Human behaviour, unreliable records: Why the UK keeps revising its migration data

In May last year, we learned that net migration to the UK (the number of people immigrating minus the number emigrating) had reached an unusually high level of 606,000 in 2022. Except it hadn’t – it was actually 745,000, we were told later in the year.

This spring, the 2022 figure was revised up again to 764,000. Third time lucky? No. Now the Office for National Statistics, or ONS, believes that net migration in 2022 was actually 872,000, 44% higher than the original estimate.

More recent figures have also been revised. While the ONS originally estimated the net migration figures for the year ending June 2023 at 740,000, this has now been revised to a record-high 906,000.

As a result, the drop from 906,000 to the latest figure of 728,000 in the year ending June 2024 is nearly 20%. But this too could still be revised in the future. Casual observers would be forgiven for being confused.

Difficulties in getting numbers right

Before the pandemic, migration statistics were produced from a survey of people passing through ports and airports. The figures were revised after an initial publication, but not very much – often by only one or two thousand. The problem was that they were wrong.

Or, as the ONS more politely put it, the survey in question had been “stretched beyond its purpose”.

Around the time of the Brexit referendum, for example, migration data greatly underestimated both EU immigration and non-EU emigration. This was partly because net migration statistics cover “long-term” migrants (people migrating for at least a year). But survey respondents often didn’t know how long they would actually migrate for, and it appears many ended up remaining in the UK longer than anticipated.

So the ONS decided – quite sensibly – to look for other data sources. The winning candidate was the immigration and border records the government already holds. I’m often asked “why don’t we count people in and out?” The answer is that we do. There are some gaps (for example people who arrive in London but leave via Ireland), but for the most part, the data is good enough to produce migration data.

Measuring long-term migration though, is still tricky. And there are two main reasons that the numbers keep being revised.

First, the methods are new and still being developed. They are officially badged as “experimental” rather than “national” statistics, although this distinction is probably lost on most people.

Initially, the ONS didn’t have access to all the information it needed. It now has better data on people who switch to new visas, such as students who stay on to work. Previously, the ONS only knew a person had switched to a new visa once they had travelled using that visa, which meant that quite a few people were assumed to have emigrated when they hadn’t. In addition, some Ukrainians were accidentally excluded from previous rounds of data and have now been added back in.

Second, when ONS publishes the statistics, it doesn’t yet have all the data it needs to work out whether people are long-term migrants or not. The Home Office now collects vast amounts of data on people coming to and from the UK, some popping in for a few days as tourists, others migrating long term. ONS sorts through them to work out who is really migrating and who is just coming and going.

However, the most recent statistics are for the year ending June 2024. If someone arrived on a 13-month visa in May, ONS can’t yet see whether they will stay the full duration of their visa (and thus be a long-term migrant) or leave a couple of months early. So they must hazard a guess about how people will behave in the future, based on what other migrants have done in the past.

And when behaviour changes – for example due to changes in the migration rules – those assumptions may prove wrong. For example, students have become more likely to remain long-term in the UK over the last few years, which has affected the numbers.

Planned revisions are thus inevitable if ONS sticks to its current timelines for publishing data. Indeed, migration behaviour may well change again this year and next, following the policy changes introduced by the Conservative government.

Migration policy doesn’t tend to stand still for very long, if the history of the past 15 years is anything to go by. This means the ONS may constantly be playing catch-up – although presumably not to the same degree we’ve seen in the past few data releases.

Why changing the stats matters

We can only speculate about whether the government would have made different policy decisions (or made the same ones sooner) if it had had access to accurate figures earlier.

Even if revisions are planned, they may still have a cost: lower public trust in the data.

Of course, ONS could have decided not to publish the data at all, given that the methods were still in development. That is what they have done in other cases, such as the now discontinued estimates of the number of foreign-born people living in the UK. This too has a cost. We can still learn a lot from imperfect data.

Indeed, while the magnitude of the net migration figures has changed, the basic picture has not changed – that unusually high migration has been driven by work and study migration, with more international students remaining long-term in the UK than in the past.

Perhaps ONS should have spotted some of the issues that have led to the biggest revisions sooner. Perhaps they could have advertised more prominently that the data are uncertain and may be revised substantially. Nonetheless, we are still in a much better place than we were before the new methods were developed, even if it has taken some time to get there.

Madeleine Sumption is Director, Migration Observatory, University of Oxford.

This article was first published on The Conversation.