Independent report

Independent Review by Sir Robert Devereux KCB, June 2025

Published 26 June 2025

Summary

The Terms of Reference for this Review summarised the areas of concern as the “performance and culture” of the Office for National Statistics, following a number of issues with core economic statistics. The full Terms of Reference have been published along with this review.

While the Review was in progress, the National Statistician Professor Sir Ian Diamond retired, with immediate effect, on account of ill health. The country was fortunate that he was at the helm of the ONS during the pandemic, during which he catalysed extraordinary work to deliver the COVID infection survey, and a range of behavioural data to help manage the pandemic. Shortly afterwards, under his leadership, ONS delivered a very successful 2021 census of England and Wales.

His interim successor, Emma Rourke, has taken two immediate steps of relevance to the review:

• extending the membership of the ONS Executive Committee, and simplifying other governance structures, to strengthen decision making;
• publishing, in parallel with this Review, a document which acknowledges the issues of quality (as requested by the Office for Statistics Regulation) and sets out how those issues can be remedied, including through re-prioritising spend in the current financial year.

I welcome both of these developments, since I expect them to make an immediate and positive impact on both culture and performance.

As I will explain, there are other – more deep-seated – issues which also need to be addressed at ONS. In my view, these would best be addressed by temporarily separating the role of National Statistician from the role of ONS Permanent Secretary, appointing to the latter someone with a track record of leading an operational business, indeed of “turning round” such an organisation. I suggest temporary separation since, with the more effort to develop evident talent within the Government Statistical Service, I think it might well be possible to re-combine the roles in due course, once the organisation’s core business is back on a more stable footing.

Having met a large number of ONS staff, and been impressed by their expertise and dedication, I am confident that with the changes already made, and those which would flow from the proposed appointments, ONS can rebuild its reputation as a leading National Statistical Institute.

Approach to the review

In order to reach robust conclusions, especially on culture, I spoke to most of the senior staff at ONS (over 65 interviews), and met a further 150 staff in group sessions, including those in the ONS field force and those involved in the day-to-day production of statistics.  I also talked to some former members of ONS, and ONS stakeholders. I am most grateful for everyone’s time, and their candour.

Economic statistics and ONS culture

Economic statistics guide critical decisions of Government, independent institutions like the Bank of England and Office for Budget Responsibility, private sector companies and individuals.

But measuring the modern economy is difficult. The economy itself changes, new data sources emerge, new technology becomes available, and international standards evolve. Like many other organisations, ONS is also wrestling with issues of recruitment, retention, and the replacement of legacy technology. And while users want the best statistics, some are unhappy when revisions are made to past series when a better statistic is developed.

ONS has a hard job managing all these challenges. And some work in recent years, not least on consumer prices, R&D, and population statistics have illustrated ONS’s capacity to engage with new data sources, develop new methods, and build modern systems.

That said, it is my view that most of the well-publicised problems with core economic statistics are the consequence of ONS’s own performance.  And that performance is affected by certain cultural issues. Three inter-related issues stand out:

• There has been a commendable interest in both new approaches to statistics (including the use of administrative data), and ensuring the relevance of ONS activity to wider political debate. Unfortunately, this has had the (unintended) effect of de-prioritising the less exciting, but nonetheless crucial, task of delivering core economic statistics of sufficient quality to guide decision making. There was an opportunity cost in ONS securing additional funds for flagship programmes, like the Integrated Data Service, and using those funds to deploy scarce human expertise. That opportunity cost was both the constraint on organisational bandwidth, and the restricted funding for core economic production teams. To give just one concrete example of this, many people reported that ONS should by now be further forward on the development of the new Statistical Business Register, both to extend it to smaller companies and to reclassify those companies in preparation for the UN’s new system of national accounts.
• This interest in the “new” might have been better managed if ONS had a stronger system of planning and budgeting. Sadly, too many people I met described a divergence between what they were asked to do, and the resources provided to do so. Of course, all budgeting rounds are difficult: demands always exceed supply initially. ONS is to be commended for its efforts to develop so-called “horizontal” planning, which sought to assess the cost of all the inputs to, say, GDP or to price statistics, and use these costs to prioritise. Every production area also has a “quality improvement plan”. Yet in the most recent annual round, while such information was fed into the planning process, the end results looked to many in ONS closer to “equal shares of misery”. Very few people can point to substantive reductions in lower priority work. Instead, the four leaders of the principal teams within ONS were invited to make their own decisions on how best to balance their final budgets, despite the fact that each core economic statistic requires collaboration across all four teams.
• This in turn reflects what many reported as a reluctance, at senior levels, to hear and act on difficult news. The organisation had established values to be radical, ambitious, inclusive, sustainable. And the experience during the pandemic well illustrates how such values can drive extraordinary behaviour and results. Unfortunately, the lesson learned seems to have been “all things are possible”, without the nuance that this is typically only true when there is an overwhelming emergency, which understandably de-prioritises many other activities. Several people suggested that the list of values was missing “realistic”.

Taken together, these factors left ONS open to a variety of risks, some of which have then crystallised in recent statistical errors.

The interplay of prioritisation and funding decisions which led to production teams being stretched, yet still relying on legacy technology, lies behind both the following failures:

• the problem with trade statistics reflected known concerns about the efficacy of the underlying software configuration; but while reported as a quality risk, was not actioned before an error occurred;
• the problem with the Producer Price Index reflected a mismatch between the methods that had been agreed and the coding which implemented the methods; the use of an older coding language, less readily documented, contributed to the problem not being spotted ahead of the error occurring.

Prioritisation and funding decisions have also affected survey data collection. To avoid repeating the underspend on ONS’s overall budget for 2021-22, a decision was made to allow some parts of ONS to spend at a rate higher than was consistent with their 2022-23 budget allocations. In practice, though, it appears that the favoured areas were all too successful spending at a higher rate, to the extent that ONS had to make significant in-year reductions to allocations. Those reductions were applied across the organisation, including social surveys, even though that had not been an “overprogrammed” area. The resulting constraint on the survey field force, enduring into the next year, directly led to the decision to remove the so-called “COVID boost” to the Labour Force Survey. Almost as soon as that boost was removed, the Labour Market Survey failed, with the suspension of the series in October 2023.

A number of other issues arose, including in early June, from incorrect data being supplied to ONS. Late, and updated, data from one business affected earnings statistics; and errors in data received from the Department for Transport and His Majesty’s Revenue and Customs, respectively affected price and trade statistics. ONS choices about priorities and funding affected the capacity of data collection and statistical production teams, delaying the identification, and also the resolution, of these issues.

All of these examples help illustrate why – for me - the failures of economic statistics which have occurred are not best thought of as isolated issues, but rather as the almost inevitable consequence of the choices made (and not made) at the top of ONS, over several years. This includes choices about what to prioritise in seeking funds in the 2021 Spending Review.

And those choices in turn reflect a reluctance on the part of some to take at face value the warnings which have been raised, apparently preferring instead to categorise those making the warnings as lacking in accountability. This categorisation seems to me to be without foundation, and it has undoubtedly made life difficult for many senior people working at ONS who are concerned about the quality of population and economic statistics. I am not surprised that so many, experienced, senior leaders have chosen to leave.

This section has summarised some of my key findings about ONS performance – especially on economic statistics – and culture, since the two are so clearly related. Before explaining how I believe ONS can turn this round, let me first address the other elements of the Terms of Reference.

Programme Delivery

The perceptions that ONS had delivered well on the census, and during the pandemic, clearly influenced the ONS settlement in the 2021 Spending Round: the budgets for the three years beginning in 2022-23 included substantial ring-fenced funding for three programmes:

• Ambitious, Radical, Inclusive Economic Statistics – ARIES - £77m over three years;
• the Future Population and Migration System – FPMS – £140m with predecessor programme;
• the Integrated Data Service – IDS - £178m.

ARIES has successfully delivered step-change improvements to consumer prices inflation, business R&D, and public sector finance statistics. The programme had originally intended also to deliver several other improvements (e.g. to household finances), but as programme challenges became clearer, these were de-scoped to protect delivery of the highest value improvements.

FPMS aimed to use administrative data to enhance population and migration statistics, and migration estimates are now almost solely based on administrative data. Considerable effort was also expended to test whether administrative data could remove the need for the 2031 Census. There are mixed views on whether this was ever a realistic possibility, and on the value of some of the ideas pursued. FPMS programme funding was also used beyond the programme itself, to subsidise other population statistical activities that were not protected by ring fenced funding (for example, production activities, data acquisition and linking, and support for devolved administrations).

IDS has proved the most difficult programme for ONS. It has now delivered some ground-breaking data linkages, while protecting personal information. Building on earlier work, ONS can now, for example, assess the impact of various forms of surgery on patients’ propensity to return to work. However, at present the benefit cost ratio is very low, and IDS has low take-up by Government analysts outside ONS. It is also the case that substantial sums initially ringfenced for IDS have – with the HMT approval – been used to fund more general technology and data costs. And despite the IDS business case originally being in part predicated on the need urgently to replace the Secure Research Service, the two services are now running in parallel.

Those I spoke to in ONS drew two lessons from these experiences:

• that ONS should have built up a stronger capability in each production team for “continuous improvement”: doing so has the advantage of keeping activity closely targeted on the issues most of value for users, being nimble, and with low overheads; in contrast, the risk of relying largely on multi-year programmes is that they take on a life of their own, sometimes with substantial programme overheads, and while they might well make a material change in one area (as with consumer prices in ARIES) there is then little if any development on other, equally important statistics.
• where programmes are required, it is better for them to co-ordinate related activity across ONS (as ARIES), rather than gather resources into the programme and act largely independently of the rest of ONS (as many people perceive IDS); and to keep governance arrangements and project management proportionate to the task.  Given that IDS also relied heavily on securing data from other Departments, several people – in and outside ONS – thought the programme initially adopted a “high handed” approach, essentially demanding data, and downplaying others’ concerns for example around security.

I commend the way these lessons are already being acted on, in the choices which ONS has been making while this review was underway. For the most part, the published plan to improve core economic statistics rests on building strong “continuous improvement” teams, with the right mixes of skills, within each production team. More continuous improvement can, of course, co-exist with traditional change programmes, where the extent of change necessitates that (as it did with developments on consumer prices).

Oversight of ONS

The governance of ONS is essentially defined by:

• the legislation which establishes the Board of the UK Statistics Authority, and its two executive roles of National Statistician and the Director General of Regulation, with the Board responsible for promoting and safeguarding the production and publication of official statistics, including accuracy, relevance and impartiality; and
• the rules governing the management of, and responsibilities of, Permanent Secretaries.

The link between the two arises because the National Statistician has traditionally also been the Permanent Secretary head of the ONS.

The Cabinet Office then plays two roles: as the Department holding the responsibility for the ONS as a “non-Ministerial Department”; and because the Permanent Secretary at the Cabinet Office is, in practice, the line manager of the ONS Permanent Secretary (who is a Crown appointment, notwithstanding his or her executive role on the UKSA Board).

These complexities were born of a commendable interest on the part of the then Government to re-affirm, and strengthen, the independence of official statistics, by introducing the relevant legislation (the Statistics and Registration Services Act 2007). The legislation is about the entire system of official statistics, not simply ONS.

Since the legislation came into effect, there have been 4 Chairs of the Board, and 5 National Statisticians, all operating within the same system of governance. In respect of the performance of ONS, this system ultimately relies on:

• accurate information being provided to the Board, including appropriate metrics;
• the Board’s approval of the Internal Audit programme, and scrutiny of its outputs;
• the Board’s scrutiny of the Regulator’s work in respect of ONS outputs;
• rigorous assessment by the Board, particularly of performance and risk;
• clear Board feedback to the National Statistician; and
• Board input to the Cabinet Office’s line management of the National Statistician.

The Cabinet Office and HM Treasury may arrange their own scrutiny of elements of ONS work, always respecting ONS’ operational independence.

Of course, all systems of governance ultimately turn on the efficacy with which each person, and body, plays their part. And sifting sometimes conflicting evidence is hard. Given the issues I have recorded above, I am not surprised that various concerns were both felt, and expressed, by the Board. With hindsight, stronger intervention earlier might have resulted in the welcome change of direction now evident at ONS.

Recommendations

Unsurprisingly, given the findings above, my first recommendation is that ONS needs a focused and consistent effort to improve core statistics: the statistical system is complex, and many elements need action, which has to be carefully sequenced; past actions appear to me to be too piecemeal, driven by individual failures.

Given this recommendation, I do want to commend the current leadership for their rapid work to develop a strategy for economic and population statistics, which has been published alongside this review today. The strategy has several important properties:

• it is focused on core economic statistics, about which most external concerns have been voiced (but includes population statistics because of their role in weighting sample surveys, and in deriving eg GDP per head);
• it is comprehensive, in the sense that it looks at all the elements of statistics production from registers and classification, through data collection and validation, to production and dissemination;
• it is honest that some necessary improvements will take time to deliver, but sets out what can be done, including improvements now to be delivered in 2025-26 and those which will follow in later years.

The strategy also acknowledges the need for a different way to deliver change, with much greater emphasis on continuous improvement, with the right mixes of skills dedicated to each production area. ONS already has examples of this way of working, and with good results. The new Executive Committee has been central to the rapid assessment both of lower priority work, and to testing the plans for improvement. All of which I commend. I have also spoken to HMT and I am confident that the new strategy can be fully resourced within the agreed Spending Review 2025 settlement announced earlier this month.

My second recommendation, to support this new direction and strengthen delivery confidence yet further, is to change the way ONS is led. Since the passing of the relevant legislation in 2007, the National Statistician has also been the Permanent Secretary for ONS. These are quite different sets of skills: they have been found in the same individual before, but the combination is rare. I believe the balance of advantage now lies in separating these roles, at least for a temporary period, in order to appoint as ONS Permanent Secretary someone with a track record of leading an operational business, indeed of turning round such an organisation.

I recommend this because I believe most of the problems with core statistics result from inadequacies in the way ONS has made decisions, planned and budgeted, and managed risks. But there are other issues to tackle too:

• the return to working in the office, post pandemic, was not handled well and one of the unions is still operating “action short of a strike”. This needs to be resolved quickly: some recognition of the past poor handling is necessary; but so too is a wider recognition of the benefits of teams spending time together in one place;
• the organisation is, like many others, highly dependent on technology, but responsibilities for this seem to me unnecessarily distributed, and as a result insufficiently effective;
• there appears to be little by way of consistent performance management of individuals;
• ONS relies on some deep technical knowledge, but it is not clear how this is developed and retained, with sufficient incentives for people to keep abreast of new technology
• the organisation’s office locations, mostly outside major cities, and with analyst pay not comparable with other departments, makes it difficult to recruit and retain the necessary skilled staff;
• while there is clarity about the importance of ONS, there is not a consistent and compelling narrative about how the organisation sees and addresses its challenges, and responds to external critiques: this is a necessary cornerstone for effective engagement of staff across ONS, and to motivate and support each person’s contribution.

These sorts of challenges occur in other operational businesses. It would be better, right now, to appoint a new leader who has strong relevant experience from running a similar operational business. The ONS context is, of course, statistical production; but there are many capable statisticians in the organisation, and more now on the Executive Committee. So, temporarily, I do not regard it as essential for the new Permanent Secretary to have deep experience of statistics: there are broader skills required right now, to turn ONS round, and put it on a more sustainable foundation for the future.

The consequence of splitting the roles is the need to find a National Statistician who could: provide statistical advice to the Board and to the Government; provide leadership to the Government Statistical Service; lead engagement with the UK’s broader statistical community; ensure the statistical system works across the whole of the UK; and lead the UK’s engagement with international bodies. This role would then be more similar to roles such as the Chief Medical Officer.                 

My third recommendation is that various parties use such a turn round period to reflect on the appropriate governance of ONS and of the wider statistical system, and if they conclude that legislation needs to change, to do so well before further appointments are needed.

I described the current governance arrangements earlier. Despite recent challenges, I am reluctant to conclude this model cannot work. It has, after all, worked well before. But it has undoubtedly become more difficult to find the person who can be both the National Statistician and the ONS Permanent Secretary.

I do think it worth considering the opportunities which might be provided to develop those statisticians already holding senior roles within ONS, and elsewhere within the GSS. It is a big leap from these roles to Permanent Secretary. In other parts of the civil service, I have seen conscious choices being made in career development, taking individuals beyond their initial speciality in order to develop broader skills. For ONS leadership succession, such career moves might well take a statistician temporarily into an operational leadership role, at a level which starts to build some of the experience necessary to lead ONS. This might well help with longer term succession, but probably not in the medium term, given the time necessary to acquire new skills.

So, a recombined role in the medium term probably relies on finding someone from a different background, or from abroad. Again, this has been done before.

But if this does not look promising in fairly short order, it might be necessary to revisit the legislation. There are plenty of options available. Personally, in this scenario, I would be attracted to a National Statistician role as Chair of the UK Statistics Authority. In this role, he or she would both carry responsibility for engaging with the users of statistics, and – through oversight of the regulator – the ability to assure both the quality of the statistics produced and how they are used. The leadership of ONS would then naturally be a separate role. There is, though, no immediate rush to decide this if the current roles are temporarily separated: but the Board and other stakeholders should be thinking now about the longer-term options, to give time for any changes to legislation, knowing how hard it will be to find Parliamentary time for this.

Sir Robert Devereux KCB

June 2025