Economists sceptical over UK Spending Review's partly AI-driven 10% budget cuts
- Reference: 1750672813
- News link: https://www.theregister.co.uk/2025/06/23/economist_question_spending_review_budget/
- Source link:
What is the evidence that you have managed to utilize AI more efficiently?
Appearing before Parliament's [1]Treasury Committee last week, Helen Miller, deputy director of the Institute for Fiscal Studies, queried the results of a zero-based review (ZBR) of budgets which led up to the Spending Review.
"Is it right that they can all make exactly 10 percent? That seems… you know… I'd be very surprised if that was the correct answer that came out of [the ZBR]," she told MPs.
Earlier this month, the government's multi-year spending review, the first since 2021, set out plans for department spending and savings, in an effort to find the right balance between raising taxes, investing in infrastructure, making cuts, improving services, and increasing borrowing. Labour founded its plans to make £14 billion ($18.7 billion) savings from greater efficiency, partly through AI and digitization.
In the [2]Spending Review 2025 document [PDF] — under the heading Creating a cost-effective, high-performing civil service — it details departmental administration budgets for 2025/26 to 2028/29.
[3]
Departments getting a 10 percent cut include the Department for Health and Social Care, Department for Education, the Home Office, Ministry of Justice, Ministry of Defence, Ministry of Housing, Communities and Local Government (MHCLG), Department for Digital, Culture, Media and Sport (DCMS), Department for Science, Innovation and Technology (DSIT), the Department for Work and Pensions, and HM Treasury.
[4]
[5]
As part of the ZBR they were required to take a [6]"digital-first approach" to consider digital, data, and technology (DDaT) spending and "involve their chief digital and information officer and internal digital functions when preparing ZBR returns"."
A number of departments have detailed how IT and digital investment will help produce administrative efficiencies in [7]accompanying Spending Review documents [PDF].
[8]
For example, the DHSC said technology and digital transformation would "optimize the use of clinical and administrative time and avoid delays caused by outdated technology."
The Ministry of Justice said it would roll out AI across its agencies to "streamline repetitive administrative tasks, enhance decision-making and improve service quality."
MHCLG aid it would "increase productivity by using AI to expedite routine tasks, quickly checking simple errors and helping with administrative tasks such as note-taking." This will reduce demands on staff time by an estimated 500,000 hours per year, it said.
UK's Darpa clone faces tough test next spring as government considers future funding [9]READ MORE
DSIT said it would "be an exemplar for modern digital government through exploring, identifying, and implementing AI opportunities across the department, to improve staff productivity and free up time spent on administrative tasks."
But Miller puzzled over how departments signing up to 10 percent administrative cuts would all come to the same figure.
[10]
"It's very stark. If you look at the departments, it's assuming the same for everywhere. It does seem hard to think that if you had genuinely gone around and looked at each of them — where some are getting big budget increase, some are getting big budget cuts, they're presumably changing programs — it is right that they can all make exactly 10 percent?" she asked.
Dr Gemma Tetlow, chief economist, Institute for Government, agreed: "it does look a little bit strange that this is the same across all departments."
The Register has asked HM Treasury for a response.
As well as the cuts to administrative budgets, the government has committed to an expectation of at least 1 percent technical efficiencies for all departments in all future years, in line with an recommendation from the Office for Value for Money, a time-limited treasury unit.
"To support this commitment, the government will publish bespoke departmental efficiency targets and plans biennially, embedding a culture of continuous improvement backed up by greater confidence in delivery," it said.
Dr Tetlow told MPs: "It's going to be incredibly hard to measure this, but the benefit of having these plans is that at least provides an area to start asking questions as time goes on. What is the evidence that you have managed to utilize AI more efficiently?"
One area the government sees as ripe for cutting is consultancy spending, which has been the [11]subject of criticism and which successive governments have tried to curtail.
[12]UK Spending Review prescribes £10B digital remedy for NHS
[13]UK govt promises digital reform in spending review. We've heard that before
[14]HMRC's Making Tax Digital scheme also made tax more expensive – by £300M
[15]IBM and Oracle to support 280,000 users after winning mega ERP govt tech contract
In the Spending Review, the government said it would bring down the costs of spending on external consultants. "After announcing savings of £550 million [$736 million] in 2024-25 at the July Public Spending Audit by stopping non-essential spending on consultancy, the government expects to save over £700 million [$937 million] per year by 2028-29, reducing spend by half compared to previous trends."
Whatever the outcome of administrative budget cuts and 1 percent efficiency savings, the Spending Review bears high expectations of digital transformation and IT investment. Over the next four years, technology professionals in central government and across the wider public sector will be left holding the baby. ®
Get our [16]Tech Resources
[1] https://parliamentlive.tv/event/index/d53eb119-2327-4d21-86ec-ada5eb45276f
[2] https://assets.publishing.service.gov.uk/media/6849172b860362efc8e78836/E03349913_HMT_Spending_Review_June_2025_TEXT_PRINT.pdf
[3] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2aFl6FF889TeecXgYWLPMWQAAA1Q&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[4] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aFl6FF889TeecXgYWLPMWQAAA1Q&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[5] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aFl6FF889TeecXgYWLPMWQAAA1Q&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[6] https://www.theregister.com/2025/06/11/uk_tech_reform_spending_review/
[7] https://assets.publishing.service.gov.uk/media/68492799d0ca5d7801e4e709/Efficiency_delivery_plans_-_supplementary_document_-_FINAL.pdf
[8] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44aFl6FF889TeecXgYWLPMWQAAA1Q&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[9] https://www.theregister.com/2024/11/13/uks_darpa_clone_faces_tough/
[10] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/publicsector&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33aFl6FF889TeecXgYWLPMWQAAA1Q&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[11] https://www.theregister.com/2024/08/24/uk_government_consultancy_spending_grows/
[12] https://www.theregister.com/2025/06/12/nhs_tech_spending_review/
[13] https://www.theregister.com/2025/06/11/uk_tech_reform_spending_review/
[14] https://www.theregister.com/2025/05/01/hmrc_making_tax_digital/
[15] https://www.theregister.com/2024/09/27/ibm_and_oracle_uk_contract/
[16] https://whitepapers.theregister.com/
Re: That is what we have politicians for
Ah, this would be the famous 87.5% of statistics that are just made up on the spot.
Re: That is what we have politicians for
... and the other 21% are plain wrong?
Management 101
Always start with the conclusion (the cost “savings” target)
Then, write a report to support your required outcome.
Or
Just make shit up.
Then backfill.
Re: Management 101
It's easy to save 10%.
Just make someone else do 10% of your job.
The difficult bit is figuring out who the patsy's going to be.
I don't want plans I want results
I was involved (on the perifery) of a project to save money on a project. It was announced to everyone so we were all "part of it". Teams to produce plans.
The manager I was working with, called a meeting, and said "I don't want plans... I want actuals". We met up again a week later and had results like
1) because I always had to approve xyz - Ive set up a script to logon once a day and do it
2) we have all these old copies of software on disk - which we think no one wants. We've archived them to tape and freed up xGB of disk space
3) We don't think anyone uses these servers. We've powered them off. We can easily power them on again if anyone complains. We can virtualise them if needed.
All small stuff - but the first steps are the hardest.
A month after the big announcement the senior manager calling for the project to save money, asked for a review on progress so far.
Departments put up nicely written presentations about what they could do. etc
My manager put up one chart with the above points and said ( verbatim ish - it took less than 2 minutes)
- these are actuals. We are not doing any planning
- in any status meeting - people are allowed at most 1 chart
- the meetings now run to time
- we all go for coffee at 1000 - and raise any issues then - saving half an hour a day * 10 people
The senior manager took the hint and realised how much time (and cost) had been wasted doing blue sky thinking, and preparing charts etc.
He gave some ground rules
- No more overtime. Schedule your work better - or ask for help.
- I want to hear about meetings which are a waste of time
- Simple charts
- Get to know people in other areas who may be able to help you ( no more silos)
- I will be walking around the building - if you see me - tell me something (good or bad)
As a result meetings shrank (they had meetings but with 3 people not 13) and they found lots of overlap and so could save duplication.
Simple stuff but it worked.
"Leading economists have questioned how the UK government's Spending Review can determine exactly 10 percent cuts to admin budgets — partly powered by AI and digital transformation — across central departments when they are starting from different places and have different projects to manage."
It's Rachel Reeves. She's lying.
You can save 10%, but as soon as consultants get involved, it'll be saving 10% by spending 120%.
Trading amortized savings for linear (or worse) increase in cost (or $/productivitiy) is a known systemic failure.
You can also just ask your employees what they think could help. A hospital did this recently in the UK, nurses suggested a new model of triaging, they tried it, reduced waiting times by a significant amount, no overhead, no consultants (+fees), and based on real world evidence, everybody gains.
If your employees come to you and show some tech can and does work, take it seriously. If your consultants do, then run fast and far.
"efficiency savings" is just a euphamism for cuts, always has been. The mention of AI is just the magical thinking some consultant has sold the politicians that they can make the cuts without anyone noticing even more potholes or your bin only getting emptied twice a year. But it really is just magical thinking, not only will there be norticable cuts to critical public services but some of those cuts will be made to pay for the magical AI that won't deliver.
That is what we have politicians for
You see, when I pull numbers out of my ass to fast talk my way out of a situation, I exhibit great care to create a plausible backstory. I'd never try something as lame as "10%".
Makes me overqualified as a politician. As do my ethics...