the sectors generally considered to be success stories in the UK.
How does the UK compare?
To see how the UK compares internationally, the authors examine 12 other countries including Japan, the US, and several European economies for the period 1998-2015. The results help diagnose the main drivers of the slowdown (e.g., do the same industries explain slowdowns in similar countries?) and explain why the UK performed worse than other countries.
Their research finds that the within-industry contribution is also the main driver in 12 advanced economies. Also in these countries, the manufacturing and information and communication sectors account for a large part of the slowdown in labour productivity growth.
Why have some of the most innovative sectors contributed most to the slowdown?
The finding that the highest-productivity sectors contributed most to the slowdown is surprising in light of the technological progress experienced post-2008, reflected in large price declines in these sectors.
One example is the telecommunications sub-sector of ICT (Information, Communications & Technology) where an improved version of the UK’s telecoms price index suggests it declined between 37% and 96% from 2010 to 2017[3]. So why does telecommunications appear as one of the biggest contributors to the slowdown in ‘within’ labour productivity growth in the UK ICT sector, and why does ICT overall appear to be one of the bigger contributors to the whole-economy slowdown?
One possibility is that technological progress in these sectors, while fast, has slowed down post-2008. This corresponds to a slowdown in research productivity as described by Bloom et al. (2020), where more and more research effort is required to maintain the same growth rate of research output. Relatedly, the association between technological progress and productivity growth may have weakened. These sectors may have been particularly vulnerable to the 2008 financial crisis, which slowed investment in the intangible capital needed to reap the benefits of improving physical capital such as computer hardware.
A further issue concerns measurement, and in particular, the weights used to aggregate across sectors. The method in the published paper uses shares of nominal value-added as weights, meaning that revenue growth in a sector affects economy-wide productivity growth. However, revenue growth may be dwindling in many of the most productive sectors, reducing these sectors’ weightings. In telecoms, for example, despite rapid volume growth caused by greater compression, more bandwidth and faster speeds, revenue fell by around 6% between 2010 and 2017[4]. The related question – to what extent have relative price changes contributed to the slowdown? – is being investigated in a working paper by Coyle, Mei and Hampton (2023).
The findings call for more detailed investigation of these issues as well as the slowdown in the sub-sectors themselves while paying close attention to how price indices are constructed. The ICT sector and software sub-sector in particular require closer inspection, as there are well-known difficulties in measuring prices and quality improvements in digital-intensive sectors.
- The Bennett Institute’s Research on Sectoral Productivity is Funded by the Gatsby Foundation.
[1] Gross value-added (GVA) can be understood as the difference between the value of an industry’s output and the inputs used in production. Double-deflation means that both inputs and outputs are adjusted for inflation.
[2] The approach used in the paper is the Tornqvist decomposition.
[3] See Abdirahman, Coyle, Heys and Stewart (2020) for more details (https://doi.org/10.24187/ecostat.2020.517t.2017).
[4] Ibid.
The views and opinions expressed in this post are those of the author(s) and not necessarily those of the Bennett Institute for Public Policy.