The adoption of advanced digital technologies like artificial intelligence (AI) and cloud computing by firms in the UK has been remarkably slow. According to the Digit Employers’ Digital Practices at Work Survey, as of 2021-22, only a third of UK businesses had invested in these.
A gap has emerged between digital “superstars” – firms who have adopted these technologies, plan to continue investing in them, and have expanded in terms of employment – and other organisations who have not adopted these technologies and have no plans to do so anytime soon.
Policymakers should be concerned about the rising concentration of technology adoption – the degree to which a small number of firms are dominating the use of modern digital technologies – since this may translate into rising concentration in product and labour markets, with adverse implications for economic inequality and shared prosperity.
Is low investment in new technologies a people problem?
In my research at the Digital Futures at Work Research Centre, I will be studying a possible reason for this rising concentration of technology adoption: digital superstars are much more likely to have invested in the complementary human capital necessary to make effective use of digital technologies. This makes them particularly well equipped to embrace and implement new technologies in their production processes.
At the same time, the high costs of such investments often discourage smaller firms with fewer resources, who may for instance face challenges raising the necessary capital to fund them. In the absence of training, workers at these smaller firms are less capable of using frontier technologies, making these firms more likely to regard such technologies as less relevant to their businesses. Indeed, in the Digit Employer’s Survey, over half of the non-adopting firms reported that there was simply “no need” for these technologies. This, in turn, makes such firms even less willing to invest in technology upgrades going forward, leading to the observed divergence in digital adoption.
How this theory accounts for some facts about technology adoption
This simple theory, summarised in the diagram below, may help explain four common findings on firm technology adoption.
First, technology adoption is concentrated in large firms. This is because adopters are likely to be firms with greater resources and sales, which allow them to make the necessary human capital investments in the first place.
Second, despite the striking capabilities of modern AI systems, many non-adopters don’t find AI technologies useful for their businesses. This may be in part because non-adopting firms lack the human capital required to realise the full capabilities of these technologies. In the absence of appropriate training and skill development investments, a firm’s workers are likely to find new digital technologies difficult to use and are unlikely to perceive the range of use cases possible for them – a recipe for disaster, as predicted by the classic technology acceptance model.
Third, firms that may look similar based on survey data don’t always make the same choices for technology adoption. This may be because, unobserved from a researcher’s point of view, adopters have made the investments necessary to realize the full benefits of technologies, and non-adopters have not.
Fourth, policies aimed to encourage technology adoption by subsidising ICT equipment costs are often unsuccessful. A recent example is the striking failure of the UK’s Help to Grow: Digital programme, which started in 2021 with the intention of helping 100,000 SMEs adopt business software by subsidising software purchases. It was shut down in February 2023 after receiving fewer than 1,000 applications. The proposed theory explains this as the consequence of not subsidising the complementary human capital required to use the new software, the costs of which can often dominate the costs of the computers themselves in the decisions taken by non-adopters.
Figure 1: High costs lead to differential human capital investments and polarised views on adoption
This theory can also help rationalise two important aggregate trends.
Second, this theory can also help explain the slowdown in “business dynamism”, which manifests in the form of slower technology diffusion in the economy and the slowdown in entry rates. This is because as large firms expand further, the rising demand for skills raises the price of human capital, making the adoption of digital technologies even more expensive.
How quantitatively important is this theory?
While the simple theory outlined above can, in principle, rationalise the facts in the surveys, it is important to establish how quantitatively relevant it is. To do this, in my Digit research, I will be constructing a mathematical model of an economy in which firms differing in (i) their capital stock, (ii) their productivity and (iii) the vintage of the technology they operate, make decisions on whether to upgrade their technologies, an activity that is costly and involves hiring more human capital. I will calibrate the model so that it can explain evidence from the Digit Employers’ Digital Practices at Work Survey.
Once calibrated, I plan to use the model to address two questions. First, I will study whether the rising productivity of digital technologies can explain the extent to which the economy has become more concentrated in digital superstar firms. Second, I will explore whether instead of subsidising technology upgrades, policymakers should focus an equal amount of attention on encouraging investments in human capital.
Whilst evidence to date remains relatively scarce, human capital investments may be the missing link in the chain between technology investments and productivity, and in explaining why policies subsidising technology purchases often result in disappointing outcomes. Understanding the complementarity between these investments and modern technologies is critical to the design of cost-effective policies to encourage technology adoption.