Fears that new technology will wipe out jobs are as old as the factory floor. But new analysis by Aniket Baksy, Daniel Chandler and Peter John Lambert shows that rather than taking jobs away, automation in British factories is linked to higher employment. It is a reminder that new production technologies often expand output and change what people do at work, rather than simply replacing them.
In 1821 David Ricardo wrote that the “substitution of machinery for human labour is often injurious to the interests of class of labourers.” Other economists, workers and policymakers have fretted about the issue ever since. Now, more than 200 years later, with the rise of artificial intelligence (AI), white‑collar workers are suddenly face‑to‑face with automation fears that have lingered in factories for decades. But our new research is a reminder that new production technologies often expand output and change what people do at work, rather than simply replacing them.
In our recent working paper we compiled and analysed a new panel dataset tracking nearly 27,000 manufacturing sites in Britain between 2005 and 2023 and their use of two cornerstone automation technologies: computer numerical control (CNC) machines (precision tools that cut or shape materials) and industrial robots (reprogrammable manipulators used for handling, welding, assembly and more). These technologies represent the two most important and widely used forms of industrial automation in recent decades.
For each plant, we observe employment across multiple occupation categories and (i) whether it uses CNC machine tools, (ii) if so, how many of these tools are installed on site, and (iii) whether it uses industrial robots.
We focus on both adoption events (the first time a plant uses a technology) and, for CNC machines, expansion events (adding more CNC machines once already in use). We estimate the causal effects of adoption of these technologies using modern econometric methods, comparing those manufacturing plants that take up new technologies with similar plants who didn’t adopt these automation technologies.
What we found from counting the robots
CNC and robot use is rising and concentrated in big plants. Between 2005 and 2023 the share of production plants using CNCs rose from about 47 to 56 per cent and robot use increased from about 4 to 8 per cent of plants between 2014 and 2023. Because robots are more common in large sites, the share of workers employed at robot‑using plants grew faster, from roughly 14 to over 27 per cent.
Automation is followed by higher plant employment. Plants that adopt CNC machines increase employment by roughly 6 per cent in the four years after installation; first‑time robot adopters grow employment by around 8 per cent over the same window (see Figure 1). These gains appear quickly and persist. In plain terms, productivity effects dominate displacement at the firm level.
Figure 1: Employment before and after the adoption of industrial robots
Scaling up CNCs delivers even larger gains – and some reorganisation of work. When plants that already use CNCs add more machines, employment rises by 9 per cent over four years and keeps trending up. At the same time, the ratio of manufacturing production to engineering, design and support workers in those plants gradually falls (by around 8 percentage points over six years). These results are consistent with firms learning how to make the most of automation technologies over time. When experienced firms expand their use of a new technology they are more likely to successfully reorganise their production processes, hence experiencing productivity gains which raise their total employment but also change the mix of workers they employ.
Industry spillovers are positive or neutral. We see no evidence that more automated firms are of systematically taking business away from others in a way that would offset firm‑level gains. If anything, competitors’ employment in the same industry tends to rise after a peer adopts CNCs or robots, and industry‑level “automation waves” are associated with positive or neutral changes in total industry jobs after four years.
How this fits the growing international evidence
Our findings align with other recent research into robots and the impact of automation labour markets. This broader firm‑level pattern is seen elsewhere, where investment in new technologies is consistently associated with an increase in employment at focal (adopting) companies. In France, for example, research leveraging event studies and an instrumental‑variables design finds that major firm-level investments in modern manufacturing capital lead to an increase in employment of nearly 20 per cent over five years. Another study comparing closely matched winners and losers for an EU technology subsidy programme for SMEs in Finland found that receiving a technology grant led to a 23 per cent increase in employment.
The literature is more divided on the impact of automation at the aggregate level. In France, the Philippe Aghion study (linked to above) found that positive effects at the firm level are mirrored at both the industry and labour market level; while Daron Acemoglu and Pascual Restrepo document negative labour-market-level impacts from robots in America. Our contribution is to add Britain’s plant‑ and industry‑level experience to this debate, and serves as a reminder that economic geography, product and input market structure, and worker mobility shape outcomes beyond the firm.
Why might jobs rise after automation?
Three mechanisms stand out. The first is elastic demand and lower prices. More efficient plants cut costs, expand output and hire workers. Positive spillovers to peers can arise if overall demand grows or if adopters stimulate complementary activity in the supply chain. The second is new and complementary tasks. Automation often frees people from set‑up, material handling and repetitive machining, shifting effort towards programming, quality, maintenance and customer‑specific customisation. The third is learning‑by‑doing. The bigger effects we see when firms expand CNC capacity are consistent with a learning curve: once processes, skills and supplier links are in place, each additional machine yields more.
What does this means for managers and policymakers?
Firms should treat CNCs and robots as platforms, not one‑off purchases. Build the complementary capabilities – programming, maintenance, fixture design and workflow integration – that let you redeploy people to higher‑value tasks. Our results suggest the largest returns arrive after you have learned how to use the technology, not on day one.
For policymakers and industry bodies the priority is diffusion and capability‑building, especially for mid‑sized plants. Support for training in CNC and robot programming and integration, better access to finance for upgrades and knowledge‑sharing across supply chains can accelerate the transition from first adoption to effective scale‑up – where the biggest gains appear.
A brief note on methods. We exploit the staggered timing of adoption and estimate dynamic effects using a local‑projection difference‑in‑differences approach that avoids “forbidden comparisons” in traditional event studies. Estimates control for detailed industry‑by‑year shocks and recent plant‑level employment trends; results are robust across alternative estimators, clustering choices and placebo tests. (Technical appendix and figures available in the paper.)
This blog is based on “Anatomy of Automation: CNC Machines and Industrial Robots in UK Manufacturing, 2005–2023” published by the Centre for Economic Performance at the London School of Economics and Political Science.
This article gives the views of the authors, not the position of LSE Business Review or the London School of Economics. You are agreeing with our comment policy when you leave a comment.
Image credit: kckate16 provided by Shutterstock.
Print Friendly
link