10
Examination of witness
Mary Gregory.
Q119 The Chair: Welcome to this ninth oral evidence session of our inquiry into settlement, citizenship and integration. We are delighted to have you with us and it would be really helpful for the record if you could just introduce yourself.
Mary Gregory: Hello. It is great to be here. I am the executive director for population, census and social statistics at the Office for National Statistics.
The Chair: As you have heard, we are fascinated with statistics on this committee so we are absolutely delighted to have you with us. Baroness Prashar will kick us off.
Q120 Baroness Prashar: Good morning, Mary. A straightforward question but with a complicated answer, I am sure: how are levels of immigration and emigration calculated?
Mary Gregory: Yes, as you suspect, a more complicated answer. I will keep it relatively simple but do ask more follow-up.
Baroness Prashar: It would be very helpful if you could simplify it.
Mary Gregory: Yes. As of our last publication in November 2025, what we now do is measure net migration using entirely administrative data, so data collected by Government for administrative purposes. For EU and non-EU nationals, that means we use visas, border and immigration data from the Home Office, and, for British nationals, we use data from DWP that looks at interactions with the tax and benefit system.
Baroness Prashar: How reliable is the Home Office data?
Mary Gregory: You have already heard from Madeleine on some challenges around that. What I would start by saying is that what we have now is a massive improvement on what we have had previously. Until a few years ago, we were using survey data, which we measured based on people’s intentions. When people arrived in a port or airport, a surveyor for ONS would say, “How long do you intend to stay?” If they intended to stay for more than a year—which is the formal definition of somebody being a migrant—then they would count in the figures.
What we can do now is look at the actual travel data to say what visa they have come in on, so we can break it down more accurately and understand whether they have actually stayed or not. But as you would imagine, there are some limitations with that data. Sometimes there will be linkage problems—for example, making sure that we see the right person arriving that is leaving, or if people leave through the common travel area we might not see that they have left. So we have to build in some assumptions around that.
The other assumption it is really worth being aware of is that you have to be in the country 12 months to be considered a usual resident or migrant but we publish estimates five months after the period people arrive. So we make some assumptions about which of those we expect to stay and then revise that data when we know more about them.
Baroness Prashar: Let me talk a little about the British citizens. How are the statistics gathered in relation to British citizens and how does this compare to pre-2021?
Mary Gregory: Excellent question. In November 2025, we changed our approach. That was the last bit that had been using what we call the International Passenger Survey for British nationals. We had long said in ONS that that was not good enough for the purpose we were using it for; it was designed for travel information, not migration. So we have moved to the new source, which is the administrative data from tax and benefits. We are confident that is better than what we had before, but there is a lot more to do. For example, from the 2021 census, we could see that the survey approach was underestimating British emigration. We now have higher estimates of British emigration based on the new data, which we think are better, but it is still based on interactions, not actual travel data, and we would like to mature that data further.
Q121 The Chair: A real difficulty at the moment is that people keep telling us there is data we would like to have and we could perhaps get it this way and so on. Is there a list somewhere of all the data that we currently do not have and it would be great to have?
Mary Gregory: Madeleine has said this in other forums before, but the more data we have, the more we suddenly decide we now know and want even more data. So we would probably never reach the point where we would say, “This is a list of everything we want”, because as soon as we got it, we would say,” Oh, now that’s really interesting; we want this as well”. But there are some things that are relatively clear that would be helpful and would include something like you discussed with Madeleine: the overstayers. It is really difficult to know who has left the country. For the vast majority, and I do not want to overstate the problem—
The Chair: Just help me. Why is it very difficult to know who has left the country?
Mary Gregory: I was just going to add that I do not want to overstate the problem because for the vast majority of people, they are relatively straightforward: they travel on the same passport through a recognised route and can then be linked; they are entering and exiting and that is relatively easy.
The problems come around the edges when people leave through the common travel area, because they do not need a passport to do that so we do not have that information. Another problem is people travelling on different passports; I know one of your previous witnesses gave the example of having dual citizenship. If they come in on one passport and leave on another, we might not match that person as being the same person. There are quite a lot of different examples that make it really difficult but the majority of the information will be really good.
The Chair: Let us leave the integration of data collection systems for a second, which I know Baroness Hughes wants to pick up. If you were to suggest to this committee that in our report there were three or four major bits of data that we are currently not collecting and should be—let us not worry about how we do it—what would they be?
Mary Gregory: The biggest thing would actually be the quality of the data. The Home Office has lots of data but if we could get even greater quality for some of that, that would be really valuable. That is things like the exit checks data and how we make sure that we have good enough quality that we can match the people leaving and the people coming in. That would be the biggest single thing that would improve the statistics.
The Chair: What is your No. 2?
Mary Gregory: Then you are probably getting into more around the characteristics of people and understanding them better. Again, as Madeleine talked about, we have some really good information—much better than we have had before—on things like what type of visa people are coming in on, and we can see if they have changed visa while they have been here and what that process is. But we know much less around who these people are and what we know about them in terms of what work they are doing or what they are earning. Again, some of that is about linking the data that already exists and some is about having new data.
Q122 Lord Filkin: If Government—leaving it vague as to which department—asked you, “We want ONS to study the degree to which there is effective integration in our society”, is that an impossible question or is it something that the ONS could address?
Mary Gregory: Nothing is impossible. It would certainly be challenging and some things that you might want to look at could already exist. A big example is the census on there. There are a number of different questions. In some of your research, you might have seen this dataset called the Refugee Integration Outcomes study, which looks specifically at refugees, but as a concept that could be extended to look at things like whether they are integrated into the labour market and what their English language proficiency is. They have matched census data with data about refugees. As long as they arrived before 2021, you can start to look at how integrated they are within the population.
Q123 Baroness Hughes of Stretford: As the Chair has already alluded to, we are also interested in how effectively different sources of data within government and potentially beyond—but certainly within government—are integrated and how far these organisations, these government departments, are communicating with one another to build a rounded picture of both individuals and the system as it is working. Can you help us with that?
Mary Gregory: There is always more we can do. What we have seen over the last five years or so is a massive improvement in the efforts to share information. I mentioned the Refugee Integration Outcomes study. That is a project that is between MHCLG, Home Office and ONS, but it uses health data as well so it is not even just relying on those departments’ data. That is a really good example of when you can bring people together and get new insights.
Equally, within the ONS, we have a lot more data from administrative sources than we had in the past. There is a huge potential in how we can use that data, but we have secondees from ONS in the Home Office and DWP to make sure we can really understand that data and develop the data as well as possible for analytical purposes.
Baroness Hughes of Stretford: We have already noted that there is limited data and limited ability to track people who should be leaving after their permission to stay has expired. Can you just outline what other sources of data are lacking in this picture? For instance, we understand that the outcomes of compliance activity are something where data sharing does not happen very well and there is better sharing with other departments, particularly HMRC, as Madeleine noted. Could you help us understand that a little better, please?
Mary Gregory: My focus would very much be on the analytical and statistical purposes of data sharing. There are challenges but I would not want to overstate the challenges compared with the progress that has been made. Some is operational complexity. Clearly, if you are the Home Office, your primary purpose is to deliver operations, not produce statistics. When we try to use that data for statistical purposes, it does not always provide exactly what we might want. But I do not think that is through a lack of good will; that is through prioritisation of the primary objective of different organisations.
Baroness Hughes of Stretford: Is that not also a lack of emphasis on establishing the effectiveness of the operations? If you are not collecting data and then not sharing it, how do you know how effective your operations actually are?
Mary Gregory: As you will all know, good practice would be to make sure you build in the statistical considerations at the start, but some systems are quite old and are being developed as they go. The world is changing and the systems are trying to keep up.
Q124 The Chair: We are well aware that a lot of improvements have been made but I am sure you will understand the frustration of members of the committee. Tell me, what is the population of the United Kingdom today?
Mary Gregory: Some 69.5 million. I cannot speak for today exactly because we do not measure each day, but it is just over 69 million.
The Chair: With what degree of accuracy are you confident?
Mary Gregory: I would not give you an exact number, but what is worth being aware of is that we use different ways to measure things. We have two different sources we have looked at recently to measure the population and they are very close to each other at a national level, so that gives some confidence. One method is using the 2021 census and then rolling that forward based on migration, births, deaths and some other small aspects. Another is trying to link data from different administrative systems to understand how many people are in the country at this point in time. Both methods give very similar numbers at the UK level.
The Chair: Many of us will obviously accept your expertise, the organisation’s expertise and so on, but we have some difficulty when you do not have accurate data about who has left the country, and that is what we are all finding difficult. Just help me out a little more. How do I have confidence when at the same time I am told we do not have a collection of data on who leaves the country?
Mary Gregory: There are a couple of things. When we do population estimates, we work closely with local authorities to understand whether they think that what we are estimating is a good reflection of their area. In some areas that is much more positive than others. Some areas are definitely more difficult and quite often that is related to international migration, especially around students. But there are a lot of processes in place for us to try to make sure we are quality assuring that data beyond just the number we have been given. Does it make sense to people? Can we compare with other sources? What are we seeing in local communities? Is there a lot of housebuilding in one area and is our estimate matching up with that? It is important to look at the broad picture as well as the very specific number.
Baroness Bertin: Can I just come in on that? There was a question from Lord Hogan-Howe to the previous witness around education, law enforcement and other touchpoints where individuals could find themselves in the system. The last witness definitely said it could be done better. Do you agree with that?
Mary Gregory: In terms of measuring the whole population?
Baroness Bertin: Yes, in terms of cross-checking and seeing—to the Chair’s point—whether this is actually an accurate figure.
Mary Gregory: There are lots of things we can and do look at, some of which we will not necessarily publish and will look at internally. As ever, we are always trying to do more. Relatively simple things are published—for example, understanding utility, where you can get data on how many domestic electricity and gas supplies there are in an area. If that is changing at a very different rate than the population, that would give us pause to say, “Does this look right?” Then we would look at other sources as well.
Another really good example is that you can see birth rates in an area and that is quite accurate data. Can you look at the fertility rate that would be implied by the population estimate we have? There are examples where that suggests maybe we have not quite got it right in the females in their 20s and 30s, so we can go back and try to understand that better.
That is just two examples but there are lots and lots of examples of other data sources we can use and look at to say, “Does this look like it is a reflection of reality? If not, what can we do to try to improve this?” For the majority of local authorities, it is going to be a really good number but there are some cases where it is particularly difficult. Westminster is a good example of that.
Q125 Lord Henley: Can we move on to the demography of the UK? What effect is external migration having on the demography in the UK? How does this compare to internal migration, as well as with an ageing population and declining birth rates?
Mary Gregory: That is a really good question. One thing to say is that there are more moves through internal migration than external migration, but obviously at a UK level that has less impact on the overall population estimate. What we see is that the majority of migration is of younger people. If you exclude the EU settlement scheme, around four in five international immigrants are under 35, for example, so that obviously changes the dynamics of the population. But as you have already referenced, there is an ageing population generally because the fertility rate has gone down.
What you typically see is that those who are migrants have a higher fertility rate and have more children. Around two in five births in 2025 were from non-UK-born mothers, so you can see that they are having quite an impact relative to the proportion of the population. It will have an impact. What we do not know is the long-term impact because we do not know how long they are going to stay. You will have a whole group who will leave again once they have finished working age so they might not add, but there will be others who become more integrated into society and then stay into retirement. It is hard to predict the impact but we can make some assumptions around some scenarios.
Q126 Baroness Bertin: Some organisations have said an interim census would be a good idea. I suppose there are several points on this. First, is it feasible and what are the advantages and disadvantages? It strikes me that if we are trying to get to the bottom of the data of who is here, who has overstayed and should not be here, does the census really help us in that regard?
Mary Gregory: The census is absolutely a phenomenal tool for understanding who is here and is probably our most accurate measure of the population. It is for Government to decide whether they want a census or not. ONS can make recommendations. You will probably be aware that we recommended a 2031 census and the Government have accepted that recommendation so we are now planning for that.
It is hard to imagine the logistical operation required in reaching every single household in the country so that is not something that could be done within a matter of months; it takes years of planning. In terms of the feasibility question, I do not think it would be feasible to do something in 2026 at a national scale. You could obviously consider boosting some existing surveys or doing other things like that that would provide something. One example that people look at is the 1966 census, which was actually a 10% sample. To put it in perspective, given the 69 million population, that is still 7 million households you would be going to, so that is still a massive logistical operation.
Q127 The Chair: One issue this committee has looked at is in relation to the common travel area between Ireland and the UK. Is data shared at the moment between the Republic and the UK?
Mary Gregory: We publish a UK estimate and then Northern Ireland tries to publish an estimate for itself, which obviously is a component but it is broken down from the UK rather than the other way around. We do not have specific individual-level data sharing.
The Chair: Is that because it is not permitted under EU rules or it just does not happen?
Mary Gregory: I do not know what is permitted and I would need to go away and check. I guess we could look at whether they are interacting but because of the movement being so frequent across the border as well, it is really hard to know which side of the border somebody is living on. They might be working in one area and living in the other. It is really difficult to get the actual migration and resident pattern.
The Chair: Given the concerns that there are—as you are well aware—about that as a route through into Britain, should data not be being shared, at least as a way of starting to address some concerns that exist?
Mary Gregory: What we can obviously look at is what the Republic of Ireland’s migration figures are, and we do that with lots of countries so that we can try to judge whether our figures mirror the figures from other nations. If we take British migration as an example, there are a number of countries where they are saying there are high levels of British immigration. Broadly, they will match the countries where we say there are high levels of British emigration. We can look at things like that—even at an aggregate level—to get an idea around quality.
The Chair: Could you just drop us a note about the legality or otherwise?
Mary Gregory: Yes.
Q128 Lord Filkin: It has been a pretty torrid few years for the ONS in terms of what one can only really describe as a pretty fundamental loss of reputation, consequent on quite a number of high-profile failings: criticisms by the Bank of England itself and by your regulator leading to the Devereux report being set up, and then finding what were effectively almost cultural problems that led to these failings and that they were failings of the ONS’s own making, not on this occasion by wicked Ministers making life more impossible for them. Clearly you will say—I am sure it is true—that the ONS is doing its utmost to turn it around because it matters fundamentally to our society that we have accurate data for public policy, business and very many purposes. It is almost a hard-wiring platform of our society. Apart from telling us it is going jolly well, do we have any very clear data that shows what changes have been made that are leading to significant improvements, robustness of data, and above all the confidence of users that the ONS is to be trusted now?
Mary Gregory: That is a good question. Much of what you describe has been in the media and very public so I will not go into that in any detail, but there are definite changes. It would be fair to say we will not see immediate changes in everything because some of it takes time to work through.
We have a new Permanent Secretary. Following the Devereux review, that role has been separated from the National Statistician, which is currently being recruited, so that will separate the operational delivery within the organisation from the statistical expertise of the person overseeing the entire Government Statistical Service, which gives the Permanent Secretary more time to focus on the organisational delivery. He has already made a number of senior appointments, myself included on census and population. We also have a new director-general for economic statistics, who is the lead on that recovery plan around economic statistics, and recruitment is currently under way for a director-general for technology.
A big challenge is actually the pace at which technology is changing, and for a number of years ONS has tried to make do with systems that are becoming increasingly inadequate for the purposes we need, so hopefully that appointment will also help with that future strategy around that. As you said, there are a lot of people in ONS who care a lot about the importance of what we do because it is absolutely critical for society. There are people who have been working hard throughout the period to try to make sure we have the best statistics available.
On migration specifically, it is a huge achievement to have moved away from using the IPS. On the outside, it can sound relatively simple but that is a lot of data-sharing agreements, changes in approaches and new data sources to work with, so that is a really massive step forward that has been achieved. You will start to see improvements across a broader range of outputs. As we think about the Budget, one of the biggest things right now is that we are focusing very much on quality over quantity. Rather than trying to spread ourselves too thin, what are the most important statistics for society and how do we make sure we are as robust as possible with those?
Lord Filkin: Those sound as if they are pretty essential changes. Do you have a process whereby you really listen to your users and hear their criticisms before they reach the journalists?
Mary Gregory: We certainly try. You might be aware that a year ago there was a statistical assembly, and that was an attempt to get a broad range of users in the same place to understand the priorities for ONS. Those priorities were very clear and there was a lot of good will behind focusing on those most important outputs.
Certainly for migration and population specifically, we have a number of user groups. You spoke to Madeleine, who chairs one of our advisory panels. That brings together a number of experts and one of their jobs is to think about what users most need to see in terms of development from our statistics. They should be helping us make sure we focus on the biggest gaps. Madeleine mentioned overstayers; that is an area we are looking at and is one I fully expect they will continue to push us on, and rightly so. We have similar user groups.
On the economic side, we have set up an economic steering group, which has senior representatives from companies such as the Bank of England, and there is a lot of very regular engagement with other government departments and key users.
Q129 The Chair: Just finally, Baroness Hughes was asking you earlier about the way in which different government departments liaise with each other about the data and so on. Can I just ask you this? Bearing in mind our interest covers many government departments in the way Baroness Hughes was describing, in terms of what you are seeking to do, are you getting equal support from each relevant government department or are some more difficult to work with and get agreement with than others?
Mary Gregory: At an official-to-official level, the relationships are very good and there is no lack of good will. What we see—this will vary by department—is that there is a tension. A lot of people have very significant resource constraints and cannot do everything they want to. Sometimes you will find that something for statistical purposes is not top of their priority list because they have an operational delivery challenge that needs to be resolved as a higher priority. I do not think there are necessarily departments not seeing the value of what we need from them, but they have some very difficult decisions.
The Chair: In terms of the areas that you are well aware we are interested in, is there a government department that appreciates the need for more data collection, more analysis and so on, but is not doing that or seeking your support to do that because of a funding constraint?
Mary Gregory: I would not say it is not seeking support, but of course the one that would make the most difference for the areas you are interested in is the Home Office. The more that can be done to support it in improving its systems and making the developments that we know it wants to make, the better that would be for it and us.
The Chair: Just so I am absolutely clear what you are saying—we are not necessarily being critical of a department that is strapped for cash and having to prioritise things—your suggestion would be that if more resources were available to the Home Office, some data and interpretation of the data would be more effectively carried out by the Home Office than it currently is, to the benefit of what we are seeking to achieve.
Mary Gregory: It is hard to understand the complexity of the data that the Home Office works with and therefore the effort that needs to go into making that usable for analytical and operational purposes. If it had more investment in that, then it could absolutely carry out its plans anyway but at a much quicker pace, which would help us all.
The Chair: That has been absolutely fascinating and thank you so much for coming in. I have no doubt that when we have reflected on some things you have told us and looked back at the evidence session, we may well come back to you with further questions; I hope you will not mind if we do that.
Mary Gregory: I would be very happy.
The Chair: For the time being, thank you very much indeed. The public session is now concluded.
