National Statistician’s Independent Review of the Measurement of Public Services Productivity

Published:
13 March 2025
Last updated:
14 March 2025

Chapter 5: Exploring drivers of public sector productivity

The level of public sector output produced reflects not only the amount of inputs but the efficiency in which labour and capital are combined to produce that output. This highlights not only the importance in understanding the type of activities undertaken by those delivering public services, but also the effectiveness with which these are managed.

It is therefore important to be able to measure a) what activities are undertaken; b) the amount of time spent on work-related activities by those delivering them; and c) how well organisations are managed. The Review commissioned two new ‘pilot’ surveys of the public sector to provide such insights. Piloting these surveys would then inform decision-making on future user need and cost associated with adding them to the Office for National Statistics’ (ONS) survey portfolio.

The potential of new drivers of public service productivity are also important. Given possibilities through use of Artificial Intelligence (AI), this chapter also explores where data can be sourced to aid understanding of how public services are making use of these new tools, as well as other examples of Research and Development within the government sector.

5.1 Public Sector Time Use Survey (PSTUS) and associated qualitative research

The ONS has a long experience of running Time Use Surveys across the household sector. This Review provided an opportunity to develop a consistent measure of the amount of time spent on different activities by public sector workers through the Public Sector Time Use Survey (PSTUS). The PSTUS gathered data (in February 2024) from public sector workers living in Great Britain through an online diary tool to record their activities (both detailed work and high-level non-work activities) during a 24-hour period.

The ONS also commissioned the National Centre for Social Research (NatCen) to conduct qualitative research with a subset of the PSTUS participants to further explore views on self-perceived productivity, impact on administrative tasks on their productivity and on possible changes to improve their productivity, for example through delegation or automation.

Particular interest was placed on collecting data on administrative tasks, to understand whether these divert resources from more ‘productive’ tasks. These were based on the following principles:

  • Subsidiarity: The activity or process could be legally, accurately and exclusively undertaken by an alternative staff member or by an automated process without detriment to the operation of the service.
  • Pertinence: The activity or process is not a key part of the professional role of the frontline worker (those who tend to work with the public directly).
  • Proportionality: The activity or process is a key part of the professional role of the frontline worker, but the frequency, length or depth of the model of delivery is disproportionate to the requirements of efficient and effective delivery of the service.

For further information about the survey and detailed findings from the survey data and qualitative research, please see the Time use in the public sector, Great Britain publications. The ONS will review the frequency that may be justified and beneficial for any future survey, for example by exploring changing use of AI.

5.2 Public Sector Management Practices Survey (PSMPS) and associated qualitative research

The importance of management for businesses’ performance is recognised in published literature. Evidence from the private sector shows that the lower the average ‘management score’ is, the lower the productivity level, where management is important in organisational efficiency.

The ONS measures private sector management practices through the Management and Expectations Survey (MES). A comparable measure did not exist for the UK public sector and is limited in scope even in the World Management Survey. Smaller scale studies have found that better structured management practices are associated with delivery of better public service outcomes.

The ONS adapted the MES to the public sector, measuring management practices in a consistent way across the UK for the first time. A pilot survey was conducted from April to July 2024. This also established a baseline performance across public sector organisations, which could be followed up in future surveys. It would also enable the ONS to compare management practices in the public and private sector.

The survey was designed to measure management quality across organisations and ‘score’ them against:

  • Continuous improvement – how well organisations monitor and adapt to unexpected situations.
  • Key performance indicators – how many, and how frequently they are reviewed.
  • Targets – how targets are set, tracked, and reviewed.
  • Employment practices – processes of promotion, management, and training of employees.

Leaders of public sector organisations from central government, local government, police and fire services, education, and health and social care sectors were sampled to complete the survey on behalf of their organisation.

The ONS commissioned NatCen to conduct accompanying qualitative research to develop understanding of public sector managers’ views on the types of administrative tasks carried out and their impact on productivity, and explore opportunities and barriers to innovation, including the use of automation and AI.

For further information about the survey and accompanying qualitative findings, please see the Public Sector Management Practices Survey pilot, UK: 2023 publication and the Public sector managers’ views on management practices, Great Britain: August to September 2024 publication.

The ONS will be making the microdata available for researchers to use via the Secure Research Service and Integrated Data Service to improve potential research in understanding how management practices might impact upon the provision of public services. The ONS plans to release further analysis of the PSMPS pilot survey in Spring 2025, which would show how management practices compare in the public and private sectors. The ONS will also review the merits of future waves of this survey.

5.3 Artificial Intelligence: research by the Alan Turing Institute

AI presents opportunities to transform public services. The UK Government has identified the potential for large-scale productivity gains from the adoption of AI across the public sector, as outlined in the 2024 National Audit Office (NAO) report.

However, to date a unified view on the feasibility or cost of delivering these improvements has not been reached. While developing such assessment is outside the scope of this Review, both PSTUS and PSMPS data collection offer the opportunity to generate new valuable insights on the applicability of these technologies to public sector activities, and the barriers and facilitators currently in place towards its adoption.

In partnership with the Department for Transport (DfT), the Review commissioned research working with the Alan Turing Institute (ATI). The aim was to develop an approach for identifying, prioritising and evaluating generative AI applications within DfT, while also measuring the resulting productivity gains. This would offer a valuable blueprint for adoption that can be used by public sector organisations.

The ATI conducted qualitative research with senior civil servants from across DfT to understand and map business processes that are both high-frequency and repetitive in nature. This research allowed DfT and ATI to focus on two business processes where generative AI tools could bring productivity gains in the short term:

  1. Accurately answering internal HR recruitment queries.
  2. Answering policy questions for policy briefs, ministerial correspondence, urgent questions from parliament or questions from journalists and the general public, and information retrieval for policy briefs.

For the first process, DfT and ATI will develop and evaluate, respectively, a Human Resources (HR) chatbot tool to automatically answer questions related to recruitment processes, based on current HR policy documents. For the second, ATI will develop a proof-of-concept of an information triage tool that can sort through large amounts of heterogenous policy documents, signposting the most relevant information to inform responses to policy questions.

ATI also carried out a mapping exercise using the PSTUS data to assess the extent to which work activities are potentially amenable to generative AI adoption. ATI built on established methodologies from the field of labour economics by adapting an ‘automation exposure’ rubric developed by Eloundou et al. (2023) to assess the potential for applying generative AI to work activities in the public sector based on their exposure to technology.

The rubric assessment was based on criteria including the activity’s contents, methods and tools of work, and the capabilities of technology. Each activity was scored against ‘automation exposure’ to indicate the potential for generative AI application to reduce the amount of time spent on it.

5.4 Key messages from the ONS surveys, accompanying qualitative research and ATI research

The research conducted by the ONS on new pilot surveys and accompanying qualitative research, and through working with DfT and ATI has generated new and complementary insights. These findings are summarised under over-arching themes.

Barriers to improvement

Findings from the PSMPS showed that the most common perceived barriers to improving the way public sector organisations are managed were cost (58% of organisations) or there was too little time to think about or implement them (41% of organisations).

Managers from the accompanying qualitative research said that lack of capacity and financial resources were barriers to innovation, particularly in sectors with more public-facing duties such as education and healthcare. Resistance to change and unwillingness of staff to learn something new was mentioned by some managers. While organisations were open to change, innovation was challenging when there was pressure to be productive in day-to-day tasks.

Impact of administration tasks on productivity

The PSTUS and qualitative research showed that while administration tasks can be viewed as time-consuming, they are perceived to be important and often essential to be completed.

Managers (from the qualitative research) also found administration tasks to be time-consuming, whereby those in more public-facing sectors (such as health, education, fire and police) found it difficult to reserve time specifically for administration and sometimes needed to work out of hours.

The PSMPS respondents reported lack of resource (66% of organisations) and ad hoc work requests (56% of organisations) as common reasons for not being able to complete administration work on time.

Use of automation and streamlining processes to save time and make administration work more efficient was a common finding across the PSMPS survey, accompanying qualitative research and PSTUS research.

Use of AI

Findings from the PSMPS showed that 42% of public sector organisations had tested, used or planned to use AI; with common reasons for AI adoption being to automate tasks performed by labour (45% of organisations) and improve quality of processes (37% of organisations).

Accompanying qualitative research on managers and public sector workers showed that they tended to respond favourably to introducing AI in their workplace and saw its potential to free up staff time, reduce time spent on administration tasks, and improve the accuracy and speed of reporting. However, participants did caveat the need for careful implementation and oversight, particularly in more public-facing sectors (such as healthcare, education, fire and police).

The qualitative research conducted by ATI identified five key areas that could benefit from the introduction of AI to support delivery:

  • Drafting correspondence – such as emails, letters, briefings for ministerial, public or internal departmental purposes, ensuring they accurately reflect governmental stances, policies, and legislation, and are consistent with previous correspondence.
  • Data and information management – gathering information and accessing data (that is often spread across different locations and in different formats) quickly, and with assurance of its accuracy.
  • Project and budgets assessment – processes related to management and oversight of projects and budgetary allocations to monitor project progress.
  • Public relations and engagement – drafting consultation questions, transcribing interviews or conversations and recording queries from the public or press.
  • Process tracking – whilst overlapping with correspondence and data information management, additional examples included maintaining logbooks to register press questions and responses, tracking where a policy brief is along the completion pipeline and tracking recruitment process (such as application status).

ATI will publish its report in Spring 2025.

5.5 Analysing Government research and development

Key to innovation is investment in the creation of new ideas and technologies. AI is just one example, but new processes, new methodologies, and new working practices are equally relevant.

The ONS conducts an annual survey of Government Expenditure on Research and Development (GOVERD) which collects information on how much the UK government spends on Research and Development (R&D). Not all UK government R&D expenditure directly relates to public service improvement as the UK government funds projects undertaken by the private sector. Equally, other types of capital investment (for example more advanced scanners in hospitals) will be equally important.

The Review compared two sources of UK Government expenditure on R&D: HM Treasury Online System for Central Accounting and Reporting data on expenditure, and the ONS GOVERD survey. This revealed that while both data sources are valuable, they serve distinct purposes and user needs. The first focuses on budgetary data, making it essential for fiscal policy and expenditure monitoring, whereas the ONS survey provides detailed survey data on R&D activities, meeting international reporting requirements.

The initial comparison highlighted that these sources are unlikely to consolidate or substitute each other because of their different definitions and purposes. The ONS will, however, continue to monitor whether the guidance, definitions and purposes of these data converge in future, which may enable the consolidation of the datasets.

Recommendation 21:

he ONS should keep under review whether there is convergence of the HM Treasury expenditure data and the ONS Government Expenditure on Research and Development Survey estimates to allow future consolidation of the two data sources.

Back to top