Research and analysis

The impact of technology diffusions on growth and productivity: findings from an AI-assisted rapid evidence review

Published 23 April 2025

Note:

This research was supported by the R&D Science and Analysis Programme at the Department for Culture, Media & Sport. It was developed and produced according to the Behavioural Insights Team’s research team’s hypotheses and methods. Any primary research or findings do not represent Government views or policy. Please note that this research was commissioned under the previous government of 11 May 2010 to 5 July 2024, and before the founding of the UK Metascience Unit.

1. Executive Summary

The diffusion of new technology is theoretically meant to translate into improved growth and productivity. Though whether and how this happens is not always clear. Different technologies have differing effects, and the underlying mechanisms driving impact are often less widely studied. There is also uncertainty over how long technology typically takes to diffuse. This rapid evidence review seeks to shed light on these questions, specifically focusing on the UK, a country experiencing a notable productivity slowdown compared to other major economies^{[footnote 1]}

The key research we focus on is: What is the impact of recent technological diffusions on growth and productivity? Rather than merely noting the quantifiable effects of certain technologies, we place a greater emphasis on identifying the underlying mechanisms that drive any impact. This particular focus is crucial for understanding how we can leverage technology to enable greater productivity and growth moving forward.

We conducted a rapid evidence review of high-quality empirical evidence. Notably, this process was Artificial Intelligence (AI)-assisted, whereby we made use of the latest Large Language Model (LLM) tools to help conduct the review. This review and the AI-assisted approach was commissioned by the Department for Science, Innovation and Technology (DSIT) and the Department for Culture, Media & Sport (DCMS) through the R&D Science and Analysis Programme.  The choice of technologies under review was informed by DSIT’s Science and Technology Framework, which emphasises the importance to future UK success of technologies such as artificial intelligence, robotics, and broadband internet.

Overall, the review finds positive impacts of technological innovations on firm growth and productivity, although these impacts vary by the type of technology. It also identifies 6 key mechanisms which determine the extent to which firms benefit from them:

• Broad knowledge gains flow from technology adoption and boost productivity.

• Technology diffusion enhances businesses’ organisational and operational capabilities, impacting productivity and growth.

• New technology can lower firms’ fixed costs and barriers to entry into markets, particularly for smaller firms.

• Advances in robotics and manufacturing technologies enable automation and improve production efficiency.

• Trade openness encourages technology adoption, improving productivity.

• New general-purpose technologies, such as AI, spur follow-on innovation throughout firms.

Based on these findings, we present the following key conclusions:

• Access to and ‘early adoption’ of general-purpose technologies is crucial. There are significant benefits of early technological adoption on productivity, which can persist for years after initial adoption. These benefits can be more pronounced for smaller firms, who may benefit from reduced barriers to entry and fixed costs. A timely new technology is AI - fostering environments that support its diffusion across UK business and sectors should ensure both equitable access to these technologies as well as productivity gains.

• Addressing local skills gaps can support less productive firms in adopting new technologies. Firms often become more productive as they adopt new technologies. However, productivity increases are only achieved if employees have the right skills to effectively use new technologies. Ensuring that any skill gaps are addressed is therefore essential. Managers can play an important role by creating an environment that encourages innovation and supporting learning among employees.

• In manufacturing contexts, firms should embrace advances in robotics for task automation. There are clear and positive impacts of technological adoption in manufacturing on productivity, efficiency, and economic competitiveness. Task automation directly reduces production costs and enhances output quality by streamlining existing processes. Secondary effects of automation include organisational shifts in business strategies and skill requirements of employees.

Table 1: Glossary of terms

Term	Definition
Innovation	The creation and application of new knowledge to improve the world.[footnote 2] The Organisation for Economic Co-operation and Development (OECD) identifies 4 types of innovation: product, process, marketing and organisational.[footnote 3]
Technology	The application of scientific knowledge for practical purposes, especially in industry.
Technology diffusion	The process by which the use of an innovation spreads and grows.
Growth	An increase in business value over time, interpreted broadly as including an increase in business revenues, employee headcount and profits.
Productivity	A measure of outputs relative to inputs used. An increase in productivity results from increased output for the same or less input.
Labour productivity	A measure of economic performance that calculates the amount of goods and services produced per hour worked.
Mechanism (that affects technology diffusion)	The key element of an explanation, which depicts the driving force that generates a certain effect or outcome in a particular condition or setting.[footnote 4]
Absorption	The degree to which individuals or organisations adopt and integrate new technologies into their practices.
Time-lags	The period between the introduction of new technologies and the point at which individuals or organisations fully learn and adopt these innovations to realise their benefits.
Digital skills	The abilities required to use digital devices, communication applications, and networks to access and manage information.
Fixed Costs	Expenses that remain constant regardless of output levels, such as rent, salaries, and investments in assets like data management equipment.
Broad knowledge gains	The process by which knowledge gained from the use of new technology enhances other aspects of work, creating broader improvements in efficiency and innovation.
Technology transfer	The process of importing technologies from other contexts, such as between different countries or across industries, facilitating its adoption.
Instrumental variable (IV)	A statistical method used to estimate causal relationships by utilising variables that influence the predictor variable of interest but have no direct effect on the outcome measure.
General purpose technologies (GPT)	Technologies that have a wide range of applications across different industries and sector, such as electricity, the internet, and artificial intelligence

2. Introduction

The late 20th and early 21st century has been a time of profound technological change. New technologies have emerged and proliferated: from the internet and mobile phones, to cloud computing and artificial intelligence. These advancements have transformed societies and economies, promising substantial gains in growth and productivity.

Despite the visible integration of these technologies into the ways we live and work, their impact on productivity is often unclear. This idea was famously captured by economist Robert Solow noting: “You can see the computer age everywhere but in the productivity statistics.”^{[footnote 5]} This reflects an apparent paradox where significant technological advancements do not directly translate into theoretical associated gains in productivity and growth.

2.1 Purpose of this review and our research question

The research question we examine is “What is the impact of recent technological diffusions on growth and productivity?” We also place particular emphasis on identifying the underlying mechanisms that drive any impact. In other words, the focus is on the ‘how’ rather than the ‘what’ – isolating the mechanisms through which the diffusion of new technologies has improved growth and productivity in a UK context, rather than merely documenting the after-the-fact impact of these new technologies.

This topic is especially relevant to the UK, given it has been experiencing a productivity slowdown. UK productivity trails that of comparable nations like the United States, Germany, and France, with its growth rate in particular notably slower since the financial crisis of 2008.^{[footnote 6]} As such, the ultimate purpose of this review is to distil our findings into clear and actionable conclusions.

The rest of this report is split into 3 parts. In the Methodology section, we outline the approach we used for finding and evaluating the relevant research. As this review was AI-assisted, we also explain which AI tools we used, for what, and why. In the Findings and Discussion section we outline the results from our evidence review, structuring findings according to the specific mechanisms through which the diffusion of new technologies has impacted growth and productivity. The report ends with a summary of conclusions from the report.

3. Methodology

We approached this project as a rapid evidence review. Although all of the evidence was sourced and compiled in a structured way, rapid evidence reviews are different to systematic reviews. They do not seek to comprehensively summarise all the literature on a topic - they are a pragmatic attempt to quickly assess the most relevant studies.

This review was conducted using the most relevant AI tools (as of January 2024) to assist with each stage of the process.

3.1 AI tools used

To assess which tools would be most useful, we experimented with a wide range of tools to see how they could perform key tasks, such as finding relevant literature or summarising the findings from papers. The long list of tools we initially tested included: Elicit, Consensus, Bard (now Gemini), Perplexity, Connected Papers, SciteAI, Claude and ChatGPT. In the interests of balancing rigour with pragmatism, our aim was to focus on a final set of 2-4 AI tools as part of this project.   

Based on:

• Our experience and findings;

• What each stage of the rapid evidence review required and;

• Input from DCMS and DSIT

We agreed on the process below. Note that we used different AI tools for different tasks, as each tool has their relative strengths and weaknesses which relate to key steps of the rapid evidence review process. Only AI tools were used to directly search the literature. We therefore did not manually search Google, Google Scholar, or other academic databases to find literature for this particular report.

Table 2: Summary of AI tools

Step	1: Sourcing Literature	2: Screening and Summarising Literature	3: Writing-up findings
AI tool(s) used	Elicit and Consensus	Claude	Claude and ChatGPT
Rationale	Elicit and Consensus both use Semantic Scholar to search for papers, an AI powered research tool for scientific literature. The Semantic Scholar database includes over 200 million publications. Elicit summarises paper abstracts to display results, while Consensus drafts a 1-sentence takeaway of each paper	Claude is a general-purpose chatbot, which also has the ability to analyse large PDFs. It can therefore be used to screen complex and dense papers according to your pre-specified criteria, to evaluate whether it is useful for your review.	Claude can assist you in summarising and writing key findings from individual papers, once you have uploaded them to your chat. ChatGPT is widely considered the most advanced AI chatbot, and is particularly effective in helping with a range of writing tasks.

Visual illustrations of how some of these tools were used are shown below.

Figure 1: Example of how Elicit was used to find papers

Figure 2: Example of How Claude was used to screen papers

Figure 3: Example of How Claude was used to write up findings

3.2 Inclusion and exclusion criteria

To ensure the literature we found was robust and relevant, we applied the following inclusion and exclusion criteria. Note that unpublished government resources were also not considered for review.

Inclusion:

• Sources grounded in empirical evidence, (e.g. papers on historic, specific technology diffusions with verifiable underlying datasets) rather than theory (e.g. papers conducting econometric modelling exercises).

• Prioritise research published since the year 2000 (not before that, although individual exceptions may be made on a case-by-case basis).

• Include ‘grey’ literature, e.g. government reports which are not technically academic publications.

• Focus on impact at the business level (productivity = measured as outputs relative to resource used / inputs; growth = increase in business value over time, interpreted broadly as including an increase in business revenues, employee headcount and profits), within the UK (but where evidence is scarce, we will look to other regions that share similar socioeconomic characteristics to the UK e.g., US, France, Germany).

• Focus more on recent (late 20th century / 21st century) technological changes

Exclusion:

• Non-English-language material.

• Studies examining multi-faceted technological / societal shifts rather than discrete technology changes.

• Macro-history texts which only cover the topic as part of a broader narrative.

3.3 List of prompts used

The list of prompts used to search for relevant literature are listed below:

• What is the impact of technology diffusions on growth and productivity in the UK?

• How has the spread and adoption of technology in the UK affected the productivity and growth of businesses?

• Empirical case studies of technologies impact on business productivity and growth in the UK

• How has the internet affected the productivity and growth of businesses in the UK?

• How have smartphones affected the productivity and growth of businesses in the UK?

• How has cloud computing affected the productivity and growth of businesses in the UK?

• How have robotics affected the productivity and growth of businesses in the UK?

• How have semiconductors and microchips affected the productivity and growth of businesses in the UK?

• How has the spread of GPS affected the productivity and growth of businesses in the UK?

• How has the rise of renewable energy affected the productivity and growth of businesses in the UK?

• How has the diffusion of lithium ion batteries affected the productivity and growth of businesses in the UK?

• How has AI affected the productivity and growth of businesses in the UK?

• What is the empirical impact of AI on productivity and growth of businesses?

• How do advances in engineering biology impact productivity and growth of businesses?

• How has the spread of semiconductors impacted productivity and growth of UK businesses?

• How has quantum technologies impacted productivity and growth of UK businesses?

3.4 Screening process

Application of the search terms across the AI tools produced an initial longlist of 35 studies. Deeper scrutiny of these studies and systematic application of the inclusion/exclusion criteria then reduced this longlist to a shortlist of 22 studies, which made up the final sample that was used in the analysis stage.

Table 3: Characteristics of studies analysed

Breakdown of final sample	Number of studies
Final shortlisted studies	22
Published in academic journals	15 (68%)
Published elsewhere	7 (32%)
Published 2018-24	13 (59%)
Published 2010-17	1 (5%)
Published 2000-09	8 (36%)
UK focused[footnote 7]	11 (50%)
Not UK focused	11 (50%)
Examining impact of technology on productivity	15 (68%)
Examining impact of technology on growth	1 (5%)
Examining impact of technology on both	6 (27%)

4. Findings and discussion

The section is structured according to the specific mechanisms discovered through which technology diffusion impacts growth and productivity. The mechanisms identified were:

1. Broad knowledge gains
2. Organisational enhancements
3. Lowering costs and barriers to entry
4. Automation and boosting production efficiency
5. Trade openness enabling technology transfer
6. New technologies, such as AI, spurring follow-on innovation

These mechanisms can overlap. When studies mention several mechanisms that could fit under 1 of the other headings, we point this out as part of our discussion.

During our review, we also encountered studies that, while not identifying mechanisms directly, provided important context. Some offered insights into why the US might realise the productivity benefits of new technology faster than the UK, for instance. Where relevant, we included these findings to present a fuller picture of how technology influences productivity and growth.

Much of the research in this evidence review (and in the broader literature) is correlational. Therefore, whether technology diffusion causally impacts productivity and growth is not completely clear. That said, the studies compiled in this review used a range of methodological techniques. Several studies, for example, used an IV approach and so more directly attempted to establish causal links – such as those measuring the impact of cloud computing and broadband in particular.^{[footnote 8]} Others collected micro-level data on technology adoption and productivity metrics across firms within a narrow industry, meaning they were more specifically able to isolate and quantify differences attributable to technology diffusion.^{[footnote 9]}

While we searched for literature on all technologies agreed within scope (see specific prompts used in the previous section), some technologies were covered in the literature more heavily than others. Based on the search, the final list of technologies covered in our review included:

• Information & Communication Technologies (ICT) - such as broadband infrastructure and computers, enterprise software etc.

• Telecommunications - covering innovations in telephony and mobile infrastructure enabling better connectivity.

• Cloud Computing - including cloud-based computing services, platforms, and infrastructure.

• Industrial Robotics - including robotic automation technologies used in manufacturing environments.

• Artificial Intelligence - encompassing AI and machine learning technologies and applications.

4.1 Mechanism 1: Broad knowledge gains flow from technology adoption and boost productivity 

The first key mechanism we identified was that new technologies enable gains in knowledge among staff which then boost firm productivity. Examples include the internet improving access to and the sharing of useful information for businesses, to increased digitisation making more routine tasks easier to complete and consequently allowing workers to focus more on knowledge intensive tasks. 1 important way this mechanism varies is that whether firms realise productivity gains from new technology may depend on whether their employees have (or can easily learn) the required skills needed for leveraging new technology into their business practices.   

A 2021 study analysing the long-term productivity impact of ‘early adoption’ of digital technologies (i.e. internet use) across local areas in the UK led to significant persistent impacts on productivity, up to 16 years later^{[footnote 10]} The authors used panel data to estimate engagement with digital economic activities (e.g. the creation and maintenance of commercial websites), and identify “knowledge spillovers” as a mechanism which may explain this result.

A 2021 OECD econometric analysis of digital technology adoption in the Netherlands found that investments in ‘intangibles’ – like software and digital skills – have positive impacts on productivity^{[footnote 11]}. The authors used firm-level exposure to sector-wide advances in intangible intensity and digital technology trends as an instrument to establish causal links with productivity growth. The results showed that initial ICT investments increased digital skills, which in turn improved labour productivity growth for services firms and younger enterprises over 2012-2017, and these investments more strongly benefited initially less productive firms, suggesting that new technology can support the “productivity catch-up of laggard enterprises.”

A comprehensive 2019 study by the OECD on European companies, including the UK, looked at how using digital technologies like broadband, enterprise software, customer relationship management systems, and cloud computing positively affected labour productivity.^{[footnote 12]} The paper combined a wide range of data: cross-country firm-level data on productivity and industry-level data on digital technology adoption in an empirical framework that factors in firm heterogeneity. The study found that companies in industries that have embraced digitalization see higher growth in labour productivity. This research also suggested that:

• Industries performing more routine tasks tend to gain more from digitalization;

• Digital technologies could lead to substitution of labour in sectors with routine tasks, shifting workers towards more knowledge-intensive activities, and;

• Digital technologies are more beneficial for already productive firms and less so when there’s a shortage of skills, pointing to the importance of matching digital tools with the right skills.

Additionally, several papers covering the impact of ICT highlighted broader considerations about how technology diffusion impacts labour productivity. A 2005 paper found that ICT capital accumulation boosted UK labour productivity growth by 13% in the 1970s, then rose significantly to 47% by 1995-2000. Similarly, a 2015 comparative study covering 1970 to 2013 across the United States, Canada, Eurozone, and the United Kingdom found that while ICT diffusion saw a long period of growth, this stabilised since 2000 across all regions – with a higher level of ICT capital stock in the US.^{[footnote 13]} The study highlighted a significant increase in ICT’s contribution to productivity growth during 1994-2004 compared to the earlier 2 decades. However, post-2004, this contribution declined, sustained only by continuous improvements in ICT performance, despite the diminishing pace of such advancements. This pattern of adoption and impact suggests a theoretical ‘n-shaped’ effect on productivity: initially, ICT had a modest effect as technologies are adopted; then, a more substantial impact as diffusion broadens and users learn to leverage these technologies effectively; followed by a tapering effect as the technology becomes fully integrated and major gains are realised.

4.2 Mechanism 2: Technology diffusion enhances businesses’ organisational and operational capabilities, impacting productivity and growth   

Closely related to knowledge gains is the notion of technology facilitating organisational enhancements for firms. These can be understood as direct improvements in existing business processes and practices such as streamlining operations – for example, manual data entry and online bookings – and facilitating easier retrieval and sharing of data – for instance, through emails and websites. A substantial share of productivity improvements are likely to also stem from complementary organisational changes and enhanced information flows.

A 2007 study using national databases on IT expenditure and firm performance analysed how IT investment and organisational changes (like decentralisation) interacted to impact labour productivity in UK firms. It found that while investment in IT did positively impact productivity, a large part of this effect could be attributed to organisational changes that IT enables, such as introducing computer-aided design (CAD) software or electronic point of sale equipment.^{[footnote 14]}

A 2003 study assessing ICT productivity impact in UK hotels similarly revealed that productivity gains did not result from ICT investments alone, but rather when ICT networking and information sharing capabilities were used effectively to transform existing operational practices^{[footnote 15]}. The research collected input/output data like staffing, revenues, room capacity and ICT integration metrics from 103 hotels to construct productivity scores, then compared these against metrics capturing technology availability, integration levels, and sophistication of use – to ultimately measure ICTs impact on productivity. Specific organisational improvements resulting from increased ICT use included automating front office tasks, consolidating databases, and enabling online booking services. The authors also emphasised the role managers can play stating all “ICT capabilities…should be managed and aligned with business strategy and operations.”

Lastly, a 2005 US firm-level analysis argued that superior ICT infrastructure boosted productivity in companies as they grew, specifically by helping firms manage complexity.^{[footnote 16]} The paper finds that high-growth firms invested more in ICT infrastructure, which then facilitated organisational changes like process improvements and automation. Superior ICT infrastructure was also linked to lower operating costs per unit of output during growth for these firms. The paper has implications for the rate of technology diffusion within a firm: namely that technology should be continually embraced as an enabler to scaling.

4.3 Mechanism 3: New technology can lower fixed costs and barriers to entry into markets, particularly for smaller firms 

New technology can reduce firms’ fixed costs - those that do not change with an increase or decrease in the number of goods and services produced or sold, such as rent, licensing fees and investment in essential data management services. 2 studies support this mechanism for broadband and cloud technologies.

A 2018 econometric study on the UK rollout of broadband infrastructure found that small and medium sized firms increased their investment in Personal Com[PCs and resultantly gained greater revenues and employment. However, no effect was observed on the productivity of these firms.^{[footnote 17]} The results are based on an IV approach exploiting local variation in pre-existing telephone networks suited for broadband upgrades. The findings suggested the mechanism underlying such growth is that broadband unlocked commercial websites and e-commerce capabilities for smaller firms for the first time. These developments effectively expanded their relative and potential market reach.

A 2023 analysis of UK companies examined the impact of cloud adoption on firm performance, examining effects on both productivity and firm growth.^{[footnote 18]} The analysis used an IV approach that exploited cross-section and time-series variation in fibre broadband speeds as instruments. The paper found that cloud computing has shifted how firms traditionally access key ICT services: transitioning from upfront initial capital investments to a more efficient, ‘pay-as-you-go’ model. The paper also found asymmetric benefits by firm age: younger firms saw significant increases in employment, sales and productivity after adopting the cloud, whereas more established firms saw comparatively weaker productivity gains. However, after adopting cloud technologies both young and more established firms spread their employee base more widely. In other words, people worked further from company headquarters.

4.4 Mechanism 4: Advances in robotics and manufacturing technologies enable automation and improve production efficiency

Robotics enable clear productivity gains by automating tasks previously conducted by humans. However, they can also have broader productivity benefits, by allowing workers to focus on new tasks or by augmenting their skills when completing existing tasks. This mechanism is therefore closely linked to mechanism 1 (technology enabling broad knowledge gains). Although the studies outlined below are primarily based on non-UK contexts, the findings carry broader lessons that can be considered when thinking of the UK’s own manufacturing industries, such as automobiles and aerospace.

A 2018 econometric analysis of industrial robotics adoption over 1993-2007 in 14 industries across 17 developed countries, including the UK, used an IV approach to measure the relationship between robot adoption and a range of economic outcomes.^{[footnote 19]} It found that robots increased labour productivity, wages and total factor productivity while at the same time lowering output prices. The adoption of robotics did not reduce total employment significantly overall, but did diminish the share of low-skilled employment, indicating potential worker displacement.

A 2007 paper focused on a narrowly defined industry – valve manufacturing in the US. The paper observed plant-level data on process efficiency metrics at different stages of valve production.^{[footnote 20]} It found that ICT systems embedded in the plant’s machines and targeted to particular production activities boosted performance across a range of productivity measures. Aside from the raw gains in productivity (e.g. speed of production), another mechanism through these improvements operated was through greater customisation of products. The authors also found that plants adopting these more sophisticated machines increased their technical training and problem-solving demands from operators. In this way, computerisation had the effect of augmenting, rather than replacing, human technicians.

A 2020 Spanish study on industrial robotics adoption by manufacturing Small and Medium-sized Enterprises (SMEs) over 2008-2015 combined survey data and multiple productivity indicators in a multivariate analysis.^{[footnote 21]} It found robots were associated with superior financial performance, productivity and higher wages. Knowledge gains also emerged: these initially complemented these automation gains but faded over time. For example, the study found that robot-adopting firms engaged in more Research and Development (R&D) activity, such as performing in-house research and contracting external R&D services and displayed higher rates of product and process innovation. Similarly, a 2021 machine learning analysis of 4,600 Spanish manufacturers over 1991-2014^{[footnote 22]} found that robot-adopting firms had significantly better economic results than non-adopters. That study also found that while both small and large companies benefited, initial productivity gains plateaued faster for larger enterprises as they matured more quickly in using the technology. These findings are consistent with the theoretical ‘n-shaped’ effect of technology on productivity, described earlier in this report. Once firms and employees learn how to use new technology effectively, initial productivity gains from automation are substantial. Gains then fade over time as firms fully integrate these technologies into business practice. This process may occur at a faster rate for larger firms, as they progress through the stages of the ‘n-shaped’ curve more rapidly, due to greater resources that can be leveraged in adopting new technologies.

4.5 Mechanism 5: Trade openness encourages technology adoption, improving productivity

Trade openness can encourage technology adoption, which in turn can improve productivity, by increasing firms’ opportunity to acquire new technologies from other countries. Relatedly, increased competition resulting from trade may also incentivise firms to adopt new technologies and innovate to stay competitive.

A 2005 econometric analysis investigated the impact of international trade on productivity growth in UK manufacturing sectors from 1970-1992, focusing on technology transfer from the US frontier.^{[footnote 23]} The authors constructed detailed productivity measures for each industry in the US and UK using Census data, adjusting for factors like labour skills and capital utilisation to enable accurate cross-country comparisons. They then calculated the productivity gap between each US industry and its UK counterpart and estimated the effect of this productivity gap on subsequent productivity growth in the corresponding UK industry. They found that UK industries which were further behind the US frontier subsequently experienced faster productivity growth. This productivity gain was stronger for UK industries with higher import shares, implying that trade openness facilitated technology transfer by enabling UK industries to access and absorb superior technologies from the US.

A 2001 paper^{[footnote 24]} examined UK innovation performance and links it to the nation’s productivity trends. The authors analysed descriptive statistics on productivity, R&D intensity, patenting, technology diffusion and other innovation metrics, comparing the UK’s performance over time and relative to other advanced economies. This data was then supplemented with historical analysis and case studies of specific industries. The authors argued that despite the UK’s strong scientific base, this often did not translate into commercial innovations and highlighted a need for greater trade openness as a means to improve the absorption of new productivity-enhancing technologies. Lastly, a 2005 paper argued that regulatory constraints related to trade inhibited the diffusion of new technologies, limiting potential productivity gains.^{[footnote 25]} The authors found that the US had higher total factor productivity growth in retail than France, Germany and the UK, despite rising ICT investment in all countries. Case study evidence further suggested that land use regulations, product market regulations, and planning laws slowed the adoption of technologies such as barcodes and electronic data interchange that enabled greater productivity gains in the US.

4.6 Mechanism 6: New general-purpose technologies, such as AI, spur follow-on innovation throughout firms

Artificial intelligence may act as a general-purpose technology which improves firm performance across a wide range of dimensions. A weakness of the studies in this section is they all focus on the US or China, rather than the UK - however, they do use sophisticated research methods and may provide some generalisable insights.

A 2023 paper exploring artificial intelligence adoption amongst German firms found that it increased firm-level productivity. The analysis consisted of 5,851 firms, broadly representative of German firms in the manufacturing and services industries, of whom 409 (7%) were AI users.^{[footnote 26]} Using an IV approach, the study found that both the use of AI and the breadth of AI applications had a significant positive impact on labour productivity, as measured by sales and value added.

A 2021 paper found that AI investments among US public firms from various sectors over 2010-2018 led to significantly better firm growth, as measured by sales, employment, and market valuations.^{[footnote 27]} The methodology involved tracking AI-skilled hiring, highlighting a rise in AI investments and its direct correlation with a range of growth metrics. These outcomes were driven primarily through AI enabling product innovation, rather than cutting costs. This growth was more pronounced among larger firms, suggesting AI’s contribution to increasing market concentration and the emergence of “superstar” firms.

An (as yet unpublished) 2024 study used machine learning to examine the relationship between over 50,000 AI patents in the US from 1990-2018 and census data on firm-level performance.^{[footnote 28]} It found that AI-related innovations did correlate with substantial firm growth: firms with AI patents saw 25% faster employment growth and 40% faster revenue growth. The paper framed AI as a general-purpose technology that boosts prediction accuracy, driving innovations across industries and leading to positive impacts on revenue per employee, value-added per employee, and total factor productivity. It also pointed towards an increase in within-firm wage inequality. A limitation of this analysis is self-selection: firms capable of AI innovation may inherently be more productive, possibly inflating the magnitude of observed effects.

A 2021 study investigated the influence of artificial intelligence and robotics on the productivity of firms, using a global dataset of 5,257 companies that filed for at least 1 AI-related patent between 2000 and 2016.^{[footnote 29]} It found that AI patenting activities had a positive impact on labour productivity, with this benefit particularly pronounced for SMEs and the services sector. The authors suggested that these entities are more adept at integrating AI innovations into their operations swiftly. Lastly, a 2020 paper examined the impact of AI on Chinese manufacturing sectors. This study focused on innovation, rather than productivity or growth, but helps clarify the mechanisms underlying AIs impact.^{[footnote 30]} Using panel data from 14 manufacturing sectors in China covering 2008 to 2017, it found that AI stimulated knowledge creation and technology spillover, enhancing firms’ learning and absorptive capacities, and increasing investments in research and development. The influence of AI on innovation was notably more pronounced within low-tech industries.

5. Conclusions

Based on our findings, we now outline a range of conclusions.

5.1 Conclusion 1: Access to and ‘early adoption’ of general-purpose technologies is crucial

Timely ‘adoption’ is important when it comes to new technologies, rather than merely letting them spread through natural ‘diffusion’. This might include speeding up investments in critical infrastructure, such as those required for broadband and 5G networks, to ensure businesses across all regions can access these technologies.

Our evidence review highlights the significant benefits of early technological adoption on productivity – even many years after initial technological adoption. This is evidenced by areas in the UK that experimented with the internet as it grew, resulting in broader knowledge gains that are captured in relative regional productivity differences.^{[footnote 31]} Early adopters can experience a competitive advantage, as they learn to integrate and leverage new technologies before others.

The literature on the diffusion of broadband and cloud computing, more specifically, also demonstrates the potential of technology to ‘level the playing field’ for small businesses. By significantly reducing upfront fixed costs, these technologies enable smaller firms to face substantially lower barriers to entry into industries. This enables entry to markets that may have been previously dominated by incumbent firms.

AI is a specific and timely technology to consider. A handful of recent studies focus on how AI adoption spurs follow-on innovation within businesses, accelerating growth and productivity. Benefits seem to vary across sectors, however, and wider implications at this stage are still speculative. For instance, some papers cited AI as potentially contributing to the rise of ‘superstar’ firms that may come to dominate markets, while others found impacts for smaller and medium sized firms. The recency of this literature however means it naturally has clear limitations.

That said, the findings highlight the importance of environments that support the broader diffusion of AI across all business sizes and sectors, ensuring equitable access to AI technologies. More empirical research is needed across diverse industries and regions to better understand AI’s impact. This will help guide future policies that not only promote AI adoption but also mitigate potential inequalities it may introduce.

5.2 Conclusion 2: Addressing local skill gaps can support less productive firms in adopting new technologies

Relationships between local businesses and further education colleges, apprenticeship programmes, and universities at the local level are important to build an ecosystem of skills that matches the needs of the local economy. The evidence review revealed how digital technologies and software enable “productivity catch-ups” for laggard firms and less productive enterprises.^{[footnote 32]} However, the research also uncovered potential ‘mismatches’ whereby skills gaps may inhibit some firms from effectively leveraging new tools. This likely exacerbates diverging productivity trends, as highly productive companies gain more benefits from emerging technologies. Ensuring that businesses have, or can easily recruit, those with appropriate skills is therefore essential.

In order to identify and address skills gaps, effective management and leadership skills are also required. The literature suggested that managers play a key role in creating an environment within firms that encourages innovation, supports learning among employees, and integrates new technologies into existing processes as firms grow.

Therefore, where appropriate, investing in leadership development programmes tailored to the unique challenges of adopting new technologies may improve adoption levels. Similar government schemes do exist (for example, the Help to Grow management course); those that focus on businesses’ ability to absorb and make use of new technology as part of their existing operations may be the most effective.

5.3 Conclusion 3:  In manufacturing contexts, firms should embrace advances in robotics for task automation

Financial and other incentives may support the manufacturing sector’s adoption of advanced robotics technology. For example, the Made Smarter programme provides financial incentives and business support to SMEs looking to invest in industrial digital technologies, including robotics.

Studies focusing on the impact of robotics and advanced information technologies in manufacturing sectors consistently showed clear positive effects on productivity, efficiency, and economic competitiveness. These benefits flow directly from the consequences of task automation. Most directly, integrating robotics reduces production costs and enhances output quality by streamlining and speeding up existing processes. Secondly, and perhaps less obviously, automation also fosters organisational shifts in business strategies and skill requirements.

1. Harari, Daniel. “Productivity: Key Economic Indicators.”   House of Commons Library Research Briefing. UK Parliament. Published November 24, 2023. ↩

2. HMG. “The UK’s International Technology Strategy.” (2023). ↩

3. Data, Interpreting Innovation. “Oslo manual.” Paris and Luxembourg: OECD/Euro-stat (2005). ↩

4. Ogink, Ruben HAJ, Martin C. Goossen, A. Georges L. Romme, and Henk Akkermans. “Mechanisms in open innovation: A review and synthesis of the literature.” Technovation 119 (2023): 102621. ↩

5. Solow, Robert. “We’d Better Watch Out.” New York Times Book Review. July 12, 1987. ↩

6. Harari, Daniel. “Productivity: Key Economic Indicators.” House of Commons Library Research Briefing. UK Parliament. Published November 24, 2023. ↩

7. This includes studies with broader data sets, where the UK is 1 country in a larger sample ↩

8. DeStefano, Timothy, Richard Kneller, and Jonathan Timmis. “Cloud computing and firm growth.” Review of Economics and Statistics (2023): 1-47. ↩

9. Bartel, Ann, Casey Ichniowski, and Kathryn Shaw. “How does information technology affect productivity? Plant-level comparisons of product innovation, process improvement, and worker skills.” The quarterly journal of Economics 122, no. 4 (2007): 1721-1758. ↩

10. Tranos, Emmanouil, Tasos Kitsos, and Raquel Ortega-Argilés. “Digital economy in the UK: regional productivity effects of early adoption.” Regional Studies 55, no. 12 (2021): 1924-1938. ↩

11. Borowiecki, Martin, Jon Pareliussen, Daniela Glocker, Eun Jung Kim, Michael Polder, and Iryna Rud. “The impact of digitalisation on productivity: Firm-level evidence from the Netherlands.” (2021). ↩

12. Gal, Peter, Giuseppe Nicoletti, Theodore Renault, Stéphane Sorbe, and Christina Timiliotis. “Digitalisation and productivity: In search of the holy grail–Firm-level empirical evidence from EU countries.” (2019). ↩

13. Cette, Gilbert, Christian Clerc, and Lea Bresson. “Contribution of ICT diffusion to labour productivity growth: the United States, Canada, the Eurozone, and the United Kingdom, 1970-2013.” International Productivity Monitor 28 (2015): 81. ↩

14. Crespi, Gustavo, Chiara Criscuolo, and Jonathan Haskel. “Information technology, organisational change and productivity.” (2007). ↩

15. Sigala, Marianna. “The information and communication technologies productivity impact on the UK hotel sector.” International Journal of Operations & Production Management 23, no. 10 (2003): 1224-1245. ↩

16. Mitra, Sabyasachi. “Information technology as an enabler of growth in firms: An empirical assessment.” Journal of Management Information Systems 22, no. 2 (2005): 279-300. ↩

17. DeStefano, Timothy, Richard Kneller, and Jonathan Timmis. “Broadband infrastructure, ICT use and firm performance: Evidence for UK firms.” Journal of Economic Behavior & Organization 155 (2018): 110-139. ↩

18. DeStefano, Timothy, Richard Kneller, and Jonathan Timmis. “Cloud computing and firm growth.” Review of Economics and Statistics (2023): 1-47. ↩

19. Graetz, Georg, and Guy Michaels. “Robots at work.” Review of Economics and Statistics 100, no. 5 (2018): 753-768. ↩

20. Bartel, Ann, Casey Ichniowski, and Kathryn Shaw. “How does information technology affect productivity? Plant-level comparisons of product innovation, process improvement, and worker skills.” The Quarterly Journal of Economics 122, no. 4 (2007): 1721-1758. ↩

21. Ballestar, María Teresa, Ángel Díaz-Chao, Jorge Sainz, and Joan Torrent-Sellens. “Knowledge, robots and productivity in SMEs: Explaining the second digital wave.” Journal of Business Research 108 (2020): 119-131. ↩

22. Ballestar, María Teresa, Ángel Díaz-Chao, Jorge Sainz, and Joan Torrent-Sellens. “Impact of robotics on manufacturing: A longitudinal machine learning perspective.” Technological Forecasting and Social Change 162 (2021): 120348. ↩

23. Cameron, Gavin, James Proudman, and Stephen Redding. “Technological convergence, R&D, trade and productivity growth.” European Economic Review 49, no. 3 (2005): 775-807. ↩

24. Nickell, Stephen J., and John Van Reenen. “Technological innovation and economic performance in the United Kingdom.” (2001). ↩

25. O’Mahony, Mary, and Bart Van Ark. “Assessing the productivity of the UK retail trade sector: the role of ICT.” The International Review of Retail, Distribution and Consumer Research 15, no. 3 (2005): 297-303. ↩

26. Czarnitzki, Dirk, Gastón P. Fernández, and Christian Rammer. “Artificial intelligence and firm-level productivity.” Journal of Economic Behavior & Organization 211 (2023): 188-205. ↩

27. Babina, Tania, Anastassia Fedyk, Alex He, and James Hodson. “Artificial intelligence, firm growth, and product innovation.” Firm Growth, and Product Innovation (November 9, 2021) (2021). ↩

28. Alderucci, Dean, Lee Branstetter, Eduard Hovy, Andrew Runge, and Nikolas Zolas. “Quantifying the impact of AI on productivity and labor demand: Evidence from US census microdata.” American Economics Review (forthcoming) ↩

29. Damioli, Giacomo, Vincent Van Roy, and Daniel Vertesy. “The impact of artificial intelligence on labor productivity.” Eurasian Business Review 11 (2021): 1-25. ↩

30. Liu, Jun, Huihong Chang, Jeffrey Yi-Lin Forrest, and Baohua Yang. “Influence of artificial intelligence on technological innovation: Evidence from the panel data of China’s manufacturing sectors.” Technological Forecasting and Social Change 158 (2020): 120142. ↩

31. Tranos, Emmanouil, Tasos Kitsos, and Raquel Ortega-Argilés. “Digital economy in the UK: regional productivity effects of early adoption.” Regional Studies 55, no. 12 (2021): 1924-1938. ↩

32. Borowiecki, Martin, Jon Pareliussen, Daniela Glocker, Eun Jung Kim, Michael Polder, and Iryna Rud. “The impact of digitalisation on productivity: Firm-level evidence from the Netherlands.” (2021). ↩