Public Administration and Constitutional Affairs Committee
Oral evidence: Governance of Statistics, HC 1820
Tuesday 5 February 2019
Ordered by the House of Commons to be published on 5 February 2019.
Members present: Sir Bernard Jenkin (Chair); Dr Rupa Huq; Mr David Jones; David Morris.
Questions 1 - 138
Witnesses
I: Hetan Shah, Executive Director, Royal Statistical Society, and Professor Guy Nason, Professor of Statistics, University of Bristol.
II: Richard Alldritt, formerly of the Statistics Commission and UK Statistics Authority.
Written evidence from witnesses:
Examination of witnesses
Hetan Shah and Professor Guy Nason.
Q1 Chair: I welcome our two witnesses to this first evidence session in our inquiry reviewing the UK Statistics Authority. Could I ask each of you to identify yourselves for the record, please?
Professor Nason: I am Professor Guy Nason. I am a Professor of Statistics at Bristol University.
Hetan Shah: I am Hetan Shah, Executive Director of the Royal Statistical Society.
Q2 Chair: Thank you very much indeed for joining us. We have two panels of witnesses today and quite a number of questions, so if we can keep the answers as crisp as possible that would help and we will try to ask short questions.
Since UKSA was established 10 years ago, what do you think it has achieved in that period?
Hetan Shah: You asked that we keep the answers crisp and I will do so but I think there is quite a lot of material to get through there, so if you will forgive me I will run through a series of points that set the scene.
The first thing is that it has been a powerful voice against the misuse of statistics. We did not have that before. I think it has established itself as a moral authority with an even-handedness between political parties. That is a very good thing.
The second thing is it has improved the quality of statistics. The Office for Statistics Regulation has been put in place, a new code of practice has been developed with a focus on trustworthiness, quality and value, which is new. This is going beyond the statistical system so we are seeing now the voluntary adoption of that code of practice, for example, with the GLA and others, and we have seen economic statistics really improve in recent years, partly spurred by the Bean review under the leadership of one of the Deputy National Statisticians.
Thirdly, there is the ending of prerelease access to Office for National Statistics stats. ONS stats used to be seen by certain Government Members, Ministers and special advisers in advance, before the rest of us got to see them. The National Statistician has stopped that practice, which I think is excellent. We are all treated the same.
Fourthly, our strategy that is in place, called “Better Statistics, Better Decisions”, links statistics to decisions, so statistics are not just there floating in the ether. That strategy was a very good move by the current National Statistician.
The fifth point, which I think is easy to overlook, is there has been real investment in technology. In particular, in the run-up to the new census, we are seeing, under the leadership of Heather Savory, one of the Deputy National Statisticians, a lot of investment in technology, which is going to stand us in good stead in the future.
Sixthly, there are some new initiatives such as the National Statistician’s Data Ethics Committee and the Data Science Campus and these are putting in place good initiatives for the future. As we talk about the challenges, I would like to come back to some of those.
The seventh point is the exciting use of administrative data and multiple data sources. We all know we are living in this big data world where data is coming at us from all sorts of places. I think the ONS and GSS, particularly in economic statistics, migration statistics, health, crime, and housing are getting much better at pulling these together from different places.
Eighth, the Digital Economy Act is a new piece of legislation that the National Statistician fed into and will give the ONS much greater powers of access to data across Government. That means a future census will be able to be based upon not just the old survey data that we used but on medical data and all sorts of other things that the records can access.
Finally, and this is my wider point, I think we have a good leadership team in place at the moment. The National Statistician has been excellent and he has put in place some strong deputies, and the Director General for the Office for Statistics Regulation. They are all strong characters and I hope, from the RSS perspective, that we can build upon what they are doing and take things forward with them.
Q3 Chair: Thank you very much. I do not mind a comprehensive reply to start with. Professor Nason.
Professor Nason: Mine is going to be quite a bit shorter. One of the things I want to focus on is independence and looking back on why this whole structure was set up. In the early 2000s, there was a perception of lack of independence between official statistics and Government. There is a great statistic—and as a professor of statistics I am going to give you one—that 17% of the public thought that there was no Government interference in official statistics in 2005. In the equivalent figures a few years ago that has increased to between 46% and 76%, depending on which statistic you are looking at. That is evidence, if you like, that the population feels that there is much less interference by Government in official statistics. I think the case for independence is much stronger now with this set-up than it was to begin with in 2005.
The other thing I would add to what Hetan said is that the system is still evolving. Ten years in some ways is a long time but in other ways it is a short time and parts of it, for example the Office for Statistics Regulation, has only been going for a couple of years in its current form. That is all I want to say about that.
Q4 Chair: What do we mean by the word “independence”? Independence is not completely autonomous. Independence is about being accountable to somebody else other than the users or the paymasters. What do you understand by the term “independence”?
Professor Nason: The statistic I quoted was asking the public, “How much Government inference do you think there is with either official statistics or, in the latest survey, particular kinds of official statistics?” You are right, that doesn’t comment on the wider structure about how that UKSA relates to Government, for example, in the whole round.
Hetan Shah: Could I add a bit on independence, if we are on this theme? The first point to make is that independence is clearly a spectrum. There is no such thing as absolute independence; it is how independent you are. We are not China. In China nobody can trust any of the economic statistics that are coming out. I have written down a quote because I thought it was so funny. Their director of statistics said, “Some local statistics are falsified and fraud and deception happens from time to time.” We are not anywhere in that situation.
There is some really strong evidence of certain kinds of independence. One is the letters to Ministers and to the Prime Minister, from time to time, saying, “Don’t do this”. Others look at our system and think, “Wow, how do you have a system that allows you to do that?” I think that is evidence of some kind of independence. I mentioned earlier about the ending of the prerelease access to statistics. Ministers, the Treasury and the Bank would like to see those economic statistics and they cannot now.
The other example, quite recently, that was passed by because of other news—and this sounds very arcane—is the treatment of student loans. In December the ONS ruled on how student loans ought to be treated on the balance sheet and because of the way they plan to treat them now, that is going to lead to a £12 billion hole in the public finances. I have seen no evidence that the Chancellor has said, “We should not do this; the ONS is wrong”. To me, there are lots of examples where we do have quite a strongly independent system.
Q5 Chair: We will come back to that. What do you think UKSA could have done better? I appreciate all the good things that the authority has done—has achieved—but where do you think it could be doing better?
Hetan Shah: I would highlight five areas. If I give you the top line and then perhaps we can explore some of them as we go on. Some of this is about the future challenges. If you think about it, there is more demand for granular data, regular data, in a way that perhaps there wasn’t 10 years ago. The other thing I would highlight is that we are now in a post-Cambridge Analytica world where the public itself is thinking about use of data in a way that it had not done before.
I would say five areas. One is leadership: how do we build leadership within the statistical system at every level? I think that we have good leaders at the top, but at the next level down you don’t necessarily have those people who want to connect to policy and have a wider vision.
The second is innovation for data science as we have all these new data sources and new methodologies. You can use scanner data to try to measure inflation, you can use satellite imagery and so on. This puts pressure on statisticians to think differently. That innovation is required.
The third is engaging more with users. Statisticians like to sit in their box and produce their statistics and unless they are challenged they will keep doing that. The value of statistics comes from their use and statisticians always need to go out and talk to their users. Not only do individual statisticians need to do that, I think the board of the Statistics Authority needs to do that too.
Four is communication. How do those statistics land with the public and the policymakers? There are various things we can do to make sure that communication gets followed through.
Finally, governance and ethics. This comes back to my Cambridge Analytica point that we are in a situation that is quite different from 10 years ago where the public really now want to know what is being done with their data. The Statistics Authority could work with other bodies who believe in the power of data for public good and make that argument.
Q6 Chair: What are you disappointed by?
Hetan Shah: Coming back to user engagement, the Bean review, which was a review of economic statistics several years ago, was an example where there had clearly been a lack of foresight by the board. Why did it take the Chancellor to have somebody else come in and say, “We need our statistics around the economy reviewed”? That was a function for the board. That is why I focused on user engagement as an example.
Q7 Chair: It is a good question. Why do you think the Bean review was necessary when what the Bean review did is what the UK Statistics Authority should be doing?
Hetan Shah: In that example they were not far enough ahead of the curve.
Q8 Chair: Why?
Hetan Shah: I would say it is a lack of foresight and a lack of engagement with their users.
Q9 Chair: How did they get away with it?
Hetan Shah: I am not sure that they did.
Q10 Chair: Who should have been holding them to account?
Hetan Shah: Possibly you.
Q11 Chair: Yes, I accept that. I accept that and I feel that.
Hetan Shah: Having said that, I would just go back to Guy’s point that this is a relatively young organisation. I am not saying that we should excuse everything but I run the RSS; we are 185 years this year and we are still pretty imperfect. Regulators have a difficult job where they are like goalkeepers, you only remember them for the goals that they let in. That is why I think it is important that we try to give a balanced perspective. The Bean review is certainly an example of where they were not on top of what they needed to be on top of.
There are clear examples now of where they are on top of what they need to be on top of. With Brexit, there is increasing interest in trade statistics and they have been quite quick at looking at trade statistics. The number of series has moved from 1,000 to 100,000 being published. The Bean review was several years ago and I think the board, to some extent, has learnt its lesson, but that is why were are pressing for greater user engagement.
Q12 Chair: Like you, we regard this as a relatively new organisation and we have been nudging it, but the Bean review and one or two other things are why we are having this inquiry, because maybe our role should be more proactive. Professor Nason.
Professor Nason: On this particular aspect, a lot of what the ONS has to do is lay down international standards, a little bit in legislation, and it gets a bit like a machine. The machine has to work and it has to produce the statistics. There was also an aspect in the Bean review where there was a statement where the thinking function or the research function of the ONS over many years gets whittled down through progressive budget cuts. The research function is always the first thing to go.
Sometimes you need something like the Bean review to come from the outside and say, “Look, we have to have a big shift”. I think that is what happened and, as Hetan says, the board reacted to that. We have things like the Data Science Campus with a lot of young people bringing in revolutionary data science methods to see if they can do things in a very different way. Frankly, if they don’t do it, I suspect a lot of outside corporations will be doing it for them, because they will have access to some of these datasets.
Q13 Chair: What could UKSA have done better?
Professor Nason: It is difficult to say from the outside but possibly more foresight, more horizon scanning, that kind of thing. That is the obvious answer.
Q14 Chair: I am just looking at the Act. It does say that it should monitor the production of publications and it will report any concerns, publish its findings and give guidance and advice to persons responsible for official statistics. Has it done enough of that?
Hetan Shah: The creation of the Office for Statistics Regulation came out of the Bean review, which is an example of how I think things have improved. One of the things that I think the OSR has done well is focus on the systems of statistics. They do systemic reviews now, for example of housing statistics or income and earning statistics. In the past, the focus was perhaps too much on, “Is this individual statistic being produced well?” rather than, “Does this family of statistics tell the people, you and me, what is happening in hospitals today?”
Q15 Chair: That leads very neatly to my next question. One of the reasons we delayed having this inquiry was that we were waiting for the House of Lords Committee to finish their inquiry into RPI and their report is not very nice reading for UKSA. They describe the failure of UKSA to deal with the RPI issue, saying that it has led to an 0.8 percentage point difference between RPI and other measures and that the authority’s error has created winners and losers. Holders of RPI-linked Government bonds are about £1 billion a year better off at the taxpayer’s expense but commuters, students, benefit recipients, tax thresholds, public sector and state pensions are all losers under this system. This would seem to fall very squarely into what the UKSA should be dealing with and the reports says: “This is clearly unsatisfactory, but why is the UK Statistics Authority unwilling to fix a statistic that it has admitted openly is flawed?” Why do you think that is the case?
Professor Nason: That is a good question and I think you should ask them.
Q16 Chair: They should ask them?
Professor Nason: You should ask them. I presume that you are going to see them.
Chair: Yes, we will.
Professor Nason: The whole thing feels a bit surreal.
Q17 Chair: They were asked by the Lords Committee why they had not taken any action and the chair of the UK Statistics Authority told the Committee, “There was no point in requesting to correct the clothing change because the Chancellor would say no”. What does independence mean if that is the answer from UKSA?
Hetan Shah: RPI is an area where the RSS has been critical of the UK Statistics Authority’s perspective. It is worth saying that the National Statistician has laid out his view of statistical indices, which we agreed with the basic framework and, forgive me because this is getting into quite dull terrain but—
Chair: It is not dull; it costs a great deal of money.
Hetan Shah: It does, but it is still dull I’m afraid. I say this as somebody who works at the Royal Statistical Society. The CPIH is what John Pullinger has said is the macroeconomic statistic, so this is a good inflation measure for the Bank of England to set interest rates and so on. He has put forward something called the household cost indices for household inflation—so how much inflation do you or I face on a yearly basis—which is a bit different from what the Bank needs to set interest rates. Thirdly, the RPI is a kind of legacy measure because it is still widely used in contracts and so on.
The RSS and the Statistics Authority see eye to eye on that framework. What we don’t really understand is why the decision of the Statistics Authority was to allow the RPI to wither on the vine and say, “We think it will fall out of use and you can use one of the other two”. They did commission Paul Johnson to do a review of statistics, so this was an area where they were proactive in getting external advice. I respect Paul Johnson incredibly but we felt that his review did not quite get it right and in a sense that set the Statistics Authority off in the wrong direction. We would like to see the RPI fixed because it is widely used.
Q18 Chair: Who commissioned Paul Johnson?
Hetan Shah: The Statistics Authority.
Q19 Chair: Why is the UK Statistics Authority treating this as a legacy measure of inflation when the Government is still using it as a primary measure of inflation?
Professor Nason: That is independence, although it may be the wrong way.
Q20 Chair: It suggests that UKSA as a regulator is not very effective at regulating.
Professor Nason: I think there were many technical issues with RPI.
Q21 Chair: Either they should be calling the Government out to stop using the measure or they should fix the measure. Why have they not fixed the measure?
Hetan Shah: Again, it is quite tricky terrain but the production of RPI is written into statute, so this is an area where, however independent you are, you can’t ignore the laws of the land, you still have to keep producing it. Certainly the National Statistician has issued warning after warning saying, “This is not a good measure of inflation, please stop using it”. I think political reality has caught up, which is that there are bondholders out there in very long contracts who are still using it.
Q22 Chair: That is clearly where it could be treated as a legacy measure but the Government still issue index-linked bonds attached to RPI.
Hetan Shah: Yes, and my understanding is that the authority said, “You should stop doing this; it is not a good measure”.
Q23 Chair: The House of Lords Committee concluded that the UKSA “could be accused of failing in its statutory duties”. How much do you agree with that?
Professor Nason: As Hetan says, it is in the legislation that thou shalt produce RPI, so I would think you would want to do that to the best of your ability or get the law changed so you don’t have to do it.
Q24 Chair: It is in the power of UKSA to recommend changes to the methodology of RPI and they have not done so.
Hetan Shah: The RSS has been public on this many times and fixing the RPI is something we would like to see.
Q25 Chair: How much should we look at the failure of RPI as a symptom of wider problems about the relationship between UKSA and its regulatory audience?
Hetan Shah: I think you can see this as probably the big failure, among other things that are going relatively well. I set out various things that are—
Q26 Chair: It is an isolated instance?
Hetan Shah: I would say so. If you take my reading that they commissioned a review by Paul Johnson, a respected commentator in this area, and as I said, from our perspective, the review got it wrong and they followed essentially the outlines of what they heard from that review. It is a respectable process, but it led them the wrong way and then that was compounded by not listening carefully enough to users.
Chair: We will come back to other instances of statistical series that we may be concerned about.
Q27 Mr David Jones: UKSA is a non-ministerial Department, so who in practice does it work for?
Professor Nason: That is a good question. It has been very difficult to find that out. I have done a lot of searching. There are a lot of indications about who it reports to. Part of it reports to you. That was said in David Norgrove’s evidence to you a while back, in his preappointment hearing. I believe that the Cabinet Office has some residual ministerial responsibility, but the lines of management and responsibility are not clear to me. The other thing that is also not so clear, as well as this Government-UKSA relationship, is the relationship between UKSA and Government Departments across the piece. It is a kind of matrix management structure where you have statisticians in Government Departments who work in the Departments, are paid for and line-managed by the Department, but there is this connection to the National Statistician, who keeps an eye on the veracity of the statistics.
Q28 Mr David Jones: This lack of clarity gives rise to a number of problems. UKSA has the statutory objective of “promoting and safeguarding the production and publication of official statistics that serve the public good”. Who determines what the public good is?
Hetan Shah: It seems to me that is the responsibility of the board and then the board should be informed by the user community. There are journalists, policymakers and the public using these statistics in different ways. The role of the Statistics Authority is to always be out there talking to them and saying, “What is it that you need from us?” You can get quite philosophical, but once you go out and say, “What do you need from homelessness statistics? Are they good enough?” I think people can answer that question. When you start asking the question at a specific level of statistic or issue, people can make sense of that. Defining the public good in a very broad sense is obviously welcome.
Q29 Mr David Jones: What if someone disagrees with the board’s assessment of what the public good is? How does that individual challenge the board?
Hetan Shah: That is a fair question. I think this is where we would say the mechanisms for user engagement are not clear enough. There are areas where it has been done well. For example, the Office for Statistics Regulation has brought together health people in a really good way. We know health in this country has multiple providers and the statistics are complicated to make sense of and the OSR has brought together the producers, the users and so on. That is a good mechanism but it has not been done—
Q30 Mr David Jones: That is your assessment of what is good. It may equally be the case—and I do not know whether it is the case—that someone thinks that is not serving the public good. I am concerned that there does not appear to be any process of determining whether or not what they were doing is for the public good.
Professor Nason: One of the routes that I have dealt with is interacting directly with the Office for Statistics Regulation if you think there are things going on that are not in the public good. They are very open and they have this mechanism where they try to get people together, the users with the statisticians and the Departments. A lot of the time that is very effective, a very positive way of working. For more extreme examples, I guess it is writing to UKSA and to OSR with serious concerns about various things and then they will be taken seriously. At the end of the day, they will deal with that; it is their responsibility.
Q31 Mr David Jones: There seems to be no means of third-party challenge to what the board determines is the public good or whether it is conducting its functions correctly. Would you agree with that?
Professor Nason: That is the role of the non-executive directors on the board, who are an independent voice on the board.
Q32 Mr David Jones: It is generally thought that the non-ministerial structure is a good thing. Is it the case, however, that there may be some deficiencies as a consequence of UKSA not working for anybody?
Hetan Shah: I have always seen that the structure is there to be accountable to Parliament. In a sense, as you say, not having a strong link back to the Cabinet Office is good. The previous structure used to be accountable to the Treasury, which is certainly not what we want. Therefore, the question in some ways is back to this Committee. You are the organ of Parliament: what are the best mechanisms for you to hold the Statistics Authority to account and to support it?
This sort of inquiry is very helpful. As it comes up to a new National Statistician and a new strategy, now is a great time for you to be giving guidance on what you think the public good should be.
Q33 Mr David Jones: Accountability is to Parliament; that is a statutory issue. Do you think that Parliament—when I say Parliament, I mean this Committee—is sufficiently conducting its functions in holding UKSA to account?
Hetan Shah: I was impressed by the way that this Committee held a series of inquiries into different areas of statistics. Your work, for example, on migration statistics, on open data, on police statistics was transformative and has been very helpful. It is completely appropriate that you are holding this inquiry 10 years in. The question for you now is to say how you build on this. Should you have, for example, a rolling annual meeting scrutinising the annual report? I am not sure.
Q34 Mr David Jones: I would like to hear from you what you think this Committee could do to improve accountability so far as UKSA is concerned.
Hetan Shah: Some kind of annual meeting with the chair of the authority and the National Statistician, Director General of Regulation, scrutinising their annual reports. That seems to me to be a kind of cycle that makes sense.
Q35 Mr David Jones: Professor Nason, do you agree?
Professor Nason: Yes, I agree with that. You all know much more than me but there are different levels of process that could be used. The NAO reports in a slightly different way from UKSA, maybe in a slightly more independent, stronger way. That is an ideal. It is difficult to know.
Q36 Mr David Jones: How transparent would you say UKSA is in its activities and its plans for the future?
Professor Nason: There is a great deal of information that you can obtain on UKSA, the annual reports and business plans, but they tend to be quite high level. If you want to get at some of the detail, that is a bit more difficult and we have noticed difficulties in getting some of the minutes of lower level committees. Minutes are missing from 2016 or 2017 or something like that. These are lower level technical committees, not the top level stuff, so sometimes it is difficult to figure out what is going on.
Hetan Shah: I certainly think formal transparency could be improved. For example, the last set of board minutes that are published from the authority board are from last July. It feels to me that that is a fairly long gap.
The other area of formal transparency that has been a bit odd is planning for Brexit, which is clearly one of the big policy changes that are coming upon us. There has been virtually nothing on the website saying what its approach would be. From a user perspective, more formal transparency would be helpful.
Having said that, I think informally they are relatively open regulatory bodies. It is quite easy—certainly for us as the RSS—to pick up the phone to them and say, “What are you doing?” and they will talk to us. Whether everybody has that experience or whether that is because the RSS is more connected—
Mr David Jones: That transparency that you are able to obtain by picking up the telephone should be readily available to any member of the public who seeks it on the website. Thank you.
Q37 David Morris: How effective is the UKSA board’s leadership of the statistical system? Does UKSA publicly intervene promptly when people use official statistics in misleading ways?
Hetan Shah: Yes, I think for the most part. This is an area where I think it has fully developed its function. In particular, with the set-up of the Office for Statistics Regulation we have seen a stronger media presence. When the regulatory function is exercised, telling off a politician or others, taking charities and the media to task sometimes, this tends to get picked up in the press. I would say it is generally pretty good.
We have also seen it move to being more proactive. It used to work much more in a way that it would respond only if somebody wrote in and said, “There is this problem”. I am seeing more and more occasions where it intervenes proactively. A good recent example was with the Department for Education, which had made a series of misuses in different ways. What we saw was a series of fairly low level letters going out and then escalating up to quite a big letter, which got a lot of press and so on. I thought that was a good example, where a Minister was taken to task over a series of failures, and I thought it was a fairly proportionate way of doing it.
Q38 David Morris: How could the UKSA board become a bit more responsive and forward looking? In that respect, how does UKSA influence leadership and culture throughout the statistical system?
Professor Nason: Could I come back on that previous question, which is kind of leading into this, with issues that may be in some sense small but important? If a politician comes up with a statistic that is dodgy somehow, a letter goes out, which is usually done very quickly and is very well written. I think that part works extremely well and we are seeing more and more of that. I think that is fantastic.
There might be an issue with what I call the slow, bigger things that go on very slowly within Government for good reason, big systems that are constructed within Government and have a lot to do with using statistical information. When concerns are raised here, as I said before, you get a very open response and a good discussion and OSR has this mode of getting people together to talk about things and that is very positive. But at some point, if the external actor thinks, “What you are doing isn’t right”, we find that the whole thing tends to slow down a little bit.
It would be nice for us if some of the fast actions you see on these simple and smaller things could somehow be applied more to some of the bigger issues. They touch on things that overlap, say the NAO’s competence, cost effectiveness and use of statistics, so there is an interaction there as well.
Q39 David Morris: Looking forward, how would you see the board becoming more responsive, given that you want to interface more with these other bodies?
Professor Nason: One of the forward-looking things is that the OSR is quite a small organisation. There are 25 people who are meant to look after the whole of Government statistics. That is a huge enterprise. One argument might be to increase the resources in OSR. That would be one way the board could exercise its duties more effectively by having that stronger—
Q40 David Morris: How do you think you can influence that leadership and culture throughout the statistical system? You are saying there are only 25 people looking after the whole of Government statistics, which does sound very low. How do you think you can influence a better responsive organisation in these particular bodies? How would you see that?
Professor Nason: I would like to see more resources going towards OSR, whether it is UKSA that can do that or they need to request more money from Treasury. I do not know how that would work.
Hetan Shah: Could I say a bit more on this? I think changing culture is the fifty million dollar question. When I started I said there were certain themes that I want to pick up and if I may I will hold forth a bit more on that.
The first point is on leadership through the system of statistics. It is particularly true in the Government Statistical Service. The ONS is a central body of statisticians. It is much easier to support those statisticians. If you are a statistician in the Government Statistical Service, you are working for a Department.
One of the things we have noticed is that with the heads of professions, over time, the status of that post has become lower and lower. We would like to see those heads of professions have a much stronger post within their Department, much more seniority, and we think that would cascade through. If you have senior statisticians in Departments they will be taken more seriously. We certainly see a correlation between those Departments where the head of profession is more senior and the quality of statistics that come out of that Department.
Q41 David Morris: You would argue that by expanding the workforce you would create managers who would trickle down to other statisticians?
Hetan Shah: It is not necessarily about expanding the workforce but how seriously they are taken within the Department. The seniority of that person within the Department I think is really important and—
Q42 Chair: How appropriate would it be for UKSA to make that recommendation?
Hetan Shah: I am not sure. It may well be but it is certainly something I hope that this Committee will, because this is a recommendation to the Government. It may well be something that the Statistics Authority ought to be—
Q43 Chair: Are you suggesting that UKSA should not be making recommendations to Government?
Hetan Shah: No.
Q44 Chair: Therefore, I ask the question again: how appropriate would it be for UKSA to make this recommendation?
Hetan Shah: It is certainly something that it could recommend.
Professor Nason: It could recommend that.
Q45 Chair: I sense that you might be feeling that it would be straying over some hidden boundary that exists.
Hetan Shah: It is just that it is not something that I have seen it do before. It is not to say it is a hidden boundary, but the organisation has innovated in the way that it has intervened and this would be a different kind of intervention.
Q46 Chair: It is meant to promote and safeguard statistics and give guidance and advice to persons responsible for official statistics, which is Government Departments, so it would seem to be perfectly within its statutory duties to make such a recommendation.
Hetan Shah: Yes.
Q47 Chair: Why do you think that you are reticent to suggest that UKSA should have done this?
Hetan Shah: It is much easier, certainly for me, to see what it already does and to talk about the incremental improvements of those things. When it is a completely new category—
Q48 Chair: Do you feel there is some constraint that would discourage UKSA from being as bold as this?
Hetan Shah: No, I don’t think it is necessarily boldness. Genuinely, when you do something different or new it is a different kind of action.
Q49 Chair: So it would be different and new. Professor Nason, do you have anything to add?
Professor Nason: UKSA could, and possibly behind the scenes it does, recommend to Departments that, “You want the best, most senior person that you can get at that level”, but I guess there are unusual budgetary questions. The statisticians in Departments are employed by the Departments. They fit into the structure. I do not see any reason why UKSA couldn’t say, “We want to change this”.
Q50 Chair: Do you think UKSA would be seen to be rocking the boat if it made a public recommendation?
Professor Nason: There was something in the Bean review along these lines. There was a small paragraph somewhere, I think. It did not actually say this, but it alluded to the fact that the status of statisticians wasn’t what it was.
Hetan Shah: I am not sure it is so much about rocking the boat. The question I would be asking is: what is UKSA’s purchase on those Departments to have that recommendation taken up? It may well be buried in some reports. It may well have said something along those lines that I may not have picked up, but what is its purchase is the question for me. What is its authority, not just in terms of being able to make that recommendation but for it to be then taken up?
Q51 Chair: There would be something rather undermining about making recommendations that are just ignored, wouldn’t there? Do you think that could be part of the anxiety?
Hetan Shah: Anybody who is concerned with influencing has to think tactically as well as strategically about how to get those—
Q52 Chair: Is not UKSA a regulator, not just an influencer? Maybe it is just an influencer
Hetan Shah: Certain things are about influence. If it has no regulatory authority over the level at which those posts are set—
Q53 Chair: Regulators that mind what the bodies under jurisdiction will do are sometimes accused of regulatory capture. To what extent do you think it is fair to suggest that UKSA has been subject to regulatory capture?
Hetan Shah: That does not really resonate with me. For the reasons I gave earlier about the independence, I do think that it is not that it has been captured. If I were in UKSA’s shoes I would be thinking, “Yes, I could say that but what is my power to make this happen, because if I say it and it doesn’t happen then I am looking pretty silly?”
Q54 Chair: How much does UKSA lack necessary powers to be a more effective regulator?
Professor Nason: When it was set up, I think the recommendation from the previous Statistics Commission was that it would be a separate body. The decision was that that did not happen and we have the structure that you see today. Also, as we say, OSR is a new body and, within the structure and the legislation that it has, UKSA has attempted to separate things out and make the OSR more separate and more independent, giving it a separate brand. There are plenty of ideas out there, for example separate boards.
Chair: That leads to the next question.
Q55 Dr Rupa Huq: I want to ask about the structure. The public sector statisticians are part of the GSS and that is a network led by UKSA, the National Statistician, but both are employed by public bodies across Government. How effective is this matrix model in resisting the imperative from public bodies to mislead the public by presenting things in a good light—“lies, damned lies, and statistics” and all that?
Hetan Shah: It is a very good question. There are some countries that have a much more centralised statistical system. If you want your independence, don’t keep your statisticians in Departments; just suck them all up into a central body, employ all your statisticians in the ONS and then you have no problem there. The downside of that is that you then lose the link to policy, so I would say that if you are concerned about independence at all costs, you lose other things.
If you look forward, my hope is that the statisticians within Government will help transform Government. We know in the private sector data is now the big asset. If you think about Facebook and Google and so on, these are the money-making companies. Government are coming around to that understanding. Where is that data? It is in the Departments and it is there to inform policymaking.
I think that while you lose a bit of formal independence by having those people in Departments, the links that you get to being in the Housing Department and talking to the housing policymaker far outweighs that. I would rather have a system where if things then go wrong the regulator intervenes and says, “You are misusing the statistics”—as I talked about with the education example earlier—than one where you are so worried about independence you pull your statisticians away from the policymakers and then you never get the link to the policy.
In a way, that is where the Bean review came from, because it was felt that statisticians had become too divorced from the Bank of England, the Treasury and so on and lost sight of what they needed from the stats.
Professor Nason: Professionally speaking, it is important that statisticians work at the coalface. If they don’t it doesn’t work. John Tukey said statisticians get to play in everybody’s backyard, and it is fundamental that we have that. But then you have this problem about professional development, possibly professional integrity, independence, and that is why we have this matrix structure. What I would like to see is somehow strengthening that aspect, allowing the statisticians within Departments to exercise their professional autonomy and integrity to a greater degree, but I still think you need the statisticians in those Departments.
Hetan Shah: One option might be to have a board member specifically tasked with thinking about the role of the GSS. That did happen in the early guise under Michael Scholar, who was the first chair of the UK Statistics Authority. He had two deputy chairs, one focused on the ONS and one on the GSS. If we think the GSS is where things are slightly weaker at the moment, there may be the chance to give a board member particular responsibility for that.
Q56 Dr Rupa Huq: Should there be a formal role? The board thing would be your suggestion? What role should UKSA play in strengthening GSS? Are there any other mechanisms that could be adopted?
Professor Nason: There is the recommendation to Government that the heads of profession have a higher status. That is a recommendation, “This is what we think should happen”. That could be made.
Hetan Shah: There are a couple of other things. One is about leadership development and that is particularly within the GSS, so in essence that contact with talent management. If you see a bright young thing coming through, who is going to make sure that your next placement gives you the kind of skills that you need? That used to happen more a long time ago than it does now, so that is one thing.
I talked a bit about leadership across the whole system. A leadership academy of some sort might be worth setting up for statisticians. I spend my life around statisticians—I am not one myself—and they are wonderful but, on balance, many of them are more inward looking, technical people, not necessarily focused on the influencing kind of skills or the, “What happens to the data after I have produced it?” bit. The statistics profession is changing and it has to be more and more about the impact, so investing in those skills and that wider leadership would be really helpful. That is not just within the GSS but across the whole stats system.
Q57 Dr Rupa Huq: Is it usually in academia where statisticians go or do they get—
Professor Nason: One thing I was going to say following on from that is this is a very big thing in academia, developing leaders for the future, especially for topics and subjects like mathematics and statistics. Yesterday the Government announced £800 million for doctoral training centres. I run one of those at Bristol and one of the things we will have is a strong leadership focus there, actually interacting with policymakers. That is part of what we are going to be doing, and we are working with the ONS and the Data Science Campus. It is one of our partners in this, so I know it is doing things on this aspect. We are working together on things like that as well as with many other companies.
Q58 Dr Rupa Huq: They are your competitors, not the private sector?
Professor Nason: There are a lot of private sector companies involved in our one, for example GSK, Shell and Trainline. As statisticians, we share these issues. As Hetan kindly said, we need more leadership development. I think that is true, but we are going to do it together across Government, industry and academia.
Hetan Shah: I talked earlier about some of the other facets of leadership. One is communications: how do you make sure that your statistics land with the policymakers and the public? In a world where we are concerned about fake news, misinformation and so on, it is a really important question. People like Will Moy at Full Fact—which is a fact-checking charity—have recommended the creation of some kind of centre of excellence and expertise within the ONS around how audiences, different parts of the public, respond to facts and how you can get them to land better. That would be a really useful thing that, for example, the Office for Statistics Regulation could pick up.
Chair: We are going to have to go a bit faster. It has been a very interesting session.
Q59 Mr David Jones: You mention the structural changes you think could be desirable. Do you think that legislative change is called for in order to improve UKSA’s efficacy? Do you think legislation is adequate—
Hetan Shah: It is good enough to get what we need. We could spend an awful lot of time trying to change structures and legislation and so on. I think it would be better to focus on the actual workings, the leadership, the many things that we have talked about already at this Committee, which could be inactive now, and I think that is the place to put the energy. I think some of the leaders that we have at the moment are open to these things and up for it.
Professor Nason: I have given this a great deal of thought. It could be possible, for example, to have a complete separation from the regulation function from the ONS function, but now is not the right time. OSR, as I said, is a very new organisation within the structure. Personally, I would like to see it continue for the foreseeable future.
Mr David Jones: In order to assess whether it works adequately. Okay. Thank you.
Q60 David Morris: How well does UKSA understand what users of statistics need now, and are likely to need in the future, so that statistics are to remain relevant?
Professor Nason: I think the perception out there is that much more could be done on the relationships between users and UKSA. Do you want to go into more detail on that, Hetan?
Hetan Shah: I would agree. This is an area where there has been some improvement under the OSR. The kind of thinking about families of statistics has been good. Creating some kind of central team within the ONS focused on user engagement and insight would be a quite helpful step forward.
The top level test is: do the public have the numbers that they need to make sense of the world? That is a very difficult test but one that we should be aiming for in an ideal world. More needs to be done around that. We are almost suffering from too much information. There are too many numbers out there. For example, is poverty going up or down? Are incomes going up or down? There are just so many things flying around that it is hard to make sense of them.
One thing you might go back to is there used to be a publication called “Social Trends” where the ONS on an annual basis would say, “These are the big trends of the day” and collate that and make it more coherent. Moving back to something along those lines, a collation and analysis function, might be helpful to the public.
Q61 David Morris: In your opinion, if you had a yearly release of figures instead of a quarterly release of figures, would it make more sense in certain areas?
Hetan Shah: I am not talking for increasing the release time. In fact, expert users want releases to be quicker and quicker, and in economic statistics we are moving more towards that and that is a good thing, but for non-experts, who are not tuned into this, hearing in the press every day about very minor changes in the trend, does not help you get the big picture. Something to counteract that, such as an annual snapshot of, “This is the big picture”, might be useful.
Q62 David Morris: How do you feel that UKSA engages with the other organisations in collating the information that is actually relevant? As you stated, there are so many figures flying around out there that it is hard to make sense of what is being released.
Hetan Shah: It does a good job of that. There is an interesting question, which is slightly speculative: in the long run will the ONS and GSS be the people producing numbers or, because of the changing landscape and more and more people producing numbers, will they be there more to pull it together and say, “What does it mean?”
An interesting example of this is homelessness statistics where it is charities that tend to produce some of those numbers, but the GSS is now looking at the evidence base and pulling it together. That may go nowhere but it might be a canary in the mine of something that will happen in the—
Professor Nason: That is going to happen more and more. A lot of corporations are collecting their own data, looking at their own datasets and coming up with their own estimates of inflation and GDP. There is a risk that UKSA will—subject to challenge and undermining—say, “Inflation is this” and some big corporation will say, “No, we don’t think it is this. It is something else”. I don’t think it is going to happen yet but you could possibly see this way into the future.
Q63 David Morris: Do you think in the future you would get the balance right in engaging policymakers with the use of the statistics?
Professor Nason: Yes. There are some good examples of good practice in Government Departments. Maybe it is because of past things, but the Ministry of Justice, for example, has fantastic processes for engaging with users, as far as I can see. Also, it is known within the academic community as a Government Department that has contributed to the research effort and been totally reproducible about what methods it has used, so it is a fantastic example. I have been very impressed by it.
Q64 David Morris: Do you feel that the methodology used by some bodies could be flawed? You just mentioned homelessness. There are so many charities feeding into a particular statistical part but we don’t really know what the calibration of measurement is, meaning what their methodology is.
Professor Nason: There is a risk across the board.
Hetan Shah: This is where the regulator needs to step in. The big challenge in the next five years is the census. The census is moving from the old method, which was you survey everyone in the country, to you survey lots of people using an online survey now but you mix this with the administrative data that the Government collects for other purposes—health records, benefit records and so on—and you develop a patchwork. Methodologically this is much tougher and this is why I think we have been talking about the innovations required in thinking about new ways of doing statistics, which is really important.
Q65 Dr Rupa Huq: It is interesting that the census is not going to be a total enumeration anymore of the population. Is it moving to a sample?
Hetan Shah: It will be, so in 2021—
Q66 Dr Rupa Huq: Is it going to be only decennial? That is way too slow.
Hetan Shah: Exactly. This is the problem. Doing something every 10 years means that by year 9 it is very out of date, so the pressure is: how do we do this on a much more rolling basis? You cannot just phase out the census and try something new in the hope that it works, so—and this was an RSS recommendation that we were very pleased to see was taken up—you run a fairly traditional census in 2021 but, side by side, you try to replicate it using the patchwork of data sources that I was talking about.
If you can pull it off that means you can do that on a fairly regular basis because you have health records, benefit records and so on all the time and then you can supplement that with surveys. Our hope is that after the 2021 census you will be able to get census-type data on a much more real-time basis.
Q67 Dr Rupa Huq: That is the last old school one?
Hetan Shah: Yes.
Q68 Dr Rupa Huq: Will it be the paper thing that comes through—
Professor Nason: No, it is meant to be online.
Hetan Shah: Completely online.
Professor Nason: But of course, not everybody will engage with that, so you need a whole host of other measures to get data submissions.
Q69 Dr Rupa Huq: My question was: how easy is it for policymakers and the general public to make use of the statistics we currently have? You mentioned social trends. When I used to look at these things, there was the British household panel survey, the general household survey—all these things have gone—and the British crime survey, which I think exists in another form. There is something to be said for longitudinal stuff that you can compare every year. You get a new Government that, for the sake of being a new broom, changes all the stuff and I wondered what you thought of that.
Professor Nason: That is my own personal thing about RPI. That is why I want it fixed—because it is such a long record. Of course, it has changed and evolved over the years but it goes back a long time. From an academic point of view, having a long history of RPI is exactly what you said.
A lot of the things you mentioned are still there but they are probably called different things and may possibly be run by different subcontracted bodies sometimes.
Hetan Shah: There is a big opportunity, though, for thinking about local community level data: what is happening in my area? We are getting the data sources now to be able to do that. There was the, “What are the crime rates in my area?” That proved extremely popular. You could now be doing that with a whole host of data sources and much more hyper-local data, so that is one of the big opportunities coming our way.
Q70 Dr Rupa Huq: What do you think UKSA could be doing to facilitate the local stuff you mentioned? Is that all verifiable?
Hetan Shah: It is coming on stream. New data sources are being used all the time. It is worth mentioning the Data Science Campus that has been set up. I am on the advisory board, so I declare an interest there. This came out of Charlie Bean’s recommendations that we need to think about data innovation. It is doing really good projects to say: how can you use new data sources? They are trying to look at, for example, satellite imagery, mobile phone records, things that have not been used so much in the past, to see if you can get good statistical outputs from that.
One of the key questions for the future is: can you use that Data Science Campus knowledge, not just to stay in-house for 70 people, or whatever they have, but to transform the whole of the ONS and GSS, because in 10 years’ time any statistician needs to be using these ways of working?
Professor Nason: To fully satisfy what you are after is going to require resources or someone else to do it, companies or academia possibly. They already do what they do. I cannot imagine that they are completely resource rich. They have duties to do all kinds of things. What you have suggested is great, fantastic, but the money has to come from somewhere to enable it to happen.
Hetan Shah: On resources, it is worth noting we are coming up to a comprehensive spending review. This Committee could be helpful in reminding the Government of the importance of data and statistics.
Q71 Mr David Jones: Would you say that UKSA is doing well in keeping up with developments in technology and data science?
Professor Nason: I think so. The Data Science Campus is bringing in the latest stuff. They are doing some very exciting things. As Hetan says, from our point of view, it would be nice if that somehow spread more into the rest of the organisation.
It comes back to what I said earlier about the thinking and the research function within the organisation. If that part has been shrunk dramatically over the years, somehow it has to be nurtured and allowed to be able to grow so it is not just a machine doing stuff but you have a chance to think and develop new things and new ideas. It is kind of there but it needs to be supported, in my view.
Hetan Shah: The leadership that the current National Statistician, John Pullinger, has shown on this has been really strong. If it had been a more traditional statistician it may have gone, “All this new-fangled stuff—that is not proper statistics”. He has done the opposite. He has been very open to it and really taken it on. I think the investment in technology—they never make the front pages of anything—has really happened. This is the first time where the platform being developed for the census won’t just get thrown away at the end. It will be used for the future work of the ONS and the GSS, which is what we all want, isn’t it?
Q72 Mr David Jones: Is there anything more you think that UKSA could be doing to stay at the leading edge?
Hetan Shah: Investing in methodology; what you were saying about the research function. As I say, now you have the Data Science Campus—we have some people who really know their stuff there—the risk is that they end up being a consultancy across Government. They do some useful stuff but they haven’t done something transformative and I think the key is to use them to transform the ONS and the GSS. For example, they are setting up a master’s in data analytics for Government, so they have some useful things there, but that is what I would push on.
Professor Nason: I would say more partnerships. You see this a lot on the security side of the house, a lot more partnerships, things like the Heilbronn Institute, which is a conduit between academia and parts of Government. The Data Science Campus is like that. There are partnerships but there could be more.
Hetan Shah: Ethics is becoming a bigger and bigger deal in this whole space because as you start using new data sources, the public are saying, “What is happening with the data about me or my mobile phone records?” or whatever it might be. John Pullinger set up the National Statistician’s Data Ethics Advisory Committee, which is doing a great job of assessing internal projects for their ethics. The stats system is doing well internally but it could now be more proactive and be a voice to the public about how the ONS uses data about you. There are a lot of voices for the privacy case, which is, “Hold your data back. Don’t let anybody use it”, whereas our view at the RSS is that data is a public good and can be used for the betterment of society.
Q73 Mr David Jones: That brings us back to the transparency issue that we were discussing a few moments ago. Do you think that that is something that should be published, that people should be able to access on a website somewhere?
Hetan Shah: The minutes of National Statistician’s Data Ethics Advisory Committee are all published and so on, so I think that is there. It is just—
Q74 Mr David Jones: But in a more digestible form than a set of committee minutes?
Hetan Shah: For me it is less about the formal transparency. I think they have the formal transparency right on that committee. It is more about: are they connecting to—there are many other organisations thinking about these issues now. In its new world of being much more of a data hub across Government, a data leader in our society, it needs to connect better into all those other bodies. There is a new Centre for Data Ethics and Innovation Consultation. The Science and Technology Committee just down the corridor is talking to its new chair right now. What is the link that the ONS has made with that sort of body and so on?
Q75 Mr David Jones: As you know, the Institute for Government identified a number of important gaps in the data that were collected and published by the Government. UKSA has a statutory duty to ensure that official statistics are comprehensive. How well is it doing in closing the gaps that were identified by the IfG?
Hetan Shah: I don’t know specifically how well it is doing on that set of gaps. I am aware of the report that you are talking about, which is the “Whitehall Monitor” and the data gaps that come out of that. The broader point is that there are many organisations out there pointing to the data gaps in their area of expertise. The IfG will have the expertise on Government. There are many others and I think the question to ask is not have they responded to a specific report, but how are they responding to the huge volume of demand across the piece? I hope this inquiry will help get to the bottom of that.
Professor Nason: I am aware of that report as well, although not the detail. This comes back to processes; what processes do you have in place to horizon scan and keep an eye on all this? It also comes back to the point that there are going to be many organisations filling other gaps, finding gaps that you never thought would be gaps. The ONS is resourced to do what it does. It could have more of a horizon scan, more of a view, but to fill them is going to require extra methods of collection. That might mean partnerships, contracting things out. They can be filled but it is an ongoing process.
Q76 David Morris: How well does the UK Statistics Authority engage with other bodies with data governance and ethics responsibilities? In that respect, how can UKSA build public confidence in statistics through its role in good data governance and data ethics? You just mentioned John Pullinger, but how do you bring this out into the public domain and give people confidence in what is being publicised?
Hetan Shah: I slightly jumped the gun in a previous answer by talking about some of those things. This seems to me to be a critical question of the future, which I think wasn’t really there 10 years ago. John Pullinger set up the National Statistician’s Data Ethics Advisory Committee, which is a good start. It has essentially focused on case law. ONS and GSS bring difficult issues to them and they rule on them. I was a member of that committee for the first couple of years. I found it very helpful and thoughtful.
What could happen next is for the chair of that committee and others within the ONS and GSS to play a very much more proactive role linking up to many other bodies that are interested in the space. The UK is a very fertile scene. Thinking about data governance, as well as the Centre for Data Ethics and Innovation Consultation there is the Ada Lovelace Institute, which is an independent body, the Open Data Institute and many others. I haven’t yet seen the ONS, GSS and OSR having a very strong presence in that space. They are clearly hearing what is being said but I think they could play more of a leadership role. They bring a lot to the table.
Q77 Chair: I do not think there is anything else we want to ask. I will just come back to the one thing about who they are accountable to. I have been looking through the Act. The word “Parliament” does not appear until section 16, which deals with something rather obscure: requests for assessment, supplementary. Yet Parliament uses the excuse that there isn’t accountability elsewhere. Well, clearly, we need to be much more proactive and maybe we are doing that. Is there anything else you would like to add about accountability? Part of the confusion, and maybe the lack of confidence in UKSA, rests on this confusion about accountability.
Hetan Shah: It is great that you are holding this inquiry. If you are able to help shape the next strategy of UKSA, that is Parliament fulfilling its role. In a sense, I do not think the legislation gets in the way of this. I think everybody I have spoken to says, “It is accountable to Parliament. Here you are, the door is open, so let’s go”. I think that is the assumption.
Chair: Thank you very much indeed.
Examination of witness
Q78 Chair: Could the new witness identify himself for the record, please?
Richard Alldritt: Good morning. I am Richard Alldritt. I was the first statutory post holder of the post of Head of Assessment under the 2007 legislation. Before that, I was Chief Executive of the UK Statistics Commission. I am five years retired from the civil service, so all my comments are through that filter.
Q79 Chair: As an ex-civil servant or as an independent person? Are they the same thing?
Richard Alldritt: Independent and ex-civil servant, yes.
Q80 Chair: Thank you very much indeed. I start with that very first question again: what has UKSA achieved in its first 10 years and what could it have done better?
Richard Alldritt: With your permission, I will answer this in a slightly different context from the previous evidence. The UKSA, which I am able to offer a view on here, is the UK Statistics Authority Board and the other elements that were brought in following the 2007 legislation, specifically what is now the Office for Statistics Regulation and the secretariat staff that support the board.
In the legislation, the phrase “UK Statistics Authority” actually means different things in different sections of the Act. In some places, it refers to the entire structure of over 3,000 staff, including the whole of the Office for National Statistics but, for very important aspects of the board’s work, the Office for National Statistics is just one of something like 45 major producers of national statistics in both Whitehall Government and the devolved Administrations.
I will answer the question in terms of what I think the board has achieved. The main thing I think it has achieved is a considerable strengthening in the management of the Office for National Statistics. The example that comes to mind is the preparations for the 2011 census. In times past, the census was managed by a very small team of middle-level staff, and it was always a struggle. For the 2011 census, the authority board looked at these arrangements critically and required very substantial strengthening in the senior people involved and then was meticulous in going through the planning.
That is just one example of where the board sat around the management of the Office for National Statistics—this is behind closed doors, and there is very little in the public domain about this part of their work because it is part of running a Government Department—and provided challenge, made very significant staff changes, brought people in and did all that sort of thing. They did that.
They introduced the code of practice in 2009, which was the first really clear codification of what good practice is in relation to official statistics, not just in the ONS, but right across Government Departments. It doesn’t just relate to the work of statisticians, it should be said. Bits of the code of practice relate to Ministers’ responsibilities for statistics. They are not terribly fully expressed, but they are in there. It says resources need to be provided under certain circumstances and so on.
The board completed the cycle of assessment reports, which was a requirement of the legislation. There were over 300 of these and I was responsible for that work. Basically, it was a systematic review of every single major area of statistics, right across the four Administrations of Government. I was trying to find an analogy the other day and it was a bit like doing a review of an old-fashioned car factory, where you would make sure that the production line was working, the supply chain was there, the finance was there and there was a marketing strategy and health and safety in place, but you could still be producing a Trabant three-wheeler or something similar at the end of the day. Code of practice compliance is not the same as you using all your resources in the best possible way. It means that the particular statistics are being done to a high standard.
That code of practice has to some extent been adopted as a model by other countries as well. It was a good code of practice, in my view. As the previous witnesses noted, the authority has become a trusted voice. When the authority—and this goes back to Sir Michael Scholar, the first chair—intervenes publicly on an issue of misunderstanding or misuse of statistics, it does carry very considerable weight. I cannot think of a case where it had to retreat from what it was saying. When it writes one of its powerful letters to Ministers or an institution, I think that is an important part of the role.
I have to say that role is a requirement of the code of practice on statisticians. If it was health, the health statisticians really ought to be taking to task anybody who misuses or misunderstands, but it was so difficult for departmental statisticians to take on that role it sort of became transferred. I was involved in that transfer to the authority for it to do that. In practice, the way we used to do it is we would work closely with departmental statisticians in deciding what we should say about a particular misuse of statistics. Those were all positive things.
Q81 Chair: Where could it have done better?
Richard Alldritt: Let me start by saying I think there was a consequence of this deep involvement in ONS management matters. I will answer that, but to just digress slightly, what is the board? The board is about 10 non-executives under a non-executive chair, plus the National Statistician, two of his senior staff and the Director of Regulation, my old post. Most of those non-executives are two days a month. That is their commitment. The chair is getting close to half of full-time.
When it got drawn into detailed engagement with the Office for National Statistics about matters like the census, the RPI issue and so on, that did rather squeeze out its capacity for engagement elsewhere. I think the practicalities of what can reasonably be expected of a non-executive board are really important going forward. While they might sign up to doing everything for everybody, in reality there is only a certain amount of capacity. Where I think they were less effective was following up on some of the messages coming out of the assessment work, for example, on the need for a cultural change.
Sir Charles Bean’s report draws attention to the fact that there is a need to change away from just producing figures to engaging with the people who use them constructively and giving them support and guidance. I think he said it better than I can. That is the change in culture that is needed because there are so many statistics now.
When I started in my career in the 1970s, there really wasn’t anything like the volume of data on any one subject that you now see. One very topical example is the press coverage of the increase in violent crime and knife crime. Are the hospital data on treatment of knife crime being published alongside them in a way that helps paint a full picture because, without going into the detail of it, violent crime is an awfully difficult thing to measure? There is a huge amount of it that never gets recorded and, therefore, when you get a sudden surge in the data, the first question ought to be: is there corroborating evidence?
When I talk about advice from statisticians, that should be their first question, “Well, hang on a minute, yes, the figures are rocketing but is this actually what is happening or is this a consequence of a change in police activity?” Sorry, that is rather detailed but it is a sort of issue of the role.
Q82 Chair: Indeed, a change to the recording practices as a result of the controversy about police-recorded crime.
Richard Alldritt: Yes. I started my career in that area and police recording practice—there is no malice here—is necessarily a very fluid thing. The world is changing all the time and how you record crime is really tricky.
Chair: The point is well made.
Richard Alldritt: I did have a couple of other points. Another thing where it has struggled—and again I believe it is a consequence of the very close engagement with the management of the Office for National Statistics—is to talk to external experts and users to identify—
Q83 Chair: They do not do that enough?
Richard Alldritt: That is right. They do not do that enough or identify what the big issues are. It is all very well to sit in the office and have a suspicion that the communication of—I will give you another example, NHS performance data, all those “within four hours” things. The way the statistics around those performance indicators are being brought together and communicated to the public is a really important issue. Is it being done as well as it could be?
I could sit here and say, “I think there is an issue there”. I do think there is an issue there, but the way you get a handle on the issue is that you don’t immediately launch a big inquiry. What you do is sit down and talk to the experts in that field in quite a lot of depth. I think that is the role that the Office for Statistics Regulation should increasingly be performing, but it cannot do it on its own. The non-executive board has to lead this itself. It cannot be delegated to officials, in my view.
Q84 Chair: When you were the Chief Executive to the Statistics Commission, you argued that the whole system needed a statutory basis, the whole system, and that is not reflected in the Act.
Richard Alldritt: I would not say that. What we argued for was that the commission had identified the value of a genuinely independent voice investigating issues of public concern relating to statistics. There was one about different Departments putting out different figures for rates of employment. I think it was something to do with—sorry, my memory is failing me—immigration unemployment. It was on that cusp.
There was a lot in the media about it, so we took it on and we brought in experts and we produced a report on it. What we were doing was acting as a completely independent look at difficult and deep issues, publishing a report and standing by the report that we had published. We would come to the Treasury Committee and we would be cross-questioned about that report. We reported directly to the Treasury Committee.
Q85 Chair: You are happy that the Act provides a regulatory framework for the whole of the statistics profession?
Richard Alldritt: Yes. I think it is there in the Act. It is but—
Q86 Chair: Why do you think there is this reluctance to bring in the outside experts?
Richard Alldritt: I go back to my point about a board only has so much capacity and a great deal of the capacity of the board has been drawn into the management of the Office for National Statistics. That is my observation. I think it is probably demonstrable. When the board started, the agenda for board meetings was essentially set by me and my staff from the regulatory perspective. After a couple of years of that, there were a lot of complaints from the other side that they weren’t giving enough attention to the ONS business.
You would get the same issue coming from both sides, both deputy chairs. One would be trying to deal with migration statistics in a sort of managerial sense and the regulatory side would want to challenge the work that the ONS was doing on migration statistics in some respects. You were getting two streams of work on the same subject coming into the top board and that was quite uncomfortable for the non-executives. I know they preferred working together on an issue rather than being split into separate functions.
Chair: That is a very interesting observation that leads on to the next question.
Q87 Mr David Jones: The structure of the board as constituted by the legislation did not meet with your approval. I think that is right. I think you have been critical.
Richard Alldritt: It depends what you mean; not so much the structure of the board as the balance of its activities.
Mr David Jones: Yes, that is right. I think you were critical of the dual roles of both producing statistics and regulating them. You have indicated the sort of tensions that could arise.
Q88 Chair: Well, you did describe it as a fundamental weakness of the legislation.
Richard Alldritt: I believe so. You used the phrase earlier of regulatory capture. When we were preparing the advice on the legislation in the Statistics Commission, we worked closely with constitutional lawyers. It is constitutionally quite a complex piece of legislation because it cuts across Government Departments and other legislation. The constitutional lawyers kept saying, “Be very wary of regulatory capture. If you put a regulatory function inside a production organisation, there is a long history of the regulatory function getting squeezed down into something less than it might be”. There is a bit of that. At the same time, the assessment work at a lower level was going on successfully and coming up with a lot of recommendations, some of which led to Select Committee hearings. It was not that the regulation was not working at all. There was one occasion where we tried to get the Department for Transport to publish road accident statistics in a different way. It ended up with a hearing in front of the Transport Committee with myself and I think it was Lord Adonis at the time and the matter was resolved there at the Select Committee and that was working. It was not that the regulatory function was not working. It was just that it felt to me like it was getting increasingly squeezed by the managerial side.
Q89 Mr David Jones: That being the case, do you think that there is a case for change in the legislation?
Richard Alldritt: In an ideal world, yes, but I am not sure that is the priority. If it was in the gift of this Committee tomorrow to reopen that legislation, I would say be a bit wary because all sorts of issues can get thrown into a statistical Act. The Act seriously needs tidying up but I am not sure that is what would happen if it was opened. I think a whole lot more things would probably pile in about data management and freedom of information and so on. It would all start growing into something else.
There is a plus side to the vagueness of the legislation. It has already been mentioned that the legislation does not say who the authority reports to in any coherent way nor does it preclude the board acting in any one of a number of very different ways. The first chair set up two structures, each under a deputy chair, none of which is in the legislation, but the legislation did not present any problem. Refocusing the activities of the authority board does not require a change in legislation, in my view.
Q90 Mr David Jones: I think you were here when I asked the previous witnesses about who assesses what is the public good. Do you think that that is a weakness in the current legislation?
Richard Alldritt: Can I present you a model of how I think it might work? I think the authority board should identify the dozen or so really big issues in official statistics at this time and do that very openly. That should be available to look at.
The first bit of pubic good is: have they identified the right things as the big issues? I am not in touch with the latest issues—my work in recent years has not been in the UK at all—but you have the issues that Charlie Bean raised in his report about the adequacy of economic statistics. You have the RPI-CPI issues. It has already been mentioned that there is a very important change to census practice coming through. There is a bunch of issues about the labour market, productivity, measurement of the gig economy and all those things. There is immigration, migration, particularly in the Brexit context. What statisticians are responsible for putting into the public domain, what advice? Why did the Home Office’s huge scheme to count emigration better never result in good statistics? The Department for Work and Pensions and HMRC are sitting on a vast database that could be used for producing a much broader range of statistics and providing much more support to the economic accounts.
There is a list like that that could be drawn up that you would expect the authority to be able to defend as being the big issues that it is concerned about. A committee, such as this Committee, could say to them, “What are you going to do about these?” and then you have a sort of structured engagement. The public interest element should come out in the wash here to some extent: are they looking at the right issues; are they approaching them in the right way?
On all those sorts of issues I would not want just civil servants to go into a corner and write a report. One of the reasons the Office for Statistics Regulation is not bigger than it is is because it is extremely difficult to find the people with the skills to do that work. My preference was probably to keep it smaller than it is. What they need to do is be put in a position where they can commission—it depends on the subject. If you are looking at NHS performance indicators, you go to a completely different set of people than if you are looking at the adequacy of economic statistics or arrangements for the census and you draw them in.
Q91 Mr David Jones: Would you say that the dual role of UKSA is impacting adversely or otherwise on the users of statistics or, more generally, the public?
Richard Alldritt: Let me answer that in this way. The use of statistics is not well understood inside the government statistical system and the reasons for that are partly historical. A lot of statistical production has a lot of history and the work of the people doing the work is just trying to keep the thing running. If you don’t understand what use is being made—it is not so much who the user is. I will go back to crime statistics. There are not many uses of crime statistics in fact. One is crime prevention. It is not so much who is using it, you can work out who would be using it, but what should they be being told? If you are trying to use crime statistics for crime prevention, what are the messages that those people need to be given about the crime data?
The Chairman mentioned changes in recording practice and so on. I don’t see how you can provide the best possible service to the country if you are not spending a significant amount of time looking at the use of the service you are providing. Throughout my time in the job I kept saying to Departments, “Find out who is using this”. Not so much the “who”: find out what uses are being made of the statistical data. Health data is used for distributing resources within local authority areas. Health people need to bear that in mind because that raises a big question of consistency of data between areas. Health, crime and so on are used as a deprivation indicator.
There are uses of statistics that are not obvious. The insurance industry uses official statistics in setting premiums. The supermarkets use official statistics in deciding where to build supermarkets. It may not be the biggest issue of public interest but it is there. I would like to see Government statisticians not just themselves understanding, but communicating what use is made of official statistics.
Q92 Mr David Jones: Could it be improved by a change in the structure of UKSA?
Richard Alldritt: Inasmuch as if they gave that their attention rather than so much of their time being taken up with essentially internal management matters within the Office for National Statistics. I don’t think there is a better board sitting out there. In my view, it is an extremely good board of very talented people. It is what they are being asked to do. Without wanting to overstate this, there is a lot of pressure from within the Government for them to run the Office for National Statistics. If there is going to be a counterbalance to that, it is going to come from the parliamentary side, I believe.
Q93 David Morris: In your time at UKSA as head of assessment, how did the board seek to resolve the tension between the two roles of production and regulation? How successful was that in managing the tensions between the two?
Richard Alldritt: We started very formally. The first chairman, Sir Michael Scholar, and I discussed this at great length and we decided we had to have two boards, in effect. Although the legislation created one board, we needed two, and two deputy chairs were each given responsibility to work with a Chinese wall between them and the non-executives were assigned to either one board or the other board. One board was responsible for the regulatory function and the other board, which was called the Board of the Office for National Statistics, met more frequently and dealt with management matters, the typical role of a board of a Government Department in effect.
It unravelled fairly rapidly as a way of operating. I can’t really remember exactly the sequence of incidents but it would typically have been that the regulatory side was very concerned about something that was actually being discussed at the ONS board side. When all the board members came together, instead of staying in their respective roles of regulator and non-regulator, they just had a big discussion about it, which they were much more comfortable with. Human nature comes into this. Being split into two camps who might be having a go at one another at a meeting is not particularly comfortable even though I did think it was quite productive.
I was always for dividing people into regulatory functions and the rest because, to contradict something that was said earlier, the split between these two was not ONS and the wider Government statistical service. It was management versus regulation, so the regulation side was looking at ONS a lot of the time. It was a difficult model to make work and I think under pressure from the non-executives it did rather unravel. That resulted ultimately in the bit that I was responsible for, now the Office for Statistics Regulation, being delegated essentially to officials, which to me is a bit disappointing. I would want to see the non-executives—these powerful people, heads of universities and people with enormous experience in the field and in administration—leading the regulatory function.
Q94 David Morris: Were there any benefits of having production and regulation under a single board, in your opinion?
Richard Alldritt: There were from the Treasury’s perspective. When the Bill was being drafted by a Treasury team, the phrase “consuming their own smoke” was mentioned in informal meetings as a concept. I think things have moved on. At that time there was sensitivity about statistical grenades being lobbed into the media. It was just felt that to have a completely independent regulatory function doing something—the old Statistics Commission would often start the day by talking to journalists or user interest groups and would do so freely and would publish the correspondence on its website. It was very open. The trouble is that is a little bit open to people riding hobbyhorses and perhaps making political points by backdoor methods and so on. We had to be a bit careful when a political party approached us. We would still look at the issue but we would not necessarily look at in exactly the way they had approached us on it.
With that background, the Treasury at the time was extremely uncomfortable with the idea of a statutory regulator making trouble for the people trying to do a good job and I do still have some sympathy with that. I don’t think a freewheeling regulator is really quite right here because it is going to stir up an awful lot of unnecessary distrust. This is the attitude of the other countries I have worked in. They like bits of what we are doing in the UK but they think there is a danger of us stirring up a lot of distrust and trust is important to statistics because if you want people to use them, they have to have faith that they are reasonably reliable. Trust can be quite easy to undermine with a few newspaper headlines.
There probably were some virtues, particularly at the time, in bringing it all together but my feeling is that the regulatory function needs a bit more structure and a bit more muscle than it currently seems to have.
Q95 Chair: You comment in your written evidence about the RPI issue and you say, “UKSA’s responsibilities to engage with Ministers and the Bank of England and to stand by decisions with which it has been involved sit uncomfortably at best with its responsibilities to set out openly what the problems are from the user perspective and to make its own independent recommendations”. You heard me quote from the House of Lords report on RPI. How much is this a commentary of general application about the functioning of these two roles?
Richard Alldritt: The RPI issue is difficult and I was slightly involved. I was involved to the extent that we looked at how the RPI was put together. We took advice, including internationally, and that advice was that the RPI is very old-fashioned and out of date in how it is constructed. I don’t remember all the details but—this is at the time—there is good deal wrong with the RPI as a technical measure of inflation judged against international standards. Under the code of practice, you don’t produce statistics that don’t meet international standards. It is pretty much an absolute. You either get them to meet international standards or you stop producing them. Unfortunately stopping producing the RPI was not in the gift of the authority because the legislation made special provision to stop it being in the gift. Therefore, we ended up in this rather tangled mess with a lot of experts thinking the RPI is not a good measure. Fixing it sounds good but you would probably end up with something that was not the RPI if you fixed it. You would probably end up with a different measure.
Q96 Chair: But that doesn’t explain why UKSA failed to make any recommendations at all.
Richard Alldritt: What explains that, I suspect, is that it has got deeply embroiled in discussing precisely those matters with the Treasury and the Bank and ONS staff and has not emerged from it.
Q97 Chair: It had been seen off?
Richard Alldritt: It could feel intimidated. Every time it tries to draft something, it could find that it is getting so much pushback. You don’t just go into a corner and write something and publish it. You do have to see how it plays.
Q98 Chair: There is a procedure laid out in the Act and it could go through the procedure.
Richard Alldritt: I don’t know to what extent it has tried to do that. I am far too far away from that.
Q99 Chair: Were you surprised when RPI was dedesignated after you left?
Richard Alldritt: No, I was involved in the dedesignation. The dedesignation was saying this doesn’t match up to—it is a requirement of the code that it meets all international standards for an index of that kind. Our assessment was it didn’t and, therefore, it didn’t qualify for the national statistics label, which means compliant with the code of practice. We judged it to be non-compliant. It was not really our job to say what should be done about it. The options were—
Q100 Chair: Do you disagree with the Lords report on this?
Richard Alldritt: I don’t think I do. They recommend fixing it. Well, the thing is if you fix the RPI you might well end up with an index that is not really the RPI.
Q101 Chair: It says, “The Authority should stop treating RPI as a legacy measure and resume a programme of periodic methodological improvements”.
Richard Alldritt: That is an option. Whether when you have done a whole stack of methodological improvements it is really the retail prices index or a new inflation index is a moot point.
Q102 Chair: It is about the Carli formula affecting clothing costs.
Richard Alldritt: It is.
Q103 Chair: That doesn’t sound like rewriting a whole index. It sounds like fixing—
Richard Alldritt: I am getting out of my depth here, but what I remember is that the mathematics at the centre of the index are very important to what the figures eventually come out at. If you change away from the Carli formula, you are getting a rather different index. It is certainly one option. I don’t know whether this has all been exemplified in terms of what the index would look like or whether that is even a valid consideration, because you don’t produce an index, look at the results and decide whether you like it.
Q104 Chair: I understand. Moving on. Sorry, just tell me one thing: you don’t think the authority has failed in its statutory duty, which is what the House of Lords report says?
Richard Alldritt: No, I don’t. I think its statutory duty here is very twisted and complex. Part of its statutory duty is to help the ONS to deal with these issues and it is doing that. If it had not dedesignated the RPI, it would have failed in its statutory duty because it had a statutory duty to test it against the code of practice, which it did. It found it deficient, I believe correctly. I don’t remember the details but I believe that was a correct decision. Certainly I know Sir Andrew Dilnot, then chair, went over that decision with immense care and decided that it was not compliant with the code of practice.
Q105 Chair: If the legislation prevents UKSA from regulating the RPI effectively, the legislation should presumably be changed?
Richard Alldritt: I believe the legislation is the problem here. ONS is being required to produce the RPI by the Government because they want it.
Chair: Or by Parliament.
Richard Alldritt: Yes. It is probably a detraction, but when RPI was written into lots of legislation many years ago, there is a view that what it meant was the Government’s preferred measure of inflation, that it didn’t need to be interpreted as the RPI at all. It was the only one and rather than write “the Government’s measure of inflation”, they wrote down the name of it and only subsequently were there alternatives created.
Q106 Dr Rupa Huq: You have commented before that some of your recommendations have not been taken on board for budgetary reasons. You said that the impact of recommendations since 2008 “seems modest when set against the resources devoted to making them”.
Richard Alldritt: Yes, only rather the other way round. A huge amount of work went into producing those assessment reports and the reports included, speaking from memory, at least 1,500 recommendations to Government Departments to make improvements. Then there was a report back from Departments on what they had done. For example, it would say that the quality of the statistical release, the published release, is not good enough. It is not giving users a clear enough steer on the strengths and weaknesses of the statistics, which is a point that Sir Charles Bean’s report makes as well. There is not enough of a steer to the user about strengths and weaknesses. They would make some changes to the statistical release in the following quarter or the one after that and then if you went back later it didn’t really look as if those improvements had been sustained.
You looked at all the work that went into the process of assessment and how much impact it had on changing culture and making improvements on the ground. There was a sense that, as a way of bringing about change, the assessment process was not it. The assessment process was probably necessary, a bit like the analogy I gave of inspecting a car factory just to make sure that health and safety and everything is in place, but assessment is probably not the way to drive change. You need the board to focus on a smaller range of bigger issues to drive change, not crawl over every single statistic.
The assessment process is in the legislation. In my view, the assessment process was designed by the drafters of the legislation to make sure that in some sense there was a fair look at every little set of statistics. As head of assessment, I never wanted to do that. I wanted to look at the ones we were concerned about, not work my way around every set of Scottish education statistics, one after the other. That didn’t seem to me the priority.
Q107 Dr Rupa Huq: What should the UK Statistics Authority do to structure and manage itself better within existing legislation?
Richard Alldritt: I have given a flavour of what I think the answer to that should be. I did raise this as a concept while I was still in post, which is that it needs to identify what the big issues are. It almost doesn’t matter how you define the big issues because you do find the same ones come out of the woodwork. There is the census, RPI, health data, the way school education statistics are published. There is a set of concerns and they should be refining those, because all of those are too broad. The adequacy of economic statistics is too broad: which economic statistics are we really concerned about at this time and what is the nature of what should be done about them?
In my view, it should be doing this publicly and at that point you have an agenda of a dozen or 16—it usually comes out at about that when you do this as an exercise—big issues in official statistics that are live at this time that require attention. The very considerable abilities of the non-executive board members should be focused on taking those forward. That is very different from how it works now. How it works now is the Office for National Statistics very largely decides what are the issues facing it that it wants the board to discuss in private.
Q108 Chair: It is a user-led organisation?
Richard Alldritt: De facto, yes, in that the Office for National Statistics can come up with more than enough management concerns to occupy the board.
Q109 Dr Rupa Huq: Are there any other changes to the legislation that you would see as necessary or desirable?
Richard Alldritt: There is the point about the RPI being all covered in the legislation. I would like to see that taken out. There is a very confused passage in the legislation about national statistics and it is a term we have not really discussed. It should simply mean compliant with the code of practice. It has no other meaning but the Act is all over the place on that. The Act talks at some length about the assessment process and says not one word about looking at the big issues. I think there is a little paragraph in the Act that says something like “and should monitor other issues”. To my mind, the Act is all out of balance. The Act should be requiring the board to get a grip on what the big issues in official statistics are and do so very openly. But the Act does not need to be changed because I do believe it is broad enough to accommodate working in that way.
Q110 David Morris: The Office for Statistics Regulation was created in 2016, led by the head of assessment. What difference do you think this has made?
Richard Alldritt: I am not well placed on that recent history because since I retired from the civil service I have mainly worked in other countries and not the UK. My concern about it is that it felt like delegating a role that the non-executives should be playing a very big part in, which is external facing, talking to people about what the issues are, pinning down what needs to be addressed, down into the OSR. I have great respect for the staff who work there. I recruited and trained most of them. They are doing their level best but they are middle to junior level civil servants. We could be asking them to address issues like the way forward on the RPI. They can do what they can do. They can do the factory inspection work well.
One of those things is about the independence integrity issue. What comes out of that inspection work is, for example, that a particular Government Department may have started putting rather politically-loaded headlines on its statistical releases, “Good news on employment this month” following some policy or something. We would come down quite hard on that and say, “It is not for you to be putting out policy responses in the statistical releases”. I would not want to be negative or diminish the work of the staff in OSR carrying out inspection kind of work. It is good stuff and I certainly would not want it to stop, but it is not what I see as the main part of the regulatory function.
Q111 Dr Rupa Huq: How well would you say that UKSA engages with the users of statistics?
Richard Alldritt: Not as well as it could or should. I have come across this in my work in other countries. There is a misunderstanding about engagement with users. It should not be about just checking whether they are happy and they know where to find things and so on. It should be about exploring with the users what the beneficial use of statistics is. In other words, talk to local authorities about the use of official statistics in their targeting of local authority resource for helping the elderly or homeless or whatever it is. In what way are statistics being used? When I was chief statistician in Wales in a previous job, we used to talk a lot to the local authorities about how they were using statistics. What you find out is that sometimes they are using the wrong statistics or don’t understand them. They are using the wrong population statistics or they are not aware that there is another entire stream of data that would be much better for their internal debates.
The goal should always be to talk to users to find out about the use that is actually being made. I used to attend a series of meetings that included the retailers. It was absolutely to find what use they made of official data. It did not lead anywhere in particular in that case, but when I worked in the Home Office, I used to talk to the Association of British Insurers about their use of statistics. A point to make here is that although Government use of statistics is pre-eminent in certain respects, it is the top, if you add up all the others there is more happening outside Government in using data to take important decisions than happens inside Government in aggregate.
Q112 Dr Rupa Huq: I had a slight spat last week in Housing, Communities and Local Government questions about cuts and I was told, “Your council has £100 million of reserves” and the council is now contesting that. It has turned into a bit of an argy-bargy and I think the issue is over whether they were ring-fenced reserves. It feels that there are always different interpretations of these things.
Richard Alldritt: For example, if the reserves had been quoted per capita of population—it would not make a lot of sense but if they had been—you are then having official statistics being used as a denominator on an issue and you want to make sure that those things are being done correctly.
I will give you an example. GDP per capita was used by the EU as the basis for allocating regional funds. GDP per capita below country level at county level is a very misleading statistic because there are some counties, like Somerset where I now live, where you have low GDP per capita but a wealthy population because of the way GDP is measured. You have to know these things and you can only get into that discussion if you find out how the statistics are being used.
Q113 Dr Rupa Huq: There is an argument that reserves should not be used. Once they are used they can’t be used again, but there was a whole thing about what was earmarked and not and what the statutory responsibilities are—you have to pay an education budget.
My next question is: how should UKSA ensure that producers of statistics do hear the voice of the people outside their own organisation who are the end-user community?
Richard Alldritt: It is not that hard. I think you have to be open. There are lots of mechanisms you can use. You can organise events. Once the message gets out that you are approachable and will pay attention, people will approach you, but also by having a set of issues that you are going to be looking at. For example if we take immigration statistics, as soon as the authority was to openly announce that it was going to have a deep dive into the collection and presentation of migration statistics, people will come to it. That is when the discussion can really take place. Public events are often quite well attended if you approach them in the right way.
You do need to hear from a wide range of people and you need to sit down with the experts. There are experts in immigration statistics. There are university centres that study this stuff and they know it backwards and a lot of them would be delighted to be invited in to talk about the issues.
Q114 David Morris: Most of my next question has already been answered. How should the UK Statistics Authority ensure that users get the right advice about the interpretation of the use of the statistics?
Richard Alldritt: That is a very big question because, as I said, there are about 45 major producers of national statistics across four Administrations. The giving of the advice is not as difficult as it sounds. In a lot of cases the use of statistics falls into quite big camps. I tend to use the example of crime statistics because it illustrates the point. There are about only five uses of crime statistics. One is managing the police resource and another is informing the public about risk. It is a slightly nebulous one but it is right that the public should know about crime in their area and so on. Another is as a basis for crime prevention initiatives and at three you are beginning to run out of uses of crime statistics. You can tailor the advice around these generic uses quite easily. When you present the statistics you can say, “If your main concern is management of police resources, bear in mind the following five points”.
It is often presented to us, the regulatory people, that we can’t possibly do this because it will take us ages to find everybody who is using the statistics. You don’t need to. You only need to find out what the broad class is. I mentioned the use of population statistics for distributing resources. It is a very common use to divide up public money according to population shares. Rather than just publish the figures and say, “Here they are”, there is a great deal of scope for saying, “These are the ways these statistics are typically used and if you are a typical user, here are some points you need to bear in mind”.
Q115 Chair: The public attach huge importance to, for example, police-recorded crime. What more should the authority be doing to ensure that—a couple of years ago the big increases in recorded crime, which were due to recording changes, were not reported as such by the BBC. They were reported as fact.
Richard Alldritt: If you have a more open engagement with the news media, it is much easier to correct these things. It would be wrong for me to mention names but in the Statistics Commission days, home affairs editors of major newspapers would phone up and say, “I’m absolutely fed up with such and such a Department. I can’t get a word of sense out of them on a subject. Can you explain why the crime statistics are doing” whatever it was. Once you have developed the dialogue with the media, they are much more amenable to being told, “What you published yesterday was just rubbish. It just wasn’t focused. You ignored completely the big disclaimer at the bottom of the statistical release saying there has been a huge definitional change”.
Q116 Chair: That was pre-UKSA?
Richard Alldritt: That was pre-UKSA.
Q117 Chair: Did the relationship not develop?
Richard Alldritt: No.
Q118 Chair: Why not?
Richard Alldritt: For all practical purposes, the Statistics Commission sat outside the civil service. Although I was a civil servant, I was effectively loaned to the Commission. UKSA, as a Government Department, works within Government Department rules on dealing with the media.
Q119 Chair: What would happen if there was a separate regulator? Would it fulfil that function more effectively?
Richard Alldritt: It would have to try. It would have to reach some kind of an agreement about how it dealt with the media.
Q120 Chair: It should be independent. This word is bandied around a lot but seems to mean rather less than it should.
Richard Alldritt: It is a very slippery word. I do know what it means in relation to statistics but it does not mean what a lot of people think it means. It means essentially decisions about statistics should be non-partisan, non-politically partisan, but it does not mean that they should ignore the wishes of Ministers, for example, in terms of their need for data on a subject.
Q121 Chair: Who are the users of the net migration statistics?
Richard Alldritt: The net migration statistics are the basis of the Government’s target on migration, so everybody who is interested in migration, from whatever side of the fence they sit, is a user of the net migration figures.
Q122 Chair: But how much use is it?
Richard Alldritt: I don’t know how it has developed. I know that several years ago we were very concerned about it.
Q123 Chair: When we took evidence on the matter, the 95% confidence interval was 37,000 either side of 100,000.
Richard Alldritt: Yes, and that probably understates the unreliability.
Q124 Chair: What is the confidence interval on net EU migration or non-EU migration?
Richard Alldritt: I don’t know the figures. What I do know is that the measurement of emigration, whether that is people who have previously entered the country or British people moving abroad, is extremely unsatisfactory.
Q125 Chair: How well is UKSA regulating this statistic?
Richard Alldritt: I would imagine it is drawing attention to that issue. I certainly remember doing so.
Q126 Chair: What is the case for net migration being a national statistic?
Richard Alldritt: It is.
Chair: What is the case for it?
Richard Alldritt: I see, what is the case for it? It is the best that can be done. The demand for that statistic exists. You pass the test of the code of practice if you do the best that can be done with the sources that are available, hence my cautionary remark that compliance with the code of practice does not mean it is a good statistic. It just means it is being done as well as it reasonably can be. We did say quite a lot about that in one of the assessment reports.
Q127 Chair: The British Standards Institute attaches its kitemarks to things that are being done as reasonably as they can be.
Richard Alldritt: We didn’t call it a kitemark. It is a code of practice. It would be different if you could put in place easily a better source but if the decision has been taken not to record details of everybody leaving the country, trying to get an accurate figure to subtract from immigration—
Q128 Chair: But what should be happening in UKSA to promote more constructive thinking in Government about how to improve that statistic?
Richard Alldritt: What the UKSA board should be doing is talking to that community of experts and commissioning a report on that topic that would then be laid before the relevant parliamentary committees and, of course, Government. But Government’s willingness to change what it collects administratively is often very much influenced by whether a Select Committee is taking an active interest in it.
Q129 Chair: Under section 8(2) of the Act, it could be giving guidance and advice to persons responsible for official statistics. Where is the evidence that it is doing that?
Richard Alldritt: I think it has done. I can’t quote where the letters and things are but I do believe that it does give advice to Government Departments, not least through the assessment reports, that they really need to address the quality of the underlying data.
Q130 Chair: The National Statistician is a permanent secretary in the civil service in a non-ministerial Department and he is chief executive of the board and the head of assessment is only an adviser of whom the board has to take note. The weight is towards those who are producing the statistics not regulating them.
Richard Alldritt: The National Statistician is not chief executive of the UKSA board but is, as you say, a permanent secretary and, therefore, one of the collective board of the civil service.
Q131 Chair: It says under section 31(1), “The National Statistician is also to be the chief executive of the Board”.
Richard Alldritt: That is one of the cases where they have used the word “board” to mean something quite different from what they mean elsewhere. What they mean there is the organisation of 3,000 people we call the Office for National Statistics, so he is the chief executive of that.
Q132 Chair: But not of the UKSA?
Richard Alldritt: Under the legislation, perhaps he is. The UKSA board doesn’t have a chief executive.
Q133 Chair: Interestingly, the term UK Statistics Authority doesn’t appear in the Act and all the Act does is establish the board, and I guess it is that board.
Richard Alldritt: What it means by the board is the entire structure of 3,000 staff.
Chair: I am glad you explained how confusing that is.
Q134 Mr David Jones: What do you say, if anything, the UK could learn from other national statistical systems?
Richard Alldritt: Quite a lot, although not perhaps in the sense of copying them. I speak for the UK statistical system here. We are seen as a leading statistical system, very much one of the top end ones, and a lot of our procedures and processes are to some extent replicated. When you look at the bits that they have that we don’t have and so on, smaller countries often have much better user engagement. I did some work with Statistics New Zealand and it is quite easy for them to talk to all the big users in the country. They know who they are. Scale is very important. It gets a lot harder when you have a country with the structure of economy that the UK has to maintain the communication.
Other countries’ legislation is often clearer. It is not true that they tend to have centralised statistical systems, and we don’t. That is historical. The reality is now that most of the more significant statistical systems, like Sweden for example, have a very large number of bodies producing official statistics around the Government. The dialogue with other countries is a very important part. Your international standing statistically is quite important. The OECD, the United Nations, the IMF, the World Bank and so on are watching whether you are statistically credible or not and what arrangements you have in place. There are a lot of arrangements in other countries that we would be doubtful about. The way the National Statistician is appointed is better in the UK than it is in the vast majority of countries I have looked at.
You get a sense that we do a pretty good job, but particularly in developing the engagement with the user community and tackling the big difficult issues, it is not that people do it better than us but they do recognise the same sorts of problems. The problem of maintaining the integrity of statistical work in the policy departments—something we have not talked much about—education, health, justice, making sure that those statistics are being produced to the same sort of standards as in the central agency is a common problem now.
Q135 Dr Rupa Huq: The final big picture, Mystic Meg question: what do you think the statistical system of the future should look like?
Richard Alldritt: Bear in mind that the system has the three big devolved elements in the three devolved Administrations. They are important and they have a very substantial measure of independence, so we are looking at four national statistical systems to some degree. I think the idea of a central body in the form of the UK Statistics Authority keeping an eye on the whole thing is very important for UK coherence. I would not move statistical functions between organisations, for some of the reasons that were stated earlier, that statisticians do their best work when they are sitting near the people, near health managers and health statisticians. I would not move the functions around. There has been a bit of that. That was quite useful but I think it is probably more or less settled where it is now. It really is about the way in which the difficult issues are resolved. There is where I feel somehow there is more needed at the centre.
Q136 Dr Rupa Huq: What recommendations from this Committee would make life easier and move in that direction?
Richard Alldritt: My recommendation to this Committee would be to ask and keep asking for a clear statement of what the dozen really big issues facing the statistical service are and what they are going to do about them. That will give you a traction on progress. A lot of these issues are very long term—the measurement of immigration has been a problem since the 1970s and the crime statistics issues have been problem since the 1970s—but focusing on a reasonable tractable group of major challenges is much better than annual reporting of, “What have you done this year and what are you planning to do next year?”
I think it should be issues-driven and that could involve other parliamentary committees as well. You do see statistical issues coming up in health and justice and environment and so on and I think there is scope for a bit of co-ordination. I know the House of Commons Library provides a lot of support to the Select Committees in this area and you do a very good job, but as a committee system, you could get more support from a statistical regulator.
Q137 Chair: Why are you so reluctant to reopen the legislation?
Richard Alldritt: I am not massively reluctant. I think what will happen is that people will start piling issues into it that are not there now.
Q138 Chair: You don’t trust the process of legislating? It is understandable.
Richard Alldritt: The trouble is that statistics is such an open subject that you can pile in almost anything on the back of it to do with privacy, freedom of information, regulating social media providers and so on. I think the legislation could be amended without a major reopening. I am no expert on this but I remember looking at it at one time and there is quite a lot that can be done to tidy up legislation through secondary legislation now, I understand. As I say, I am not an expert.
Chair: We might take some advice on that, but thank you very much indeed. It has been a very interesting morning and a great start to our inquiry. I am grateful to you and to the other two witnesses.