A saving can be real and still be incomplete.

An organisation can reduce one cost, record the benefit, and later spend money somewhere else to keep the same service functioning. Overtime may sit in a different line. Recruitment may arrive years later. A failed contract may be stabilised or terminated under another programme.

That does not mean every later cost was caused by the original saving. It means the saving should not be judged from one favourable account alone.

The prison and probation systems provide two unusually clear public records of this problem.

The prisons that became cheaper to run

The Prison Unit Cost Programme aimed to reduce the operating cost of public-sector prisons between 2013 and 2016 while maintaining safety, security and decency.

Official programme data recorded £306 million in annualised savings—15% against the 2012–13 baseline. The 2015 Spending Review described approximately £300 million a year in savings from 2015–16 and said public prisons were being kept as efficient as private ones.

The staffing changes were substantial. Between March 2010 and June 2014, total full-time-equivalent staffing in public-sector prisons fell by 28%, a reduction of 12,530 posts. The ratio of prisoners to staff rose from 3.8 to 4.9. Annual staff turnover rose from 7% to 15%.

The decline began before the formal Prison Unit Cost Programme and was not caused by one policy alone. A parliamentary committee recorded several possible explanations for shortages: benchmarking, voluntary redundancies, resignations, recruitment freezes, sickness, an unexpectedly high prison population and a stronger labour market in some areas.

Prisons were also dealing with psychoactive drugs, overcrowding, ageing buildings and changes in the prisoner population. Those pressures matter. They prevent a clean claim that reduced staffing caused every later failure.

But the later spending still belongs beside the original saving.

As staffing pressures continued, prisons also spent substantial sums on Payment Plus, which paid officers for additional hours. The bill was £37.2 million over 11 months of available data in 2016–17 and £50.3 million in 2017–18.

Government then announced an additional £100 million a year and 2,500 more prison officers. By March 2018, it reported that the target had been exceeded, with 3,111 additional full-time-equivalent officers recruited since October 2016.

Violence had also increased. The House of Commons Library found that assaults by prisoners more than doubled between 2010 and 2019. Officer numbers fell by about a quarter between 2010 and 2014 and recovered after 2017, but headcount alone does not restore experience immediately.

No honest calculation can subtract the overtime and recruitment figures directly from the £306 million saving. The dates, baselines and accounting boundaries differ.

The point is simpler. The saving was announced in one place. Later costs of covering shifts and rebuilding the workforce appeared somewhere else.

Probation and the price of a failed contract model

Transforming Rehabilitation reorganised probation in England and Wales. A public National Probation Service managed people assessed as posing higher risks. Twenty-one privately owned Community Rehabilitation Companies managed lower- and medium-risk cases.

The reform did something important: it extended statutory supervision to people released after short prison sentences, who had previously received no probation support. The new contract model was supposed to bring innovation, value for money and lower reoffending through outcome-linked payments.

The Ministry of Justice implemented the change quickly and without a full-scale pilot of the national structure.

The National Audit Office later found that the companies had underinvested, hoped-for innovation was limited, and friction between the public and contracted services persisted. Lower-than-expected caseload volumes placed companies under financial pressure because the Ministry had overestimated how easily their costs could fall when income fell.

By 2018, the contracts were being ended early.

At the time of its 2019 review, the National Audit Office projected that the Ministry would pay £296 million above the original contract terms through changes intended to stabilise the providers. Minimum termination costs added £171 million. The combined projected amount was at least £467 million above the original terms.

Those figures must be read carefully. The Ministry’s August 2018 forecast of payments to the companies was £2.3 billion through December 2020. The original expectation had been up to £3.7 billion through 2021–22. The periods, workloads and service assumptions were not the same, so the lower total does not settle whether the model provided value for money.

The National Audit Office’s conclusion was nevertheless clear: the reform had delivered poor value, required expensive stabilisation and had made little progress in transforming probation services.

Probation was reunited as a public service in June 2021. The Promise Ledger records that unification commitment as delivered. That correction did not make every problem disappear. Later audits found continuing staffing shortages, high workloads and weak performance.

A structure can be repaired without immediately restoring the capacity that repeated reorganisations have consumed.

The missing side of the account

The prison and probation cases are different.

Prisons show a staffing and operating-cost programme followed by heavy overtime and fresh recruitment spending. Probation shows a contract model that required additional support and early termination.

Neither proves a universal formula linking every reduction to every later cost. Together, they expose the same accounting weakness.

A staffing reduction appears under efficiency. Additional shifts sit in the pay bill. Recruitment is treated as a later operational decision. Complaints, delays and safety failures have separate systems. Costs carried by workers and families may never enter the organisation’s accounts.

The question is whether a cost was removed or merely moved.

If a process genuinely needs fewer resources while maintaining the same service, that can be a real efficiency. If permanent staff are removed and the remaining workforce is later paid to cover missing hours, the cost has at least partly changed category. If a contract has to be stabilised and terminated early, those costs belong in any serious assessment of the original model.

Count the whole result

Major efficiency programmes should be required to publish one account containing both sides of the change.

Using the same baseline and reporting period as the financial saving, that account should show what happened to:

  • permanent staffing and workload;
  • sickness absence and departures;
  • overtime, agency and temporary staffing;
  • errors, complaints, delays and restricted services;
  • later recruitment, stabilisation and recovery spending.

The reporting should continue for several years. A saving declared in year one may depend on costs that do not become visible until year three.

This would not prevent staffing changes, digital reform or contracting. It would make the claim testable.

The responsible organisation could still argue that outside pressures caused the later problems. It could show that demand rose, the workforce changed or separate events intervened. The public could then see the full explanation rather than one favourable number.

The Ministry of Justice and HM Prison and Probation Service were asked whether they regard the Prison Unit Cost Programme savings, later overtime and recruitment spending, and Transforming Rehabilitation stabilisation costs as connected—and how they assessed the full human and service cost of each programme. No substantive response had arrived by the 14 August 2026 deadline. That non-response is not evidence for the claims in this article.

A saving is not complete while its costs are being paid somewhere else.