John Pullinger sits down with Mark Smulian to discuss sharing UK expertise globally, fake news and how data abundance changes policymaking
Asked to simulate being in conversation while being photographed, John Pullinger shows off skills that should make him a Just A Minute panellist. Launching immediately into an extemporised monologue about the excellence of his office armchairs, Pullinger shows that if statisticians have a boring image this shouldn’t be applied to him.
Pullinger is chief executive of the UK Statistics Authority, of which the Office for National Statistics (ONS) is the executive arm, and heads the newly-created government analysis function. His mission is to educate civil servants to understand data as evidence, and make policy choices demonstrably based on this. He must also try to do likewise for politicians, and to use his powers to publicly call them out when they do not.
Pullinger’s plans will touch everyone in the civil service, as he aims to reach “the 1%, the 50% and the 100%”. This means specialist training for the 1% who are statistics professionals who handle huge data sets, help for the 50% with significant policy or commercial roles “who need to understand sophisticated evidence and be able to make sense of that”, while, for the 100%, “to get everybody to ask hard questions, like: ‘Is that really what it seems to be? Who is telling me this, and why?’, and just to be a bit more sceptical about things”.
Indeed, he sees the new analysis function both working with statistics professionals and seeking to skill up policy colleagues and increasingly operational colleagues – those for example at the front line of collecting tax or managing borders. “They can have figures to think about but not the skills to make sense of them, so we can help them.”
That will be welcome for any civil servant who struggled with mathematics at school, but the ONS has plenty of people who can crunch numbers. More important than numerical skills is making people aware of why evidence must underpin decisions, and showing them where to find and how to interpret that evidence.
“Big data” is widely claimed as the route to better policy decisions, but this term can be casually deployed by those who understand little about it. Pullinger is determined that civil servants who need to act on data analysis should know the value of evidence and that politicians should respect facts.
He says: “What we have done in statistics over the past 10 years is progressively make it unacceptable for people to either not use evidence, or to misuse it.
“Interventions by the chair of the UK Statistics Authority and the head of the Authority’s Office for Statistics Regulation have made people think twice about using numbers in ways that they shouldn’t.”
Those caught out have included Dominic Raab, when he was housing minister, over a statement that ONS data showed immigration had increased house prices. Indeed even prime minister Theresa May has been reprimanded – over a comparison of Accident and Emergency waiting times in England and Wales and over spending on local police services. (However Pullinger says that both politicians were “very gracious about it”.)
More recently, last month the UKSA pulled up the Department for Education over a graph on schools spending and a comparison with education spending with OECD countries – both of which it said had been inaccurate.
Pullinger makes a wider point about current controversies over “fake news”: “It possibly means a greater role for [statisticians], as I think there is a backlash. A lot of journalists are realising that unless someone stands up for proper news, the whole value of news is diminished.
“I think we are part of a community that says proper facts have a value in our society, and we should not let this other agenda corrode that. We have a statutory duty to intervene – typically when a senior public figure says something – and we want to create a society that cares more about choices based on evidence rather than spin or vested interests.”
Standing up to the prime minister requires one to be very sure of one’s ground, and Pullinger has been immersed in data all his career.
He joined the civil service 38 years ago with a degree in geography and statistics from the University of Exeter and worked in several Whitehall departments with his current post being his third ONS stint, having taken the role in 2014. Immediately before this he was librarian at the House of Commons, which he recalls as “a good reminder that everything we do in statistics is for the customer, who can be very demanding, and they are using your data in a contested environment often on decisions that are very important for all of us”.
“What we have done in statistics over the past 10 years is progressively make it unacceptable for people to either not use evidence, or to misuse it”
There are around 4,000 ONS staff, some of whom are interviewers working from home to question members of the public about issues where the ONS is conducting research.
One of Pullinger’s priorities is to encourage diversity of thought among them, which has led to an unusual arrangement under which any civil servant can join the ONS board. He explains: “There is always a risk of ‘group think’ because you become acclimatised to that organisation, it is inevitable and you have to work hard to keep different ideas in the loop.”
Pullinger tried inviting people from other departments onto the ONS board but found only limited success. “I now advertise across the whole civil service saying ‘do you want to serve a term on the executive group of ONS?’” he says. “We’ve just got two started, one a social survey interviewer and one an analyst at Sport England, and they bring a different perspective to our discussions and that is exactly what I what.
“We are at risk if we do not get that broadest idea of diversity into our decision making. No level of seniority is required, we had recently a young fast steam project manager who was fearless in intervening. I think they have all got something out of it for career development, but giving my leadership team that extra layer of credibility is what I wanted them to do.”
The main ONS operation long ago moved to Newport, south Wales, but the London office has been beefed-up following a 2016 review by Sir Charles Bean.
Pullinger says: “I’ve been very clear our head office is Newport and that is the centre of economic statistics in the UK, but we need a complementary function making connections with the organisations that sit in London – the Bank of England, Treasury, think tanks, politics generally. We’ve got economics and social analysis groups and communications and most of regulation, so we are heading to 100 staff in London.”
Pullinger has also headed the Government Analysis Service since its launch last January. It has drawn together professionals as diverse as actuaries and engineers.
He explains: “The need for sound analysis has been thrown into pretty sharp focus with Brexit, so we’re trying to get statisticians and economists to come together to answer these tricky questions. I expect we will need people expert in different disciplines. If you’re doing complicated statistics you need a statistics degree, but you can only make it count in a civil service context if you are blending it with other colleagues and other knowledge. So I don’t see analysis being a new profession; it is a function.”
The analysis community is, he says, well spread both geographically and across departments. Successful recruitment in Manchester and Liverpool, in particular by HMRC, has shown “they can build a strong analysis community in north west England with colleagues who are already there, like the Health and Safety Executive,” Pullinger says. “You get a community of people who build good careers in analysis in that part of the country.”
He admits that the civil service may not be able to compete on salary for some of these professions, but says it wins on the interesting problems it can throw at people to solve.
“We can compete on interest with the best of the private sector,” Pullinger says. “To be working on problems like how we understand trade as it flows across the Northern Irish border, or how we make sense of the immigration arrangements needed after Brexit, these are interesting and difficult problems and people come to do those jobs because they are fascinated by them.”
One such vexed question is the future of the Retail Prices Index (RPI) and how far this remains useful. “It’s an active point of discussion, but there are three different things people want it to do,” Pullinger says. “One is a reliable measure of inflation, the second is an index that tells you how prices affect individual households, and the third is that RPI represents an accumulated set of expectations and if you change it that gives windfall gains and losses – in the gilts market many tens of billions of pounds are denominated in RPI.
“If we change it we need carefully choreographed discussions between ourselves, the Bank of England and Treasury about consequences.”
“We can compete on interest with the best of the private sector. How we understand trade as it flows across the Irish border, or immigration post Brexit – these are difficult problems, and people come to do those jobs because they are fascinated by them”
Pullinger has written about moving to an era of “data abundance”. For those who might think there is already more than enough, he explains this is about using data in creative ways. “Twenty years ago, if you wanted to find something out you had to design a survey, send questionnaires out, wait for results then analyse and publish them,” he explains. “Most of our systems are based on that idea, yet now there are so many sources of real time data – whether government or satellite images or transactional data they are all sources we can use.”
Examples include adding VAT into calculations of national accounts and Home Office exit checks into immigration data.
“Data abundance turns things on their head, starting from the question then thinking what are the data sources that can answer that question, rather than ‘here’s a survey, what can it tell you about the world?’” Pullinger says. “It’s a fundamental change in our way of thinking. For example, in the industrial strategy government is not interested in average productivity but in how we could design a regime that encourages investment in those likely to succeed.”
One counterintuitive example of this way of thinking came when ONS was asked to determine the value of investing in urban greenery as a means of carbon capture. “We created a machine learning algorithm that uses Google Streetview to identify greenery and then work out the carbon capture potential and value of investing in green spaces in our cities,” Pullinger explains.
“I had not immediately thought of Google Streetview as a data source, but it is good for that kind of thing. Another case is we’re working with maritime organisations to look at trackers in shipping containers and so can get much better analysis of trading patterns.”
Pullinger is also a figure in international statistics and thinks the UK has an important role to play sharing its expertise.
He has been chair of the UN Statistical Commission, which is devising measures to show whether or not the 2030 sustainable development goals are being met.
ONS is working on a programme sponsored by the Department for International Development with the governments of Ghana, Kenya and Nigeria on using data from satellites and mobile phones for work on crop yields. “We could learn from some of the passion those countries show to drive the better policy and service delivery we will need in the post-Brexit role,” he says.
Statistics may look intimidating, but Pullinger argues that in the modern world they are both inescapable and can be the friend of civil servant trained to use, rather than fear, them.