Please see the below article from EPIC Investment Partners detailing their discussions on the UK Public Sector. Received this morning 09/10/2025.
The UK has a big public sector productivity problem
As the Chancellor does her sums, she will see that UK public sector productivity is still 4% below the levels it reached in 2019 (ONS estimate), with the NHS lagging by rather more. This is a major extra cost to the taxpayer and probably is larger than the whole of the overrun in the deficit the OBR will report to her. If only she could magic it away, harder choices like tax rises or cuts in spending that people value would be avoided.
The previous government presided over a big recruitment into the civil service and into NHS Administration during covid, and those extra numbers have stayed in post. The current government has added further to them.
What is current policy to boost productivity?
The present budgets assume there will be a squeeze on staff numbers and an increase in output to tackle this problem. The spending figures on administrative overhead, which is running at £14.08bn this year, are for it to rise to £14.23bn next year, before falling away to £13.2bn by 2028-9 and tumbling to £12.6bn the following year, after an election. This implies reducing numbers, as there will be pay awards and promotions to cater for.
However, if you read on in the public accounts, you discover that a Transformation Fund has been set up to spend £3.25bn on the application of AI and other improved processes to achieve some productivity gains. Adding those back into the figures, the administrative costs rise from £14.33 bn this year to £16.23bn next, an increase of 14% before falling to £14.89bn in 2027-8. That is still 4% above the current year level. The “Transition costs” are said to be one offs and temporary so are not included in the main figures.
It may well be that AI offers plenty of scope for economies in administration in government. The expensive welfare system can be extensively computerised and already uses a lot of computing power to distribute benefits to millions of people. The NHS might be able to handle patient records, bookings, workforce planning and other central tasks better with more AI. These gains should, however, be additional to recovering the lost productivity since 2019, as 2019 was a pre-AI age.
It is also the case that spending more money on digital systems only delivers productivity gains and cost savings if sufficient staff posts are removed to more than offset the extra computer cost. The government needs to explain both to its staff and to the public what it has in mind in terms of lower staff numbers and how it can handle that without compulsory redundancies. Natural wastage can achieve a lot but it requires more discipline over external hirings than the government has managed so far.
Trends in civil service numbers and gradings
Civil service numbers fell to 384,000 full time equivalents in 2016, following a long period of controls on recruitment and replacement. It has now risen to 516,000, an increase of a third. The public sector as a whole employs 6.1m workers, of which 1.2m are designated as being in administration.
Between 2014 and 2024 there was substantial grade inflation or promotion. In 2014 58% of the civil service was at Executive Officer or higher grade, whilst that rose to 74% ten years later. Some of this reflects the automation of some lower paid jobs and the need for higher paid supervisory. Some of it probably was caused by the wish to retain people without being able to give them an above average pay rise, unless they could redefine their responsibilities and be promoted. Either way the civil service is now more top heavy and has higher average pay.
What are the likely chances of getting a decent boost to growth and to cost savings from public sector productivity?
In theory the chances are good. We know the government can be run with far fewer civil servants without new technology, as has been done in the recent past. It should be possible to recapture the lost productivity by better management of staff numbers as people move jobs or retire. AI should be an ideal technology to make a big reduction in civil service administration costs. Much of the work is drafting briefs, policy papers and administrative documents. AI would be well placed to do the first draft and to give officials easy access to the stock of established government policies, positions and data to inform the latest brief or decision.
In practice it looks as if the transformation will be delayed and the costs will be front loaded. The second year of a new government would be the ideal time for Ministers, when they have learned more of their departments, to work with the senior officials to get some savings. The Transformation fund, with a very round number of £2bn to spend next year, clearly needs a lot more work on what the money will buy, how long it will take to put in and where the savings will come from. The forecast of bigger savings in a future year that will definitely be after the election is not a very bankable proposition in this fast-moving world of AI. The Chancellor will need to look elsewhere for savings or put up taxes if the AI programme remains on its current leisurely timetable.
Please continue to check our blog content for advice, planning issues and the latest investment, market and economic updates from leading investment houses.
Alex Clare
09/10/2025