Author: Jay Lindop
Since David Cameron became UK prime minister in 2010, the British government has capped skilled migration, become more sceptical of student visa applicants and introduced its infamous “hostile environment”. All were policies adopted with one eye on the net migration statistics, on the back of Cameron’s pledge to reduce immigration into “the tens of thousands”.
In the midst of those dramatic changes, the survey which underlies that official net migration number was tweaked. A simple new question introduced in 2012 was meant to garner more detail on the different kinds of immigrants coming to the UK.
Instead, in the years that followed, it exposed problems in the UK’s immigration statistics. They could mean the overall net migration figure has been repeatedly overestimated by tens of thousands of people.
• For more like this, see our refugees and migration newsfeed.
It’s prompted the UK’s Office for National Statistics to launch an ambitious new effort to overhaul the figures, integrating data from official border checks and the tax authorities. But it’s no easy task — because, as the experience of other countries shows, getting a reliable count is a technical challenge second only to the difficulty of deciding what “immigrant” actually means.
The kids aren’t alright
The UK’s quarterly statistics on long-term international migration are drawn from a survey introduced in the 1960s primarily to track tourists’ spending. The new ONS project aims to supplement that survey with administrative data created in the course of running government programs.
That’s because of problems with counting emigrants, highlighted by the new question. The 2012 change meant that foreigners leaving the UK would be asked what their original reason for coming had been.
It was meant to allow the ONS to produce finer-grained data on the movement of different types of immigrants. But the new figures immediately raised eyebrows, because they suggested that the number of international students leaving the country was much lower than the number arriving.
“It’s arguable that the overall figures might be right. But it would be a fluke”
The gap implied as many as 100,000 people were overstaying their student visas. But that number — which helped drive a series of changes to visa rules — didn’t seem to be consistent with other sources of data on students.
In 2016, after enough time had passed that the difference could no longer be viewed as an anomaly, the ONS launched an investigation. “We did some work using Home Office administrative data,” said Jay Lindop, deputy director of migration at the ONS. That report, based on information from the checks immigration authorities conduct when travellers leave the country, was published last August.
The data revealed that there was no significant number of former student overstayers, Lindop said. Instead, the new question wasn’t identifying departing international students — their changing plans and frequent trips back and forth meant the intentions-based question wasn’t reliable. The result was a sizeable underestimate of how many students were leaving.
The ONS says that it thinks the overall net immigration number is still reliable, because the overcount of students is balanced by undercounting in other groups. Marley Morris, an expert on immigration at the Institute for Public Policy Research, was sceptical. “It’s arguable that the overall figures might be right,” he said. “But it would be a fluke.”
In any case, the discovery of the problems with international students in the figures prompted the ONS to launch its migration statistics transformation project. It aims to put administrative data, collected by government agencies for their everyday functioning, at the heart of the UK’s migration statistics.
The power of administrative data
It’s an approach that many other countries already take. The Canadian statistics body, for example, bases its immigration figures primarily on microdata given to it by the country’s immigration department.
Canada’s official statistics draw directly on the number of permanent and temporary resident visas granted each month, according to Hubert Denis of Statistics Canada. But emigration is still more complex, since departures from Canada aren’t logged by border authorities.
“Administrative data has potential — but it’s not a miracle”
Instead, Statistics Canada draws on other administrative data to estimate how many people have moved out of the country. When someone stops filing tax returns, for example, or a family stops claiming child benefit, that suggests they may have left the country.
That’s the kind of approach the ONS is looking to adopt, using data from tax and health authorities as well as the Home Office to inform migration statistics. But it comes with difficulties. Tax information, especially from people who are self-employed, may not be available for many months.
And information gathered for administrative purposes generally needs to be adapted for use in statistics: people may have multiple records, or none at all, for reasons irrelevant to migration. Changes in tax and benefit law can affect the continuity of data.
“Of course administrative data has potential,” said Denis. “But it’s not a miracle. We have to use them carefully. It takes time to do properly.”
Australia, by contrast, has been able to avoid many of the difficulties of combining different data sources thanks to the comprehensive information it gathers at the border.
“Every time someone arrives or departs, a record for that movement is created,” explained Myles Burleigh, of the Australian Bureau of Statistics. And all an individual’s movements are associated with an anonymous unique identifier.
As a result, the ABS doesn’t rely on surveys, estimates, or even information about the type of visa someone has. Instead it simply counts the number of days each individual has been in and out of the country, and counts them as an immigrant once they pass a certain threshold.
That kind of system is unlikely to be possible in the UK, where there are minimal immigration checks for EU citizens or anyone coming to or from the Republic of Ireland. “It’s a pretty fortunate situation to be in, to have the level of completeness that we do,” said Burleigh.
So the ONS will have to draw on a wider array of administrative data. It’s conducting a feasibility study to assess the suitability of different sources for use in the revamped migration statistics. “Nothing worth having is ever simple,” Lindop said.
What is a migrant?
Establishing what that threshold should be, though, is more difficult.
The UN’s standard definition of a long-term immigrant is someone who changes their country of residence for a year or more. But as travel becomes cheaper and easier, putting that definition into practice has become more difficult.
“If an Australian moves to London but comes back every Christmas, in the past our system wouldn’t ever count them as leaving Australia,” Burleigh said. “They hadn’t ever left for 12 months in a row.”
Canada’s approach is to assume that immigrants are resident for the duration of the visa that gives them residence.
Australia’s more comprehensive records have allowed it to develop a different system. The ABS operationalises the UN’s one-year rule, Burleigh said, by using what it calls a “12/16 rule”. Someone is counted as usually resident in Australia if they’re in the country for 12 months over a 16-month period.
That captures international students living in the country, even if they usually leave during university holidays, but excludes backpackers who travel and work in the country for a year before leaving.
The ONS is looking at Australia’s approach, as well as similar methods used elsewhere in Europe. “There’s lots of different variations in the way that people migrate now that are not conforming to the traditional definition,” Lindop said. “We really need to understand the user need and the migrant journeys, by looking at the patterns in the data.”
But the definition it uses will ultimately depend in part on what data sources the ONS can viably draw on. The questions of who to count and how to count them are challenging enough individually, but in practice they’re also interdependent.
“The more you delve into this, the more complicated you find things are,” Lindop said. “But we’re also finding out how much administrative data can tell us, especially when you combine different sources.”
The importance of reliable statistics on not just the size of migration flows but also their composition, Morris suggested, makes that complexity worth grappling with.