Home / publications / commentaries /

Upward mobility? Earnings trajectories for recent immigrants

20 Sep 2023

by Ben Brindle, Madeleine Sumption & Jonathan Portes

This analysis uses new data from HMRC to show how the size and earnings of the migrant employee workforce has changed over time in recent years.

  1. Understanding the Evidence

    There is extensive analysis of migrants’ outcomes in the UK labour market, summarised here. This evidence is largely based on the Labour Force Survey (LFS), or the related Annual Population Survey, and to some extent the Census. However, the LFS contains only a very limited longitudinal component, and while it does, in principle, ask how long migrants have been in the UK in practice this variable is patchy and of limited use. This means it is not possible to use the LFS to assess in any detail either how the labour market characteristics and outcomes of new migrants have changed over time, or how migrants, once arrived, progress in the labour market. Analyses produced using the LFS are therefore generally “snapshots” of the entire resident migrant population, examining those who arrived relatively recently together with those who have been here for years or decades. Moreover, the recent results of the Census suggest that the LFS undercounts the migrant population, particularly those of EU origin, an issue that was severely aggravated by the pandemic. Meanwhile, longitudinal surveys, in particular the UK Household Longitudinal Survey, can shed light on the trajectories of different groups over long periods of time, but because of timeliness and sample size issues are less useful in assessing recent development in migration flows and labour market outcomes.

    However, HM Revenue and Customs (HMRC) have recently begun publishing data on migrant workers from the Pay As You Earn (PAYE) Real Time Information (RTI) system. Unlike the LFS, it covers the whole population rather than a sample; it is based on a count of all payrolled employee jobs that were active in a given month using HMRC’s PAYE RTI data (unfortunately, this means that it omits the self-employed who operate as sole traders). This information is then combined with the HMRC Migrant Worker Scan (MWS). This uses input from the Department for Work and Pensions (DWP) which issues National Insurance numbers to those arriving in the UK aged over 16 years through its adult registration process, which usually includes an interview to verify identity and right to residency in the UK. Combining this information allows the data to be disaggregated between those who were either born in the UK or arrived here before the age of 16, and those who were registered as adults when non-UK nationals. Within that, the published data disaggregates between those who were EU nationals and those who were non-EU nationals. The data are therefore divided into three groups we will call EU origin employees, non-EU origin employees, and UK-origin employees, where the latter also includes those who moved here as non-UK nationals before the age of (approximately) 16. While this is not a standard definition – analysis using the LFS distinguishes by country of birth, so a “migrant” is anyone who was born abroad and subsequently moved to the UK – looking at “adult migrants” who came to the UK as non-UK nationals is arguably more relevant for analysing migrants’ labour market outcomes (with the omission, as noted above, of many self-employed).

    In response to a series of Freedom of Information requests, HMRC have further disaggregated the data, providing information on the year the individual first appears in the PAYE dataset (if after 2014, when the data series begins), and earnings at the 25th, 50th and 75th percentile of the distribution for each subgroup. The year the person first appears in the PAYE dataset will not necessarily be the same year that they arrive in the UK or register for a NINO but should be the first time that they enter the UK labour market as a paid employee. In other words, for each year after 2014, we know how many employees in each of our three groups first took up payrolled employment in the preceding year, the year before that, and so on back to 2014; and for that group we have some basic information on the distribution of earnings. This means that, for the period since 2014, we can look at both changes in the composition of new migrant entrants to the (payroll employee) workforce over time, and the earnings trajectories (in aggregate) of different migrant cohorts, as well as measuring how many remain in the payroll employee workforce.

Background and broader trends

The period for which we have data saw very large shifts in migration flows, driven both by policy and external events, as well as dramatic developments in the labour market more generally. In 2014, transitional controls restricting free movement with Bulgaria and Romania ended, resulting in a sharp surge in EU migration; this reversed after the Brexit referendum, but non-EU migration began to rise. The pandemic led to a significant reduction in the number of EU resident workers, and then in January 2021 the introduction of the post-Brexit immigration system ended free movement while liberalising migration for work from outside the EU, which grew sharply in 2021 and 2022, especially in the health and care sector. These developments are summarised in both Portes (2022) and Sumption (2022) and are reflected in our data (Figure 1) as well as in published ONS migration statistics.

Figure 1

The figures show that the number of EU workers appearing on payrolls for the first time fell steadily over the period and then even further during the pandemic, but still remained close to 100,000 in 2021 and 2022, despite the introduction of the post-Brexit immigration system and the relatively low number of work visas granted to EU citizens. This is not necessaily an indication of the level of new arrivals to the UK as the number will include people who have the right to reside and work in the UK under the EU Settlement Scheme but had not previously worked as employees (e.g. people who were students, had caring responsibilities, or were self-employed)

The data also show the very sharp post-pandemic rise in new entrants to the employee workforce from outside the EU. While the number of work visas rose sharply during this period, this cannot account for the entirety of the rise, which will include people who were already resident here but had not worked before as employees, students, dependants of those arriving on student and work visas, as well as Ukrainian refugees and new arrivals from Hong Kong. In other words, even though most non-EU migrants coming to the UK in recent years have not been on work visas, they have nonetheless made a substantial contribution to the size of the labour force.

Attrition from the employee workforce

The HMRC data provide a crucial new insight into how long migrants remain in the employee workforce over time, because we can track how many people who first appeared in, say, 2015 were still in the dataset a few years later. If people disappear from the dataset it will not always mean that they have left the UK: some will have stopped working, for example to care for children, or become unemployed or moved to self-employment, although return migration is likely to be the major reason over the longer term.

Of the EU-origin migrants who registered in 2015, only about half were still on payroll in 2022 (Figure 2). Attrition of non-EU origin migrants was lower, but still significant. For UK-origin workers, it was unsurprisingly much smaller, but still represented about 1 in 6 workers—most likely due to moves out of employee status. For established workers—that is, those who first appeared in the dataset in 2014 or earlier—attrition rates were lower, with more than 60% of EU-origin workers still in the data in 2022, while three-quarters of already-established non-EU origin workers remained.

Figure 2

Earnings of new entrants to the employee workforce

The HMRC data also provide information on migrants’ earnings and how their distribution has changed over time. In theory, we should expect the end of free movement and the introduction of the post-Brexit immigration system to affect the composition of people coming to the UK, and thus their earnings. For example, the new immigration system requires EU workers to meet salary requirements, while free movement did not. On the other hand, skilled EU workers who are eligible for work visas under the post-Brexit immigration system might find the UK less attractive due to Brexit or the end of free movement. For non-EU workers, the increase in the number of people coming to the UK could in theory result in either higher or lower earnings among new entrants. The non-EU origin group includes larger numbers of skilled workers than in the past, but the Skilled Worker route considerably lowered salary requirements, which were further reduced by the extension of the Health and Social Care visa to care workers. Meanwhile, large numbers of people have arrived on other visas with no earnings criteria (such as Ukrainians or international students working part-time during their studies).

Since we have earnings data for the calendar year, earnings data in the year of entry is incomplete and is typically very low for new registrants, so we measure earnings in the first full year after the year the person first appeared in the dataset (for example, calendar 2016 earnings for those who first appeared in the dataset in 2015). We also show earnings for the whole sample for comparison, which includes UK and longer-established non-UK workers (Figure 3).

Figure 3

From 2016 to 2022, the earnings profile of migrant “new entrants” from outside the EU improved compared to the overall workforce. At the median, non-EU new entrants (who arrived sometime in 2021) earned 97% of the overall workforce median in 2022, up from 80% among new entrants in 2016. This is despite the very rapid expansion in their numbers; there is at this early state as yet no evidence that this has led to lower earnings among new non-EU migrants on average. But while non-EU new entrants are now entering the workforce somewhat higher up the pay distribution, EU-origin ones are slightly lower. This is more obvious at the 25th percentile, where new EU-origin workers were entering at significantly lower relative wages in 2022 compared to before the pandemic.

Does this mean that the post-Brexit immigration system has failed to attract a higher skills profile of new EU migrants, as it was intended to do? It is probably too early to say. The 2022 figure for “new entrants” to the labour market only covers one year of the new system, and will include some workers who arrived under free movement but have only entered the labour market more recently. However, the limited change in new entrants’ earnings profile may also result in part from lower migration of highly skilled EU workers who previously came to the UK under free movement rules. After all, visa grants to EU citizens have been surprisingly low.

Earnings progression over time

Our data also allows us to look at earnings trajectories over time, for the same cohort of migrants. A key caveat to all of the analysis is that because many migrants leave the employee workforce (as shown above), changes in average earnings over time may partly reflect changes in the composition of the workforce due to emigration.

On the one hand, some theories of migrant self-selection suggest that for obvious reasons migrants who succeed in the labour market are less likely to emigrate; however, some studies find the opposite, particularly for highly skilled workers, who may simply be more mobile. For example, there is evidence that recent out-migration of EU workers has been particularly marked in relatively low paid sectors like accommodation and hospitality, which might bias estimates of earnings growth for EU-origin workers upwards. However, this is less obviously the case for non-EU origin workers. Existing data suggests that migrants who come to the UK from outside the EU for family reasons or to claim asylum both have much lower earnings and are much less likely to leave, while most work migrants leave within a few years; this argues the reverse.

With these caveats in mind, we begin by looking at “established” workers, meaning here those (of whatever origin) who had already been recorded on the database by the end of 2014 (Figure 4). The data show that since 2014, median monthly pay grew faster among both EU and non-EU origin workers than among UK-origin ones.

In 2014, non-EU origin and UK origin workers who were already in the HMRC dataset by 2013 had similar earnings, while EU-origin workers lagged behind. By 2022, however, median pay in the non-EU group was 6% higher than for their UK counterparts, while median pay for EU workers was broadly similar to the UK level.

Meanwhile, non-EU workers were more concentrated at both the bottom and top of the income distribution. This is likely to be driven by those on skilled work visas at the top, and students, dependants and refugees at the bottom. Surprisingly, the earnings distribution for EU-origin workers was smaller than for other groups. At the lower end of the distribution, EU workers had higher earnings than either non-EU or UK workers. This is harder to interpret, since many at the 25th percentile of monthly earnings are likely to work part-time, and it is this is less common among EU-origin workers, according to the Labour Force Survey.

Figure 4

Overall then, we see some evidence that the earnings of both EU and non-EU employees increase over time relative to UK employees. As noted above, this simply represents a series of snapshots of the workforce at any one time. The data do not show whether this is due to the same migrants getting faster pay increases than their UK counterparts (which may be expected given that they will on average be younger) or due to selective out-migration of lower-paid workers—or some combination of these factors.

New entrants’ earnings progression over time

How do the average earnings of newly arriving migrants change over time? Here we are to some extent limited by the relatively short period for which we have data. However, even this limited period is an advance on previously available data. Here we show the median earnings for those entering the labour market in a given year, relative to the median earnings of the entire workforce in that year (Figure 5).

Figure 5

The data suggest that non-EU origin employees are entering the workforce at a higher point in the wage distribution and are (possibly) progressing somewhat faster than EU citizens. The net effect is that while it took 2015 non-EU entrants around six years to reach the overall median wage, 2019 entrants had already exceeded it after two years. This was a period of very considerable turbulence in the labour market, particularly during the period of the pandemic, and then more recently a period of historically high nominal earnings growth, but nonetheless suggests that more recent non-EU arrivals, as well as earning more immediately on arrival, are also seeing strong earnings growth relative to the overall workforce.

By contrast, there appears to be less rapid improvement in earnings for EU-origin workers at the median. Nonetheless, the 2015 and 2016 cohorts had reached the UK average monthly earnings within five or six years.


The analysis presented above is very preliminary and based on aggregate data only – we follow groups whose composition changes over time, and do not have information on important factors which impact earnings, such as gender and age. However, our initial analysis suggests that:

  • Substantial shares of migrants leave the UK’s employee workforce within the first few years, and this was particularly the case among EU workers who started working in the UK under free movement.
  • The earnings of non-EU workers first entering the employee workforce increased from the mid-2010s to 2021, suggesting that the average skills profile of non-EU migration increased over this period, at the same time as their overall numbers increased rapidly. Whether as a result of earnings growth or selective out-migration from the UK (or at least from the employee workforce), the monthly earnings of non-EU origin workers exceeded the UK median earnings within a few years for all cohorts for whom we have data.
  • Entry earnings for EU-origin workers were lower, suggesting that EU workers are entering somewhat lower down the wage distribution, even after the end of free movement. We only have very limited data for those arriving after 2020 – that is, those who will have arrived under the post-Brexit immigration system – but this may reflect the fact that most new EU-origin entrants are not actually using the post-Brexit system for work visas but are entitled to work here via other routes. Alternatively, it could also reflect self-selection in outward migration.

These conclusions are very tentative, but overall suggest the migrant workforce is gradually improving its position in the wage distribution and that the recent shift from EU-origin workers to non-EU origin workers, driven both by Brexit and broader economic and labour market developments here and in the countries of origin of migrants to the UK, may be accentuating these trends.


Thanks to Michael O’Connor for comments on a previous draft. This analysis was produced with the support of Trust for London, one of the largest independent charitable foundations in London, which supports work that tackles poverty and inequality in the capital. More details at www.trustforlondon.org.uk.

Press Contact

If you would like to make a press enquiry, please contact:

Rob McNeil

+ 44 (0)7500 970081

 Contact Us 


This Migration Observatory is kindly supported by the following organisations.

  • University of Oxford logo
  • COMPAS logo
  • Esmee Fairbairn logo
  • Barrow Cadbury Trust logo
  • Paul Hamlyn Foundation logo