Category Archives: Uncategorized

The limits of utilitarianism in health care

Lord Sumption has come under fire for comments he made in BBC a debate about whether lockdown was “punishing too many for the greater good”, where he told a woman with stage 4 cancer that her life was “less valuable” than other people’s.

He was saying this as part of his argument against lockdown, which is broadly that it punishes the young to save the old. Others have pointed out that this argument doesn’t work, because the young are hardly going to benefit from an overwhelmed NHS, dead parents and collapsing consumer spending among the wealthiest age group. Covid policy is not, in general, a trade-off between wealth and health, and it’s certainly not a zero sum game between generations.

But I want to set that aside and look at the argument on its own merits. Do we, and should we, consider some people’s lives less valuable than others when setting health care and wider government policy?

Within health care, there are places where we are very explicit about how we value lives. When NICE looks at the cost-effectiveness of new drugs, it will only approve them for use in the NHS if they cost less than £20,000 to £30,000 for each QALY (quality adjusted life year) they deliver[i]. A lifesaving intervention for a younger and healthier person will, other things being equal, deliver more QALYs than one for an older, sicker person, so we are willing to pay more for it. When it comes to how much we pay for drugs, we are explicitly utilitarian.

When we look more broadly at health care activities, however, things get messier. While health economists and policy-makers do try to assess the cost-effectiveness of interventions in a broadly utilitarian way, the reality of delivering health care often works out differently. For all the talk of prioritising community care and prevention, funding and activity continue to be sucked into hospitals and emergency care, and the last two years of someone’s life are by far the most expensive for health care. This is partly about the difficulty of managing change in a complex system, but also because there is a moral framework at play here that isn’t strictly utilitarian – we feel obligated to help people who need it and to “save someone’s life”, even if we might get more QALYs using the money somewhere else.

The allocation of government resources beyond health care diverges even further from a utilitarian ideal. While the cost-effectiveness of policies is assessed, the concept of QALYs is largely limited to health policy evaluation and there is no grand model that ensures a pound doesn’t get spent on roads if it could deliver slightly higher marginal utility in schools. Moreover, spending isn’t really allocated according to cost-benefit analysis when it comes down to it. The Treasury coordinates a fairly rigorous spending review process with other departments, but decisions on what money goes where will ultimately be made by ministers, based on what they think the public want and what they think will win them votes.

And so we end up with our public resources distributed through a messy combination of utilitarianism, public opinion, politics and accident. While this does lead to some indefensible decisions and pork-barrel politics, the fact that we have such a mixed approach is not a regrettable failure to implement our utilitarian principles, but a reflection of what we actually believe. There is a tension between our utilitarian sensibilities and our ethical and democratic ideals that all people are equal. Although we try to formalise this to some degree by looking at equalities alongside cost-benefit analyses, “equalities impact assessments” are often a bit of an afterthought, and we don’t have a way of trading of net utility against equalities – they are two separate frames that remain in tension.

This is where Lord Sumption gets it wrong – or at least voices a fringe opinion. There is a place for utilitarianism in public policy. It’s an important part of what people believe is fair, and can be implemented pretty explicitly in narrow applications like NICE’s drug assessments. But if we let it become our whole approach for allocating resources or deciding on policy, we ignore a widespread and fundamental belief that all people are equal, with equal rights, including the right to life. The next thing you know, you are on the BBC telling people who are old or disabled (both protected characteristics in equalities law!) that they are worth less than others. It’s not clear that this is ethically sound or sustainable in a democracy. We’re better off with the messy compromise that we’ve got.


[i] The fact that the threshold is in fact a range might suggest that the process is not as strictly utilitarian as it first appears.

Do the Covid vaccines vindicate our model of pharmaceutical development?

The development of three highly effective vaccines in record time was not only a rare piece of good news in 2020, but an incredible scientific achievement – and one delivered by our much-maligned drug companies. So often seen as the bad guys, Big Pharma has ridden to our rescue when we needed them the most. So does this mean that our system of drug development is working just fine after all?

It is said that war is a major driver of innovation. The idea is that an all-consuming threat brings a clarity of purpose and focuses a nation’s resources on solving specific problems. This leads to technological jumps in certain areas, which then have wider applicability when the war is over – in aviation, radar or nuclear physics. The Covid-19 pandemic was the pharmaceutical industry’s war. While in normal times research and development is spread across a range of diseases, suddenly it was absolutely clear what the priority was – and, crucially, absolutely clear that if you developed an effective treatment or vaccine you would sell bucketloads of it. With such a singular clarity of purpose, our drug development system has delivered, and in some style.

But what does this impressive system do in peacetime? Like any other market, the answer is that efforts are directed towards the things that are most likely to make money. Unlike many other markets, however, the ability to make money is entirely dependent on government regulation – and specifically patents. This is because pharmaceutical R&D is very expensive, but the marginal cost of manufacturing a drug is very cheap. The value of R&D is held in the intellectual property, so for there to be any incentive to develop new drugs, this needs protecting. While drugs are under patent they are hugely profitable because the manufacturer is granted a government-sanctioned monopoly, but as soon as the patent expires anyone can manufacture a generic version and the profitability largely evaporates.

The problem with profits relying so heavily on regulation is that the nature and specific detail of those regulations – rather than patient demand or social benefit – start to be the main factors driving behaviour. This has resulted in the development of a range of “me-too drugs”, designed to be just different enough from an existing product to get a fresh patent, with little regard to whether they provide a significant benefit to patients over and above what’s currently available. In normal times, a large slice of pharmaceutical R&D effort goes into these sorts of products, which are a less risky way to deliver profits than pursuing genuinely novel products.

In stark contrast to the response to Covid-19, these incentives leave some of our biggest public health concerns woefully underserved by pharmaceutical R&D. More than 700,000 people are estimated to die each year as a result of antimicrobial resistant infections, which the WHO describes as a global crisis. But this crisis – or rather the volume of drugs that you would sell if you fixed it – just isn’t big enough to convince the industry that it’s worth going through the expensive and risky R&D process required to develop genuinely novel medicines. Many drug companies have simply stopped working on new antibiotics. Of those that are in development, many are me-too products with limited benefits over what’s already available and only two out of 50 are active against the multi-drug resistant Gram-negative bacteria, which the WHO considers a top priority.

This year has shown the incredible scientific capability that we have been able to maintain across our universities and the pharmaceutical industry, and how effectively they can work together to develop innovative techniques and crack difficult problems. Whatever else our system has or has not been doing, we can be thankful that it has achieved this. But we also know that in normal times, the current system of regulations and incentives fails to target this capacity at our biggest public health problems.

There are two different lessons we might take from this. One is that, in the absence of anything better, we at least have a system that maintains an impressive capacity that can deliver incredible results if we ever find ourselves in a truly all-consuming crisis. Poorly targeted R&D in normal times might be a price worth paying for this, and allowing companies to play the regulations for huge profits a necessary evil. Or we may take the view that this year has shown us what could be possible under a different system. If we were able to focus pharmaceutical R&D on novel discovery in areas of genuine priority, perhaps we could deliver astounding results every year, without the need for a global pandemic.

Pull the lever

Why you should probably vote for the least bad party that can win in your constituency

It’s election day and we’ve got choices to make. Of course, we have choices to make every time an election comes around, but for many people – seemingly more than usual – none of the options are very appealing. As ever, there are only two parties capable of forming a government, but the realisation in recent years that both have serious problems with institutional racism has left many people feeling that voting for either is morally unacceptable.

So what should you do in this situation? Some of us are lucky enough to have an easy answer. If you live in a seat where another party (that you actually like) can win then you can simply vote for them. If you live in a very safe Tory or Labour seat then you can also vote for your favourite party knowing that it won’t affect the election result. But if you live in a Tory/Labour marginal, then I’m afraid that you are stuck in a trolley problem.

The trolley problem is a thought experiment where a trolley (or a train) is hurtling along and five people are tied to the track ahead. If you do nothing they will be killed. There is a lever to divert the train down another track to which one person is tied. If you pull the lever you will save five people but kill one – what is the morally right thing to do?

The point of the though experiment is to explore the moral equivalence between action and inaction: is doing nothing morally neutral, or are you responsible for five deaths? If you pull the lever are you a murderer or have you saved four lives? While this has been the subject of years of debate among moral philosophy undergraduates, empirical research suggests that nine in ten of us would pull the lever.

What this means is that most of us don’t consider abstention a morally neutral choice. If both options are unpalatable, but one is worse than the other, failing to choose the less bad option allows the worst to happen. If you find both the Tories and Labour unpalatable, but believe that one is worse than the other – for example if you oppose racism but believe in strong public services / think that nationalisation would be a disaster – then voting for a third party in a Tory/Labour marginal means you are not using your vote to prevent the worst outcome.

I’m not here to tell you what’s morally right, and the fact that nine in ten would pull the lever in a trolley problem means that 10% of us don’t think that’s the right thing to do. But if you are in the majority that would pull the lever, and you find yourself in a trolley problem today, then you should vote for the least bad candidate with a chance of winning, however unpalatable you find them.

Gotta get yer gotchas right

The gotchas around the Conservative pledge to recruit 50,000 more nurses are wrong – but the truth is actually worse.

The Conservative announcement that they will “deliver 50,000 more nurses” has not gone down as they would have hoped. Labour has picked up on the fact that some of this increase will be delivered by better retention of existing staff. Nicky Morgan had a particularly uncomfortable experience trying to explain her way out of this on Good Morning Britain, as presenters (and probably viewers) became increasingly incredulous.

The problem is that this criticism is wrong-headed. Increasing the size of the nursing workforce isn’t just a case of saying so – it’s a stocks and flows problem. If we want a larger nursing workforce, we need to increase the flows in (through more attractive, better-funded training, or recruitment of nurses from overseas) and decrease the flows out (through better retention). The Conservative manifesto refers to policies that are designed to do this, such as reinstating the bursary for student nurses. We might think that the 50,000 figure is overegging the effects of these policies – the Health Foundation has said that in reality half will need to come from overseas – but you wouldn’t want to attempt this without looking at retention.

That said, there’s a bigger problem with this commitment – or at least how it is being sold and understood. The manifesto costings earmark £759m for this policy in the first year. You may have noticed that this is much less than the cost of employing 50,000 nurses (which Full Fact has estimated as at least £2.8bn). That’s because this money isn’t for paying nurses salaries at all. It is just for the policies, like the student bursary, that the government hopes will improve recruitment and retention. It means that NHS England may have a better supply of skilled nurses to employ, but it doesn’t pay their wages.

So what does this mean for the number of nurses that will actually be working in the NHS? Well, what it certainly does not mean is that there will be 50,000 more nurses working in our hospitals and communities than there would be in the absence of this announcement. That’s not to say nurse staffing levels won’t rise: the government has already announced increased NHS funding for the coming years and some of that might be spent on providing more nursing. But this new policy is about developing the nursing workforce, not about funding more nursing care.

Almost everything I have read about the policy has misunderstood this point – and I doubt that would have been lost on the authors of the manifesto. Much cheaper to announce some measures that you hope will lead to better recruitment and retention and let everyone think there will be 50,000 more nurses working in the NHS than to find the £2.8bn that would actually cost. There has to be a strong suspicion that this was a deliberate attempt to mislead.

The correct gotcha is therefore this: will there be 50,000 more nurses working in the NHS as a result of this policy? If so, the numbers don’t add up, since £750m can only pay for around a quarter of that number of nurse salaries. If not, can we be clear that this is just about training, recruiting and retaining people to fill existing jobs, or jobs that the NHS is already planning to create with previously announced funding?

Although, to be fair, that would not have been nearly as effective on Good Morning Britain.

Prevention, longevity and health system costs

Kailash Chand has a piece in the Guardian arguing that unless the NHS fundamentally changes its approach to focus on prevention rather than cure, it will not be financially sustainable. This reminds me that many well-informed people hold inconsistent views on prevention and health spending.

Ask most people why health spending is rising and they’ll tell you it’s because people are living longer. But why are we living longer? According to the Lancet, the main causes are falling tobacco use in men – the result of a hugely successful public health campaign to prevent smoking-related diseases – and falling cardiovascular mortality – which is closely related to how much exercise we do. So prevention is a major driver of people living longer, and people living longer is leading to higher health care costs. Does that mean that prevention is bankrupting the NHS?

We can begin to unscramble this mess when we realise that mostprojectionsof health spending estimate that the contribution of ageing populations to spending growth is quite small, because as people live longer lives they are also healthier at any given age. This is reflected in the fact that a large proportion of health care costs relate to the last few years of someone’s life, however long that life is. An 85-year-old has on average a third of their health care costs still to come, but someone who dies younger doesn’t avoid these costs, they just incur them earlier.

So dying is expensive, but not all deaths are equally dear. A recent US study found that dementia is the most expensive way to die, once social care and out-of-pocket spending are factored in. This is bad news because we don’t really know how to prevent dementia. There are some promising interventions, and recent evidence suggests that age-specific rates might be falling, but the evidence that we can prevent dementia is currently fairly weak. The risk of developing dementia rises rapidly with age, so it’s plausible that if we prevent people from dying of other things then they will live long enough to develop dementia and end up costing the NHS more.

Now I’m not really trying to argue that prevention increases costs. It’s very complicated to work that out and, at a whole system level, empirical evidence is hard to come by. Moreover, there are some areas where prevention does seem to save money. A recent evaluation of diabetes prevention in the NHS found that it was likely to be cost saving (after 12 years) – and to be fair, diabetes is one of the examples that Kailash cites. Targeted interventions like this, especially those focusing on diseases like diabetes that don’t kill us but cost money, could help with sustainability. But it’s far from clear that a system-wide shift towards prevention is going to stop health spending from going up.

This is no reason not to do more prevention, of course. Living longer and with fewer diseases is a great thing in itself and if preventive interventions can help us to do that then we should implement them, whether they are cheaper or more expensive than our current approach to health care.

But acting as if prevention (or integration, or anything else) is the answer to all of our concerns about the sustainability of health spending is not helpful. It reinforces the idea that rising health spending is a symptom of failure and always bad – but if what we get for the money is longer, healthier lives, it might well be worth it. And in doing so it lets the current government off the hook for their persistent underfunding of services and distracts from more important questions. If we want the NHS to meet the needs of an ageing population and provide us with the latest treatments, we might just have to accept that this means putting in more public money and paying for this through higher taxes.

Austerity and Brexit in England

Ever since the UK voted to leave the EU, there has been a steady stream of articles and analysis trying to figure out why. Clearly there is more than one answer: different people voted for Brexit for different reasons. Nonetheless there are some patterns. By analysing the vote share by local authority, the Resolution Foundation found that areas with higher employment rates, larger student populations, more people with degrees and higher social cohesion were more likely to vote remain. Areas with more old people, more homeowners and those that have only recently seen an increase in immigration were more likely to vote leave.

But one possibility has proved controversial: was austerity partly responsible? Chris Dillow thinks it’s possible. Austerity contributed to stagnant incomes, which may have increased resentment towards “elites”, and to a decline in public services which the leave campaign blamed on immigration. Chris’ thesis received a bit of stick on Twitter from Giles Wilkes and Rupert Harrison.

In one sense, they have a point. The Resolution Foundation’s analysis looked at how average incomes in different areas were related to the share of votes for leave. While the level of income was important, recent changes were not, suggesting that the income effect isn’t related to austerity. But in another way Chris might be right. Stagnating incomes may be an indirect effect of austerity, but a rather more direct effect (which is not included in the Resolution Foundation’s analysis) is the deterioration in public services.

Austerity has led to cuts in many public services, but local councils – who take out the bins, run the libraries and provide social care – have been hit particularly hard. Local government spending power[i] in England fell by nearly 15% in real terms between 2011/12 and 2015/16, but the impact wasn’t felt equally in all parts of the country. Areas that collect a lot of council tax relative to their total spending got off lightly – Surrey’s spending power fell by less than 5% in real terms – while those that rely heavily on central government grants have been hammered – Liverpool City Council’s spending power fell by nearly 23%.

Big drops in spending power mean closing libraries, fewer bin collections and cuts to social care. It seems plausible that in areas where public services have deteriorated further, the argument that immigrants are overwhelming these services – as championed by the Faragist wing of the leave campaign – may have more traction.

The chart below shows how changes in spending power in local authorities in England[ii] between 2011/12 and 2015/16 are related to the share of votes cast for leave.

Change in local government spending power (horizontal axis) versus share of votes case for leave, upper-tier local authorities in England

Change in local government spending power (horizontal axis) versus share of votes case for leave, upper-tier local authorities in England

You might look at this and think there is no clear correlation – but the distribution is far from random. It looks to me like there are two things going on: a negative correlation for most areas, plus a cluster at the bottom left that seems to behave differently. There are no prizes for guessing where most of these outliers are located: they are London boroughs.

The next chart shows the same data with inner (blue) and outer (red) London boroughs highlighted. London voted differently to the rest of the country. Inner London (and some “outer London” boroughs such as Newham) saw big cuts in local government spending, but voted overwhelmingly for remain.

Change in local government spending power (horizontal axis) versus share of votes case for leave, upper-tier local authorities in England
Blue dots are inner London boroughs, red dots outer London

Change in local government spending power (horizontal axis) versus share of votes case for leave, upper-tier local authorities in England

Not all London boroughs followed this pattern. Havering had a relatively small drop in local government spending, but voted heavily for leave. This shows the limitations of using “London boroughs” as a sociological grouping. Havering is the most easterly London borough and surrounded on three sides by Essex. It is just a half hour’s drive from Newham, but a very different place.

Just as not all of London followed a “London-like” voting pattern, not all other areas followed an “unLondon” voting pattern. If we exclude London from the chart, there are still a few stray dots hanging around in that bottom left area – areas that, like many parts of London, voted remain despite large council cuts. Again, there are no prizes for guessing where these places are: successful cities like Liverpool, Manchester, Bristol and Brighton.

Change in local government spending power (horizontal axis) versus share of votes case for leave, upper-tier local authorities in England excluding London

Change in local government spending power (horizontal axis) versus share of votes case for leave, upper-tier local authorities in England excluding London

So it seems that we can divide England up into two groups: “London-like” areas, which include most London boroughs and some other successful cities; and “unLondon”, which is everyone else. Many London-like areas have seen big cuts to local services and still voted remain. But when we look only at unLondon[iii], we see a different pattern: areas with bigger cuts to local services cast a greater proportion of votes for leave.

On the basis of this, it seems quite plausible[iv] that austerity was one of the drivers of the Brexit vote – but this effect was mediated by cuts to local services, rather than stagnating incomes.

Change in local government spending power (horizontal axis) versus share of votes case for leave, upper-tier local authorities in unLondon

Change in local government spending power (horizontal axis) versus share of votes case for leave, upper-tier local authorities in unLondon


[i] Calculating trends in council funding is tricky, because responsibilities of councils change year-to-year. When responsibilities are added, extra money might be attached to them but this doesn’t ease the pressure on other services. Luckily, the Department for Communities and Local Government publishesspending powerestimates which (for any two adjacent years) try to take account of these changes. By cumulating the year-on-year changes, and adjusting for inflation, we can get a reasonable estimate of the changes over time in funding for local services.

[ii] The data are for upper-tier authorities. For two-tier areas (the shire counties) the spending power of the districts within each county has been included to make the figures comparable with unitary authorities.

[iii] For the purposes of this analysis, only Liverpool, Manchester, Brighton and Bristol have been excluded from unLondon, since they are the most obvious outliers.

[iv] There are two important caveats here. First, to believe in this correlation, you have to believe that the London/unLondon split makes sense and isn’t just a convenient choice to generate a spurious correlation. For me, the story works, but you will make up your own mind. Second, this analysis only looks at one variable, so it’s possible that the pattern is actually driven by something else, such as difference in average incomes. The Resolution Foundation’s work deals with this problem by including a wide range of variables – but nothing on cuts to local services. I’d like to see them add this to their analysis.

Past the first post

It’s just over a month until the UK goes to the polls in what looks like being one of the best or worst elections in living memory.

The best, perhaps, because we might be seeing the end of an era of politics characterised by two parties fighting it out to occupy a centrist territory delineated by the views of focus groups. The worst, because the art of misleading and withholding information from voters has reached new heights, to the point where we are being asked to vote for or against £12bn of mystery spending cuts. (The two may or may not be related.)

So I suppose that starting a policy blog at a time like this inevitably means writing about things that are at least loosely related to the election. However, as far as possible*, these posts will stick to facts and figures and avoid the unsubstantiated speculation of the previous paragraph.

More generally, this blog will try to shed some light on public policy issues by looking at relevant data and evidence, using interesting charts and graphics wherever possible. So lots of pictures – like a sketchbook (geddit?).

Right, so that’s the first post done, even if it’s not hugely inspiring. I hope you enjoy the blog, or at the very least read it.

* As far as possible, but no further