Week in Public Services: 3rd April 2024
This week: public sector pay trends; AI myth-busting; and a reality check for Welsh education policy
General
Things seem quiet in Westminster, maybe too quiet, as we emerge blinking from a four-day weekend. There is still no general election date, but local elections are quickly bearing down on us, as I’m reminded by a seemingly constant barrage of electoral commission adverts. If anyone’s wondering what’s going on, the Institute for Government devolution team have got you covered. They’ve published two explainers: one on the metro mayor elections from my colleague Briony Allen and another on the London assembly elections from Millie Mitchell and Akash Paun. Likely more to come so watch their twitter feeds for updates.
Back on firmer public services territory, the IFS have a report out about recent trends in public sector pay. There’s a slightly nuanced picture. Both public and private sector pay improved substantially between 2001 and 2008. That was followed by a decade of declining performance in both sectors, with the public sector doing less badly. Then the pandemic threw a spanner in the works and private sector pay ended up in a higher place compared to 2001 than public sector pay.
The other key takeaway is that there has been substantial compression in public sector pay. The 25th percentile saw their pay grow by roughly 16% in real terms between 2007 and 2023. Meanwhile the 90th percentile — which includes professions such as doctors and teachers — saw their pay fall by roughly 10%. Maybe part of this is due to the large rise in the national living wage in that time? Whatever the cause, it does help to explain why doctors and teachers have arguably had the most acrimonious industrial disputes since the end of the pandemic. It also probably explains some of the recruitment and retention issues in those services; as Paul Johnson says in his write up in the Times (£) “If we want high quality, highly qualified people delivering vital public services then we can’t let their pay fall ever lower relative to what is available elsewhere”. Couldn’t agree more.
Health and social care
The King’s Fund and Nuffield Trust released their annual report on the public’s view on the NHS and social care. The headline is that satisfaction with the NHS has never been lower — only 24% of respondents in 2023 said they are satisfied or very satisfied, down from 29% in 2022. The ratings for social care are worse still; only 13% are satisfied or very satisfied. Simon Bottery pulled together two good threads diving into greater depth on the social care trends. The leading reason for dissatisfaction with NHS services is difficulty getting GP and hospital appointments — a data point which shows that waiting times is the key way that people experience the service. Another interesting result is that only 6% of people wants the government to reduce taxes and reduce spending on the NHS. Which is good news for a Conservative government that’s cut taxes twice in the last year and plans to do so again.
The new financial year starts next week so obviously NHS England released operational planning and guidance — basically the system’s targets for the year — months ago, right? Right? Nope. They published it on March 27th, with one working day before the start of the financial year. What’s actually in the document? There’s the usual (and relatively unambitious) increases in key targets — e.g. pushing 4 hour A&E waits up from 76% in 2023/24 to 78% in 2024/25 — and a lot of self-congratulation about how successful 2023/24 has been. But I found annex “3D — Use of resources” the most interesting. In that section, buried right at the bottom of the document, the NHS more or less admits that there is a productivity problem in the service. Of course, that is not news to anyone who’s been paying attention. But it has been very difficult to get NHSE to admit it publicly. Amanda Pritchard flatly denied that it was an issue in front of the health and social care committee last November. If acknowledging a problem is the first step towards solving it, then this represents progress.
Swift on the heels of the planning guidance, came this Times report (£) that the NHS faces a deficit of £4.5bn in 2024/25, before the financial year even starts. That has led the NHSE CFO, Julian Kelly, to claim that the next year will be about “consolidation” in trust finances, potentially in the form of staff cuts and a reduction in the use of bank and locum staff. Some have argued that since 2019, the NHS has basically just hired too many staff. That might well be the case; it’s hard to see the benefit of all those extra staff in NHS performance. But it does seem odd that last year the NHS launched its Long Term Workforce Plan, which aims to expand staff numbers by more than 300k by 2036/37, and then immediately encourages trusts to cut headcount. Does this just mean that some of those cuts will be gradually unwound? If so, that seems remarkably short-sighted and could lead to a similar situation as happened in the police between 2010 and 2023 — cutting experienced staff only to go through the expensive and haphazard process of re-hiring replacement staff. Struggling to square some of these circles. And it brings me back to a point that we made in Performance Tracker last year: there is a vast gulf between announcing plans to hire more staff and actually providing the funding to pay them.
The BMA renewed its mandate for junior doctor strikes. If the government hoped that there’d be a decline in support among doctors, they’ll be disappointed: 98% voted in favour, pretty much identical to the previous two votes. A silver lining for the government is that this vote had the lowest turnout: 62% down from 77% and 71% in February and August 2023 respectively. A couple more votes on that trend and the strikes will end lack of turnout. The BMA hasn’t announced any new strike dates yet, but now can do until mid-September. Interestingly, that might mean that they have to renew the mandate again if they want to strike during an election campaign towards the end of the year.
Disclaimer on this one: I am massive AI sceptic. The word “luddite” has been thrown around the office. So maybe there is a tiny bit of confirmation bias going on here. But I loved Jess Morley’s presentation at the Nuffield Trust summit. She argues that while there are reasons to be optimistic about the potential for the use of AI in the NHS — particularly in relation to algorithmic clinical decision support software (ACDSS) — there are also considerable barriers.
First, existing NHS IT systems are notoriously creaky. It is therefore not at all clear how you can run complex AI programmes on top of them. As Jess memorably puts it, it’s like “running HS2 on rotten wooden railway track”. Second, there is very little evidence that AI works in a clinical environment. Third — and I think this point severely undermines many of the claims that AI will revolutionise NHS productivity — Jess argues that using AI image recognition software to speed up diagnoses is great, but does nothing to solve the slow flow of patients through the rest of the pathway. Finally, it could create the “inverse data quality law”, wherein better off people generate a lot of medical data which is then used to train AI models, but ignores more deprived people who tend not to generate the same quantity or quality of data. So good to hear a discussion about AI in the NHS rooted in practical considerations rather than magical thinking.
Max Warner at the IFS has done some really interesting analysis of elective waiting lists since 1987. He finds that charts which stitch together different waiting list metrics — like these from the Financial Times and the Labour party — are somewhat misleading. First, while the various metrics overlap in the time that they were published, they are often not that comparable, changing at different rates. Instead, Max finds a far more nuanced picture. For example, while the waiting list might have grown between 1987 and 1997, median waits for care fell. Similarly, even though median waits rose during the 2010s, there were still far fewer people waiting a year or more than in 1980 and 1990. Definitely not as straightforward as some of those stitched together charts imply (shock!).
We’ve argued in previous work that hospital staff are more inexperienced now than they were before the pandemic. But it’s difficult to quantify the extent of this. Lucina Rolewicz at the Nuffield Trust tackles the question in this piece and shows that on a number of metrics, this is certainly the case. One thing that shocked me from this is that the proportion of the nursing and midwifery staff aged under-25 has increased from about a quarter in 2015 (27.6%) to more than a third (33.6%) in 2023.
The BMA is surveying members on their views about the 2024/25 GP contract. If a different survey by GPOnline is to be believed, it looks like GPs will vote overwhelmingly against the government’s proposals. It’s worth noting that that BMA survey is largely for show; the government will impose the contract on GPs regardless of its outcome. What’s interesting is what happens next. The BMA has warned that it will ballot for industrial action, with walk outs potentially occurring during an election campaign. We have, however, been here before; GPs have threatened industrial action for the last two years and nothing has come of it.
I loved this thread from Craig Nikolic which argues that the NHS’s focus on same day care in general practice is actively endangering patients and the doctors who have medical responsibility for them. He argues — rightly, in my opinion — that the lack of care available anywhere else means that people are repeatedly pushed back through general practice. In that context, the massive recent expansion in GP appointments is effectively useless as it is not expanding access to general practice, but is instead just dealing with the increasingly complex need of patients who access the care they need elsewhere in the system.
Politicians, journalists, and anyone else who wants a low-effort way of demonstrating their tech savviness have long trotted out claims that robots will revolutionise care work. Despite the hype, robots have been a complete flop. This Atlantic article examines why. First, robots are expensive and care providers don’t tend to have a lot of spare cash lying around. Even when providers do buy robots, they need a lot of maintenance. Second — and far more importantly — they’re just not very good. Research has shown that they work best (which is still not very well) when they’re used in conjunction with staff members, thus creating more work for already overstretched employees. On a personal level, I find the supposed utopian vision of people being cared for by robots unspeakably bleak.
Children and young people
PISA results are a few months old now (we wrote about the headline findings at the end of last year), and the IFS has published a really interesting paper arguing that Wales’s declining performance should be cause for concern among policy makers. The author, Luke Sibieta, controls for factors such as deprivation, ethnicity, and funding and still finds persistent differences in outcomes compared to other UK countries. Luke instead lays the blame at the door of Welsh education policy, specifically the move towards a skills-based (rather than knowledge-based) curriculum, pupil assessment and school accountability. He finishes by urging policy makers to pause and consider the direction of travel.
The Education Policy Institute (EPI) examines the outcomes for suspended pupils in this paper. They find that (maybe unsurprisingly) peoples that are suspended are on average 12 months behind their peers who have not been suspended, resulting in worse GCSE grades. Even after controlling for other characteristics, the gap persists. Despite this, the EPI backs away from the assertion that the suspensions cause poor outcomes, instead arguing that there may be other confounding, unobserved variables.
Another EPI report. This one, published in partnership with Renaissance, concludes their series looking at the impact of the pandemic on learning outcomes. They find that outcomes for secondary reading, secondary maths, and primary maths are lower than before the pandemic. Only primary reading bucks the trend, being slightly higher. The disadvantage gap widened in all but secondary maths.
This long read marking ten years since the introduction of education, health and care plans (EHCPs) reveals the permacrisis in the SEND system. The government launched EHCPs with good intentions. But the simultaneous decision to cut other statutory local authority services, including services for young people, meant that increasing, pressure was piled on EHCPs. The resulting gap between parents’ expectations and what councils could actually deliver has led to a vicious circle: financial pressures cause councils to constrain funding for mainstream schools; parents of children with SEND requirements then embark on a months-long legal process to obtain an EHCP and secure a place at a specialist school; this then increases the SEND costs for councils. Repeat ad infinitum.
Law and order
A compelling argument from Rick Muir about the case for reforming policing. He outlines three major challenges. The first is capacity. The nature of demand for police services has changed — not least the massive explosion of online fraud, the subject of a recent event we held at the IfG. Next is capability. The police do not have the necessary skills or mix of staff. As an example, he highlighted a shortfall of 7,000 detectives — a deficit that is obscured by the hiring of 20,000 additional general officers. He also argued (probably uncontroversially at this point) that the service needs to take far more drastic steps to root out bad officers and improve recruitment and vetting. The final challenge is organisational. The separation of the service into 43 forces means that there is little national collaboration. He therefore recommends some kind of strong, central coordinating body. He also argues that there has been massive underinvestment in capital, particularly in IT systems. That kind of talk is catnip to us at the IfG who have been arguing the same for a while.
The Times have a story from a journalist who took an undercover job as a prison officer in HMP Bedford. The insights about the quality of the estate are bad, but relatively old news at this point. The parts that shocked me the most was the lack of vetting when the prison hired him as an “operational support grade” escort — basically a contractor. With very little training, he was allowed to work in close proximity to prisoners from day one, often with few security checks when entering the prison. This lack of vetting reminds me a lot of the police, where a sudden need to massively expand the workforce incentivised the hiring of people with very few background checks. Not an ideal way to manage the workforce.
Local government
An annual highlight for those of us who love some road maintenance data: the Asphalt Industry Alliance’s published its annual report on the state of roads. The big headline is that the one-off cost of repairing the country’s roads rose from £14bn in 2022/23, to £16.3bn this year — a 16.3% increase and a record high. I do wonder with though how much of the increase is driven by rising costs of conducting the same work, rather than a substantial deterioration in road quality.
