Treasury Committee
Oral evidence: Economic statistics, HC 682
Tuesday 4 February 2025
Ordered by the House of Commons to be published on 4 February 2025.
Members present: Dame Meg Hillier (Chair); Dame Harriett Baldwin; Rachel Blake; Chris Coghlan; Bobby Dean; John Glen; John Grady; Dame Siobhain McDonagh; Lola McEvoy; Dr Jeevun Sandher; Yuan Yang.
Public Administration and Constitutional Affairs Committee member also present: Sam Carling.
Questions 1 - 127
Witnesses
I: Professor Sir Ian Diamond, National Statistician, Office for National Statistics; Mary Gregory, Director of Population Statistics, Office for National Statistics; Richard Heys, Deputy Chief Economist, Office for National Statistics.
Examination of witnesses
Professor Sir Ian Diamond, Mary Gregory and Richard Heys.
Chair: Welcome to the Treasury Committee on Tuesday 4 February 2025. Today, we are looking at economic statistics and are pleased to have in front of us Professor Sir Ian Diamond, the national statistician and head of the Office for National Statistics. He is joined by Mary Gregory, who is the director of population statistics at the Office for National Statistics, and Richard Heys, who is the deputy chief economist at the Office for National Statistics.
We were keen to have this session because there have been some issues about the return rate of the labour force survey and the impact, therefore, that that has on forecasting by bodies such as the OBR, the Treasury, the Bank of England and others. Normally, the ONS is scrutinised by our sister Committee, the Public Administration and Constitutional Affairs Committee, and we are delighted to have Sam Carling MP join us today from that committee to aid us in our deliberations.
Q1 John Grady: Good morning, Sir Ian. The publication of the labour market data was suspended in October 2023, when the responses dropped to a level that made it unreliable. Did that come as a surprise to you?
Professor Sir Ian Diamond: It is worth saying up front how privileged we are to be able to give evidence this morning. We hope that this is not the last time, because it is fantastic for this Committee to show real interest in, if you like, the heart of what we do as statisticians.
Was it a surprise? Response rates to the labour force survey have been an issue, more or less, since it started. I am sure that all of you will remember the issues of unemployment statistics in the early 1980s, when the Government were producing a number of different statistics. The idea was, “The only way that we can really find out about this is by undertaking a survey”. That survey started off relatively short, but, even then, there were concerns about response rates, and they have been going on for a long time.
I would have to say as well, though, that, as with all surveys that go on for 40 years, they grow organically as different Departments want different information, and so I do believe that, by the time of the pandemic, the survey had become rather longer.
During the pandemic, we had to very quickly pivot and stop face-to-face interviewing. Clearly, during the pandemic, our interviewers could not go into people’s houses, sit down, have a cup of tea and interview them for 45 minutes. We then pivoted very quickly to telephone interviewing and managed, with a bit of weighting, to keep things going during the pandemic.
As we came out of the pandemic, we expected—and there is always the benefit of hindsight in this—that things would go back to normal. Treasury was taking away Covid funding, and people were getting out and about. What did not happen, though, was people returning to being happy about people coming into their houses to spend 45 minutes interviewing them. In the period post the pandemic, we were seeing a steady reduction in response rates.
As an aside, let me just say that response rates are, in and of themselves, not the critical issue here. The issue is bias that you are not able to control for. Of course, as your response rates go down, you increase the probability of bias. In August 2023, we did not see a sudden crash. We had seen a steady decline, but the issue that really hit us in August 2023 was that our sample that we obtained of young men was, frankly, too biased towards those who were unemployed.
That was why, when the analysts looked very carefully at the numbers that were coming out, had we just blindly gone and published those data, we would have made estimates that, frankly, had too much unemployment, and that would have just been wrong. People would have responded, and that would have been a very poor place to be. We saw this. Our analysts were able to look at it. We shared it with our stakeholders and we all agreed, and that is why we delayed that publication and immediately put into place a recovery process.
In a simple answer, we were already at a low level. We believed that we could manage it. We were managing issues of financial constraint and having the right number of interviewers, and also aiming to move into the new transformed labour force survey that I am sure we will go into a bit later. We acted quickly to respond to that. We did not publish data that were wrong.
Again, as a historical artefact, the Turnbull report into the issue in 1998, when the Office for National Statistics published average weekly earnings that were going in the wrong direction, said that the numbers were just published. Analysts and stakeholders were not engaged. We have learned from that, and it was also something that Sir Charles Bean raised in his report in 2017. Being really good with your quality assurance is very important.
John Grady: I am mindful that we have quite a lot of questions to cover today.
Professor Sir Ian Diamond: I am sorry. I just thought that you would be really interested.
Q2 John Grady: That is very helpful context, but, as a more direct, Aberdonian-style question, the board looks at this in October 2023, but the issue comes to light in August 2023. My first yes/no question would be this: before August 2023, were there material concerns within the ONS about this survey?
Professor Sir Ian Diamond: The simple answer to that is that we were monitoring very carefully what we knew was a low response rate. As I indicated my previous response, the curve was relatively flat and, therefore, we believed that, in the context of fiscal constraints and of recruitment and difficulties with response, we were managing it. Clearly, as soon as we did not, we acted.
Q3 John Grady: In August 2023, there is real concern.
Professor Sir Ian Diamond: For the data in 2023. For clarity, I was informed in mid-October by the then director of economic statistics, Darren Morgan. My request was, “Can we do something to use some other data?” His advice was, “No, we cannot”. That is when I informed the chair in my weekly bilateral, and it was discussed at the next meeting of the board, which was in October.
Q4 John Grady: There is concern among staff in August, September and October, but this comes to your awareness only in October. Have I understood that correctly?
Professor Sir Ian Diamond: I repeat that these were the data collected in August 2023. It is not real-time analysis. I was informed more or less as soon as the issue was really clear. You collect the data. It then has to be processed and cleaned. It takes a couple of months. That is why we report the August data in October.
Richard Heys: Can I give an example? I publish productivity statistics using the labour force survey. We published the April-June 2023 wave on 24 October. That gives you an idea of the time lag. That is the day after the LFS publication was suspended. In that publication, looking at that data, we were suppressing industries, agriculture, and mining and quarrying, because the responses in those sectors had fallen to a level where I manually intervened and said, “We will suppress that”, according to the code. That is something that occasionally happens. What you then expect to see is for the data to recover in subsequent periods to allow you restore the data series. In this instance, however, it was the start.
Q5 John Grady: In summary, what I take from that is that this really crystallised as an issue in October and not before.
Professor Sir Ian Diamond: Yes.
Q6 John Grady: Are you confident that the preceding data, having got to this position in October, was reliable?
Professor Sir Ian Diamond: We were watching them very carefully and were controlling for one or two biases. For example, during the pandemic, when we increased our sample size, we identified that we had recruited too many owner-occupiers and were putting weights in to control for that. The whole time, we are looking very carefully at the data. We weight data in order to make it representative at all times. We look, even now, for bias, and are confident that the trends that we were reporting were those that were happening.
Q7 John Grady: The data comes in in August. You get to know in October. In retrospect, should it have come to your attention earlier, given the materiality of what followed?
Professor Sir Ian Diamond: With the benefit of hindsight, I cannot really answer that question. You would need to ask Darren Morgan when he first became aware. Darren was a very experienced director of economic statistics and he came to me, as I understand it, as soon as he was absolutely clear that this was an issue.
Q8 Sam Carling: Good morning, Sir Ian. There are no issues with the labour force survey discussed in the minutes of the meetings at the UK Statistics Authority throughout 2023, up until the publication was suspended in October, and we just heard a bit about that. Could you just tell us a bit more about why this would not have been escalated to the board earlier than that?
Professor Sir Ian Diamond: The board receives a report at least quarterly, because the labour force survey was part of what we called our ARIES programme. That included seven projects and was a major ringfenced programme that reported, typically, quarterly.
The December 2022 meeting report was very clear about the challenges that the labour force survey was showing. My reports to the board, informal or formal, would have concerned those issues as they went along. There was not a formal report to the board from the labour force survey, but it would have been included in the quarterly reports of that ARIES programme. Subsequently, from early 2024, we have made a labour force survey monthly report to the board, as either a formal paper or an annex to my chief executive’s report.
Q9 Sam Carling: That is really helpful and covers a question that I was going to ask about lessons learned. Even then, when it was raised in October, it says in the board minutes that “members had not been made aware of the extent to which the LFS response had fallen recently”.
Professor Sir Ian Diamond: I respect the board saying that. As I indicated, it was a slow drop. The board members did raise that question, and we respect them, which is why we said, “We will absolutely make sure that we give you the response rates for the whole time”.
You mentioned lessons learned. With the benefit of hindsight, I would not have had this programme as part of that overall ARIES programme; I would have had it as a programme in its own right. That is what I have done since then, and we now have it as a programme in its own right.
Richard Heys: This was a long-term issue. The Bean review that has been mentioned gave us a list of areas of recommended improvement. Immediately post the Bean review, we prioritised those that related to the national accounts, working with our external stakeholders—double deflation, implementation of that administrative data into the national accounts, and new surveys. Essentially, as a second-wave project, as we came to further spending reviews, we looked at those economic statistics that had not been tackled in that first wave related to the national accounts.
We had been working, as the economic statistics group, to tackle this as part of a sustained programme of work, which is why the TLFS project had been brought through into the ARIES scheme and raised through our spending review bids with the board at that time, but that did not recognise the pace at which the ground changed underneath our feet with response rates post Covid. This was part of a long-term plan, but it overtook us before we could get to the TLFS, essentially.
Q10 Dr Sandher: Professor Diamond, can you talk me through your sampling and your weighting procedure for the LFS?
Professor Sir Ian Diamond: Do you mean the LFS or the transformed labour force survey?
Dr Sandher: Let us start with the LFS.
Professor Sir Ian Diamond: What happens is that we sample and send out to addresses saying, “You have been sampled. Please will you ring up and make an appointment for an interviewer to come?” A little under a third of respondents do that. If you say, “We will make an appointment”, clearly you get an appointment. For the remainder, we go to what we call knock to nudge, and we are doing 3,000 of those a week. People knock on the door and say, “You have received this letter. Will you take part?” We are finding very high levels of flat refusal compared with pre-pandemic issues. Indeed, our head of survey methodology tells me that it is taking two times the level to get an interview as it would have done pre-pandemic.
Once you are in the survey, you agree to take part for five quarters—although, to be clear, there is dropout between waves—and the reason for that is that, rightly, the Bank of England and other stakeholders need both stocks and flows, and you get those flows from the repeated waves. What that means, of course, is that, with that problem that I have highlighted to John Grady, although we immediately put into place more interviewers and more action, you do not get the full benefits of it for five quarters. That is why our March data this year, published in May, will be, I intend, really good labour market data.
Q11 Dr Sandher: Could you pick up on the weighting?
Professor Sir Ian Diamond: I apologise. We have a sample, but that sample will not necessarily be representative of the population, so we weight using the most recent population figures to bring a sample to a population estimate.
Q12 Dr Sandher: What are the criteria by which you weight? What are the cells that you are using?
Professor Sir Ian Diamond: We are using age and sex, but I indicated in my response to Sam Carling that, during the pandemic, we also brought other weights in. Particularly, we weighted by tenure type, and that is pretty important, because you then need to make sure that you reweight when you get new population figures. At a time of quite rapid net inward migration, those figures increase and you need to reweight by age and sex.
Q13 Dr Sandher: During Covid, face-to-face surveying declined by quite a lot.
Professor Sir Ian Diamond: It did not decline. We stopped it.
Q14 Dr Sandher: You stopped and then restarted it in October 2023. How did you change your weights during the Covid period?
Professor Sir Ian Diamond: That is what I have indicated to you.
Q15 Dr Sandher: You added tenure. The reason why I am asking is that, in the 2020 US presidential election, we know that the reason why the polls were off was because they had differential response rates, depending upon various characteristics. What I am really trying to get at is, when you changed the weight, whether the weight change was to reflect a differential response characteristic. In terms of bias, were you sure that you were not picking up a certain set of people who, for example, were working at home or were out of work, and not overweighting for reasons that were pandemic-related?
Professor Sir Ian Diamond: Let me be clear. That is a really good and the right question, and one that we worry about enormously. The indication that I have given you around tenure was because it was pretty clear that, as we tried to increase our sample size during a period when we had had to take face-to-face completely out and just go on, we had far too many people who were owner-occupiers. What that would mean is that there would then be bias, because owner-occupiers tend to be older and a different type of person.
What we did was weight by tenure. We looked very carefully at what the best thing to do would be, because, at the end of the day, if you move to the more disadvantaged areas of our country, you are perhaps more likely to perhaps to be in private rental, and you really do need to make sure that you are not biasing your data. While there are still biases, there are biases in every survey. I do not know of a survey that has ever been done that does not have some bias. You try very hard, through weighting, to eliminate that bias, and that is what we tried incredibly hard to do.
Richard Heys: It is worth saying that the challenge that we had was because of the change of mode. When you move from interviewer to telephone, you are assuming that every house has a landline or a telephone that you can get a number for. For short tenure, there is a lower propensity to have landlines, and a greater reliance on mobile phones. Therefore, part of the reason why we made the change in weights was to account for the impact of this change of mode.
Professor Sir Ian Diamond: It is worth saying that, ideally, we would love there to be a mobile phone directory or an email directory. We do not have those. All we have is an address list to write to people, and then landlines.
Q16 Dr Sandher: This is a nudge that you had during the pandemic. If I was a renter who did not have a landline—and lots of people around this table would not have a landline at home—you sent out letters. You nudged by telephone only to those with landlines, and so there was a whole set of people who you might have been missing. Did it ever occur to you to weight by the prevalence of a landline, or something that indicates that, since that seemed to quite an obvious bias in the sample?
Professor Sir Ian Diamond: We did try to experiment, under deputy national statistician Iain Bell, with knocking to nudge, where you did not try to have a conversation. It was a knock and then, “Can we ring you from the car?” for example. We tried an awful lot of new, innovative techniques to improve responses. We did not use the weighting process that you propose.
Q17 Dr Sandher: Can I also pick up on some of the groups that might have been harder to reach then and now? One of the big concerns in this country is the rise of those who are inactive due to ill health, and particularly young people. How are you accurately capturing young people in the labour force survey?
Professor Sir Ian Diamond: To be absolutely clear, we identified earlier than just about anyone else that increase in inactive young people aged 18 to 24, largely due to—“largely” sounds too big, but certainly the key reason being due to mental ill health. It was a presentation to Baroness Neville-Rolfe by the then director of economic statistics, Darren Morgan, that highlighted that earliest.
We have also been very early in highlighting issues of inactivity and the flight, if I may say so, post Covid from the labour force of a large number of people aged over 50. We have really got there and put a lot of work into making sure that our sample gets young people, and people from more disadvantaged areas. We knock to nudge much more in what we would say are disadvantaged areas than in advantaged areas.
Q18 Dr Sandher: You are capturing more people who are inactive due to ill health. The question is whether the sample is capturing them correctly and completely, because the bias could be in either direction.
Professor Sir Ian Diamond: We try very hard to recognise potential biases and to make sure that our samples are big enough to minimise those biases. That is why we knock to nudge much more in disadvantaged areas, where we are really trying to make sure that we get high numbers of people who are at different ages, and not end up with, shall we say, a particularly high number of people of a particular age and sex.
Q19 Dr Sandher: The thing that I worry about, which you caught on to, is the differential in non-response by characteristics that are not age, sex and tenure, which means that you might be over-capturing and under-capturing certain parts of the population. You point out that renters are much harder to reach. We have spoken a bit about how it is much harder to reach people who do not have landlines. There are specific parts of the population that might be missed out. Similarly, the young people who you are capturing, who are inactive due to ill health, may or may not be representative of the population as a whole.
Richard Heys: Those long-term, strategic issues are exactly what the TFLS is designed to target, which has an enhanced, significantly larger sample size. It is moving to a digital design, which is more accessible for some of those groups. I have forgotten my third point.
Professor Sir Ian Diamond: You are making some really good points, if I may say so, Dr Sandher.
Dr Sandher: That is very kind of you.
Professor Sir Ian Diamond: No, you are. We worry hugely about all of those issues. I would be delighted, if you had time some time, for you to come down, so I could take you through this in much greater detail.
Q20 Dr Sandher: I have never been to Newport, so off I go. Let me fire off a final question. Apart from more resources, what do you need to ensure that the sampling and weighting procedure is better, or is it just more cash?
Professor Sir Ian Diamond: In the main, it is. The other thing that is really important is the potential of administrative data to support, and that does require a culture of data-sharing in Government, which is something that so many people have commented on as not being in the right place that it should be. We need more resources.
We can talk about ways to improve response rates, and there is a long history of those. For example, when I was an undergraduate, I wrote essays that said that you could never give people incentives to fill in your questionnaire, because you could never pay them enough to make it worthwhile. Now, we do it routinely, because the research shows that, over time, that has changed. We are researching all the time to maximise response rates, but resource is what you need, as well as clear data-sharing.
Q21 Dr Sandher: We need more cash and better data. There is a linkage.
Professor Sir Ian Diamond: Yes.
Q22 Chair: I was interested in your points about landlines, and I thought it might be worth doing a quick litmus test in the room. Could colleagues put up their hand if they have a landline at home? It is a generational thing. I would just hazard, Sir Ian, that those of us who knock on doors partly for a living, or whose jobs rely on talking to our voters, have long known that it is very hard to collect landline telephone numbers. We have had to find ways of contacting people differently. Why has it taken the Office for National Statistics so long to catch up?
Professor Sir Ian Diamond: We did not use landlines until the pandemic to that extent.
Q23 Chair: Landlines replaced the door-knocking.
Professor Sir Ian Diamond: It was multiple knocks on doors.
Q24 Chair: Was it a surprise to you that people did not have landlines?
Professor Sir Ian Diamond: No, not at all, but when we hit the pandemic, we were faced with a challenge where we could not have people knocking on doors and had to do what we could through writing to people. I have to say that, as with many things that happened during the pandemic, the extent of civic duty rose and we got a lot more responses than we might have expected in normal times, and that is something worth thinking about.
Q25 Chair: That does not counterbalance the fact that you did not go in person.
Professor Sir Ian Diamond: No.
Chair: We can perhaps get a bit more into that later.
Q26 Yuan Yang: Sir Ian, you mentioned that, during the pandemic, you found high levels of flat refusal.
Professor Sir Ian Diamond: That is post-pandemic.
Q27 Yuan Yang: I was wondering if you could talk a bit more about what two times the level of effort means in terms of the ONS’s resource. Might this be a steady state, or is it changing?
Professor Sir Ian Diamond: The bottom line is that we have very much tried to increase the number of our interviewers. We have increased the number greatly since the pandemic. Because of the importance of the labour force survey, we transferred extra interviewers to it. That comes at a cost to some of the other surveys. At the same time, we are recruiting and increasing the number of overall interviewers who we have.
We also greatly increased the number of households that we contacted, from about 16,700 to about 28,000. It is about increasing the interviewers and the size of the overall survey, as well as the extent to which people go back and knock. That is, basically, what we have tried to do. We have also increased the incentive to encourage people to take part.
Q28 Yuan Yang: What is the incentive?
Professor Sir Ian Diamond: We have moved from £10 to £50.
Q29 Yuan Yang: Do you have any theories about the reason behind this increase in effort? Will it be the steady state level of effort or will people change?
Professor Sir Ian Diamond: I am a statistician, so I always come back to colleagues around the table. We are moving back to a normality, but are we going to move to a time when people will go back to feeling comfortable and willing to have someone in to talk to them for 45 minutes over a cup of tea? We are not seeing that at any length.
There are one or two other issues. While I have mentioned the difficulties of getting interviews in some of the more disadvantaged areas, not least because people are busy, our interviewers are also finding it harder in some of the more advantaged areas, where there are increasing numbers of gated communities that you cannot get into. Some of the flat refusals come because of Ring doorbells.
Chair: We all know about those.
Professor Sir Ian Diamond: Dame Meg and I had a conversation. These are the sorts of things that you will also face when you are canvassing. I am not trying to teach you to suck eggs, but these are things that are happening to us day by day. It is one of the reasons why we are so keen to transform some of our surveys to a more online basis in order to enable people to answer the questions when they want to and in a manner that they wish to. For example, “Can you answer the question on your mobile phone on the bus on the way home?”
Q30 Lola McEvoy: I just wanted to pick up on your confidence around the data on inactivity. There have been a lot of concerns across every quarter, not least for those in Government, around how to tackle this problem that has come out of your statistics. I just wanted you to explain a bit more your understanding of how accurate that data is around inactivity.
Professor Sir Ian Diamond: We are very confident. They hang together against all of the things that the Government are facing at the moment around benefits. Yesterday, I was in a meeting with the permanent secretary at the Department for Work and Pensions, and we were talking about this very subject. In the Department for Work and Pensions, the amount of money that is going towards older people is really high.
Let us just be clear. What we have reported and what we stay with is an increase in inactivity linked, not always but quite often, to mental ill health among young people. It is a bit more towards young women than young men, but we are not talking about enormous differences there.
We are also seeing quite a large flight from the labour force of the over-50s post pandemic. When we work with our colleagues at the Department of Health or, again, the Department for Work and Pensions, what we see is that, while there are multiple reasons, the three main ones would be cardiovascular disease, musculoskeletal and mental health. Sadly, if you get cardiovascular disease, you have it. There are many other reasons, but those are the three. I could send you a lecture that I gave on it a couple of weeks ago.
Q31 Lola McEvoy: I would be very interested. Before we move on to the next set of questions, forgive me, but how can you be so absolutely confident that this dataset is correct, when the larger points around the data that you have highlighted have been such a problem and we have been downgraded to “in development” as a statistical resource?
Professor Sir Ian Diamond: Please do not think that I am being complacent. I am not. I lie awake at night worrying about this the whole time. We work very carefully with the administrative data that we are able to get, and particularly data on people on payrolls, as well as our own data that come from employers, which includes workforce jobs. They hang up pretty well.
We also have an advisory committee on labour force, which includes some of the best academic labour force economists and statisticians, chaired by Professor Jonathan Portes, and also includes our stakeholders—the Treasury, the Bank and the OBR. We are being completely transparent and all believe that these numbers hang up, particularly when you hang them up against the data from DWP and what is happening in the health service. We are all very confident in those numbers.
Will we be more confident in the transformed data going forward? We are in a dual run and working on that, and there are reasons, which I am sure we might want to go into, for why we think that that is a good way to go.
Q32 Sam Carling: Just to touch on a couple of other things that Dr Sandher was speaking about, you said that, at the moment, you sample and then weight by age and sex. Presumably, you are limited from stratifying the sample in advance because you do not have enough data on the people living there until you have surveyed them. Is that a fair point?
Professor Sir Ian Diamond: That is right. You sample but you cannot guarantee that your responses will be exactly within the stratifying facts that you are interested in. You are constrained to the responses that you can get.
Q33 Sam Carling: That is a fair point. You have talked about how you are weighting afterwards. It does surprise me a little that, even before this became an issue, there was not more discussion of how you weighted for different demographic groups, because there are different ways of classifying people into demographic groups. You can argue about the specific ones that are used, but is that something that has been discussed previously?
Professor Sir Ian Diamond: Weighting in labour force surveys is a subject of enormous international debate. We are part of that debate. Richard has been part of the debate, and we are always looking for the best ways to stratify or post-stratify and to weight.
Q34 Sam Carling: In which case, as a follow-up question, are we in line with what a lot of other countries do?
Professor Sir Ian Diamond: Yes, 100%. We are always in conversation with our international colleagues. Indeed, I am meeting with my counterparts in Canada and Australia at 9.00 this evening, virtually, because that works for everybody. One of the things that the Australians and the Americans have done is to move to a much shorter questionnaire that is really focused just on the key labour force issues rather than the longer ones. That is where we have worked with them to look at exactly what they are doing.
Indeed, that has led to us now shortening our online questionnaires to six minutes per person. We really are only at the first wave of collecting the information that is absolutely critical to getting these labour force estimates. To go back to the example that I gave to Lola McEvoy, we want people to be able to, for example, fill in the questionnaire on their mobile phone on the bus on their way home, if that is the best way for them to be able to support us.
Q35 Sam Carling: That is helpful. Continuing on the theme of international collaboration, the decline in our response rate has been much worse than equivalent surveys in other countries. Do we have much indication as to why that is?
Professor Sir Ian Diamond: There are multiple reasons. Some would say that, although they have had a decline, it is mandatory in Australia. Some would say that our decline was because we had the longer questionnaire for a longer time. There are also cultural factors as to whether you respond to Government in a way that perhaps, over a period of time, we eased off.
I come back to the point that I made that we experienced, post Covid, a slower return to answering questions than we expected. That is why we have had to get more interviewers, and we are now back to a much higher level than previously. I have to say that, just last week, we held our first ever statistics assembly for stakeholders, which 600 people came to. At the session on labour force, one of the country’s leading labour statisticians said—and I was not in the room, so let me stress that this is as reported to me—that he feels that we are now where we were in 2008, which is good news.
Q36 Bobby Dean: Just to return to Ms McEvoy’s points when she was asking you about your confidence in your data around inactivity, you referred to the data backing up incapacity benefits, and your conversations with DWP. John Burn-Murdoch looked at this in the Financial Times and said that, in country after country, all the studies show that the alignment is more to do with changes to incentives and the stringency of benefits rules rather than any direct link to people’s health.
Furthermore, you are probably aware of the Resolution Foundation’s analysis that there has been a sharper bounce back from inactivity during the pandemic than your data shows. Can you explain to us the discrepancy between what your findings are and those of the Resolution Foundation in terms of their administrative data approach?
Professor Sir Ian Diamond: Let me just say that I really love John Burn‑Murdoch’s work. It is absolutely brilliant, but he is making a comment there. I do not comment. I look at numbers.
With regard to the Resolution Foundation, it published in, I think, November 2024. What they may or may not have been aware of was that, in December 2024, we were going to produce what I would call rebased estimates using the latest population data, which, because of net migration, were bigger. The Resolution Foundation had access to those data, entirely reasonably, and used them before we brought them out. When they produced their data, they were using a different population base to the one that, at that time, we were using. It is bit like trying to compare Torquay United with Manchester City. It is not a very sensible comparison.
The other thing that I would say about the Resolution Foundation work is that they make—and it is their decision—assumptions about self-employment and being out of work. We do not believe that a national statistics institute should be simply making bold assumptions about that. We respect what they have done and I do not have a problem with their calculations, but they are based on assumptions that we do not believe are right for national statistics institutes to make.
Q37 Bobby Dean: The John Burn-Murdoch thing was less commentary and more reference to research.
Professor Sir Ian Diamond: I love what he does, by the way.
Q38 Bobby Dean: Coming back to the Resolution Foundation, its central claim is that this stickiness of inactivity after the pandemic could be illusory, and you are still saying that, while you might be comparing apples with oranges, your data shows that it is not illusory and that we do still have a sticky inactivity problem.
Professor Sir Ian Diamond: We believe that there is inactivity among people over 50 and among young people, which we are continuing to see.
Q39 Chair: I just wanted to pick up on a couple of quick points, one of which is about safeguarding. Under the old regime, you would send someone round. They would sit for 45 minutes and have a cup of tea. We all have to deal with safeguarding issues. If you talk to other people who do home visits, it is relatively rare now for people to go on their own. Perhaps midwives do, but a lot of people are not lone workers now. Is that a consideration in home visits?
Professor Sir Ian Diamond: We do worry about safeguarding. Our interviewers all have emergency devices.
Q40 Chair: But they still go on their own, so you are not having to double up for that reason.
Professor Sir Ian Diamond: No.
Q41 Chair: A number of times, you have mentioned Darren Morgan, your former head of economic statistics, and said that we need to ask him those questions. He is not in front of us, and you are. He left the ONS about a year ago. Why did he leave?
Professor Sir Ian Diamond: He spoke to me—and I can report only what he told me—and said that he had decided that he could afford to retire. He chose to retire. He wanted to spend more time in retirement.
Q42 Chair: We do not need to have the whole personal conversation laid out. He retired, basically.
Professor Sir Ian Diamond: That is right. Let me be clear: I was very disappointed that he did, and I did try to persuade him not to, because he was very valued.
Q43 Chair: Has he been replaced?
Professor Sir Ian Diamond: Yes, by a person called Liz McKeown, who is doing a great job. She has led two things. She has led all of our media work, so you may have heard her on many occasions on the radio. She has also led all of our stakeholder work extremely brilliantly. That is not to underestimate the work that she does day in, day out, looking at the numbers.
Q44 Chair: Darren Morgan has another job, though.
Professor Sir Ian Diamond: As I understand it, he has a part-time job as the external affairs person for the Economic Statistics Centre of Excellence.
Q45 Chair: So he went of his own volition.
Professor Sir Ian Diamond: Yes, 100%.
Q46 Chair: There was no conversation between you and him. It is just that you have mentioned him a number of times.
Professor Sir Ian Diamond: I have mentioned him twice, because it seemed to me that it was quite important.
Q47 Chair: But he knew things that you did not, so I am just wondering if there is an implicit suggestion in there that you think that he should have been telling you things that he was not.
Professor Sir Ian Diamond: No, not at all.
Chair: I just wanted to be clear for the record, because he is not in the room to ask.
Professor Sir Ian Diamond: I was just making it very clear that I knew when he told me.
Q48 Chair: You are not suggesting that he should have told you sooner, just to be really clear.
Professor Sir Ian Diamond: No, not at all. Let me be very clear: I tried as hard as I could to persuade him not to go. I found him an excellent colleague.
Chair: As he was mentioned, I felt it was important to get clarity.
Professor Sir Ian Diamond: He is a really talented statistician and I was sorry that he left.
Q49 Chair: You have touched on this before, but, in retrospect, should you have been more curious about the day-to-day activity of the head of economics in terms of what they were finding?
Professor Sir Ian Diamond: That is a really good question. We are producing 700 outputs a year. I have what I believe to be a brilliant group of deputies, who have a brilliant group of directors. I am very confident in everyone.
Q50 Chair: Is there a process or system in place whereby, if one of your excellent deputies finds an issue with the datasets—and Mr Heys gave an example about how you suppress data at certain points—that triggers a report to you or to the board?
Professor Sir Ian Diamond: There is not an algorithm, but there is a judgment, and there have been many occasions. Mary Gregory, who is sitting next to me, came to me about some migration statistics in late October last year and said, “Ian, I need a meeting”. In terms of the conversation that I have already had with Dr Sandher about weighting by tenure, Jonathan Athow, who was then the deputy national statistician—
Q51 Chair: Just to be clear, there is a judgment issue that people make on a professional level, and you are confident that that is enough.
Professor Sir Ian Diamond: There is a judgment issue. I show interest; do not get me wrong. Richard and I meet and talk about productivity statistics on a regular basis, but I do rely on people raising issues.
Chair: I was just asking about the process.
Richard Heys: There is a system. There is proportionality. Sir Ian has a deputy head of the statistics profession. On productivity, I will publish several hundred lines of data to accompany an article going into division or region, and these sorts of things. If I have a detailed issue, I take it to the deputy head of profession, who reports back to Ian at a collective level. For more substantive conversations, they are escalated via the line management chain.
Professor Sir Ian Diamond: We have a culture whereby I start most of our team meetings with, “It is speak-up day, speak-up week and speak-up month”. We have a culture where you must feel comfortable to speak up. It is incredibly important.
Q52 Chair: That is very helpful. Can I then move on to the transformed labour force survey? This has been incredibly delayed. What is critical here is that, up to April 2024, your updates were saying that the transition was still being planned for July of that year, and yet, in December that year, it got pushed back to 2027. I know that you had the external reviews, but can you explain, in short order, the reasons for that delay?
Professor Sir Ian Diamond: I will say a couple of things very quickly. First, we still believed—and it was the original plan as a result, partly, of fiscal constraints—that we would dual run for two quarters, and that led to the decision that, if all went well, we would run for two quarters and transfer fully in July or September 2024.
A number of things happened. First, I commissioned a review of the—
Q53 Chair: That is the Chambers-Brown report.
Professor Sir Ian Diamond: Yes, the Chambers-Brown review. They did a really good review. They commended the design, which is important, but their recommendation, which I then received again from Huw Pill, the chief economist at the Bank of England, when I went for a conversation with him, was that they wanted four quarters of dual running. Because of the seasonality in the labour force, if you do not do four quarters, you cannot look at that seasonality so effectively.
That persuaded me and my colleagues about listening to stakeholders, and what is the point in getting external review if you do not take it seriously?
Q54 Chair: Should you have had more input at an earlier design stage? In your minutes of November, all major milestones had been achieved since January 2023. A few months later, in February 2024, there were a number of challenges in relation to delivery, though not so much the shape of it. Then it was the external review that started the changes. Were you late in getting that input externally?
Professor Sir Ian Diamond: We believed that the survey and the design were going well. It is a really good design and has not really changed. The issue was—and you know it only when you go fully out into the field—that we had one or two challenges with some of the questions. Most notably, we initially had some challenges with self-employment. We have also had some challenges with disability.
Most importantly, we have had some challenges with occupation. The reason for that is that, as Richard will be able to tell you, there are people who require what we call a six-digit standard industrial classification. That makes it harder, because we cannot just ask you what you do and you write down, “I am a politician”. That would get us one box, but we want to know, “Are you the Chair of a Select Committee? Are you the Prime Minister?” It becomes much harder, particularly with, for example, nurses. You need to ask more than that.
Q55 Chair: You have been doing these surveys in a different form, so this is not something that is new to you.
Professor Sir Ian Diamond: We thought that we could do this better. What we found was that, to get the really detailed six-digit industrial classification, we were going to need new methods and that our plans did not work as well as we were expecting them to.
Q56 Chair: Thank you for sending us the lessons learned information, which says that you “inherited a complex set of user requirements which were not always well articulated and understood” and “the scope of the project lacked clarity”. I am just puzzled as to why the scope was off when you have been doing this data collection, albeit in a different form. What was the difference? What went wrong between moving from paper to the digital transformation?
Professor Sir Ian Diamond: Fundamentally, however you trial things—and we did many focus groups to get the answers as well as we could—when you go into the field, you quite often find some issues and, as I said, we found that with disability.
The occupation one was perhaps not as defined as we wanted it to be, and that is why we have worked really hard. Rather than just trying to get ever more complex dropdowns, we have started to use artificial intelligence, and that is working very well in trials. We also took the opportunity to shorten the questionnaire even more, and that is working very well.
Q57 Chair: That is the one that can be done in six minutes on the bus.
Professor Sir Ian Diamond: Yes, that is right. In summary, it is not ideal that things do not work right first time, but, when you do surveys, they rarely do. It is only when you go into the field that you absolutely see what the challenges are.
I would say that the reports that I am getting at the moment—and I look at them very carefully, and I am leaning in a lot—are looking pretty good. We are going to take a decision on the future of the transformed labour force survey on 3 March, and I am hoping that that will include us publishing a paper showing some of these results.
Q58 Chair: Will that paper be published on 3 March?
Professor Sir Ian Diamond: 3 March will be when there is a go/no go decision. It is a team decision.
Q59 Chair: Does that mean a board-level decision?
Professor Sir Ian Diamond: It is a decision that will be taken by the executive on 3 March and then taken to the March board meeting. What we have done with our prices data, where we are also in a dual run, is to publish papers that show the numbers and explain the difference, because we want our stakeholders to know. That is what I am hoping to do.
Q60 Chair: You have parallel running. The current scheduled date, according to your December update, is 2027. Could you explain what that will mean?
Professor Sir Ian Diamond: What I said to you in that letter was that it is my intention, if at all possible, to do this earlier. I am still of that intention. Let me be clear about that.
Q61 Chair: But 2027 is your backstop.
Professor Sir Ian Diamond: It is a backstop. I would very much like the number to end with a 6. I would also like it to be in the earlier part of that year, but I am not going to do it until our stakeholders are onboard and we are completely confident. As I say, at the moment, I am encouraged by what I am seeing.
Q62 Chair: Just finally, before I pass on to colleagues, you have one form of survey that is long form, over five periods, and then a short form that takes six minutes on a bus.
Professor Sir Ian Diamond: It is still over five periods.
Q63 Chair: For a lay person, can you explain how you make sure that there is read-across from data from one set that has had problems through to the new regime, so that anyone in a policymaking position or using this data can be sure that they are able to get that trend that they so often need?
Professor Sir Ian Diamond: I would love to tell you that you just press a button and it tells you, but it is not like that.
Q64 Chair: It is not that, so what is it?
Professor Sir Ian Diamond: This is really hard, forensic statistics, looking at the numbers very carefully. I probably do not need to tell you, but, at the moment, the larger your sample, the smaller the confidence interval around it. Therefore, as you go down, you want to go down to lower levels of aggregation and to be looking very carefully, coming back to the points that were made about tenure.
It is real hand-to-hand, forensic analysis, looking at the two. That is why we produce a paper that explain the differences that we are seeing. What I can say is that the numbers will be different. There is no question about that.
Q65 Chair: Colleagues will go into this in a bit more detail.
Professor Sir Ian Diamond: If you do two surveys using different bases—
Q66 Chair: This is my point. Some organisations will have the capacity to do that kind of number-crunching, but a lot of people will not. They will just read and perhaps mis-compare statistics.
Professor Sir Ian Diamond: Our job is to bring accessibility to that, and that is why I want to produce a really thorough paper and want it to happen earlier rather than later, so that people get used to those numbers. It is exactly what we are doing on prices.
Q67 Chair: Just to summarise, you are still optimistic that you could achieve a 2026 switchover completely, so no parallel running. We know that you have 2027 as the backstop, but that means that you switch one off and move straight to the other.
Professor Sir Ian Diamond: My professional view is that I am super hopeful.
Richard Heys: Take the concept of actual hours worked. This is your usual hours, minus holiday, plus overtime and these sorts of things. With the move to the digital survey, we used to have to make adjustments for people who were on holiday, because they were not available to answer the face-to-face question. They can now answer the survey on their phone on the beach on the Caribbean, if they want to, which means that the number of people reporting that they are on holiday in the reference week has changed. To continue to have an aligned actual hours series, we are having to make sure that we have treated that effect and ensured that we have a consistent time series. That is just a little worked example of that type of forensic analysis.
Chair: Colleagues, depending on where their questioning goes, may come on to some of that in a moment.
Q68 Chris Coghlan: Let me just ask a follow-up on some of Mr Dean’s points, just because it is a fundamental issue about whether the UK has a chronic long-term health problem in the labour force. If I am right in my understanding, you are very confident that it does, not only because of the robustness of your methodology and data collection, but also because you are seeing it in incapacity benefit data and NHS data.
Arguably, you could discount the incapacity benefit data, according to John Burn-Murdoch, because there is potentially a correlation between changes in incapacity benefit data and long-term health issues. Are you saying that you cannot discount the NHS data because that proves that it is real?
Professor Sir Ian Diamond: We use the NHS data. There is no right answer here where we can just say, “There is the answer”. That is why we look at all other sources of information that we can. In consistent surveys—remember that it is not one survey—across different waves, with different people and large sample sizes, we are continually seeing this as a factor of the post-Covid years in our country. We have estimated that the exit from the workforce of the over-50s is approaching 500,000 people. I have not seen any information that leads me to discount that.
Q69 Lola McEvoy: Members of the Monetary Policy Committee have highlighted to us that seasonally adjusted data on flows in and out of employment are still suspended due to problems with weighting. Is that still the case? Does that not affect what we are talking about on this point of analysing inactivity?
Professor Sir Ian Diamond: In terms of what we are doing, I would not say it if—
Q70 Lola McEvoy: That is what we are trying to get at, if you are saying something that is very different to what we have heard from everybody else.
Professor Sir Ian Diamond: You are right. We have not been looking at those so much, but we have been able to look very carefully at the stocks over time.
Q71 Chris Coghlan: By “NHS data” what exactly do you mean? Do you mean that, if I am applying for incapacity benefits, I then get my doctor to sign me off and that, arguably, it is the same thing, or do you mean more than that?
Professor Sir Ian Diamond: I am using DHSC data that were provided to me for a lecture that I was giving. As I have indicated to you, I am very happy to let you have a copy of those data.
Q72 Chris Coghlan: That would be helpful. Thank you. In your letter to the Committee, you highlighted that budgetary constraints have been an issue. To what extent did budgetary constraints contribute to the problems with the labour force survey?
Professor Sir Ian Diamond: We are not going to hide from that. Let me be clear that, in 2024, when we asked the Treasury for some more money, it gave it to us, but we had a number of challenges. The first was that 40% of our budget was ringfenced. The second was that we then had to make £34 million of efficiencies.
Q73 Chair: What percentage of your budget was that? If you cannot tell us now, you can write to us.
Professor Sir Ian Diamond: Can I send you a letter?
Chair: I should be able to do the maths quickly, but I do not have your full budget in front of me.
Professor Sir Ian Diamond: The third issue was that, during these years, we had a number of inflationary pressures. Most notably, you may recall that colleagues were given a £1,500 award, which cost us £9 million. Pay rises for the last two years have been above the 2% per annum that we had in our 2021 spending review agreement, and we had to find those. Therefore, the amount of money that we had lost has led to £30 million to £40 million of savings. That has to be found somewhere. We are reducing the number of outputs that we do. There are no outputs that have no stakeholders, so a number of stakeholders are pretty unhappy about that, but I have been very clear with all of my colleagues that we are not going to reduce on quality. Quality is all that should matter to us.
Q74 Chris Coghlan: What exactly are those ringfenced budgets that have now been released?
Professor Sir Ian Diamond: We had a ringfenced budget for some of our transformed economic statistics, a ringfenced budget around the future of population and migration statistics, and a ringfenced budget to produce an integrated data service.
Richard Heys: As well as public service productivity and overseas development.
Q75 Chair: Who decides on the ringfencing—you or the Government?
Professor Sir Ian Diamond: The Treasury does.
Q76 Chris Coghlan: Is that now being released?
Professor Sir Ian Diamond: To give the Treasury its due, the budget for overseas development is relatively small. For example, the chief statistician of Côte d’Ivoire is in the offices today. That is a relatively small amount of money. We still have small ringfences around the integrated data service and around the future of population statistics, but it is really good that we have lost the great majority of those ringfences. In the business planning exercise that we are just finishing for 2025-26, whereas it is tight for everybody, it is not tight for surveys. We are increasing, as we have to, the number of interviewers. I wrote to you, Dame Meg, about the number of interviewers I was expecting by the spring.
Chair: Yes, 150.
Professor Sir Ian Diamond: Mike Keoghan, who is our deputy, did a review for us last summer. He estimates that, to get great statistics now across all of our surveys, we would need 1,500 interviewers. We do not have the money for that now, but we are increasing towards 1,000 as we move forward. We are spending much more money on getting interviewers.
Q77 Chris Coghlan: In 2016, Sir Charles Bean found that the move from London to Newport had a detrimental effect on quality due to staff turnover, which could be permanent. Is that still a factor today?
Professor Sir Ian Diamond: No, I do not think that it is. It was and is the case—and Sir Charles Bean was entirely right—that, when you transfer any organisation from place A to place B, a large number of people will not wish to transfer, for all kinds of domestic and professional reasons. That certainly had an impact on our economic statistics in the first part of the last decade, but I believe that we now have stability.
That does not mean that it is not difficult to recruit the very best people. There are a number of reasons for that. I do not need to tell you that there is a market within the civil service. For historical reasons, we are relatively low players. Alongside the Treasury, Defra and others, we opened an office in Darlington. It is a hugely successful office, and we are very proud to be there, but it is sometimes difficult. They do not have Departments in Darlington, so you just go in, and it is a brilliant idea.
Chair: We are visiting Darlington next week.
Professor Sir Ian Diamond: I just wanted to make it clear that you can sit at the same desk, move to a different Department and, effectively, do the same sort of job and get paid some thousands of pounds more. That cannot be right, in my view.
Q78 Chris Coghlan: You told me that you are over a third down in terms of the number of data collectors that you would like, and that you have had huge budgetary challenges, yet you are still very confident in terms of the robustness of the data that you are collecting.
Professor Sir Ian Diamond: We are trying super hard. As I have said to you, I am really confident around the labour force survey, because we have transferred large numbers of interviewers to build the numbers up there. We still have biases, which is why I very clearly said to you that those numbers will improve, because it is five quarters with our May publication, but we have really focused on getting the right number of interviewers.
Let me also say that, with regard to interviewers, they do not get paid a lot of money. They are also not tied to an office. They work from home. I have given you an example of where we compete against the civil service, but they are competing against the private sector. To be honest, if your job is to knock on doors and do it very well, if somebody is going to give you more money to knock on similar doors, it is not surprising when they leave.
We have done a number of things. We have tried to increase the grading that we can pay, by saying to experienced interviewers, “We can put you up a grade”. We also have a pay request in with the Treasury at the moment, which I am hopeful it will be positive about and which will enable us to incentivise and pay some more money to our interviewers, because, frankly, that is what we need to do to maintain the very best interviewers.
To come back to the question that you and Lola McEvoy raised, we are also seeing not just ill health as a reason for people leaving the labour force, and that impacts on us. Some of our very best interviewers tend to be people who may have had a career in the police or the military or something like that, and we are then offering part-time employment of 21 hours a week, which boosts their pension. They are brilliant interviewers, but we have observed that not all of them have come back to the labour force after the pandemic.
Q79 Bobby Dean: First of all, it is an interesting idea for politicians that we might knock on doors for the highest bidder.
In terms of some of the other surveys that are out there, we got news in December that a further 14 datasets from the annual population survey have been downgraded in terms of their status. Can you tell us what the timetable is for those APS-derived datasets?
Professor Sir Ian Diamond: The annual population survey feeds off the labour force survey. As we get the labour force survey back into shape, I would subsequently expect to have better estimates from the APS. The APS is also part of the transformed survey, and we will be producing results of those when we feel able to do so.
Q80 Bobby Dean: Just to be really clear, the problems being experienced there are exactly the same problems, with the same impacts, and you are trying to address them in the same way. If I could also clarify, you talked about getting more interviewers on to the labour force survey. That will also benefit this. It is not going to be a zero-sum game between different surveys.
Professor Sir Ian Diamond: Initially, when we first responded, it was. We were initially faced with headcount restructuring, and we were managing that as well. We have greatly increased the number of interviewers. We are moving to work with an agency to provide more interviewers to enable us to stabilise all of our surveys by getting more of them on to more surveys.
Richard Heys: Just to add to that, we often use the APS for some of the wider statistics such as wellbeing. Because we have established principles that surveys that ask those questions ask them in the same way, we are able to work with other Departments to find other datasets to combine and test whether we can make those deliverables without needing to rely just on the APS. That requires certain data standards to be hit, but it is something that we are exploring.
Professor Sir Ian Diamond: The APS will also be reweighted—and Mary and I were talking about this yesterday—using new subnational population projections, I hope, later this summer.
Q81 Bobby Dean: You mentioned—and I joked about it—competition with the private sector. Are you finding that the problems that you are experiencing in data collection are also happening in the private sector, or is public sector data collection particularly being affected because of the competition faced from the private sector?
Professor Sir Ian Diamond: Everybody is experiencing challenges. I met the chair of the National Centre for Social Research just a couple of weeks ago, and he was very concerned about response rates. We are clear that this a national and international challenge. As I indicated, though, we have a problem whereby we cannot just say, “We will increase the wage rates”. That presents us with real problems in recruitment, though I am hoping that we are overcoming them now.
Q82 Lola McEvoy: First, I would like to say that, as the MP for Darlington, I applaud your advocacy for higher wages.
Professor Sir Ian Diamond: Darlington is a lovely place, is it not?
Lola McEvoy: It is a wonderful place.
Professor Sir Ian Diamond: I had never been there before.
Q83 Lola McEvoy: Thank you for advocating for my hard-working constituents, who do deserve higher wages.
On the point of the impact of the problems with the LFS, I and the Committee are concerned about the wider impact that this has on other really crucial surveys that are used to work out things like GDP. We have the living costs and food survey, which is really important, and I know that you have made some changes to that. I just wanted you to take the Committee through what you can do to reassure us that that survey is as tight as possible.
Professor Sir Ian Diamond: In April 2024, we boosted the sample on the living costs and food survey to address the challenge of falling numbers of interviews. That was really successful in increasing them by one third between quarter 1 and quarter 3 of last year.
We have also prioritised extra field resource into that survey. It is also a really important survey to think through digital developments there—and we are—just because one of the things that it asks is, “What did you spend money on?” Those diaries take a bit of time, and there can be some memory issues over time, so we are looking at digital ways to improve them by you sending us your supermarket bill, or something. That is something that is absolutely at the beginning of experimentation, but certainly something that some of the commercial-sector companies are looking at in similar areas. We are trying to make sure that we are building developments as much as we can.
Q84 Lola McEvoy: But you are confident that, as it stands, that data is reliable.
Professor Sir Ian Diamond: Yes.
Lola McEvoy: In my previous life, I did quite a lot of work with the ASHE dataset, which I loved. It is a fantastic bit of work.
Chair: Could you explain what ASHE is?
Q85 Lola McEvoy: Sorry, it is the annual survey of hours and earnings. There were always concerns among those of us who worked with low-paid workers that they were sometimes underrepresented. Certainly, low-paid self-employed people were underrepresented, as were trades. Is ASHE a concern of yours in general? Has it been impacted by these changes? What are you doing to make sure that everybody is represented in that survey?
Professor Sir Ian Diamond: It is incredibly important. Inclusivity is absolutely critical. In fact, when I came into post, I focused on inclusivity, because I am with you on those issues.
We are confident in it, but we are, at the moment, really transforming the way that we bring the data in, using modern pipeline methods, improving the quality of it and, therefore, reducing the potential for any error. It is something that we use an enormous amount and that we are always concerned about.
It is also really important to be on top of who is responding. You only need a very big company not to respond, and you get those biases, so we are looking the whole time and imputing if that happens. We have a large client unit that works closely with the larger and more complex industries to make sure that we are not missing anything.
Richard Heys: One of the things that we have established over the last 10 years or so is the Economic Statistics Centre of Excellence, which gives us a researcher base who are continually looking for and exploring where issues are arising that we need to begin to think about. We have seen research through there on ASHE recently, and I know that the team have been engaging and reviewing that, so we do try to keep at the cutting edge of understanding where issues are arising.
Q86 Lola McEvoy: Are there any datasets out there that you would like to get your hands on? There are existing datasets. We talked about the private sector operating, and a bit about polling companies. If you had a magic wand, given that we have a mission-led Government that are trying to work cross-departmentally, are there any datasets that would transform the way that you can work if you were able to get hold of them?
Professor Sir Ian Diamond: I have mentioned the challenges of data-sharing. One of the things that our engagement with the public says is, “Why are you asking me this? You already know it”. I really believe that we should be in a position where, if you fill in your tax form, it tells you a few things and you just have to check them. I find that incredibly helpful. We believe that we should be able to get there.
What are the sorts of datasets that we would like? We would certainly like better DWP data. We have been working with them for some time. I am hopeful that those data will be able to be used later this year, although, to be absolutely honest, I might have said exactly the same last year. Those data are really important.
Coming back to the question that you and Chris Coghlan raised about NHS data, we commissioned, along with Chris Whitty, Professor Cathie Sudlow to do a review of health data-sharing. It is a brilliant review, which was released in November last year, and I am really pleased that NHS England is responding to it in a very positive way. That will enable us to get better access to NHS data in an absolutely privacy-enhanced and ethical way. It is only ever used for the public good, but those datasets would be good.
I would also point to some things that we are using. We are about to dual run using scanner data from almost every supermarket. That is billions of pieces of information, and bringing that in has been a challenge, but an enjoyable one. Another is around some of the issues that Richard has been looking at with regard to public sector productivity and how people are spending their time, and time-use surveys.
Richard Heys: One of the things that makes the LFS so valuable is that it gives you a way to get to the self-employed, which is a dataset that you cannot get from big firms, et cetera. HMRC’s Making Tax Digital programme will replace self-assessment for the self-employed with a digital record, which will be uploaded quarterly. That is for those earning more than £50,000 from April 2026, and more than £30,000 from April 2027. That will be a hugely useful administrative dataset to calibrate the self-employed data that we collect and to put alongside the PAYE records that give you what people have been paid as employees. That wider administrative data perspective then gives you a real opportunity to triangulate the surveys that we get from businesses and households.
Professor Sir Ian Diamond: A final thing is single digital sign-on, which DSIT is leading on. If done properly—and I should not use the word “if”, because it will—how good will that be?
Chair: Some of us have been knocking around Whitehall for a while, and “if” is always a good word to have with data and IT projects in Government, but that is the cynic in me.
Q87 Yuan Yang: I have some questions about the future of data-gathering. Sir Ian, you mentioned the increased cost of getting people to respond on the doorstep. Is the current balance between administrative and survey data the right one for the long run, or will there be a tendency towards more use of administrative data?
Professor Sir Ian Diamond: I am a total fan of a system where we use both administrative and survey data. I absolutely believe that we need to get more administrative data and to have it in a way that it flows regularly, but we should not believe for a minute that administrative data are perfect, clean and wonderful, because they are not. They need cleaning.
Will there be a need for surveys? Yes, 100%. For example, I would imagine that business sentiment is very important for you. I cannot take my mobile phone out, point it at you and know what your sentiment is on anything. Fundamentally, I need to ask you, and there are many things that we will always need to ask people, to get their views on, and to find out what they are doing. Therefore, a system is critical.
Richard Heys: The national accounts are produced by all countries in line with a manual that is called the System for National Accounts. This comes up for revision every 15 years. Sir Ian and I will be in New York to sign off the latest changes in March of this year. The new SNA 2025 revision will be a substantial step forward in that manual.
Within that is a chapter that looks at the development of something called labour accounts. At the moment, we use individual surveys, such as the labour force survey, to give us our best estimate from one source of these data. Labour accounts are designed to bring together multiple datasets to allow you to triangulate. In the same way that we balance between production output and income on the national accounts, we will be balancing between administrative, business and household data around our perspective of the labour market. The Australians have a model of this that they already run, and we can provide you some more information on that, if that would be helpful.
Q88 Yuan Yang: Do you feel, Sir Ian, that, given the budgetary pressures that you were describing, there will there be a tendency in the long run towards having fewer interviewers than you have right now? Do you see a tension there?
Professor Sir Ian Diamond: My view is that we will continue to make sure that we reduce the number of surveys to an optimal number, but we will still need interviewers. At the moment, I do not see us having fewer interviewers in the long run, until we can find ways to know what is going on without asking people. We will move things online and we will use administrative data, but I still believe that there is going to be a real need to talk to people, if not face-to-face then certainly by telephone interview.
Q89 Yuan Yang: Just to be clear, in terms of some of the datasets and some of the variables that you currently gather from survey data, is there no plan at present to move from using survey data for those variables to using administrative data?
Professor Sir Ian Diamond: Let me be clear. We are looking all the time at ways to use administrative sources. That was the point that I made.
Q90 Yuan Yang: What I am trying to ask is whether you are planning to use administrative data to replace some survey data sources.
Professor Sir Ian Diamond: At the moment, I would say to accompany sources. In other words, “We will tell you what we know, and then we have to ask you fewer questions”. I am not seeing enormous numbers of areas where we will totally stop doing things in economic statistics, but we need to boost them with administrative data.
Q91 Yuan Yang: You said that there are not an enormous number of areas, but are there some areas that you will stop?
Professor Sir Ian Diamond: There are some areas. We are already doing really good work using administrative data to improve our migration estimates and our population estimates. This year, for the first time, we will produce two set of population estimates, in June and December, that are totally administrative-based, which is great for local authorities. We are doing and thinking through this the whole time, but I am just not convinced that we will not be doing surveys in the future.
Richard Heys: It is to do partly with the sophistication of the data that people want. Bringing that into the national accounts gives us measures of output by firms, but, given the time lag after which we receive it, we still need to run the survey to get the fastest estimates out. Once we get that data, we get a far more sophisticated picture.
Professor Sir Ian Diamond: Administrative data are collected for administrative purposes and, therefore, do not always contain the information that you particularly want.
Q92 Yuan Yang: That is also my concern, which is why I wanted to know if there would be a long-term deterioration.
Professor Sir Ian Diamond: There was a consultation with HMRC a couple of years ago to get them to ask about occupation, but they were not minded at that time to do that.
Q93 Yuan Yang: Can I just move on to another exciting area? You mentioned the mandatory collection of survey data in Australia. The Lievesley review also mentioned this as an option. In your letter to us, you say that you would need support to enable that. Is that a positive resolution for you? Is mandatory data important?
Professor Sir Ian Diamond: It is a really good question, and the simple answer is that I do not believe that it is a magic bullet. The second issue is one of public sentiment. There are certainly those who say that Australia is fine, but it is a place where, for example, voting is mandatory. At the moment, we have a piece of work going on, led by our director of strategy, June Bowman, which is looking at all of the issues. That piece of work will be ready reasonably soon.
We will then consider whether we want to think through the legal issues. Our legal team are part of that review, but we will be looking to ask the question as to whether this is something that we would want to do. It is a fine balance, but one that could help with some surveys.
Q94 Yuan Yang: If the answer to that was yes, would you need more powers from Parliament?
Professor Sir Ian Diamond: Yes, 100%.
Q95 Yuan Yang: Have you already embarked on that discussion with Ministers?
Professor Sir Ian Diamond: No, we have not. Our first place was to get our own views and our board’s views, and to decide exactly what was needed. If we were to decide that we wished to move in that direction, I would ask immediately for a meeting with Minister Gould.
Q96 Yuan Yang: When do you think you will have an outcome from that review?
Professor Sir Ian Diamond: We have to make that view during this calendar year.
Q97 Yuan Yang: We will be very excited to hear about that. You mentioned the difficulty—and this is something that we also experience—of finding mobile phone numbers, emails and other forms of contact details that are not landlines. Are there additional powers that the ONS would need in order to access these directories?
Professor Sir Ian Diamond: I do not know the answer to that, because I do not know how to get a directory of mobile phones.
Chair: You might want to talk to some of the political parties. It is above our pay grade.
Professor Sir Ian Diamond: I will take that away and write to you, but I do not know the answer to it.
Chair: We will be keen to hear that.
Q98 Dame Harriett Baldwin: My questions are for Mary Gregory. Our constituents and public policymakers really need to have good information about net migration for the UK, and yet the former special adviser to the House of Commons Public Administration Committee said about the migration statistics that they were “all but meaningless”. Is that fair?
Professor Sir Ian Diamond: No.
Dame Harriett Baldwin: I was asking Mary.
Q99 Chair: Ms Gregory, perhaps you could elaborate.
Mary Gregory: Yes, I can. That is in the context of the revisions that we have seen recently, and I will explain a bit more why those have happened, but a really important thing that has been covered in some of the other contexts is that we are using administrative data now. That is really positive in so many ways.
We do not just use that blindly. We look at other sources as well. For example, if you wanted to look at the previous publication in figure 9, we compare our estimates based on administrative data with other sources, so you can look at visas granted. As you would expect, we are a bit below that and a bit above the national insurance numbers that have been allocated. That is exactly where you would expect net migration to be as a general trend, which gives us confidence on immigration.
There have been revisions. That is almost inevitable at this point, because we are developing a new approach. As you may well know, we used to use the international passenger survey for migration estimates. There is a lot of information about why there were some challenges using that, and the new approach is a much better way to measure. Not only is it better at the headline, but it gives us new information that we could not look at before, which is so important for policymaking.
We can look at students who have come in on student visas and see whether they move on to a graduate visa and how long they stay for. We can understand much more about the differences between nationalities. We know that, typically, if students come from China, they may well stay for less than a year or are much more likely to leave quite quickly, whereas, if they come from Nigeria, for example, they are much more likely to bring dependants and to stay for longer.
This is information that we did not have as clear a sight on when we were using the survey data, so there are a lot of benefits in what we are doing. There have been revisions. We think that we are nearly there, and we have made a lot of progress.
Q100 Dame Harriett Baldwin: You mentioned the revisions, some of which are vast and are certainly more than the size of a constituency. Are you saying that, now that you have started using more administrative data, such as on things like people’s national insurance numbers and so on, you are less likely to see revisions of that size?
Mary Gregory: The revisions have happened as we have improved the way that we are using administrative data. We have talked a bit about how, in an ideal world, you would have run the old and the new methods in combination. We did not have that luxury in migration. We stopped using the international passenger survey at the beginning of the pandemic and moved very quickly to the administrative data. We have been developing that as we have been publishing the estimates.
Q101 Chair: You have been doing it in flight.
Mary Gregory: Yes, exactly. We have said that they are statistics in development, which is what you have heard in terms of the LFS. We are very close to not needing to be in development, and we hope that, by the end of this year, we will meet the standards of accredited official statistics. We will then ask the Office for Statistics Regulation to accredit our statistics; it will be up to them if they choose to do so.
Q102 Dame Harriett Baldwin: My question is whether we should continue to expect to see revisions of the size that we have seen, or was that an anomalous period when there were these enormous revisions?
Mary Gregory: We should definitely expect the revisions to reduce in scale. There will always be some revisions, but they should be much smaller. Some of the work that has led to those revisions is improving the methods in order to make sure that, in the future, the revisions are smaller.
Professor Sir Ian Diamond: It is worth saying that one of the reasons for revisions is that you do not want lagged data to an extent. We were with the Home Office yesterday, and they were very clear that they would like us to give early estimates. We do not think that the revisions will be anything like what we have seen in the past, but they want those early estimates rather than just waiting until we are absolutely sure.
Mary Gregory: It might be worth briefly saying that there are two sources for the revisions. One is that we publish estimates five months after the period that we are talking about, but, to be a migrant, you have to stay for a year, so we are publishing estimates before we know whether people have stayed, and we have to make some assumptions about how many are going to stay. That is one source, but those should be relatively small, because we are improving how we do that.
The other source is where we have improved methods or got new data, and they are the ones that have been bigger in recent years but should stabilise a lot as we go forward.
Q103 Dame Harriett Baldwin: Can I just back to the subject that we have been discussing for the whole session, which is the labour force survey and the interaction between that and the net migration statistics? You had a letter from the chief economist of the Bank of England about how often you update the population weights. Can you tell us how you have replied to that letter?
Mary Gregory: We talk to the Bank of England a lot. In population estimates, we talk with colleagues in economics statistics to try to make sure that we can provide what they need. The challenge, of course, is that, if you update migration, you need to update population and, if you update population, you need to update the survey weights.
Typically, the LFS uses what we call population projections, because they are more timely than our population estimates. Historically, we have updated them every two years. We have done a set every year for the last three years, the most recent being just last week, and it is those that are then used to do the reweighting for the LFS and are of particular interest to the Bank, because of that purpose.
Professor Sir Ian Diamond: We also enable the OBR to have early sight of them for their projections.
Q104 Dame Harriett Baldwin: That is for the ones that are coming out on 26 March. Are you expecting these changes to the weightings to have a big impact on the headline employment figures?
Mary Gregory: The population estimates are higher than they were before, because the migration numbers are higher than had been anticipated, so that will make a difference to the employment figures.
Q105 Dame Harriett Baldwin: When we will see that?
Mary Gregory: The decision for when the reweighting happens is for colleagues, but we know that they need the subnational population projections, which will come out for England in May, and for other nations later this year. That will then allow them to update the weights.
Q106 Dame Harriett Baldwin: But the Bank of England will not have it in time for tomorrow’s decision.
Mary Gregory: No, but it does have the projections that show what the population estimates are.
Q107 Dame Harriett Baldwin: In terms of the challenge around this data, you have recently published some population projections, which I imagine are subject to similar levels of uncertainty. Is it realistic for politicians and political parties to say that they are going to target a specific level of net migration to the UK? Give your statistician’s reply as opposed to your political reply.
Mary Gregory: Obviously, it is not for me to give a political reply.
Chair: We would not tempt you into the politics of it.
Mary Gregory: What is really important to understand is that net migration is made up of a lot of routes. It is important to understand the make-up of the top-level figure, rather than just focus on that top-level figure.
Q108 Dame Harriett Baldwin: Political action, though, can have an impact on many of these independent drivers of this uncertain statistic that you then report.
Professor Sir Ian Diamond: I just want to make the point that, for the first time, our May data will really split things down, so that you will be able to see the characteristics. One thing that I have said a number of times, and will say again, is that, if you are looking at net migration, for example, I would expect it to go down in the next couple of years, for a very simple reason. This is not rocket science. During the pandemic, not many students came to this country. What that means over time is that there are not many people to leave. Post pandemic, lots and lots came in, because we are very attractive place to come and study in. As they start to leave, net migration will go down, because you will have a steady state.
It is important to recognise that, which is why the work that Mary and her team have been doing to follow up students in particular is incredibly important. We cannot just be assuming what the proportion of students who are going to go is, but those May numbers are super interesting.
Q109 Dame Harriett Baldwin: Will we see this super interesting set of data later this year?
Mary Gregory: Yes, on 22 May. To expand slightly, in our population projections, we have to make assumptions about future migration. At the moment, the principal variant assumption is 340,000 per year, so it is for politicians to decide whether that is where they think it should be.
Q110 Dame Harriett Baldwin: That is a really important point. Does the figure of 340,000 per year take account of this distortion?
Mary Gregory: It looks at the historic average and assumes that we will have something much more similar to the last 10 years than the last two or three, when it has been particularly high.
Professor Sir Ian Diamond: To be clear, projections are not forecasts. Projections are put together by a group of experts.
Q111 Dr Sandher: I want to talk about how well the wealth and assets survey is capturing the top 1%. The wealth and assets survey shows that the top 1%’s wealth share fell from 14% to 12% between 2010 and 2020, which was during QE. If you combine that with the Sunday Times rich list, you would find that the top 1%’s share of wealth has increased from 20% to 25%, and the wealth-to-GDP ratio from 600% to 700%, and that the number of billionaires doubled from 80 to 160. How well is the wealth and assets survey capturing top wealth shares in this country?
Professor Sir Ian Diamond: It is a super good question. We are doing everything that we can to look at those numbers, but it is a challenge. Both ends of the spectrum are a challenge here.
Richard Heys: That is also recognised internationally, and we face very similar challenges there to many of our peer countries.
Q112 Dr Sandher: On the international comparison, you look at the bottom 90%, the top 10% and the top 1%, and then apply a standard weighting factor to each. This goes back to our first discussion. In the US, they are doing it by percentile instead, so they are checking differential non-response in each percentile—i.e. “Those in the 97th percentile answered this many times”. By doing that, you will get a differential non-response within the intervals that you have at the moment. Would you be willing to consider that in future iterations of the wealth and assets survey? You could also do that retrospectively as well.
Professor Sir Ian Diamond: We are absolutely prepared to look at it. The Office for Statistics Regulation is going to review the wealth and assets survey. We are going to work with them very carefully, and international comparisons are part of that.
Dr Sandher: That is excellent. The academics are Ben Tippett and Mr Wildauer.
Q113 John Glen: Over the last two hours, Sir Ian, you have shown us the absolute commitment that you have to get the best out of the data sources that you have, and you have explained in some detail the challenges that you face.
None the less, the outputs that you come up with collectively drive Ministers’ decision-making and drive commentary on quite fundamental aspects of the way in which we run our country. If we look at the international trade data, which lost its accreditation 10 years ago, we are at a critical moment in terms of looking at our trading relationships. What level of confidence should we have in the data on the decline in exports, for example, and the differences between services and goods?
Professor Sir Ian Diamond: You are right that trade asymmetry is a well-known international problem. We are concerned about it. Have you read Richard’s piece on it?
Chair: We are now prompted to.
Professor Sir Ian Diamond: If you have not, now is your chance to, because Richard wrote the major piece on it.
Richard Heys: Work that we did over the last few years with Thomas Baranga at Harvard sits alongside a programme of work that we have done on international negotiations with partner countries. For every export that we estimate, there is an import that they estimate.
There are many reasons why these can differ. If you are doing surveys, you are going to get normal variation, but you can, within those, see patterns whereby, if a country estimates its imports to be lower than all of the imports of the countries that say that they are exporting to it, you can begin to make adjustments that help us to understand what some of the drivers here are and how some of those change through time.
Historically, it was generally perceived that your import estimate would be stronger, because, if you were charging tariffs or other charges, you wanted that data to be accurate, whereas what you perhaps paid less attention to was what was going out. Increasingly, we have found in the modern era that the export data appear to be dominating for lots of countries, and their attempts to understand where they are trading globally seem to have a little more accuracy than some of the import data.
Some of those changes through time mean that we are having to alter our perspective. As Ian said, we did lose the accreditation in 2014. One of the challenges here is that, in trying to get it back, because OSR flagged these trade asymmetries as a question mark, you have to make an assessment of who is right and who is wrong. If you take the US and Canada, for example, they both estimate their import number and then swap their datasets so that Canada uses the American import data as its export number.
Q114 John Glen: I get the concept, but what I am trying to get at is that, in the end, numbers are presented. They drive policy decisions or directions of travel. What sort of quantum out could we be? How do we assess the overall level of comfort and confidence that we should have as it drives pretty delicate balances of where we orientate our trade policy to?
Richard Heys: The key thing for us is that we look to our policy users. The Department for Business and Trade and other users have regularly given us positive feedback. During the journey that we have been on since 2014, we have had a really large investment in our trade data. It is now more granular and you can better see what is happening at the country and industry level. We have not been able to get the badge back, because of the philosophical challenge around whose data is best.
Q115 John Glen: But we should not be fundamentally concerned about the numbers that you are coming up with.
Richard Heys: I do not think so, no. Again, in parallel to the System of National Accounts update, the balance of payments manual from the IMF is being revised on the same timetable, so the international community is collectively looking at some of the issues around measurement in this space. As we implement those changes over the next five or six years, we will be at that cutting edge with the other leading developed countries in implementing those.
Q116 John Glen: I wanted to ask one other question about the alignment of the collection of data between you, as our national authority, and the devolved Administrations. My understanding was that, essentially, they have rights over what data is captured. Can you say something about how that dynamic works and whether it is in the best place?
Professor Sir Ian Diamond: We work very closely with our colleagues in the devolved Administrations. I meet regularly with the chief statisticians of Wales, Northern Ireland and Scotland. Statistics is, in the main, devolved, although some surveys, such as the labour force survey, are UK-wide.
Statistics being devolved means that, quite often, they are collected for policy purposes. In the last couple of years, we have put an enormous amount of effort, as have our colleagues, into trying to make UK-wide numbers. That is not as easy as it may seem in many areas. We have made real progress on some NHS statistics and will continue to do that. We have a team that works on coherence. At the end of the day, it comes down to priorities.
I take a very strong personal view that we have missed a trick in the past. There was a natural experiment to have different social policies, and we did not collect comparable data from 1999 onwards that would enable us to have done that. That is why I have made it my challenge to do what we can to improve it, and I remain of the view that it should be a high priority.
Q117 John Glen: What more do we need to get that right? You are absolutely right that devolution gives different discretion to different devolved Administrations, but unless you collect common data, you cannot make meaningful comparisons.
Professor Sir Ian Diamond: We need political commitment.
Q118 John Glen: What does that mean practically?
Professor Sir Ian Diamond: Practically, it means that the three devolved Administrations all agree that this is a good thing to do. That has not always been an easy conversation, in my experience, but that will seems to me to be important. We can then make things happen. There are resource issues as well, because you are still trying to collect data for particular policies and, if it is additional data, there is a resource issue.
Q119 John Grady: I just want to follow up on one thing as someone with an interest in devolved statistics. Is the political will, and the will of the devolved Governments, all there to have these harmonised statistics, or is it patchy?
Professor Sir Ian Diamond: I am not able to comment fully on that. That is a political decision.
Chair: We will not tempt you, at this late stage, into the politics.
Q120 Rachel Blake: You have done more for enthusiasm about statistics in the last two hours than many of your communications colleagues across Government have done on this topic for a long time, so I would like to congratulate you on that.
Chair: You are speaking to an interested audience.
Professor Sir Ian Diamond: You do not realise how much of a thrill this has been.
Q121 Rachel Blake: I am probably at one end of the curve on this, and it has certainly significantly increased my capacity for this. One of the areas that we have been skirting around is confidence in the ability of decision-makers in the statistics. I am really interested to get to the bottom of your assessment of risk in the statistics and the numbers that you are putting out there.
We know that the Governor of the Bank of England wants us all to answer our phones more. You want to get hold of more phone numbers. Members of the Monetary Policy Committee want to understand flow into unemployment in a lot more detail, regardless of whether you have confidence. What assessment of risk have you taken in terms of the numbers that you are putting into the public domain?
Professor Sir Ian Diamond: We spend a lot of time talking about risk. No data that we produce is absolutely without error. That is why we always produce uncertainty estimates, and it is critical that, as a statistician, you do that.
We also spend a lot of time, not only as ONS but also with our board, discussing risk. We assess whether, for example, our quality risk is within or without appetite on a regular basis, and what that means. For example, in the last year or so, we have brought electronic used car prices, electronic train prices and new rents data into our calculation of inflation. We are about to start dual running grocery scanner data.
We are clear that policymakers need to make decisions based on the best data. For example, we could have just said, “We will dual run for a month and then bring in the scanner data”, but our inherent conservativism, plus our conversations with our stakeholders, which are really super important, mean that we are going to dual run for a year on that, making clear that it is both the new data using scanners, and the traditional clipboard collected data. On those data, we have zero risk.
We also produce, every December, baby names. Frankly, it is really fun, which is great, but, if we were, one year, to suggest that Gary was now the most popular boy’s name, it would be an error but the world would not be about to end.
Q122 Rachel Blake: Baby names are a relatively low-risk area. Coming back to the labour force survey, you said in October 2023 that you just could not put it out because it was too biased around one particular cohort. You took the decision there that the risk to policymakers from using this was too significant to put out the data. How are you communicating to policymakers the risks that are inherent within the data?
Professor Sir Ian Diamond: We talk to policymakers the whole time. For example, we have an advisory committee on prices and on labour force. We are about to advertise for a new advisory committee on national accounts. We have regular stocktakes with the Treasury. We are talking to our major stakeholders all the time.
It is incredibly important. As I said in my response to John Grady, we were criticised back in the 1990s for just putting the numbers out there. We now put them out only following conversations with our stakeholders. Most importantly, throughout economic and population statistics, no data comes out without what we call curiosity sessions. This is about not just the statisticians who are producing the data but others too getting together and really getting underneath and understanding those data, so that, when we put those numbers out, they are the best that we can do at the time.
Q123 Rachel Blake: I am tempted to ask you about the meteor news last night, because there are some statistical decisions to make about our activities around that.
If you believe that you are communicating risk levels to policymakers, while you have acknowledged that there is always risk, are there any particular outcomes to making the wrong decisions around the economic statistics, particularly in the labour force survey, that you would be particularly concerned about and would want to use this opportunity to share with us?
Professor Sir Ian Diamond: We have engaged with our stakeholders to the extent that we understand what they are most concerned about, and we are providing them with the best answers that we can.
Q124 Rachel Blake: What are they most concerned about? Where do they think the variations might come, and which are they most worried about?
Professor Sir Ian Diamond: They are most worried about what is going on in employment and in unemployment, and whether they are going up or down. If things are not changing, it is not difficult. The real challenge in statistics, in my opinion, is estimating turning points. Because it is a survey, it could go up or it could go down a bit. That could just be some kind of artefact in that collection, and it is going to go up again, so being clear about turning points is incredibly important.
During the pandemic, we were producing a weekly survey of the incidence of Covid. We were up all night worrying about whether it was a change or an artefact of the sample. There were big decisions around lockdown and things such as that. We were absolutely aware of the risks and the critical decisions that had to be made, and that we needed to do everything that we could to understand and communicate them.
Would it be lovely if we had better data literacy in our country? Yes, 100%. We have a real responsibility to make our data as accessible as possible. I am not convinced that things are perfect. I am not complacent at all, but we do have a role to play and we will continue to try to do that.
Richard Heys: On your precise question, as well as measuring short-term volatility in things such as employment and unemployment, the labour force survey also gives us data to look at some of the longer-term trends in the UK economy, and particularly things such as productivity.
We published an article last week around productivity trends that we discussed widely with pretty much all of the economists who work on productivity in the Government to try to look at the issue of whether there has been sufficient data post Covid to look at that issue differently than the message that we had seen sustained since the financial crisis of slow growth but weak compared to other countries and previous history.
What we can see there is that we get enough commonality between the business and household surveys and the administrative data to have some confidence that they are telling a joint story, but there has not been quite enough time or enough change away from long-term trend rates to suggest that there is yet that movement.
Q125 Rachel Blake: There are some real policy consequences to thinking, for example, that unemployment could be growing because of movement out of inactivity. There are some significant choices that Ministers would have to make. If they can be confident that there is a relatively low risk of this mistake, that gives more confidence throughout the policymaking process.
Professor Sir Ian Diamond: Yes, 100%. If you look at the data on productivity in the last decade, that was driven by labour force rather than by innovation, although there was some innovation—do not get me wrong If you come back to the conversations that we have had this morning about inactivity, the way in which that is impacting on productivity has to be really important, which is why we have to be as confident as we can in those data.
Q126 Sam Carling: Thank you for coming in today. It has been a really helpful discussion. We have talked a few times about end-users of the data, their reasons for using it and the way that they are working. What consultations do you have with your data’s end-users when you are making changes to methodology? Particularly with the movement from the LFS to the TLFS and the data that could be lost, how are you ensuring, through talking to them, that the valuable bits are being retained and perhaps the less valuable ones are not?
Professor Sir Ian Diamond: I will come to your direct question, but it is incredibly important to say that the ultimate end-users are every one of our fellow citizens. I take a very strong view, and one of the reasons why I am passionate about this job is that data can shine a light on inequalities in our society, and it is then for you to take the difficult decisions about what to do about them.
Part of that end-user issue is that, as I just described, we have a number of advisory committees. We work very closely with all major Departments that use our data, including DWP, HMRC, the Home Office, Treasury, the OBR and the Bank, and we have very regular conversations with them. Those advisory committees often meet every month.
When we make changes, we do so from day zero—“This is what we are intending to do”—and work with them. That is why everyone is comfortable with what we have done on scanner data, but our inherent conservatism, and that of the Bank, around getting it right says that we will dual run for a year. If any of the Committee would like to have a field trip to the sunny climes of Newport, you would be really welcome.
Chair: We will take that invitation and have a discussion about it.
Professor Sir Ian Diamond: Some of the numbers are so exciting, but we would have to show them to you.
Q127 Chair: You are probably preaching to the converted on that. You have a willing audience here. Can I thank our witnesses very much indeed? We have had a useful discussion about the challenges of delivering the statistics that you produce for policymakers.
In particular, we discussed the actions that the Office for National Statistics has taken to boost response and to counteract the bias that you just touched on, the difficulty in reaching a sample by landline, and why responses have fallen in relation to other countries.
We also note, Sir Ian, that you expressed confidence in the data, including the numbers on inactivity and long-term illness. We have heard about the ONS’s wider efforts to make use of administrative data that Ms Yang particularly focused on to complement surveys, and the co-operation that it requires to do so from Government Departments. It would be quite interesting for us, as we discuss this with our sister Committee, where we take some of our thoughts. Thank you very much indeed. We will be producing our thoughts on this, whether we write to you or do a report.
Professor Sir Ian Diamond: Thank you very much. I and my colleagues very much hope that this is not the last opportunity that we have for such an interesting conversation.
Chair: Our sister Committee is your sponsor Committee, so Mr Carling’s appetite is whetted by this session. Professor Sir Ian Diamond, Richard Heys and Mary Gregory, thank you very much indeed for your time.
Professor Sir Ian Diamond: Thank you very much.