Did a million migrants really leave the UK during the pandemic?

We’re hearing strikingly different estimates about what’s happened to emigration during the pandemic, so what do we actually know? Ben Chu investigates

Some studies suggest the decline in the number of migrants in the UK might have been rather less extreme (PA)

In January, a bombshell study suggested that more than a million immigrants might have left the UK during the pandemic, representing the sharpest fall in the UK population since the Second World War.

The results were shocking in themselves, but also potentially highly economically significant since a migrant exodus on this scale would imply a major contraction in the UK’s potential labour force.

If true, it raises the risk of severe shortages in sectors that rely heavily on overseas workers, such as food production and hospitality, when the economy reopens.

But some studies since then – including one by the Office for National Statistics (ONS) on Tuesday – suggest the decline in the number of migrants in the UK might have been rather less extreme.

A London School of Economics analysis suggests the fall in migrant workers might be just 235,000, around a million less than earlier estimates.

So what can we say with confidence about emigration during the pandemic?

Read more:

• Follow Brexit news live: Latest Boris Johnson updates 
• Government acquires second, brand new ‘Brexit jet’ for Boris Johnson 
• Bid to save leaseholders from huge bills to remove cladding defeated by government

And why do we have such a fuzzy picture of something of such social and economic importance as the number of people in the country?

To understand the issue it’s instructive to examine how it was raised.

The January study by the researchers Jonathan Portes and Michael O’Connor which first flagged the possibility of an exodus of migrant workers was based on data from the Office for National Statistics’ Labour Force Survey (LFS).

This survey, which gives us headline official estimates of UK-wide employment and unemployment, is compiled by asking a very large sample of people (around 100,000 per quarter) from across the country about their work status.

The statisticians then extrapolate from the results to produce a nationally representative picture, in a similar way in which opinion pollsters produce a picture of national voting intentions from a survey of 1,000 or so randomly chosen people.

The LFS also asks respondents about their nationality and country of birth, and the raw calculations suggested that the non-UK born population fell from 9.2 million in the third quarter of 2019 to 8.3 million in the third quarter of 2020 – a decline of around 894,000.

That’s an astonishingly large fall in itself. But the ONS also made an adjustment calculation that suggested it could be even larger.

Despite the estimated fall in the number of non-UK people, the statistics agency did not revise its overall estimates for the size of the overall UK population.

That decision essentially required the ONS to assume a very large increase in the UK-born population in the country over the past year to compensate and to make the numbers add up.

Portes and O’Connor argued that such a UK-born population rise was simply implausible and pointed out that if one assumed, instead, that the UK-born population stayed broadly level through the pandemic, the logical consequence was the emigration of around 1.3 million non-UK born people from the UK over the past year.

Yet that’s not the end of the investigation.

Some have suggested that there might be problems with the Portes/O’Connor calculation, pointing to changes in the way the LFS survey has been collected during the pandemic.

Because of the risk of infection from face-to-face contact, ONS researchers have switched from going to people’s houses to get information to collecting it by telephone.

“If migrants are less likely to participate than non-migrants with the new method of data collection, this means their numbers will be underestimated,” says the Migration Observatory research group.

And during the pandemic the response rate to the LFS survey has slumped dramatically, which might be consistent with fewer migrants being reached and therefore undercounted.

The Migration Observatory suggests that migrants might be less likely to have a landline, making it harder for the ONS reach them.

There’s also been a fall in the share of LFS respondents who rent their homes, down from 32 per cent normally to 21 per cent in the second quarter of 2020. If migrants are overrepresented among renters, which is quite possible, this could also bias the results.

Separately to all this, the ONS has now looked at company “real time information” (RTI) payroll data collected by the HM Revenue and Customs.

Because a person’s national insurance (NI) details, collected by HMRC as part of their payroll information, also include their nationality this can be used as another indicator of what has happened to the number of migrant workers in the UK workforce.

It’s important to stress here that HMRC payroll data cannot show the full labour market picture because it only covers employees, leaving out the seventh of the UK workforce who are self-employed, a group in which migrants are probably overrepresented. It also doesn’t tell us anything about those not in work.

Yet it’s notable that the decline in non-UK workers from the payroll data is considerably smaller than indicated by the LFS data (-4 per cent vs -15 per cent)

It’s worth stressing at this point that neither the LFS nor the HMRC through their payroll data collection activities are designed to measure the size or composition of the population. So what is?

The answer is the International Passenger Survey (IPS), a large face-to-face survey of people passing through airports and other international departure and arrival points, and the census.

But there have long been problems with the reliability of the IPS related to its sample size and it does not have the official “National Statistics” kitemark. And, moreover, the IPS was suspended during the pandemic for safety reasons.

As for the census it only takes place once a decade. By coincidence Census 21 for England and Wales fell last weekend (21 March), but the results take a long time to be processed and, because it provides a snapshot of a day, they won’t tell us anything about what has happened to the size of the migrant population over the course of 2020.

One might reasonably argue the ONS and the government should invest more in collecting robust population statistics.

But, given where we are, the best hope we have is probably to continue to use multiple surveys in conjunction with government administrative data such as payrolls. The estimates these extrapolation techniques yield will never be perfect, but they should hopefully enable us to guess better.

What we can say with reasonable confidence is that no one thinks that the UK-born population has risen considerably over the past year. And it seems very likely that the number of non-UK workers has declined. But there remains a range of views about the extent of that decline.

Work published this week by the Resolution Foundation think tank suggests that the number of migrants in the UK has fallen by around 500,000, based on the assumption that migrant workers are proving harder to reach for the LFS. That’s less than half than the January estimate of 1.3 million.

Yet it’s worth remembering that this would remain a significant decline with potentially large consequences for the labour market if those workers do not return when the economy reopens.