Why are the latest net migration figures not a reliable guide to future trends?

by Tessa Hall, Alan Manning & Madeleine Sumption

This report examines how different types of immigration contribute to net migration in the long term and what levels of net migration the UK might expect in coming years.

Note: this analysis was originally published in October 2023 and has been updated following the release of new official migration statistics in November 2023.

1. Introduction

Net migration has been unusually high in the past two years. The Office for National Statistics (ONS) estimates that net migration to the UK was 745,000 in 2022, up from 184,000 in 2019 before the pandemic (ONS, November 2023). The ONS methods are experimental and provisional, but the basic picture—that net migration has increased sharply—is unlikely to change as the official data are revised.

High net migration figures have intensified debate over UK migration policy and the future outlook for migration. Politicians of both major parties have said that they think net migration is ‘too high’ and would like to see lower numbers. At the same time, ONS population projections assume that net migration will fall to 245,000 per year. The Office of Budget Responsibility (OBR) has applied the same assumption in its economic forecasts.

Why is the gap between current and projected future net migration so large? Is it plausible that net migration will fall by as much as 67% over the next decade or so, to around the 245,000 that ONS assumes? This short report draws on analysis from the Migration Observatory and the Centre for Economic Performance at the London School of Economics to examine this question.

Net migration in the UK is defined as the number of people who immigrate to the UK long term (for at least 12 months), minus the number who emigrate long term. But often discussions of net migration overlook emigration, and specifically the fact that most migrants come to the UK for periods of only a few years, before emigrating again. This means that while net migration may look unusually high over the next couple of years, emigration should eventually rise too, causing net migration to fall even if the number of people immigrating to the UK remains high.

To illustrate this, we present a ‘bottom-up’ approach used to project net migration from 2023 to 2030 based on a set of assumptions about levels of immigration and the share of migrants in each immigration category who subsequently leave the UK. Using past data on visa grants and expiries, we estimate the contribution of each type of migrant to overall net migration. The model does not attempt to forecast immigration itself; past research shows that this migration is inherently difficult to predict. Instead, we simply assume that current immigration levels continue in most (but not all) categories, and examine what happens to net migration if migrants stay in the UK at the same rate that they have done in the past.

The analysis is thus not designed to be a confident prediction of future net migration. Forecasts are invariably wrong and this one will be no different. Different assumptions will lead to different outcomes and reasonable people will disagree about which set of assumptions is most plausible. Instead, we aim to illustrate some basic points about the dynamics of net migration during periods such as this one. In particular, the analysis suggests that:

• High immigration leads to high emigration, but not immediately – there is a lag of two to three years. Unless there is a large change in emigration behaviour from what we have observed in the past, it is reasonable to expect that emigration will increase between now and 2025, bringing down net migration, even if the number of people arriving in the UK remains high by historical standards. Of course, behaviour may change and this would make the projections in our model less accurate.
• In our baseline scenario, net migration falls to around 350,000 by 2030, roughly similar to pre-Brexit levels. A majority of this decline results from an increase in emigration, particularly of international students. A decline in humanitarian immigration also plays a role in the decrease.
• However, there is a wide range of plausible scenarios for future net migration. The projection is sensitive to assumptions about future immigration levels and how long migrants will stay in the UK. For example, if instead of assuming a decline in asylum applications we assume that 2022 levels continue indefinitely, the 2030 net migration projection becomes 390,000. However, the conclusion that net migration is likely to fall from its 2022 peak holds true under a wide range of assumptions. Even if immigration for all categories—including Ukrainians, whose numbers have already declined sharply—remained at 2022 levels, projected net migration in the model still falls to 500,000 due to rising emigration.

To examine the potential future trajectory for net migration, we need to make assumptions about future immigration, i.e. long-term arrivals (Section 2). We must then understand what share of migrants in different immigration categories are likely to emigrate (Section 3). We then examine what this means for future net migration (Section 4). In Section 5 we explore how our 2030 net migration projections change under different assumptions about future immigration levels and stay rates. And in Section 6 we describe various future net migration scenarios in the UK context.

2. What might future immigration look like?

Immigration into the UK can be broadly subdivided into different categories depending on people’s migration status. Key categories include immigration on mainstream visas (for work, study, family, and other); people in new humanitarian visa categories (BNO route and Ukrainian visa schemes); other humanitarian routes (asylum seekers and resettlement refugees); and immigration of those with free movement rights from before Brexit (EU citizens with pre/settled status under the EU Settlement Scheme and British citizens).

The future outlook for each category is uncertain. However, we can make transparent assumptions about what immigration might look like for each group. Table 1 in Section 5 shows how our future net migration projection reacts to each assumption being changed within reasonable bounds. The assumptions are described in more detail in the LSE technical paper that develops the model. They are necessarily speculative. In practice, some will turn out to be too high and others too low. The model is based on the assumption that:

• Future immigration of people on work, study, family and other visas except Ukraine and BNO schemes continues at the ONS’s estimated December 2022 level. These categories made up the majority of immigration in 2022. In practice, future levels are hard to predict. Work visa grants have already increased from this level, although recently announced restrictions on care workers’ family members may reverse some of that increase; student visa grants are also likely to decrease following the government’s decision to restrict students’ ability to bring dependants. Neither of these changes are factored in.
• The Ukraine and BNO humanitarian visa categories decline somewhat in coming years, as there is some reason to believe they will not continue indefinitely. In particular, the crisis in Ukraine has been an important driver of increased visa grants. The numbers of visas granted was highest in the early weeks after the Russian invasion and have since tapered off. The number of BNO visas issued has been declining since the route first opened on 31 January 2021, from 29,311 in Q2 2021 to 8,647 in Q2 2023. In our projections we assume that BNO and Ukrainian visas issued will continue to decline in future.
• Future asylum immigration settles at 50,000, based on the 2019-2022 average level, and total refugee resettlement is 5,000 per year, in line with previous government commitments. Asylum related migration is volatile and particularly difficult to forecast. In the past, upward spikes in the number of asylum applicants have often been temporary. However, we also include a scenario in Section 6 in which asylum applications continue at 2022 levels.
• Net migration of British citizens returns to and continues at its pre-pandemic (2015-2019) average level of -30,000. Estimated immigration of British nationals was unusually high in 2020 – 2022 and emigration unusually low.
• Net migration of EU citizens who are not coming on visas under the post-Brexit immigration system (i.e., who have settled or pre-settled status under the EUSS) increases to zero over the next few years. That is, we assume that EU citizen net migration will not be negative forever and that EU citizens will start contributing to overall net migration due to their use of the new visa system.

Figure 1 shows the overall effect of these choices on total assumed immigration from visa and humanitarian routes. We assume that long-term immigration from these categories will fall by 18% from 1,060,000 in 2022 to 875,000 in 2030. This is primarily due to fewer people entering from Ukraine, as well as a small decline in asylum applications and immigration from the BNO route. We do not factor in the impacts of recent policy announcements restricting legal migration.

Figure 1

3. What might future emigration look like?

The main purpose of our model is to project future emigration (rather than immigration, which we assume mostly remains at 2022 levels), and thus understand what may happen to net migration.

Every year some migrants leave the UK. The likelihood of this happening will depend both on the reason for migration and how long the person has been in the UK. For example, international students are much more likely to leave than people coming to join British or settled family members. People who have been in the UK for a long time are more likely to stay permanently. In practice, Home Office Migrant Journey data suggest that most non-EU citizens who leave do so within 8 years.

To project future net migration, we need assumptions about what share of migrants who arrive in a given year and in a given immigration category will leave in the subsequent years.

• For work, study, family, and other immigration categories, we calculate these migrant ‘stay rates’ using Migrant Journey data, which shows which migrants still have valid permission to remain in the UK at the end of each year after their initial visa grant (refer to the LSE technical paper for details). We use Migrant Journey data for migrants who arrived between 2010 and 2014, for which we can observe their visa status up to 8 years after entry to the UK. Our model assumes that migrants who have been in the UK for longer than 8 years will stay indefinitely.
• A separate stay rate is calculated for each visa category and year of arrival cohort. These are adjusted to account for the changing characteristics of migrants in each cohort, including the rising share of Skilled Worker route grants and the declining share of intracompany transfers and change in the countries which migrants are coming from. For the stay rate of future arrival cohorts, we use the most recent adjusted stay rates (2022).
• We adjust the work visa stay rate upwards substantially from past levels to account for the rising share of Skilled Worker route grants and the declining share of intracompany transfers.
• We adjust the student stay rate upwards, since Migrant Journey data shows that recent cohorts of students are more likely to stay in the UK.
• The likelihood of migrants staying in the UK varies by country of origin, and the countries which migrants are coming from has changed over time. Our region-of-origin adjustment leads to higher stay rates in our projection, as a greater number of migrants in 2022 originated from countries with high stay rates than in the past (Figure 2). The region-of-origin adjustment also addresses the increasing number of dependants per main applicant over time. More dependants generally mean higher migrant stay rates. So, the stay rates calculated using the 2022 ratio of dependants to main applicants will be higher than they have been historically. This is accounted for the region-of-origin adjustment. In fact, even if we further adjust stay rates to account for the ratio of dependents to main applicants, the impact on estimated stay rates is not significantly different from those calculated using country mix adjustment alone.
• Data on stay rates are not yet available for BNO and Ukrainian refugees, because these routes are new. We assume the majority (90%) of BNO migrants will remain in the UK after 8 years. Ukrainians currently have 3-year visas and the government has not said whether it will extend these visas in future. Some Ukrainians may prefer to stay in the UK if the option is available. We assume an 8-year stay rate of 70%, which we recognise is necessarily arbitrary.
• For asylum seekers, we assume a stay rate of 90%, based on the fact that most currently receive refugee status and in recent years removals of refused asylum seekers have been low. We assume that all resettled refugees remain in the UK indefinitely.

Figure 2

Emigration depends on a combination of immigration levels and the rate at which migrants who have immigrated leave the UK over time, as the LSE technical paper describes in more detail.

Because immigration levels have increased, emigration levels are expected to increase in future too. The projected path of future emigration, based on our future immigration assumptions and assumed stay rates, is shown in Figure 3 for visa and humanitarian migration. The projected increase in emigration results primarily from larger numbers of international students leaving the UK, followed by an increase in emigration of work-visa holders.

Figure 3

Students play a large role in projected future emigration. However, it is important to note that trends in international student migration have changed enormously recently. More students have come to the UK, from different countries than in the past. A policy change towards students’ dependants has also been announced but not yet implemented, at the time of writing. Students’ migration patterns are thus particularly difficult to predict and it is possible that the timing or scale of emigration could differ from what we see in past data.

4. What does this mean for future net migration?

Subtracting projected emigration (Figure 3) from our assumed immigration levels (Figure 1) and adding the assumed net migration of British and EU citizens with pre/settled status gives a projection for overall net migration, shown in Figure 4.

These figures should not be used to predict what the next set of official net migration statistics will be. The precise timing of emigration and the levels of immigration will likely change. The purpose of this analysis is simply to illustrate a plausible scenario for the overall trend in coming years.

Figure 4

In the model, past immigration is taken directly from ONS estimates, but emigration is not—it is projected based on previous immigration flows and can thus differ from published ONS figures, which have a more complex methodology and use non-public information about entries and exits at the border (and not just visa grants and expiries).

Indeed, for 2022, our model predicted higher emigration than ONS believes actually took place. As a result, our model also predicted lower net migration—712,000 for 2022, compared to 745,000 in the ONS estimate. This underlines the fact that the model is not expected to predict net migration precisely. While the model predicts that emigration increases substantially between 2022 and 2024, if the ONS figures are correct it suggests that the increase in emigration was lower than the model anticipates.

With these caveats, the main scenario shows net migration falling rapidly, to around 350,000 by 2030—although there are many factors that could change this course, which are outlined more in the next sections of this report. Of the total decline in net migration from 2022 to 2030, 190,000 results from a fall in immigration, mostly because of fewer Ukrainians coming to the UK. Another 300,000 results from rising emigration, primarily of international students. However, the decline is partly offset by the net migration of EU pre/settled status and British rising by 130,000. This is because we assume EU pre/settled status net migration goes from its current negative value to zero, more than offsetting the small decrease we’ve assumed in British net migration.

5. Net migration in the longer run

Using assumptions about the level of immigration and the rate of emigration, we can project long-run net migration in a ‘steady state’. This is the long-run emigration level in each category multiplied by the fraction of those who stay indefinitely. Table 1 shows how much different types of immigration contribute to long-term net migration, assuming that emigration rates remain roughly similar to what they have been in the past. Readers can adjust the assumptions in the table to see how long-run net migration might respond in different scenarios.

Future immigration could differ from our baseline assumptions for a variety of reasons. Some of these are explored in more detail in Section 6. These include the possibility of increased or reduced demand for overseas workers, changes in stay rates, a possible decline in international students due to recent policy changes, and continued high levels of asylum migration.

Note also that the model uses official ONS immigration figures as an input, and these have been substantially revised over time. Any further revisions from ONS would change the data used in the model, and therefore also the projections.

In our baseline scenario, people on work visas contribute the largest share of net migration in the long term. Students are the second-largest contributor. Even though students’ stay rates are lower than any other group, the number of students is sufficiently large that a small share remaining permanently can make a meaningful contribution to long-term net migration. (Note that most students who remain in the UK long term do so by switching to skilled work visas, so work visa policy is likely to affect students’ stay rates more than student visa policy.)

Note that the long-run immigration estimates in Table 1 are lower than the number of visas issued. This is because some people do not take up their visas and others stay for less than a year and so are not considered long-term migrants in the official net migration estimates. Readers should not adjust the table using visa grant numbers as this will greatly overstate projected net migration. For guidance on plausible levels of immigration by reason for the main categories, readers can consult previous ONS estimates for non-EU citizens in Figure 3 of the May 2023 ONS migration publication.

Readers should also note that just because it is possible to use this tool to produce a long-run projection at a given level—high or low—does not mean that it is plausible or likely.

Table 1

6. Future net migration scenarios

Future net migration trends are highly uncertain. If immigration levels are different from the baseline scenario laid out above, or if stay rates change over time, this will affect long-run net migration—as Table 1 shows. Indeed, it is extremely likely that at least some of the trends projected in Table 1 will change substantially in future because immigration patterns tend to evolve over time. Below, we discuss some of the main uncertainties in more detail.

Changes to student stay rates: The Graduate Visa route, introduced in July 2021, allows international students to stay on in Britain 2 years after they have graduated. This may increase student stay rates if it enables students to find a job that allows them to stay in the UK, and hence it may increase net migration from this category. The availability of long-term visas for social care jobs may also affect student stay rates because it is relatively easy for people to get these jobs if they want to remain in the UK. Care workers are exempt from the £38,700 salary threshold the government has said it will apply from Spring 2024 onwards. Due to high immigration in this category, long-term net migration is very sensitive to stay rates. An increase or decrease in the student stay rate of 5 percentage points changes our long-term net migration estimates by 21,000.

Decline in student numbers: Government restrictions on the ability of international students to bring dependants, which come into effect in January 2024, could cause a fall in student immigration. This stems from the mechanical effect on student dependants as well as a potential fall in student numbers themselves, as studying in the UK becomes less appealing. For example, a fall in long-term immigration of people on student visas (including dependants) of 100,000—a 23% decrease—with no change in the return rate, would reduce the long-term steady-state projection for net migration by 22,000.

Changes to worker immigration: In 2023, work visa grants had already risen from 2022 levels, driven by larger numbers of health and care workers. As vacancy rates in the UK labour market have started to ease off over the course of 2023, there was some sign of work visa grants in the private sector starting to decline. In December 2023, the government announced restrictions on work migration, including a higher salary threshold for the minority of skilled worker route jobs that are not in health and care; a ban on partners and children of care workers; and more regulation of licensed sponsors in the care sector. Earlier in 2023, the government released an NHS workforce plan, which aims to reduce reliance on overseas recruitment of doctors and nurses in the medium term.

It is very difficult to predict the impact of this package of changes. While it is reasonable to assume that they will reduce work immigration (including work dependants, who are included with workers in the ONS figures), this is not guaranteed if demand for care workers remains very high and the government continues to take no action on care worker pay. If changes in policy or labour market demand decreased worker long-term immigration by 100,000 (a 35% decrease), this would reduce overall net migration by -57,000. A decrease of 100,000 (including dependants) would still leave skilled work migration well above 2019 levels.

Higher worker stay rates: Future migrant worker stay rates might be higher than those observed in past Migrant Journey data, even after adjusting for the country of origin of migrants and the rising share of skilled worker visa grants. One reason is the change in settlement policy, so that people no longer need to meet the settlement salary threshold, which was introduced in April 2011. Also, immigration has been rising rapidly for health and care workers. The number of health and care worker visas issued increased by 153% between Q2 of 2022 and Q2 of 2023. This group is more likely to bring dependants and thus may be more settled in the UK. We know from the Migrant Journey data that dependants (and hence also the main applicants they come with) have higher stay rates.

A worker stay rate of 70% would mean an increase in our long-term net migration projection of 37,000.

Asylum applications at 2022 level: Asylum applications can have a significant effect on net migration due to high stay rates. We assume that asylum immigration will fall from its 2022 level. If, however, the 2022 levels of long-term migration resulting from asylum applications (92,000) continued, the steady-state net migration would be 38,000 higher.

The scenarios described above could occur simultaneously. If, for example, the graduate visa route increased student stay rates to 27%, worker long-term immigration remained at 2022 levels, and asylum applications remained at their 2022 levels, our resulting long-term net migration projection would settle at around 415,000. If, on the other hand, we saw the student stay rate fall to 17%, student numbers and demand for migrant workers fall by 100,000 each, the long-term net migration projection becomes 260,000.

7. Conclusion

There is a wide range of plausible trajectories for future net migration, making the precise future level extremely difficult to predict. But although the future is uncertain, there is good reason to believe that net migration will fall over the next few years—assuming there is no new major shock that affects immigration to the UK (as the Ukraine war did in 2022, for example). The most important reason for this is that the recent shift to a higher immigration level should cause emigration to rise, offsetting the long-run impact on net migration.

The model presented in this report suggests that the largest category affecting the projected level of net migration in the coming years is the work route. In our central scenario, work-related migration (including dependants) accounts for 45% of net migration in the long term. Work migration has increased sharply, largely driven by the health and care sector (including the opening of the care worker route).

Even though most students eventually leave the UK, they still make a meaningful contribution to net migration (26% of net migration in the long-run model).

Of the categories in our model, work and study are the ones that are most amenable to policy changes. Policymakers interested in shaping net migration levels in either direction are thus likely to find more impact in these categories. Indeed, work-visa policy is likely to be the main factor affecting the contribution of both work and study to long-term net migration, since students usually need to switch to work visas in order to remain in the UK permanently.

Before Brexit, projections suggested that migration levels would fall due to the post-Brexit policy regime’s restrictions on EU migration. While the future outlook is necessarily uncertain, our model suggests that the current post-Brexit immigration system will not necessarily lead to lower net migration than the figures of roughly 250,000 to 350,000 seen in the mid-2010s. The largest reason for this is the increases in non-EU work and study migration. For example, non-EU long-term work migration increased from 99,000 to 285,000 between 2019 and 2022, according to the ONS estimates. This increase of 62,000 per year, if sustained indefinitely with a stay rate of 57%, would imply an additional 35,000 of net migration per year in the long term. Similarly, non-EU long-term study migration increased by 300,000 over the same period. With a projected stay rate of 22%, this implies an extra 22,000 in annual net migration in the long term, compared to 2019 levels.

Finally, many factors could throw off the projections presented here off course. For example, the model is based on the assumption that past patterns of behaviour will be a reasonable guide to the future. However, if stay rates—particularly of students or workers—change significantly beyond what we already assume will result from the changing nationality profile, this would affect overall net migration. Perhaps most importantly, the number of people arriving in the country is very difficult to forecast. This is why for the most part we do not attempt to do so, but simply examine what would happen if 2022 immigration levels were sustained (with some small exceptions including humanitarian migration and EUSS net migration). This uncertainty underlines why it is difficult for the government to commit to delivering a specific level of net migration, even though it controls the visa system.

Acknowledgements

This briefing was produced with the support of Trust for London. The Trust is one of the largest independent charitable foundations in London and supports work which tackles poverty and inequality in the capital. More details at www.trustforlondon.org.uk.Thanks to everyone who gave comments on this analysis, especially Carlos Vargas-Silva and Georgina Sturge. All errors are the authors’ own.

Note on changes made to this report

December 2023:

• Updated immigration figures and ONS net migration comparison (YE Dec 2018 – YE Dec 2022) with November 2023 ONS estimates.
• Adjusted student stay rates in central scenario upwards (LT stay rate changed from 18% to 22%). For each cohort of arrival (2010 – 2022), used the conditional stay rate for the years after arrival 1 – 8 from the most recent available years of MJ data (2022 report). For example, for the 2022 arrival cohort (and all subsequent years) used the % of the cohort that left the UK after 1 year from the 2021 arrival cohort, the % of the remaining that left after 2 years from the 2020 arrival cohort, etc.

Data sources

• Home Office: Asylum and resettlement datasets (2023): Asy_D01: Asylum applications raised; Asy_D02: Outcomes of asylum applications at initial decision, and refugees resettled in the UK
• Home Office: Entry clearance visa applications and outcomes (2023): Vis_D02 Outcomes of applications for entry clearance visas
• Home Office (2020): Fifth report on statistics relating to exit checks: 2019/20
• Migrant Journey: 2022 report: MJ_D01 End-of-year category and immigration status of all applicants issued leave by year of initial grant, category of initial visa, and country of nationality
• ONS LTIM (2023): Long-term international migration, provisional: year ending December 2022
• ONS LTIM IPS estimates (Aug 2020): Provisional long-term international migration estimates (Discontinued after August 2020)
• ONS Methods (2023): Methods to produce provisional long-term international migration estimates