Knowing the unknown

I wrote something in March about the announcement that HMRC would be providing data from their records to help the Office of National Statistics (ONS) assess the reliability of the official statistics about migration to the UK which are primarily derived from the International Passenger Survey (IPS). The idea was that as HMRC can establish the nationality of people from their National Insurance number (NINO), they could count up the number of non-UK NINOs that are actually being used to pay taxes or National Insurance contributions or to claim tax credits or child benefit, as this would be a good indicator of someone being in the UK, and comparing with the official migration statistics.

With my usual caution, I concluded

Certainly the HMRC data will not provide an actual count of how many are working here and some judicious interpretation will be needed, which will rely on a very clear description by HMRC as to what they have done and what exactly their figures represent. It is to be hoped that this is what they provide ….

As it turns out, there wasn’t actually a public release of data by HMRC, but instead a report of some length by the ONS which took account of data that HMRC had provided to them. There was some description of what had been done but even I am left a bit baffled by what the figures are supposed o tell us. A set of charts was presented for the three standard divisions of the EU into EU15 (Western Europe), EU8 (Eastern European countries joining in 2004), and EU2 (Romania and Bulgaria). This is the one for EU8. What it shows is the number of people the IPS estimated arrived for the long term from these countries (orange), the number of people from these countries who were issued a NINO in the year (blue), and the number of these NINOs that were used in that year for paying something to HMRC or claiming something from HMRC.

To the extent that the purpose of the exercise was to cross-check the IPS estimates of numbers of people arriving, this seems hopelessly unlikely to succeed. You can see by squinting at the top of the picture that ONS say ‘these findings suggest that not all those who register for a NINO interact with HMRC in the year in which they arrive’. Well it’s more than just suggest — it proves that not all those who register for a NINO interact with HMRC in the year in which they arrive.

This isn’t entirely unsurprising. One of my first blogs was on some ad hoc data released by the ONS that showed that in 2012 only half the people who had arrived in the previous two years were working. This alone would suggest that the number of NINOs that see HMRC activity in the year in which the NINO was allocated will be a very poor guide to the numbers arriving, as people who aren’t working won’t interact with HMRC at all unless they have children.

This ad hoc data didn’t provide any breakdowns by country, but even allowing for quicker entry into work by some than others, certainly overall it seems that the HMRC data on interactions in the first year of arrival can’t really tell us anything at all about the level of flows of migration into the UK.

My own interest isn’t in flows anyway but in stocks: how many people from abroad are actually in the UK — for the primary reason that I’ve outlined before that this can obscure what is happening in the UK labour market, for example in my blog here which concluded with this picture

So does the information from HMRC used by the ONS in their illustrations tell us anything useful about that? No, it doesn’t.

But, and it’s a very big but, HMRC do have exactly the information on how many people from abroad are in the UK and who are interacting with them. Yet this much more interesting stuff is relegated to a rather thr0w-away paragraph at the end

While this refers to information about those who arrived in a four-year period, it shows that HMRC can say for any tax year how many people originally from abroad were interacting with them in that year either as payers of taxes or as recipients of benefits.

In his most recent blog on this subject, Jonathan Portes notes that the one million people from the EEA referred to seem to be much higher than the comparable numbers recorded as working in the UK in the Labour Force Survey and says

First, it seems implausible the LFS numbers are correct. I think there clearly are more recent EU migrants present and active in the UK labour market than suggested by the official statistics. By how much, we don’t know. Nor do we know why. But, on the basis of what we know now, this seems to be a bigger discrepancy, and one that is more difficult to reconcile, than that found in the migration statistics themselves. Note that in itself this doesn’t mean the overall numbers are wrong — perhaps new migrants are misclassifying themselves as having been here longer — but it is cause for concern.

I broadly agree with that, but a key issue is what ‘correct’ means. The LFS deliberately covers only what it calls the resident population, defined as people who have been in the UK for over six months. Obviously this will not include people who come for spells of work of less than that. The reason this matters is that there might be a considerable number of jobs in some sectors that are permanently filled by a rotating cast of people from abroad who individually do not stay in the job (or indeed in the UK) for long enough to be counted as resident in the UK. So if the person carrying out the LFS knocks on the door of a house with six migrant workers and is told that no one has been in the UK for more than six months, they will put away their questionnaire and move on to the next house on their list, as the occupants aren’t part of the resident population. But the house will be permanently occupied by six people throughout the year, it’s just that their identity changes over the year. Similarly, six jobs will be permanently filled by them, and so in practice there are effectively six migrants working throughout the year in the UK. However, they will not appear in the official labour market statistics, and it is as if this employment simply does not exist when the numbers of people working are reported by the ONS.

Jonathan continues …

this doesn’t in itself change the broad assessment of the labour market impacts of EU migration, summarised by my LSE colleagues here. Although much of the research they discuss is based on LFS data, by no means all is — in particular, as they note, recent analysis by me using National Insurance registration data confirms the broad picture of no significant impact on employment, and small if any impacts on wages. But it does perhaps mean we need to pay more attention to the impacts of short-term migration — which the LFS, by construction, cannot tell us.

It’s true that it doesn’t in itself change those assessments, but clearly it can be a confounder for researches based on the LFS that look for example at the relationship between changes in wages and changes in the number of migrant workers if potentially large numbers of the latter simply aren’t being factored in because they are excluded from the underlying data available even though they are a real presence in the real world.

It’s clear that there is a real issue here, and the release of more detail of the data the HMRC have collated will help establish just how big it is and what difference it might make.

Like what you read? Give Michael O'Connor a round of applause.

From a quick cheer to a standing ovation, clap to show how much you enjoyed this story.