Treasury Committee
Oral evidence: Review of the Office for National Statistics, HC 694
Thursday 7 January 2016
Ordered by the House of Commons to be published on 7 January 2016
Members present: Andrew Tyrie (Chair); Mark Garnier, Helen Goodman, George Kerevan
Questions 1-78
Examination of Witness
Witness: Professor Sir Charles Bean, Department of Economics, London School of Economics, gave evidence.
Q1 Chair: Good morning, Charles, or is it Sir Charlie?
Professor Sir Charles Bean: Last time I was here we had a debate about this.
Chair: What have you settled on now?
Professor Sir Charles Bean: You told me that whatever was decided at that meeting would go down for ever more, and it was Charlie.
Chair: I have obviously had a lapse in memory. What would you prefer?
Professor Sir Charles Bean: I am quite happy with either.
Q2 Chair: Can you first of all tell us: how bad is this problem with UK statistics? Let me ask you more directly. Let us take our GDP statistics; how would you describe the quality of those compared to other countries?
Professor Sir Charles Bean: The first thing to be said is that the trigger for the Chancellor and Matt Hancock commissioning my review was doubts about the productivity numbers, of which GDP, of course, is a key component.
Q3 Chair: Actually, the trigger was that this Committee began to get a bit worried about these statistics and started asking the Governor of the Bank in several evidence sessions.
Professor Sir Charles Bean: That is true. Of course, I was here.
Chair: And in the end, the Governor, I suspect—although he never told me this—went to have a word with the Chancellor and said, “I think you need to do something about it”. So you were fingered with the task of trying to sort it out. Is that a fair description of what has happened so far?
Professor Sir Charles Bean: It is certainly true that I was fingered. Going back to your question on the GDP numbers, one can recognise that there are some known deficiencies in these. For instance, they use something called single deflation methods, which are discussed in detail in the report. Several other countries have already moved to the more appropriate double deflation method, like the Americans, for instance. Potentially, using single deflation can be misleading when the structure of production is changing quite a lot, and in the UK, we have very out‑of‑date information now about the structure of production; the input‑output coefficients.
Q4 Chair: What I am trying to get at is that out there there is a wider public that might find this stuff quite dry and recondite. What we have to do is work out whether this is important. I think you would agree that it is extremely important that we have the best possible data; otherwise public policy becomes “rubbish in, rubbish out”. How can you possibly take good decisions if you are using ropey data? We have to make sure that we are best in class. Are we best in class for the GDP statistics?
Professor Sir Charles Bean: I certainly would not have said we are best in class.
Q5 Chair: All right, strip out Japan and just look at the major advanced Western economies without Japan, which has historically been a statistical disaster zone for a generation. How would you rate us?
Professor Sir Charles Bean: We are certainly short of the United States. As I have already said, you have this example of double deflation technology being used there. Some other countries also make much more use of administrative data than we do. That potentially allows for more timely and more accurate information. I also want to add something that is a little bit more forward‑looking about the GDP numbers, which is very important. What I have been talking about so far is just about where we are, but the nature of economies has been changing very markedly with the digital revolution.
Q6 Chair: We are behind the curve, are we not?
Professor Sir Charles Bean: All statistical authorities are probably behind the curve in measuring some of the—
Q7 Chair: I am asking you for a comparative assessment, compared to other countries. We are not in the vanguard of trying to keep up with this, are we?
Professor Sir Charles Bean: I would not have said we were the class leader in this. We certainly could do better; there is no doubt about that.
Q8 Chair: When Sir Charles tells us, “I would certainly say we are not a class leader”, I think it is fair to say that the wider public should interpret that as, “We are not doing very well”.
Professor Sir Charles Bean: We certainly could do better, and we should do better.
Q9 Chair: Can I just come onto a second issue, which I hope, at any rate, you are going to look at in your final report a bit more, because your interim report, which has some very interesting recommendations, scarcely addresses this. Am I right in thinking that the quality of UK statistics, and assessing them and keeping an overall qualitative eye on it, is the responsibility of the UK Statistics Authority, which was created in 2008? Why do we need your review? Do we not have a body that is supposed to be doing this already? Why have they not blown the whistle on these ropey stats a long time ago?
Professor Sir Charles Bean: You are absolutely right that the UK Statistics Authority does have a role in maintaining the quality of ONS statistics, and, indeed, it has commissioned various reviews and things like that—most recently, the Barker‑Ridgeway Review of the national accounts, and Paul Johnson’s review of price indices. They are examples of reviews commissioned on particular areas where there may be concern, and they also commission internal reviews where things have gone wrong, and so forth.
It is reasonable, though, for the key stakeholders—who have an interest in the quality of these statistics—if it is appropriate, to commission a review. It is worth stressing that the review I have been asked to do is not focused narrowly on deficiencies in a particular series, or anything like that; it is trying to be forward‑looking, to identify not just current statistical needs but also where future statistical needs might lie, and the ONS’s capabilities in delivering those, identifying gaps and where things need to change. The nature of the ONS does need to shift from being a factory that is oriented towards the processing of sample surveys to something that is more oriented to providing a service to users, which exploits other sources of information and which is much more proactive about thinking about the deficiencies of its statistics, and the consequences of change in the economy. That is quite a big change in the vision for the organisation and, in terms of future‑proofing UK economic statistics, it is important that that is the direction in which the institution heads.
Q10 Chair: That is a very interesting set of well‑put remarks, which is no less than I would expect from someone like you, Sir Charles. You have done a super job over many years in government, and we are very glad to have you doing this work, but my question was really much narrower: what is the point of the UK Statistics Authority if we need to wheel in Charlie Bean in order to sort out the stats? Is it not their statutory responsibility to have done this already?
Professor Sir Charles Bean: They certainly have a statutory responsibility to set their direction.
Q11 Chair: How do you mark their performance, out of 10?
Professor Sir Charles Bean: As yet, I have not really engaged in the detail of the governance process.
Chair: Could you give them a score in your final report, please, with a recommendation for improvement if you think it is less than 10?
Professor Sir Charles Bean: Certainly, if the governance processes are deficient or could be improved, that will be something that I expect to address under the third bullet of the terms of reference.
Q12 Chair: Getting on to the terms of reference, you are quite right; that is the third bullet, and I have them in front of me. The second one is the ONS, but the lead one, which I will read out, is: “Assess the UK’s future statistics needs in particular relating to the challenges of measuring the modern economy”. That ranges much more widely than your report, does it not?
Professor Sir Charles Bean: Yes, although the report covers quite a lot of areas.
Q13 Chair: Why is it that you have not looked at Government departments, and the way that they are collecting data?
Professor Sir Charles Bean: I have tended to focus on the key economic indicators and series that will be of use to users. That is true. Inevitably, I cannot cover every single possible economic statistic that somebody might be interested in and there are a lot of particular areas that were flagged in the call for evidence. I chose to focus on what I thought were the most important ones. Obviously there are the national accounts, and things like flow of funds, which is very important for analysing financial stability risks, and regional data—which has been the subject of past inquiries, like the Allsopp review, but frankly the regional data do fall far short of where one would like them to be. Because they were the ones that I thought were most important to policy‑makers, those were the ones I chose to focus on, but you will find in the report some brief discussion of land prices, house prices, and things like that.
Q14 Chair: Your final report is at an advanced stage already, is it not? After all, you are intending to get this out before the Budget.
Professor Sir Charles Bean: Yes, but I am intending in the final report to say more on some of the more forward‑looking elements. This includes implications of the digital economy and the consequences of the evaporation of national boundaries, as far as the location of business is concerned.
Q15 Chair: Sir Charles, I want to convey one general point. Opportunities do not come very often to sort an area out of this type. You have it; we need you to be robust in your findings and to be clear and unambiguous in your description of deficiencies where they exist. Otherwise, the bureaucratic waters will close over these things, shortly after your report is done. There will be some quiet changes to a few pieces of the bureaucratic jigsaw, and life will continue as before, with, as I put it—and I am sure you disagree with this, because it is far too crude—“rubbish in, rubbish out” statistics in a good number of areas that matter to policy‑makers and the British economy.
Professor Sir Charles Bean: I have a lot of sympathy with that and that is precisely why in the report I decided that instead of having 50 micro‑recommendations—if you look at a lot of past reviews, like the Atkinson Review, there are a lot of very small individual recommendations, which sometimes can miss the big points—what I want to do in the review, and what I have tried to do in the interim review is set out some really big themes, which are important for ONS management to take forward.
Chair: This morning, we need to examine a bit further the first part of your terms of reference, which I have just read out. We also need to look at the move to Newport, and maybe examine the regional data that are very important and that you raised a moment ago. Helen Goodman has some questions on the terms of reference and the implications for Government departments.
Q16 Helen Goodman: I have some questions about particular areas that highlight some of the problems. The first set of data that I am interested in is labour-market data. I am extremely concerned at the way this is presented; I am extremely concerned at the way the JSA numbers are treated as proxies for unemployment and I wonder whether you have looked at this. The reason I am concerned about this is because work at Glasgow University suggests that nowadays, a quarter of people are being sanctioned, and at the point at which they are sanctioned, they do not get JSA. Half of those people do not claim again; they are discouraged from claiming. You can see that there is considerable under‑reporting of unemployment by relying on that. It is fine to say that this is the claimant count—of course, that is perfectly okay—but there is a mis‑description in this. What do you think can be done about that?
Professor Sir Charles Bean: Most policy‑makers would not focus on the claimant count as the main measure of unemployment. They would use the Labour Force Survey measure of unemployment. It relies on asking a sample of people whether they are in work, whether they are looking for work, whether they are out of the labour force, and so forth.
Q17 Helen Goodman: Is that the ILO measure?
Professor Sir Charles Bean: It corresponds to the ILO measure, yes.
Helen Goodman: Is it what is reported as the ILO measure?
Professor Sir Charles Bean: Yes. It is worth saying that even the Labour Force Survey measure does not necessarily capture under‑utilisation of labour, because people may be working part‑time when they want to work full‑time. There has been quite a lot of work on this in recent years, both in the US and in this country. My former colleague on the Monetary Policy Committee, Danny Blanchflower, together with David Bell, has done quite a lot of work on developing alternative measures of unemployment that capture the under‑utilisation of labour, because workers are working part‑time when they want to work full‑time and so forth. Certainly, when I was at the Bank, we would look at those sorts of measures, and if you read the inflation report, you will see those sorts of measures reported.
Q18 Helen Goodman: I did ask the Bank about that and I was not wholly satisfied with the response I got, which is one of the reasons I am asking you about it now.
Professor Sir Charles Bean: Going back to your question about the claimant count, the one useful bit of information you get from the claimant count is that because it is one month ahead of the Labour Force Survey information you can use the change in the claimant count from one month to the next as a leading indicator of where you think the more appropriate measure, the Labour Force Survey, is going. But I really do not think any serious policy‑maker in the Bank or the Treasury would use the claimant count as a measure of something like the pressure of demand in the labour market.
Q19 Helen Goodman: That is very interesting. We have also seen a loss in the amount of data that we get at a constituency level. One of the things that we used to get was the vacancies‑to‑claimants ratio at a constituency level. This is only published at the national level. Was it a policy decision to stop providing this or did the ONS stop providing this because it was under‑resourced and no longer had the capacity to do it? Do you know?
Professor Sir Charles Bean: I could not give you chapter and verse on that. The one thing I would say about the vacancy data is that it had gradually become a less reliable indicator of the extent to which firms were looking to hire, because so much hiring does not take place through formal vacancy offices and things like that. This is an example of a generic point that actually links with your original question about the claimant count: often, you may have economic statistics series that start out as quite good measures of whatever you are interested in.
Helen Goodman: Of whatever it happens to be.
Professor Sir Charles Bean: But then over time, because of changes in the structure of the economy, they become less and less representative. That has been the case with the official vacancy data. Now, it may well be the case—but you would have to take this up with the ONS—that the reason they stopped collecting the information was simply that it was no longer sufficiently reliable to make it work using the resources to collect it. But that is, as I say, simply conjecture on my part.
Q20 Helen Goodman: Do you think it might be more reliable in some places than in other places?
Professor Sir Charles Bean: Quite possibly, yes.
Q21 Helen Goodman: That is quite significant. I want to ask you about the child poverty statistics now. In the 2010 Child Poverty Act, we had four measures of child poverty: relative poverty, absolute poverty, persistent poverty, and material deprivation. This Government have decided that they do not want to keep targeting those. That is fine; it is perfectly free to target whichever it thinks is right for its policy purposes. We will still get the measures of relative poverty and absolute poverty, and from those we can calculate persistent poverty. Do you know what has happened to material deprivation? I will just remind you about material deprivation. This was based on work, regularly updated by the Joseph Rowntree Foundation, on what a child needed—so a child needs to have some celebration on its birthday, a child needs to have a week’s holiday every year, and that kind of thing. That is obviously not being done in the ONS, but where is it being done and are we still getting that data?
Professor Sir Charles Bean: That is something I really could not tell you.
Q22 Helen Goodman: I am a bit bothered about this. Now we are not getting the vacancy ratio; now we are not getting material deprivation data. One of the things about statistics, which you know better than me—and I would be grateful if you could either reflect on this in the future, or tell me what you have done about this up until now—is that snapshots are all very interesting, but what we want is a long time series to see how things are changing. I do not know whether you have addressed that at all with them?
Professor Sir Charles Bean: No, that is something that is generically important. Lots of researchers, policy‑makers and so forth want to be able to track a series over time. It is not particularly the series that you are referring to, but there were certainly a lot of complaints from users during the call for evidence that you now could not get a long time series of the key economic indicators before 1997, because the ONS had not reconstructed the earlier historical data and people were having to do all sorts of splicing of time series to try and create something. There was a lot of demand for the ONS to rectify that, and to reinstate long time series for GDP, consumption, and all the usual national accounts indicators. That generic point will be true in lots of areas. You will want to look and see how things have evolved over time; possibly quite long periods of time, so it is important that if a statistical agency decides to change the definition of a series or use a different methodology, it tries to construct something that allows you to have a long, consistent background.
Q23 Helen Goodman: I want to ask you about the wellbeing statistics. This is new and there has been some discussion about whether the ONS is sufficiently intellectually curious and prepared to take on new ideas, one of which is the wellbeing statistics. I will not go into why that is a good idea because I am sure you know the arguments. Do you think that the questions in the British Household Survey will give us adequate data on this? Do you think the ONS has been sufficiently responsive to this debate?
Professor Sir Charles Bean: They have been sufficiently responsive within the constraints of the resources that they have. If you put resources into that it means they cannot use them for other things. Personally, I would not have argued that they should have been putting more effort into that. Inevitably, if you had more resources you could do more with it. I should say that my own view in this territory is that I am not a fan of trying to aggregate lots of indicators into some single, aggregate wellbeing indicator. It is more meaningful to use a scorecard approach and look at the range of individual elements.
Q24 Helen Goodman: Like the UNICEF model on childhood wellbeing, for example. They have 10.
Professor Sir Charles Bean: Yes. That is the way, for instance, the Bank of England, when I was there, would approach looking at financial stability. You would look at a range of indicators of the vulnerability of the system. The reason that it is sensible to do that is that different people may have different value judgements, and weight the different elements in different ways. You also want to be able to understand how the different elements are evolving. It is more useful to use something like a scorecard approach rather than thinking, “Let’s have a single alternative to GDP that measures wellbeing”.
Q25 Helen Goodman: The questions that I have asked you indicate that, for many users of statistics, there is a slight anxiety that the things that the Bank and the Treasury are interested in are too often the things that are being concentrated on, and the ONS is not doing enough broader work on other things. Do you think that is fair?
Professor Sir Charles Bean: I am not sure I do. It is worth saying that the national accounts, I think, are 13% of the ONS’s expenditure. It is a remarkably small component. It is true that the Bank and the Treasury are big users of macroeconomic statistics, and the ONS is conscious of that, but there is an awful lot of effort put into the production of other sorts of statistics. Inevitably, there has to be some balance struck between resources in one area and resources in another. That is almost inevitable, given that budgets are finite, but I do not take the view that there is excessive focus on economic statistics relative to social statistics.
Q26 Helen Goodman: You spoke a little earlier about the impact of the economy moving to more digital platforms. One of the big problems has been that the crime statistics have not measured ID theft online, so we have had the impression that crime has been plummeting while all this online crime has been absolutely ballooning. Now, you might say, “Well, that is a social thing”, but it has massive economic implications; I will not bother to spell them out, but it is absolutely clear that it does. Have you looked at that? Can you think about that? How we might gee that up?
Professor Sir Charles Bean: Again, this is an example of this generic point that the meaning of a statistic may change over time because of a development. The really important thing is that the statistics agency and users generally are conscious of that. One of the problems with the ONS in the past has been that it has been a bit locked in the mind‑set of “Turn the handle; here are the numbers”, and not asking itself enough, “What do these statistics mean now? Do they still mean what we used to think they mean?”, and explaining the deficiencies of the statistics to users. That is an important thing that the ONS needs to start doing that it does not do enough.
Q27 Helen Goodman: I agree with that, but we also want to have that thing that we thought we were getting before. Presumably now, as well as explaining the deficiency of what they are doing, they need to move into another area.
Professor Sir Charles Bean: Indeed, absolutely. One of the things that I have argued that they should do is more one‑off studies of the consequences of these sorts of changes. Certainly, the economic statistics that they produce are defined by international agreement—UN statistical standards as implemented by Eurostat—so they have to fit a template. Those templates evolve slowly.
What the ONS can do, or should do, is be much more agile and on the ball, thinking about where the template or the statistics are not keeping up with reality and doing one‑off studies—possibly in conjunction with outsiders, experts who have knowledge about these areas—to say, “Yes, this could be important”, or maybe, “It is not important”. Sometimes, they will find out that something that people think is important actually is second or third‑order importance. Those sorts of exploratory bits of research on statistics can be extremely useful in helping users to understand whether the statistics are likely to provide a reliable guide to what they are interested in.
Q28 Helen Goodman: I would now like to turn to the press release that the ONS released just before Christmas on the Wealth and Assets Survey, which is something that a lot of people are interested in. It is quite good. It disaggregates by region, property, finance, physical and pension wealth. I wanted to look to see how much of the nation’s wealth and assets was tied up in property in London and the south-east. That is not in here, but they must have it in the ONS. That raises two questions: one, what can we do to improve the quality of regional statistics? Two, when you get something like this, there is no way for the ordinary person to interrogate it. There is no “If you want more detail on this, email Joe Bloggs at www.ons.gov.uk”, or something. Do you think they could be a little bit more responsive, so if somebody wants to dig down in a bit more detail, they might be able to get that?
Professor Sir Charles Bean: Yes—not just could; they should. There are two dimensions to this. One, they ought to be better able themselves to dig down into the underlying data sources.
Helen Goodman: Yes. I do not want the data; I want my question answered.
Professor Sir Charles Bean: Absolutely, but part of it is about improving their own systems and capabilities to be able to do that easily. Secondly, a point that several users made to us is it is not always easy for them to access the microdata that underlies the sort of aggregate information that you are referring to there. In some cases, individual researchers might want to play with it themselves, in a suitably anonymised form. In other cases, it might be somebody like yourself who just basically wants a bit of information about the cut of the data in a different way.
Helen Goodman: Exactly.
Professor Sir Charles Bean: What one ideally wants to be able to move to is a statistical institute that can deliver that and enable users to cut the data in whatever way is most useful and relevant to them, and not be constrained by the particular framework, whether it is regional data fitting into NUTS 2, NUTS 3 boundaries, or sectoral information. You may want to cut it different ways, and all those sorts of things. That is something that the ONS should aspire to be able to provide. There may be limitations to this because there might be a pretty high demand for it, with people saying, “I want this, I want that”, and so forth, but it should be at least made easier than it is at the moment. Typically, what tends to happen is that the ONS is quite good at providing that sort of information to the big users, like the Bank and the Treasury, but smaller users tend to be further behind in the queue.
Helen Goodman: Lower in the hierarchy.
Professor Sir Charles Bean: Certainly, in the call for evidence, there was a little bit of a distinction between the big, key stakeholders, who always say, “The ONS may take their time, but they are very helpful in providing this information”, and some of the smaller users, who have said, “They are not very good at coming back and giving us the information. It takes a lot to get it out of them”. That may be finite resources, of course.
Q29 Helen Goodman: The House of Commons Library gave some evidence and, not surprisingly, they particularly picked up this loss of regional statistics. There used to be people embedded in the RDAs, and we have lost that. We do not have RDAs now, so how do you think we can re‑establish that link so we get better regional data? Again, the British economy is changing. What is going on in London and what is going on Durham, where my seat is, are worlds apart.
Professor Sir Charles Bean: The regional data issue is an important one, which is covered in the report in some depth. There was a push for expanding the provision of regional data with the Allsopp Review and, as you correctly say, more than run into the sand, it has almost gone backwards a bit. The problem with providing regional data is, if you are going to have data that is sufficiently accurate at a fine degree of regional disaggregation, which you might well want, you need very large samples. It is potentially very expensive if you are going to rely on using the conventional survey approach to collecting it. That is why the only practical way of unlocking better information about regional developments, regional economic activity and so forth is use of admin data.
If you think of the VAT database, that contains information on 1.8 million businesses. If they are single-establishment businesses, you know exactly what they are doing in that particular region. Similarly, the PAYE data gives you information on what somebody is earning in the particular place where they live. Effective use of the admin data potentially allows one to get a better handle on some of the regional questions and, in particular, it allows users to cut the geographical boundaries in different ways that are appropriate. They do not have to stay restricted to the traditional regional administrative boundaries.
Q30 Helen Goodman: That is true, but it also leaves it open to the Government to manipulate it more easily, does it not? They run the administrative systems. Let us take the example of PAYE: the Government put up the tax threshold, fewer people pay income tax, so it becomes less useful. That is one tiny little example.
Professor Sir Charles Bean: This is why it is important that when you use administrative data, the users should be critically aware of where there may be limitations and things left out. You have to use it intelligently, and there are plenty of examples of people coming a cropper with the use of administrative or Big Data, because they do not think enough about what is going on. Nevertheless, in terms of giving you a rich enough data set, it is the only really viable way forward, unless we are prepared to spend a lot of money on collecting regional information through surveys.
Helen Goodman: Okay. I sense some impatience.
Q31 George Kerevan: I wanted to continue with that theme. One of your grand issues is the lack of use of departmental admin data to create statistical series. You go further, on a something that I think will be close to the heart of most MPs, when you say that individuals and businesses are asked to provide the same data to multiple departments that then do not talk to each other. If it is commonsensical to use the departmental data more, why is it not happening?
Professor Sir Charles Bean: The processes that are supposedly in place to facilitate it, these information‑sharing orders, are cumbersome.
Q32 George Kerevan: Could you elucidate how that works? Who lays down the orders?
Professor Sir Charles Bean: Yes. Suppose the ONS wants to access some data that another department has. It has to say pretty specifically what data it wants, how it wants to use it, and so forth. The ONS does not have a right of access; it is basically up to the data owner to decide whether to provide access or not. Then, as I understand it, it is typically quite a legalistic process that is gone through.
Q33 George Kerevan: Who is the data owner? Is it the Chief Secretary?
Professor Sir Charles Bean: It will be the department. Implicitly, it is the Minister at the top, but it may be delegated down to a civil servant.
Q34 George Kerevan: Ministers could release this data, then. That would be one process.
Professor Sir Charles Bean: Certainly, if there was a push from Ministers at the top to say, “Be very open”, that would help. As I say, as I understand it and as I have been told, there is quite a cumbersome legal process. It takes typically six months minimum to go through this.
Q35 George Kerevan: Is that tied up with data protection?
Professor Sir Charles Bean: I do not think it is specifically tied up with data protection, because the Data Protection Act allows the transference of data for statistical purposes.
Q36 George Kerevan: So it could just be departments wanting to withhold the data.
Professor Sir Charles Bean: I do not know the exact history of how the legislation came to be framed the way it was. It was clearly intended to make it easier for departments to share information by laying down legal gateways, but the reality of it is that it has proved to be quite cumbersome. One of the problems is that, because the department that is requesting the data has to be quite specific about what it wants, how it is going to use it and so forth, it makes it quite difficult to experiment. Anybody who has used data knows that you have to do a bit of digging around, look at it, see if it makes sense, and try some other things. There is a certain amount of experimentation that goes on to see what works. The process militates against that, so that is one leg. Then you perhaps want to marry that with an innate caution on the part of the Civil Service sometimes about providing information. It is often easier to say “No” than “Yes”, frankly. Thirdly, because it is a bit of a pain getting this data, the ONS has shown less drive in getting access where it can. It has been doing some recently, and the current management team is trying to push more in this direction, particularly using HMRC data, the VAT data and so forth.
However, you have these three things together which means that the process is much more cumbersome as compared to some other countries. For instance, if you look at Scandinavian countries, virtually all of them rely almost exclusively on admin data for constructing their main economic statistics. The Canadians, who are often held up as being a bit of a paragon of statistics, have essentially an automatic right of access to this sort of information; Ireland is another country that does so. We are odd, in many ways, in having this rather cumbersome and awkward process. Having better, smoother access potentially enables you to economise on sample surveys, which reduces the burdens on businesses and households. It potentially enables you to generate more accurate and timely statistics. To me, it is almost a no‑brainer that one ought to be making better use of admin data. In fairness, I am not the first person who has made this point; it was made in the Pickford Review.
George Kerevan: Remind us of when that was.
Professor Sir Charles Bean: It was in the late 80s or early 90s.
Chair: Stephen Pickford.
Professor Sir Charles Bean: Yes. This is going a long way back.
Q37 George Kerevan: I might be tempting you too far, but from the ONS’s point of view, are there particular Government departments that are more recalcitrant in supplying data than others?
Professor Sir Charles Bean: You would have to ask them whether there are particular ones that they have found more difficult than others. Often, it may well be the case that if it is difficult lower down, you escalate things and maybe you can unlock it, but the whole process absorbs a lot of energy.
Q38 George Kerevan: Just for balance and fairness, if we did find a way of making access to departmental data easier, the issue of confidentiality would arise.
Professor Sir Charles Bean: Yes, absolutely.
George Kerevan: It is quite a large public interest issue in the UK, particularly with NHS data. From the point of view of the other countries you have mentioned and that you have looked at, are there specific ways in which you could build in processes for ensuring confidentiality?
Professor Sir Charles Bean: It is certainly very important that you do maintain confidentiality. One of the key points that is worth making is that the ONS does not have interest in individuals qua individuals, or individual businesses qua individual businesses. They are just units of observation, so potentially you can anonymise the information and that is certainly one thing that is important.
Systems need to be secure. The ONS has been carrying out the census for 200 years and it keeps personal information secret for, I think, 100 years. To my knowledge, there have not been any leaks of that information, so the ONS has quite a good record in this space. I do recognise that this is something where, if you are going to make greater use of the administrative data, you have to be very conscious of the need for confidentiality and that information that should not be in the public domain does not leak out and that systems are developed for that. There may also be cases where there are difficult questions about, “Actually, should this information be shared?” One of the ideas that is in the review is that you might need an ombudsman to deal with any difficult issues.
Q39 George Kerevan: I will move on, because there are some other points to consider. I wanted to add a rider to Helen’s questions about “regional statistics”. We are moving into a new era where the devolved Administrations will have more taxation powers, and that is leading to a realignment of the Barnett formula, in particular in Scotland at the moment; I think the process will flow on to the other devolved Administrations. The Scottish Government and the Treasury are in discussion about how to reformulate Barnett so that the residual Barnett consequentials can be calculated when the Scottish Parliament is raising more of its own resources in taxation.
There is a healthy debate, which is not yet resolved, as to the formula that will be used to recalibrate the Barnett residuals, but when it is operational, it will all depend on access to statistics from the ONS regarding future revenue streams to central government, GDP, population growth, and a whole raft of statistics. In anticipation of this, I am interested in your views as to how robust the ONS statistics are in being able to facilitate that kind of highly politically charged debate over the allocation of resources. It depends on forward estimates of where UK Government funding will grow, how taxation will grow, how these will grow in Scotland, and how you balance the two. Is the ONS prepared for this?
Professor Sir Charles Bean: There are two separate issues here. There is the actual data, the past data, and it is very important that the collection of that information is independent and is not subject to political interference and so forth. You seem to be raising a rather separate question, which is about projections of the future. For the most part, I would not want to be reliant on the ONS to be in the forecasting game. That is not what their role should be. The sort of exercise that you are suggesting may be better carried out by something like the Office for Budget Responsibility, and I would not want the ONS’s mission, which is basically to produce the necessary range of statistics to as high a quality as necessary, to be compromised by them getting into territory that involves them making predictions and judgments that other people could clearly take different views on, which has to be the case in forecasting.
Q40 George Kerevan: That is very reasonable. Thank you. Finally, I want to raise the question of the move to Newport, which had an impact on the staffing base. Perhaps you would just set the scene for us. A significant number of the London‑based staff did not go to Newport. How did that impact on the quality of the work at the ONS?
Professor Sir Charles Bean: The key thing is that it meant that they lost a significant cadre of experienced staff who decided not to go down there. It does appear that several of the errors in the construction of data—which are distinct from revisions to data, which occur because new information becomes available—are, in part, a reflection of a lack of experience and nous, and the ability to sense‑check data, which is something that comes with experience and so forth. While it might be difficult to directly trace a particular error to the loss of a particular staff member, it is very difficult not to come to the conclusion that the loss of such a large fraction of experienced staff probably played into the increase in those sorts of errors that has materialised in recent years.
One thing that I wanted to stress in the report is that this does not imply that this is a permanent deterioration. Some of the press reporting of my interim report was a bit off the mark, in implying that the move to Newport permanently lowered the quality of UK statistics. That certainly should not be the case, because over time you expect the new staff there to build up capabilities. Newport is likely to become a hub of statistical expertise, as links with local universities develop and so forth. This is very much a transitional cost.
Q41 George Kerevan: I understand that, but do you have an estimate of how long recovery will take?
Professor Sir Charles Bean: It is not easy to hire equivalently skilled and experienced staff. National income accounting is something you learn on the job, frankly; you might do one lecture on it during your university degree and that is about it. Really, you learn it on the job, which means that you are talking about 10 years or something like that before you have really adjusted to it. The move took place in 2007, and over the two or three years afterwards, so we are getting to the point now where those transitional costs should be starting to disappear. It is worth also saying that ONS management has recognised the issue, and is trying to increase the economic capability within the organisation so that there are more people around who can do this kind of sense‑checking—“Do the numbers look sensible?”— standing back from them, and not simply saying, “Well, we have turned the handle and this is what comes out. Take it or leave it”.
Q42 George Kerevan: Do you think 2020 would be a cut‑off date?
Professor Sir Charles Bean: Yes. There is bound to be always the odd error, but frankly, if there were quite as many, and the consequent erosion of trust in statistics on the part of users, in three or four years’ time, I would be worried. It would say that there is actually something deeper than just that loss of experienced staff associated with the move to Newport; there would something else wrong in the organisation that needed to be addressed.
Q43 George Kerevan: There was one point in your recommendations on which I needed a little bit of clarification. You recommended that the ONS should increase its London profile. What does that mean?
Professor Sir Charles Bean: That basically means having more presence in London, engaging with London‑based users. London is where a lot of major users are; it is not just the obvious stakeholders like the Bank and the Treasury, but others such as the financial institutions and their economists, and so forth. The principle of the appeal, really, is that the ONS needs to engage widely with users across the country; it cannot just sit down in Newport. It is worth stressing as well that I do not see this as an either/or. People sometimes think, “If you are saying that you need more of a London presence, that means moving people back from Newport and downgrading Newport”.
George Kerevan: That is what I was teasing you on.
Professor Sir Charles Bean: It would be a mistake to talk about moving operations from Newport back to London. That would lead to a whole lot of staff who have been acquiring experience down in Newport saying, “I cannot afford to live in London”, and leaving. You would just recreate the problem. However, it does make sense to have this operation in London that can engage with London‑based users.
Q44 George Kerevan: So a larger London office of the ONS, with staff there?
Professor Sir Charles Bean: It may mean some staff, particularly the more developmental research staff, spending some time in London. They do not have to be based there all the time; it is easy enough to get from one place to the other.
Q45 George Kerevan: But you are envisaging a facility with computers.
Professor Sir Charles Bean: Yes. One of the things that I am recommending is to have stronger links with outside users. In fact, just before Christmas, the ONS responded pretty quickly by announcing that they are setting up some fellowships to bring in outside academics who have interests in statistics, and they want to be able to interact with people. I could imagine that it would be highly desirable for some of the Bank of England economists to be seconded to the ONS for a year or two, but some of those people who might be willing to come to the ONS for the experience might not be quite so keen to relocate their families down to Newport. Having a base in London might make it easier to make that possible; they can still do trips down to Newport occasionally. We have to recognise that having this outpost in London, even though the main operation is in Newport, is a sensible way of increasing engagement with the user community.
Q46 George Kerevan: Would any senior staff be at the outpost?
Professor Sir Charles Bean: It is already the case that UKSA operates part of the time out of there, and the Chief Statistician has to interact from time to time with people like yourselves, Ministers and so forth. It represents a base that they can operate from. However, I do not think that it should be the case that they spend all their time up here. It would be a bad situation if Drummond Gate was the head office, and the operation was just down in Newport.
Q47 George Kerevan: You can see a logic here. I have seen it in some of the banking institutions, where you have a base in Edinburgh and you have a London office, then suddenly, 10 years on, it has moved on because of the pull of the City. All the senior staff are sitting in London and the backroom staff are sitting in Edinburgh.
Professor Sir Charles Bean: It would be very important for management not to let that happen. Senior management needs to be seen on the ground where the main operation is, in Newport. They have to spend a good part of their time there, which is what they do at the moment. On one of the visits that I did down to the ONS I met with groups of junior staff and I thought one of the things that they might complain about was not seeing enough of senior staff. I should say that even at the Bank of England, where my office happened to be on a different floor from where most of the economists were, the economists used to moan about not seeing the senior people enough. I did think that there might be a real issue with the ONS staff not seeing senior management enough, and I probed really hard in several different meetings, and it was not something that they thought was an issue. At the moment, certainly, it is working really quite well; senior management are visible.
Q48 Chair: That can be interpreted two ways. Either people are happy with the guidance that they are already getting, or people are very happy not to have anybody shaking up their quiet life out of view.
Professor Sir Charles Bean: No, I was not asking the question, “Do you want to see more of them?” I was actually asking them, “Do you see them there?” It was quite clear that they are very visible and interacting in the appropriate way with staff, and providing leadership to that staff.
Q49 Mark Garnier: Good morning. I just had a quickie on the revisions we have in terms of GDP. The ONS is criticised, as you know, for revising GDP on a fairly frequent basis, and yet when you look at it compared with other countries, it does it on a relatively infrequent basis. Do you think the criticism is unfair?
Professor Sir Charles Bean: Yes, some of it is unfair, and that is why I wanted to include some of the statistical information in this report to get across that the revision record of the ONS is not unusual. There is no doubt that users can sometimes get frustrated with revisions and sometimes there are some quite high‑profile ones; the much‑vaunted double/triple dip has now been revised away to just a single dip. There are examples in the past. If you go back to 1997/98, the initial releases of estimates of growth actually suggested that growth slowed quite sharply at the time of the Asia crisis, which might not seem terribly surprising. Current vintages of data suggest very little slowing, so there is quite a big change there in the picture of what was going on in the economy.
This is a reflection of the fact that the initial estimates are based on an incomplete set of information. The stuff is not properly balanced until two years down the road and, even then, you can argue that it is not properly balanced because the information used to do that—the so‑called supply use tables—is out of date. Then, you can have methodological changes further down the road. As a policy‑maker, you accept that there will be revisions and one of the things that I always wanted to stress to people that I was talking to, and one of the things that annoyed me about some of the press coverage of the data, was that they would treat the number as precise and fixed, and that is it, whereas it is an estimate.
Q50 Mark Garnier: To be fair, actually, politicians treat them as absolute indicators, and, depending on what side of the argument they are on, it can be positive or negative.
Professor Sir Charles Bean: Quite ridiculous attention is paid to what are really quite small variations in the numbers. As a policy‑maker, you have to accept that and you look at a range of indicators and try and form a picture that pulls them all together. These can be different indicators of the same thing, so we would look at business surveys as well as the GDP numbers and the information we are getting from our agents about activity, but then you would also be looking at what was happening in the labour market as well. You look at other things that you would expect to move in line with output, and then you would be asking yourself, “Does it all fit together? Is there something not quite right here?” Actually, a lot of what the Monetary Policy Committee would spend its time talking about in the first part of its meeting is puzzles in the data, where things do not quite fit together.
Q51 Mark Garnier: That is very interesting; it sounds like there is a job for us to do in Parliament and with the media to educate people that this is just a best guess, although the other side of this is that the MPC is absolutely packed with some of the pointiest‑headed boffins known to humanity. They do their best to understand all of this data, and even then, they do not necessarily get it right. I wanted to go through the way we collect data. There is this debate about double deflation/single deflation on value added. Currently, I think we only use that on electricity and agriculture, whereas almost every other developed country in the world uses it, and we are not planning to extend it until 2020. Is there a problem about it not being brought in for such a long time? The other thing I wanted to get you to answer, if you could develop this, is the extent to which this could be one of the reasons why we have this productivity conundrum. Is it possible that this type of statistical analysis could explain why American productivity is getting better—they do use this double deflation measure—and it is not working for us, where we do not use it. Could you expand more on this? It will probably be a long answer.
Professor Sir Charles Bean: Yes, and indeed, that is the purpose of the box in the review that discusses this issue. It also tries to do a back‑of‑the‑envelope calculation about how big these effects might be. I do not want to say that that is the explanation, or even a significant part of it. It certainly would not be an explanation for all of the productivity shortfall—that is too big—but it could explain part of it. As you correctly say, if you look at the US, the post‑crisis productivity performance is still slow relative to pre‑crisis growth—and there is a lot of discussion in US circles about why productivity growth might have slowed, and I will come back to that in a minute. The fact that we look even worse may, in part, reflect these measurement issues. The only way we will know that is by doing it properly, so to speak, and then seeing what difference it makes. However, there was an interesting study in the US—the exact reference is in the report—where they do do double deflation, but also work out what the numbers look like if you just do single deflation. Indeed, there is quantitatively significant discrepancy between the two.
Q52 Mark Garnier: Is it more volatile?
Professor Sir Charles Bean: Yes. So although it is something that sounds pretty arcane, it may be material. What you want to do is measure value added but, in fact, you are effectively using proxies of gross output, and if the structure of production is changing over time, you end up getting biased estimates.
The other thing I wanted to say in connection with the US slowdown and some of the literature in this space is that there is a debate about whether some of that might be attributable to new economy effects of the sort that are mentioned in the interim report, and the fact that certain sorts of activities have essentially been dis‑intermediated. People are booking their holidays on their smartphones, rather than doing it through a travel agent; there is the Uberisation of the economy, and all of those sorts of things. There are genuine questions about whether that sort of economic activity is properly being captured in the way we measure things.
Q53 Mark Garnier: Do you think it is? That is a very interesting point.
Professor Sir Charles Bean: There are some real issues here because the business models that are being used in many of these spaces do not rely on charging the user very often. You get advertisers to pay for space and things like that. There are genuine questions about how well some of that stuff is being picked up.
Q54 Chair: That would point to GDP being higher than the published measures. We are not capturing part of GDP.
Professor Sir Charles Bean: Yes, and there is an expansion in what you might think of as being home production. Smartphones and advances in IT have enabled users to do their own thing much more, whereas previously they would have paid somebody to do it. GDP, which is a measure of market activity, would correctly pick that up as a fall in GDP if people switch to doing it themselves rather than paying somebody else to do it. GDP is not a welfare measure, so I do not want to get into that, but if you think of it as in some sense picking up what is going on in the economy, you would be missing this. Our whole frame for thinking about statistics is not well‑framed for capturing those things.
Q55 Mark Garnier: Is there not a problem, though? The nature of the economy is so dynamic at the moment. We probably have no idea what type of job my 13‑year‑old child will have when he comes out of education. He could easily be very successful at something that has not yet been invented. How do the statistics organisations keep up with this incredible fluidity in the way people are economically active?
Professor Sir Charles Bean: Statistical institutions actually do not keep up very well. There is something about national statistical institutes; it is not quite a sociology, but because the economic statistics that they produce are to an internationally agreed template, that template inevitably lags developments in the economy and, in many ways, the framework that we use for the national accounts was born in the Great Depression, when Keynes developed basic macroeconomic ideas and so forth. We put a lot of effort into measuring manufacturing and relatively much less into services. That all reflects the fact that the frame that statistical agencies use to measure the economy evolves slowly, and it eventually responds to changes, but it is always lagging behind.
Q56 Mark Garnier: It is going to be a lot further behind, if you look at the Industrial Revolution—
Professor Sir Charles Bean: Absolutely. If the pace of change is relatively slow, it may not matter too much, but the faster the pace of change, the more out of date the frame of reference might be. That is why it makes sense for statistical institutes to become more agile and more questioning about how their statistics might be becoming less representative of what it is they are aspiring to measure, and why it makes sense to encourage them to do more one‑off studies with academics or expert users like Google or whatever that might be more aware of how some of these things are happening.
Q57 Mark Garnier: Is this where the Big Data concept comes in?
Professor Sir Charles Bean: It is certainly potentially useful.
Q58 Mark Garnier: One of my next questions was about Big Data. £145 million is spent on the ONS per year; of the 2,300 people who work at the ONS, just 11 are involved in Big Data. It seems that what you are discussing is an area where there is what is described as a sandpit, where you have clever ideas that you are trying to identify. Are we just not resourcing that part of the ONS as much as we should?
Professor Sir Charles Bean: It is certainly pretty small‑scale. There are about a dozen people working in this experimental sandpit, developing some of these data science techniques that you need for analysing Big Data. Just for comparison purposes, the Bank of England has an equivalent unit with 15 people.
Q59 Mark Garnier: 15? I think Google has hundreds, has it not, or Bloomberg?
Professor Sir Charles Bean: Bloomberg, which we went to talk to—and this is for the organisation as a whole, not just the London operation—said they had about 4,000 data scientists, coders, and data engineers. The figure might not be directly comparable.
Mark Garnier: Their turnover is $9 billion.
Professor Sir Charles Bean: It just gives you an idea of the difference in scale of the operation. I do not think there is any doubt that if the ONS is to make more use of both administrative data and, potentially, outside Big Data sources, it will need more capability in this territory. That is something that it needs to build quite consciously.
Q60 Mark Garnier: There is a lot of new technology coming in in terms of collecting this data. Of course, we have the traditional method—ultimately a bloke with a clipboard standing outside a supermarket, working out what the cost of a basket is—and now, of course, we have what I think is called web‑scraping, which sounds like a fairly antisocial habit. We have had one or two errors in this already, but do you think that there is a possibility that, with these new types of data gathering, we could end up with more errors; there could be more problems with going to people like Zoopla, Google and Twitter, which at the moment may quite happy to allow their tax data to be mined, but may, at some point in the future, try to stop it? Do you think it is a bit of an uncertain area? Again, of course, with everything changing so incredibly quickly, how do the technologies keep up with that?
Professor Sir Charles Bean: That is one of the reasons why the report puts more weight—at least at present—on exploiting the administrative data that the public sector already holds, rather than looking to make extensive use of lots of different, outside data sources. In many cases, for those outside Big Data sources, the information is acquired incidentally as a result of some other activity and then the question becomes, “Is it a reliable indicator of whatever it is the ONS might be trying to measure?” This is territory where you might very easily have a relationship at some time, and then that relationship changes, so you have to be careful about using that information.
There might be some sorts of information, like information on credit card transactions and things like that, where you could envisage it being embedded within the regular statistical production processes. However, I would see a potential role in another form, which is that you might be able to use some of this external information not as the prime source of information, but maybe for filling gaps that are there temporarily because information is slow to come in; it is called “nowcasting”. It would not then matter too much if that data ceased to be available in the future or it changed its nature, as long as you are conscious of it. The other place where I could see private data sources being potentially useful is where the ONS wants to undertake work in trying to assess whether some new area is important. It might not go into the business of setting up some new survey to investigate it, but it can exploit some information that is out there already that enables it to get a handle on the level of some sort of digital economy transaction, or something like that.
The key thing about this is that it means you need a statistical agency that is much more intelligent about how it is looking to use information. It has to be appropriately critical, and you are absolutely right to ask whether there is a danger of just naively using—and ending up misusing—the data because you do not really understand it. It is important to use it critically and not just turn the handle on it, but with that proviso, it is potentially a useful additional source of information to the ONS. Web‑scraping, which you mentioned, is quite a good example. As it happens, when I was on the MPC, the staff used to do a web scrape for prices, the elements of the RPI, on exactly the same day as the RPI collectors would collect the information.
Mark Garnier: In order to check it.
Professor Sir Charles Bean: To get a real‑time estimate of what the inflation numbers would be. They were reasonably accurate.
Q61 Mark Garnier: Just coming back to this Newport question and whether it is the best place to be, it sounds like you need some pretty intelligent free-thinkers, and it strikes me that the office would be better placed nearer a university, for example, which specialises in this type of innovative mathematics.
Professor Sir Charles Bean: That is why it is important that the ONS builds up relations, particularly with the local universities. Fortunately, where it is located, there are a lot of good universities—you have Swansea, Cardiff, Bristol, Bath, and several others—as well as some other Government departments down there, which have big elements of data. So I could see, over time, that region of the country becoming a hot spot of statistical expertise.
Q62 Mark Garnier: Fantastic. One last question, if I may: can I just drag you into the debate about Europe? As you know, it is something that is preoccupying quite a few people here at the moment. The Royal Statistical Society has come up with a submission saying that quite a large proportion of the ONS’s output of statistics is what is demanded by Europe, and in many cases this requirement is not necessarily that useful to us, but it is apparently useful to Europe. The net result of this is—given the fact that the budget of the ONS is not getting any bigger, but the requirements of these European statistics are getting bigger—what is useful to us is being compressed, and therefore there is less resource for statistics that we in this country find important. Is that a fair criticism?
Professor Sir Charles Bean: Frankly, I would say that it is probably not a first‑order issue.
Mark Garnier: Are they right?
Professor Sir Charles Bean: A large fraction of the statistics that the ONS is producing—95%, or that sort of number—is being produced according to internationally agreed templates. The reason for that is so that numbers are comparable across nations. It is valuable that statistical bodies get together, agree the methodologies that they are going to use for measuring GDP, what to do about underground economy transactions, and all of those sorts of things. They have agreed protocols, and the United Nations is the vehicle that does that. Eurostat is simply the body that is then implementing that in the European setting. It is worth remembering that Eurostat includes non‑EU countries in it; it is not particularly associated with the EU. There are some particular statistics that need to be produced for European Union purposes—things to do with Maastricht definitions of the deficit and things like that—but that is a relatively small fraction.
Q63 Mark Garnier: Could you give an order of magnitude? 5%? 10%?
Professor Sir Charles Bean: Less than that.
Mark Garnier: So not quite vanishingly small.
Professor Sir Charles Bean: Yes. You would not be saving a lot of resources. You would only save significant amounts of resources if you said you would stop taking part in Eurostat/United Nations agreed figures.
Q64 Mark Garnier: That would be counter‑intuitive. You want to be able to compare your economy with that of others.
Professor Sir Charles Bean: Indeed. We, the UK, want to be part of the international fraternity there, producing statistics that are internationally comparable. We should be at the forefront of encouraging all statistical agencies to develop statistics to keep in line with changes in the economy. I would suspect that even if we were outside the European Union—
Mark Garnier: We would be doing same thing.
Professor Sir Charles Bean: The ONS will want to be doing pretty much what it is doing at the moment. There might be some marginal differences, but I would be very surprised if there were any significant ones.
Q65 Mark Garnier: Just to be clear, the confusion that the Royal Statistical Society may have is that Eurostat is the organisation that effectively delivers what the United Nations requires in Europe, and so to call it a European directive is actually a misnomer.
Professor Sir Charles Bean: Yes. There are some directives; there is no doubt about that.
Mark Garnier: But 5% would be your maximum estimate.
Professor Sir Charles Bean: Yes.
Q66 Chair: So the Royal Statistical Society has this wrong.
Professor Sir Charles Bean: I think so, yes.
Q67 Mark Garnier: You also mentioned something that is quite important: our ability to influence what the global standards are like. Given the fact that we are involved with Eurostat, do you think that our ability to influence these global standards is enhanced by the fact that we are part of the EU or that it does not make much difference?
Professor Sir Charles Bean: Whether we are part of the EU or not is not really that material. The US and other countries in the world, like Australia and so forth, are equal players in this. Maybe at the margin it might make a difference, but at the first-order I really do not think it is that material.
Q68 Mark Garnier: I am just going to ask one question that I was overhearing your muttering to the Clerk, Chair, so I will ask a question that you were probably about to ask, if that is all right. Do you think the Royal Statistical Society actually knows what Eurostat does?
Chair: You are choking.
Professor Sir Charles Bean: Remember that the Royal Statistical Society is a number of different individuals; it is not one person, so I suspect some people have a better idea than others. Frankly, if you had asked me when I was at the Bank if I knew what Eurostat did, I would have probably not given a very accurate answer.
Q69 Chair: They have just expressed a pretty strong view on quite an important subject and you have told us it is wrong. Maybe you have got it wrong, but if they have got it wrong, it is a pretty poor show.
Professor Sir Charles Bean: Eurostat is important in setting down the templates. That is absolutely true. However, it is implementing things that are agreed on a broader international stage. This key point that Eurostat is not the same as the EU is relevant here.
Mark Garnier: That is really helpful.
Q70 Chair: I am going to end with a series of assertions based on your interim report and on some of the things you have said, and ask you to challenge those statements, but before I do that, I want to clarify just one second‑order point. You have said that sharing of administrative data is a no‑brainer. Would you extend that to HMRC’s data on individual taxpayers, or do you think they have a point when they argue that confidentiality trumps the sharing benefits?
Professor Sir Charles Bean: Yes, I would extend it to that. The key thing is that this this information is anonymised.
Q71 Chair: Their point is that sooner or later it will not be, or there will be a mistake and then it will become evident. They have been making that point persistently—not that if it was fully anonymised, they still would be against it, but that it puts taxpayers at risk, and once that is done there will be a decline in trust, which would have a deleterious effect on the yield.
Professor Sir Charles Bean: As I have said earlier, security is very important. However, it is worth stressing that other countries seem to manage this. I talked, in particular, to the head of the Danish statistical office about this and he stressed that they are very conscious and very rigorous about maintaining confidentiality, precisely because they are conscious that if there were a failure, they may well find that they were no longer able to gain access to that information; there would be public disquiet about it.
Chair: That is helpful.
Professor Sir Charles Bean: I absolutely see that this is an important issue, but I do not think that it is such that you say, “Because there is a small probability”—and I believe it should be made vanishingly small—
Chair: You mean not that it should be, but that it can be.
Professor Sir Charles Bean: That it can be, yes.
Chair: That is the key issue. Everyone is agreed that it should be. The problem is: can it be?
Professor Sir Charles Bean: Yes. HMRC data, in principle, could be leaked at the moment.
Q72 Chair: But at least they are controlling their own futures. That is their point.
Professor Sir Charles Bean: That may be their point, but as I have said before, the ONS has shown a remarkable ability to keep census information confidential over 200 years, so it has a pretty good track record.
Q73 Chair: Could I end with a few assertions, and then you can just see whether I have this wrong. First of all, statistics matter a lot. Poor statistics lead to poor public policy decisions and that, in the end, results in a loss of welfare.
Professor Sir Charles Bean: Yes, and not just public policy.
Chair: And others.
Professor Sir Charles Bean: From your point of view, yes, I do very much agree with that.
Q74 Chair: Secondly, it is clear that the UK lags other countries in the quality of its data.
Professor Sir Charles Bean: We are not best practice.
Q75 Chair: It is a phrase that you have used. You have used the word “lags”; you have written it down and published it, even if you are coming back to “not best practice”. Frankly, some of the statistics that we have been trying to use have yo‑yoed around so much that they have seemed scarcely fit for purpose at times. The ONS has fallen a long way short. You have described it as lacking intellectual curiosity and being unresponsive to consumers—we had an extensive exchange on that point—among a number of other serious weaknesses. Perhaps I am going further than you would want in this conclusion, although I seemed to sense it in one of the replies you gave me at the start of the hearing. We have a watchdog that is meant to be keeping an eye on this, the UK Statistics Authority created in 2008, and they seem to have been asleep while this deterioration was going on. That sounds like a lot going wrong, does it not, Sir Charles?
Professor Sir Charles Bean: It is certainly saying that there are gaps relative to where we should be. The one thing I would say on your last remark, UKSA being asleep at the wheel—which I suppose is the way you might want to put it—
Chair: We will put it like that, then.
Professor Sir Charles Bean: You should recognise that they have actually been quite active in instigating reviews where things go wrong. In many ways, there is more openness now about where there have been failings.
Q76 Chair: To mix a metaphor, is this not locking the stable door after the horse has bolted? I come back to the point that I began with: we had several years of hearings here, before this Committee, with people responding to questions about the odd statistics that we were working with by saying, “Yes, we are very worried and we are a bit concerned about the way the ONS is doing its job”. This was all in public session. The UK Statistics Authority was presumably capable of hearing that, and it has got so bad that the Governor and the Chancellor decided that they had better cook up the Bean inquiry. You should not have been necessary, should you? You should be having a quiet g&t somewhere, rather than having to sweat it out trying to find out what is wrong with UK statistics.
Professor Sir Charles Bean: Certainly, all is not well. If all had been well, I would not have been commissioned.
Chair: Very careful. You do not need to answer the question.
Professor Sir Charles Bean: No, but I think that is a fair reflection.
Q77 Chair: All is not well here, but it is well in the state of Denmark. That seems to be what you told us a moment ago. Is there anything that you have not had an opportunity to say that you would like to say, Sir Charles?
Professor Sir Charles Bean: I do not think so.
Chair: We will await your final report.
Professor Sir Charles Bean: I will say more in the final report, particularly on the governance questions, which I have not particularly engaged with. Some of your questions regarding UKSA are ones that I have not particularly thought about.
Q78 Chair: We are very concerned indeed about this area, in case you have not noticed; I am sure you have. Furthermore, we would also like you to bear in mind a number of the questions that were posed with respect to the first tier of your terms of reference, which certainly seems to be, if not relatively untouched, lacking the cover we would hope for a report of this type.
Professor Sir Charles Bean: Can I actually add a request of you? I know that normally at these Committees, it is you asking the witness.
Chair: We will grant a dispensation, and you can ask questions of us.
Professor Sir Charles Bean: We have moved on to governance questions a little bit and this is something I want to come to in the final report, and which I quite deliberately put to one side. I would welcome any thoughts that you and your colleagues on the Committee have about governance questions. It is something that I have been asking other users in the call for evidence and so forth, but it is something that parliamentarians have a particular interest in. From the tone of some of your remarks, I can produce some of your thoughts on the governance question, and equally I know that in the past, you have been on record as saying that you think that the Treasury maybe needs more of a role in the oversight of the production of economic statistics. If there are any particular views that you or your colleagues have, I would be very happy to take them on board in the process while I am working on that particular strand of the final report.
Chair: We are unlikely to say anything substantive formally before your final report, except possibly with respect to one issue, which is which part of Parliament should be looking at this and how Parliament can have a role. You have the substantive task of looking at this, and you have heard the reservations that we are putting, and we would like you to go away and think about them. We are very grateful to you for being prepared to allow yourself to be fingered, as I put it at the beginning of the hearing. I hope I speak on behalf of the rest of the Committee in saying that there are few people we could think of who are better suited and qualified for doing this quite difficult job, and thank you very much for doing it on behalf of the country.
Professor Sir Charles Bean: Thank you very much, Mr Chairman. I am very happy to.
Oral evidence: Review of the Office for National Statistics, HC 694 25