A new working paper by Coyle et al. developed a novel framework to show vast differences in productivity between firms even within the same sector. Lucy Hampton explores this research and the trends differing across manufacturing and ICT.
By the end of 2019, nearly 11 years after the financial crisis, UK labour productivity was about a fifth lower than if the pre-2007 trend had continued. The implications for current income and wellbeing are stark.
The manufacturing and ICT (Information, Communications and Technology) sectors were the main contributors to this slowdown, according to research by Diane Coyle and Jen-Chung Mei. However, numerous questions still remain about the patterns of slowdowns within different sectors.
For instance, is the current slowdown a widespread phenomenon, or is it concentrated among a subset of firms? And do patterns differ when we compare labour productivity with total factor productivity (TFP), which measures the efficiency of all measured inputs (not just labour)?
While labour productivity can be enhanced simply by increasing the capital stock and thereby providing more equipment per worker, TFP encompasses a broader range of factors such as technological change, regulation, institutions, and management. Understanding TFP is particularly important as there is evidence that the UK’s productivity puzzle is primarily a TFP puzzle.
However, there are well-known challenges in the measurement of TFP, such as the absence of information on specific firm-level prices and quantities, and the difficulty of accounting for changes in quality. A new working paper by Diane Coyle, John McHale, Ioannis Bournakis and Jen-Chung Mei develops a novel framework that allows researchers to derive firm-level quality-adjusted TFP from revenue data. In doing so, they are able to identify different patterns between firms. The results point to contrasting explanations for the puzzle in the ICT and manufacturing sectors.
Differences in productivity between firms are vast, even within the same sector
Focusing exclusively on sector-level evidence can obscure the large differences in productivity between firms. Export-intensive firms, for example, tend to be more productive as exposure to foreign competition forces them to focus on developing a core range of their most productive varieties. Differences in productivity have also been found for characteristics such as firm size, skills, innovation output, and foreign-ownership status. At the same time, a growing body of evidence suggests that productivity growth is mainly occurring at the top of the productivity distribution within large ‘superstar firms’.
How big is this variation? TFP distributions are shown below for the manufacturing and ICT sectors respectively. In these charts, a difference of five means that firms can produce 150 times as much output with the same measured inputs. The incredible variation in productivity displayed here illustrates the importance of combining insights about the whole economy with firm-level evidence.
Figure 1: Manufacturing and ICT TFP distributions. TFP in logs. Source: Coyle et al (2023)
How do trends differ across firms?
The charts above are also informative about changing patterns over time. There is a shift in the distribution left in manufacturing in 2019 (grey) compared to 2008 (red), indicating broad deterioration in productivity, and the opposite pattern in ICT, which now has fewer low-performing firms and more high-performing firms. In both sectors the proportion of firms with extremely high TFP (above 5 in the graphs) has increased.
Contrary to the idea that superstar firms are pulling further away from their competitors, their paper finds evidence of catch-up. The graphs below show TFP growth against the previous period’s TFP, for manufacturing (left) and ICT (right). Since the correlation is negative, this means that high-productivity firms are growing more slowly and vice versa.
Figure 2: TFP growth rate against current TFP, manufacturing (left) and ICT (right). Source: Coyle et al (2023)
How do trends differ across manufacturing and ICT?
A key finding is that there are diverging trends in TFP for both sectors, illustrated in the charts below. Manufacturing shows a consistent decline since 2011, with a small improvement between 2015 and 2018, before declining again in 2018. In contrast, ICT shows a consistent upward trend, except between 2011 and 2015 when it flatlined.
Figure 3: TFP over time, manufacturing (left) and ICT (right). Source: Coyle et al (2023)
Breaking this down further, the authors distinguish between changes in TFP owing to changing TFP within firms and reallocations of economic activity between firms, such as less productive firms employing more workers. The story in manufacturing is one of declining productivity within firms (the black dashed line in the figure below) combined with reallocation towards less productive firms (the dashed grey line).
In contrast, in ICT there is a small decline in productivity within firms with reallocation towards more productive firms. In ICT, these positive reallocation effects more than compensate for the decline in productivity within firms, creating an overall upwards trend. This suggests that manufacturing has a bigger role in the overall UK TFP slowdown.
Figure 4: Cumulative TFP growth, manufacturing (left) and ICT (right). Source: Coyle et al (2023)
New puzzles: why is TFP falling within firms?
The UK’s TFP puzzle then becomes about explaining the decline in within-firm TFP over this period. One possible explanation other than outright technological regression is declining quality. In this work, quality is treated as a ‘shift-factor’ for demand, i.e., something that reflects consumers’ increased willingness to pay for products. It is possible that UK firms are failing to compete in international markets, causing decreased willingness to pay for UK products relative to those of international firms and hence declining quality-adjusted TFP. Further work should investigate this international dimension.
Other possible drivers to explore regarding the productivity slowdown include the adoption of digital technologies, changing management practices, changes in foreign ownership, and difficult-to-measure inputs like intangible capital or human capital (which according to one measure has grown more slowly at the national level since the early 2010s).
Investigations of TFP growth are complicated by the fact that TFP is a ‘measure of our ignorance’: it is essentially unexplained output growth. As such, a bottleneck for better policymaking and research is the quality and availability of data on the price and quantity of firms’ inputs and outputs. Obtaining more comprehensive data, as well as delving into specific drivers like international competition, may prove to be crucial pieces in solving the productivity puzzle.
Working paper: Recent trends in firm-level total factor productivity in the United Kingdom: new measures, new puzzles
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.
