The last chair of the UK Statistics Authority was a fierce defender of the impartial and objective use of statistics. His replacement, Andrew Dilnot, tells Joshua Chambers how he intends to operate in this sensitive and important role.
Is Andrew Dilnot going to be a “powerful and aggressive regulator?” That’s the key question this interview should answer. After all, that’s the role that Parliament – in the shape of Public Administration Committee (PASC) chair Bernard Jenkin – has said it wants him to play as the new chair of the UK Statistics Authority (UKSA).
Jenkin made the comment last year, adding that “Parliament’s enforcer of statistics” must be “a back channel for people to blow the whistle when their arms are being twisted to do something that you know isn’t right”.
The government had first put forward a different candidate for the job, Dame Janet Finch, but PASC made clear in its pre-appointment hearing report that it didn’t think she’d be tough enough. Finch dropped out of the running, and the appointment process was re-jigged in order to encourage more appropriate candidates to apply. Dilnot subsequently emerged as Britain’s top stats bod.
What to expect
Certainly, he’s powerful. UKSA oversees the Office for National Statistics, and is the independent watchdog for monitoring the quality and use of official statistics. Dilnot can publicly embarrass both government and opposition whenever they step out of line.
But is he going to be aggressive? Sitting in his salubrious study, the softly-spoken Oxford don seems thoughtful, cautious and meticulously precise. Will he pursue the role in the manner Jenkin envisages?
“It’s very important that all who work in this field should be aware that the authority cares deeply and believes passionately in the value [of statistics]; and if there’s flagrant misuse, we’ll say so loudly, and I hope with the kind of authority that over its first few years it has built up,” he says.
According to Jenkin, Dilnot’s predecessor Sir Michael Scholar “set a standard in that role of extraordinary rigour and devotion to the cause and complete courage and integrity”. Scholar criticised politicians publicly and repeatedly for “abusing” official statistics – especially knife-crime statistics in the run-up to the general election. Will Dilnot follow the same model?
“Of course, I hope there’ll never be an opportunity to do so,” he says, laughing. “I know that there’ll never be such an occasion but, yes, I feel very strongly about the importance of integrity in the production, the use and the dissemination of statistics. These are rich and beautiful things and we mustn’t have them abused. We just mustn’t.”
As UKSA chair, Scholar repeatedly got on the wrong side of powerful figures, even being dubbed a “Labour stooge” by London mayor Boris Johnson. Does Dilnot think he’s taking on a public role, or will he try to operate behind the scenes?
“I’m absolutely committed to doing whatever’s necessary to maintain the integrity of the whole statistical operation,” he replies. “I think it’s very important to continue as Sir Michael did, with such extraordinary good judgements throughout his time. There was never a sense that he was looking to take a position for himself, and that’s something I hope we’ll be able to continue; but there was a clear sense that if something needed to be said, Sir Michael would say it without any concern about the consequences for him, and I absolutely hope to continue that.”
Seduced by statistics
Dilnot is a keen statistician, and spent 21 years at the Institute for Fiscal Studies before becoming a journalist and academic. Ask him about his background, and he soon gets distracted and starts to burble enthusiastically about the importance of numbers. “I’m not only unashamed: I’m proud of the fact that I just love numbers. I really, really, really love data,” he says, before sighing. “I think in this country we’re lucky to have an astonishingly rich array of data; there’s data on almost anything you could think there ought to be data on. We can find out a lot about most of the central parts of the way that our society and our economy, our environment, our geography works and then use them for making public policy.”
He sees two sides to his new role at UKSA: one as a tough-talking scrutineer and public figure; and a second as a champion of the effective use of statistics across government. The latter role is “something that I’m keen the authority does more and more with. I think there’s a way of developing its role, working alongside Parliament, often through PASC, but also working with producers and consumers of statistics.”
In particular, he wants to win civil servants over to the value of data. “The work of the civil servant always requires an understanding of the world, and you’ve got to start with understanding the bit of the United Kingdom or the wider world that you and your ministry is responsible for – and so data should be the first thing asked for in any policy area.” From the distribution of schools in areas with growing populations to the implications for social care of changes in demographics, and from traffic patterns to average temperatures, Dilnot thinks that the statistics should play a greater role in civil servants’ policy and delivery work.
Better policymaking
Too often, civil servants fail to use statistical evidence as the foundation of the policymaking process, Dilnot thinks. “This is something that it isn’t easy for politicians or senior civil servants to talk about. But I think it is the case that something that the whole statistical community could do better is make sure that we’re at the beginning of the policymaking process, not the end of it.”
There’s a tendency for civil servants and ministers to set out their beliefs and expectations, then later look for the evidence to back up their claims, Dilnot says. Statistics are “a little bit too close, too often in this country, to being thought of as something that comes in after you’ve decided what you’re going to do, instead of being the crucial tool for deciding how you should frame it and what you should do.”
Using statistics appropriately is time-consuming and difficult, he adds – but it’s worth making the effort. “It’s not just: ‘What’s the number?’, but: ‘What does the number mean?’ It’s hard, and sometimes it will challenge preconceived notions but, in the end, if you don’t do that, you’re going to get it wrong.”
Culture change
This problem isn’t endemic to government; Dilnot decries a systemic ignorance of statistics across the whole of British society. “We live in a country where it can sometimes be sort of alright to say: ‘I’m not very good at maths, I don’t do numbers,’ and that’s not embarrassing.” Numeracy isn’t considered as important as literacy, he adds: “None of us could ever say: ‘I don’t read very well’.”
Further, preconceived notions are often allowed to take root without being challenged, Dilnot thinks. “It’s terribly easy to believe that you know what the world looks like, and we’re all rather tempted by that. It turns out that, in fact, we don’t. So my mental image of myself is of somebody shorter than the average. I’m not shorter than the average – I’m massively shorter than the average!” This poses a rather obvious, if awkward, question; it turns out that he’s five foot seven. “We all tend to think about ourselves as being quite near to the mean, and that feeds out into the way we think about the world,” he says.
Over the last 20 years, he’s quizzed permanent secretaries, the Number 10 Policy Unit, and senior journalists about the key statistics in their fields, such as the average income of people in the highest-earning 10 per cent, or GDP growth over the last 60 years – “and the answers to these questions are systematically gotten wrong; and wrong by a very, very large amount.”
So are there any particular policy areas where statistics are most commonly misunderstood or not included in the policymaking process early enough? He laughs: “You wouldn’t expect me to give you an answer to that.” But I do, because it’s now his job to talk about these problems – and with 30 years of experience examining public policymaking, he’ll surely have noticed variations in performance. “I think the answer is: all of them,” he responds. “There isn’t an area where it seems possible to read the newspaper or listen to what’s said in policy debates where you think: ‘Gosh, yes, people here are really on top of what’s going on’… Every single department can do better here.”
The trust deficit
People have noticed that government doesn’t always use statistics properly, Dilnot argues. “Public trust in [use of statistics] isn’t as high as it should be, and so there’s an expectation, a feeling that somehow there’s something funny going on,” he says – although he’s keen to stress that he’s no reason to believe that official statistics are falsified: “I’ve never seen any evidence of any actual manipulation of the statistics.”
Nevertheless, one major concern is the pre-release of statistics to ministers and special advisers before they are seen by the general public, leading to the fear that “statistics may be being politically manipulated”, he says. He adds that as long as pre-release survives, “there will continue to be accusations that other announcements are made because somebody has seen sight of something that’s going to be embarrassing”; in other words, that spinners will try to minimise media coverage of awkward statistics by pushing other stories to the press. Hence, he argues, “I just think we should stop it”.
Scholar was keen to reduce the amount of pre-release, and PASC was disappointed that Finch didn’t think it an issue; so Dilnot is making a strong stand by suggesting it be stopped altogether. While he acknowledges that the practice may be valuable in a couple of areas where data releases can affect markets, even here he wants to reduce access.
The use of crime statistics has often caused controversy. Indeed, a report by the UKSA in 2010 noted that “public criticism of the statistics and mistrust of the way they are used and quoted, as far as the authority can tell, exceeds the level of criticism and mistrust in most other countries.” Why does Dilnot think this is? “We have two different ways of reporting crime statistics [a UK-wide survey of people’s experiences; and crimes reported to the police] and we can get data that’s seen as, if not inconsistent, at least different. Managing the understanding and presentation of that is really quite tricky and requires a community that feels confident that games are not being played,” he says.
Dilnot therefore wants to see the use of statistics become more “sophisticated”, and hopes “there’ll be a growing recognition that any one statistic will typically be an attempt at a simple summary of a complex reality” – though he admits that “that’s quite a big ask.”
GDP statistics are one example of a complex debate becoming over-simplified, he thinks. “The GDP measurement is done really rather well, but what we don’t always get right is people’s understanding of those estimates: they’re trying to capture an amazingly complicated and rapidly-changing world, and to give a sense of what it looks like very, very shortly after the time period that we’re measuring.” It shouldn’t be surprising when the statistics are later revised, he says, but there “can be a tendency to assume that something has gone wrong if we change the estimate.”
There’s also a broader cultural problem with growth statistics, he says. “Our response to statistics in the UK really focuses very, very heavily on the number that came out at 9.30 this morning – the very latest. Those are important – but in some ways, even more important is the long story, which contains much more information.” Beyond GDP there are other good metrics for looking at the economy – for example, prices and earnings growth, unemployment levels or retail sales. “So given how much data there is available, it’s not often going to be the case that a single number will tell you everything you want to know if you’re trying to work out the complicated answer to a complicated question, like: ‘How should we manage the economy?’”
Fragmented data in future?
Some fields have better statistics than others, and social care is particularly weak on this front, Dilnot thinks. He knows the area particularly well, having just chaired an official government commission charged with finding a solution to the problems that come with an ageing population, and complains that “there’s not as much or as rich data as I would have liked, partly because it’s been a local rather than a central government responsibility so there’s a diversity of structures, and partly because a lot of social care is provided informally by individuals looking after their family or friends, or by the private sector, so again we don’t have the data.”
Given that the government’s localism agenda is set to expand local responsibilities, is fragmentation of data likely to be a growing problem? “I don’t think it will be a problem, but it’s something that government will want to keep an eye on,” he says. “The reason it was an issue in social care was that the data hadn’t ever been built up, and there was a very large private sector and informal involvement. It wasn’t just a local authority responsibility: there were at least three players, and we hadn’t found ways of bringing the data from those three together. In itself, increasing localism doesn’t present challenges that are in any way insuperable, but it will mean keeping an eye on making sure that we continue to collect good data.”
By the ‘we’, he means the Office for National Statistics, which he stresses costs only £205m a year in non-census years: “Compared to annual public spending of £800bn, much of which is being allocated using data that’s generated by the statistical service, it looks like pretty good value and is very, very important.”
No country for old men
For some of the later parts of this interview, it feels like Dilnot is tip-toeing away from anything that could prove controversial. He’s made his stand on pre-release; but despite his decades of experience of the UK public sector, he doesn’t want to criticise specific departments or areas of concern. Take the use of targets in public services. At the IFS, he spent some of his time skewering them; but asked for advice on how the public sector can set better targets, he refuses to comment.
I turn to social care, hoping to get a glimpse of him speaking out. Some commentators have called his solution, which proposes a lifetime cap of £35,000 on social care costs, “regressive”. What does he think?
“I know there have been some suggestions,” he says; and then he methodically pulls the criticisms apart. First, what’s meant by regressive? The term only makes sense when considering the overall impact of all taxation and public spending on individuals, he says, because most taxes are not hypothecated.
Then he picks apart the argument that his proposals are regressive because they would help the better-off as well as the poor. The NHS isn’t dubbed regressive, he says, “despite the fact that significant amounts of money are spent on the wealthy.”
Finally, he argues that his solution would be “highly progressive”, because wealthy people pay much larger sums in tax than the poor, and thus – assuming that care is funded out of taxation – they will on average end up subsidising the care services of the majority.
Here, he shows his mettle. Meticulously, he pieces together the evidence to defend his argument. So will he lead a powerful and aggressive regulator? Dilnot is a quiet man in a public role, but he can speak with a steely authority when needs be – and he certainly says that he is prepared to do so.