Official figures for eastern European net migration could have been underestimated by more than 50,000 a year in each of the last five years.
Migration Watch, in a recent analysis.
The independent research body Migration Watch has claimed that the government is significantly underestimating the level of immigration from the EU. Its headline claim was cited on the BBC by Nigel Farage, leader of UKIP, who is campaigning for the UK to leave the European Union.
If true, the statistical error could have a significant political impact. After all, Prime Minister David Cameron claimed that part of the deal he negotiated for the UK’s “special status” in the EU was all about reducing “the very high level of migration from within the EU”.
There is nothing much actually new in Migration Watch’s report, but it does exploit a genuine and puzzling discrepancy between different sources of information that help us to estimate the UK’s immigrant population.
Where migration data comes from
The four main sources of international migration data are the International Passenger Survey (IPS), Labour Force Survey (LFS), National Insurance (NI) numbers, and visas. The problem is that the gap between IPS and NI data has become quite striking: in 2015 there were 828,000 NI registrations, 76% from the EU, while migration statistics for year ending September 2015 show a net (immigration minus emigration) figure of 323,000, with 53% from the EU.
These figures are dis-aggregated into “old” EU countries such as France and Germany, those countries including Poland and Hungary which joined the EU in 2004, and Bulgaria and Romania who joined in 2007. As Migration Watch points out, the discrepancy does seem to be larger for citizens from the newer EU members, some of which are from eastern Europe.
This is not “news” as such because this particular discrepancy has been known about for some time, but it does seem to be increasing. A few years ago, when ID cards for foreign nationals were being considered, it was revealed that there were millions more NI numbers created than were needed. The linkage between the different databases is the subject of a review being carried out by the Office for National Statistics, due to be published in May, that will also include the most recent data.
The intervention by Migration Watch is typically opportunistic, using a well-known anomaly to, as its chairman says, cast “serious doubt on the accuracy of our immigration figures”. Ahead of the publication of the ONS review, the report has thus been released to generate maximum impact – although some care has been taken not to make any unusual, or statistically questionable, assumptions.
Counting national insurance numbers
The key calculation made by Migration Watch involves the difference between the IPS and NI data – they have taken this to mean that there has been an under-recording of 50,000 EU migrants a year, and that they are from the countries that joined the EU in 2004. This is certainly possible, but the claim is rather confident considering the lack of any fine-tuned NI data in the public domain.
Other analysts have been more cautious, and take more seriously the possibility that EU nationals might be getting NI numbers but not staying long enough (over 12 months) to be considered an immigrant on the IPS figures. See, for example, written evidence from the economist Jonathan Portes on this issue.
Considering that the IPS is a sample-based survey, Migration Watch may be right to prefer the NI data as a more reliable source. But we only know the “top-line” data for NI registrations as the underlying figures are not made available. Recent releases of the data by the Department of Work and Pensions also warn that comparisons of NI registrations over time should be viewed with caution because of a change to the process of recording them in 2014.
As the ONS pointed out in a briefing on this topic in early March, the latest data on short and long-term migration would seem to offer some support for the alternative hypothesis that the higher number of NI numbers are recording a group of people that includes a proportion that would not be classed as immigrants by the IPS.
This report stated that data for the year ending June 2013 showed long-term immigration from the EU at 183,000 and national insurance number registrations (to mid-2013) were at 398,000. Short-term immigration recorded for work, study and business (including those where a NI number is not necessary) was at 239,000.
Ultimately, each data source has its own strengths and weaknesses and each tells us something slightly different. The problem was covered in a recent Radio 4 programme.
It remains to be seen what the outcome of this particular issue is, and how accurate Migration Watch have been, but it will surely impact on ongoing discussions regarding governance arrangements for national statistics. The Bean report on UK economic statistics recommended that migration data be improved. It complained that data was mainly based on “surveying individuals in air, sea and tunnel ports which can be subject to measurement and sampling error” and suggested supplementing it with administrative data – precisely what is now supposed to be happening.
The conclusions of the ONS review, and the way these are presented, could affect both trust in immigration statistics, and the way the UK thinks about freedom of movement in the EU.
Verdict
The veracity of Migration Watch’s headline finding depends on levels of short-term immigration from the EU during the period in question and the extent of sampling errors through the IPS. If the former have risen recently, then the under-estimation could be less significant than Migration Watch suggests.
If the IPS is accurately recording immigration then the underlying data for NI numbers should explain why there has been such a discrepancy. Otherwise, the Migration Watch numbers are certainly one reasonable guess, and recent history does suggest that in the context of free movement, EU migration to the UK can be higher than expected.
This phenomenon looks more significant for EU nationals from newer EU members states – referred to in the British debate as eastern European immigration despite also including countries from central and southern Europe. But as Migration Watch admit, the gap between the two sources of data is now growing for all EU citizens.
Review
Emma Carmel, Senior Lecturer, Department of Social and Policy Science, University of Bath
This fact check is very well set out and makes very clear the key issues in generating accurate data on migration. It also shows that we need to be cautious when trying to understand these figures, and sceptical about simple and politically convenient interpretations of all migration data.
Although there is a difference in the two figures, this is not the same as a discrepancy, as suggested by Migration Watch. This is because the figures do not measure the same thing. Differences between them must be interpreted, but this difference need not imply a problem of inaccuracy in either source of data.
For example, in addition to the points above, while the increase in NI numbers are for new registrations, some of these registrations might be for adults already resident in the UK for some time, rather than for “new arrivals”.
It would be useful to have more detailed data about the people registering for NI numbers, such as their migration history, work history, education and age. All of these would help us interpret whether the difference in the two sources of data can be explained by increases in mobility, increases in NI applications from existing residents, or for other reasons.
Overall, Migration Watch should not interpret a difference in figures in one year as representing an annual difference in inward migration. Net inward migration varies year-on-year, and because we do not have an explanation for the difference between IPS and NI data, we certainly should not make any assumptions that the difference is ongoing over a period of years.
For example, the February 2016 Migration Quarterly report explains the increase in net inward migration in the preceding 12 months, as being due to decreases in emigration, rather than increases in inward migration.
Sometimes even in those countries which have compulsory registration, such as Germany, Sweden or the Netherlands, data on who is arriving, leaving and staying, is still not always comprehensive, reliable, or directly comparable, due to difficulties in the timing of data collection or problems with measurement, as well as increases in mobility.