2.1. Brexit: The Impact on Migration

Although our research focuses chiefly on international student mobility, it is important to understand more broadly how Brexit and the end of free movement impacted migration into the UK. The UK voted to leave the EU in a referendum held on June 23, 2016. Although the referendum to leave the EU occurred in 2016, free movement ended on December 31, 2020 when the UK officially withdrew from the EU single market and the transition period ended. Prior to withdrawal, EU nationals were able to freely live, work, and study in the UK without additional visa requirements and vice versa. However, from 2021, new and stricter requirements were placed on those who wanted to move between the UK and the EU (UK Parliament, 2020).

While the UK had previously been a popular destination for EU migrants, research indicates a decline in its popularity as a destination after 2016. Before 2016, EU citizens comprised the majority of net migration to the UK. According to figures from the Office for National Statistics (ONS) and the Migration Observatory, EU citizens accounted for between 59% and 77% of net migration in the year ending 2015 (The Migration Observatory, 2024). However, following the referendum, evidence shows a decline in EU migration to the UK.

Using a DID approach, Di Iasio and Wahba (Reference Di Iasio and Wahba2023) demonstrated a decrease in net migration of EU nationals from the UK after the EU referendum, particularly among those whose primary motive was to work in the UK. Additionally, Sîrbu et al. (Reference Sîrbu, Goglia, Kim, Magos, Pollacci, Spyratos, Rossetti and Iacus2024), utilizing a combination of traditional and digital trace data, reported a decline in movements between EU countries and the UK from 2016 to 2020. Further, Rampazzo et al. (Reference Rampazzo, Bijak, Vitali, Weber and Zagheni2024), using digital trace data, identified a decrease in migration to the UK from the EU post-referendum, especially among those aged 20–29. Both Sîrbu et al. (Reference Sîrbu, Goglia, Kim, Magos, Pollacci, Spyratos, Rossetti and Iacus2024) and Rampazzo et al. (Reference Rampazzo, Bijak, Vitali, Weber and Zagheni2024) also highlighted that Eastern and Central European countries were particularly affected by these declines.

Less work has focused on the post-withdrawal period beginning in January 2021. This is in part due to the added complexity of the COVID-19 pandemic, as well as the availability of timely data. That said, Official statistics from the Office of National Statistics (ONS) show a decline of around 40% in EU migration to the UK between December 2020 and December 2023 (Office for National Statistics, 2024). Also, the ONS shows that although there were declines in EU migration from the 2016 referendum onward, the steepest declines occurred after the first COVID-19 lockdown in 2020 (Office for National Statistics, 2021). Further, data from the Migration Observatory indicates that declines in migration were the greatest for Eastern European countries such as Poland, Romania, Lithuania and Bulgaria (The Migration Observatory, 2021).

Given these marked declines in migration and mobility from EU countries to the UK, it makes sense that student mobility would also decline. It is important to understand how student applications can be contextualized within broader changes to EU migration into the UK post-Brexit.

2.2. Brexit: The Cost of Ending Free Movement for International Students

For EU students, the cost of studying increased due to the end of parity with UK student fees, reduced opportunities to secure loans, and tightened visa requirements. There is an undoubtable economic impact of Brexit on EU students, making the costs associated with studying in the UK much higher for cohorts beyond the 2021/22 academic year than for their predecessors.

Literature on the influences on international student mobility often discusses the impact of financial costs on the decision to study abroad. Mazzarol (Reference Mazzarol1998) and Mazzarol and Soutar (Reference Mazzarol and Soutar2002) include the cost of fees, alongside the cost of living, as a key influence on international student mobility. There is also evidence that EU students have responded negatively to increased fees in the past, with applications dropping in 2012 following an increase in fees from £3375 to £9000 in England, Wales and Northern Ireland. The 2012 increase saw a 13-percentage point decline in the number of EU students studying in the UK after its initial introduction, although this did recover (QS, 2019). Issues of financial burden have arguably prevailed as an enduring concern, with a survey conducted on prospective international students in 2019 finding that 36% of students from the EU who would have considered studying in the UK became “less interested”, with 65% of those stating that this is because the UK is now a less financially viable option (QS, 2019). Although fees are the only one deterring costs for international students, it is an important one. The increase in fees will have played a role in declining numbers of applications from students from EU countries.

In addition to increased fees, the end of free movement limits EU students’ ability to remain in the UK to work after graduation. In an analysis of international students’ intention to return after graduation Falkingham et al. (Reference Falkingham, Giulietti, Wahba and Wang2021) showed that, after the triggering of Article 50 to leave the EU, EU students were much more likely than non-EU students to plan on leaving the UK after graduation. In addition, they found that this effect was greatest for those from countries who joined the EU after 2004 enlargement.Footnote ² In the literature more generally there is evidence that international students will seek to stay in the host country post-graduation. Rosenzweig (Reference Rosenzweig2008) highlights how international students are likely to stay in the host country once they have completed their studies. Likewise, Robertson (Reference Robertson2011) identified how students who come from less wealthy backgrounds may be motivated to convert to a long visa after graduation. However, given that the right to remain for EU students was revoked by Brexit, the impact of having to apply for a visa and the extra costs associated with this are likely to negatively impact the attractiveness of the UK as a destination. In a survey of prospective EU students conducted post-referendum but before the end of free movement or the pandemic, of those who were deterred by Brexit from studying in the UK, 47% were worried that it would be harder to get a job when they graduated, and 42% were worried it would be harder to get a visa (QS, 2019). Such responses point to an unease around prospective job opportunities post-Brexit.

Literature on international student mobility often points to the relative wealth between countries as a driver for students studying abroad. Many authors have explored the relationship between per capita income of the origin country and outward flows of international students. The negative impact of per capita income in the origin country on outward student flows has been well documented. When considering the UK as the sole destination, countries with lower GDP per capita tend to have smaller flows of internationally mobile students (Zheng, Reference Zheng2014; Neville et al., Reference Neville, Rowe and Singleton2024). In relation to Brexit specifically, in a study considering changes in applications in EU students immediately after the 2016 referendum Amuedo-Dorantes and Romiti (Reference Amuedo-Dorantes and Romiti2021) found the largest drops came from countries with a lower GDP per capita and higher unemployment rate. These results could suggest that in a post-Brexit context, we can expect to see the greatest declines from those countries where GDP is lower.

2.3. Brexit: Negative Sentiments Towards Immigration

Further to financial and economic effects, other confounding factors that have been shown to deter international students include the wider socio-political environment. The decision to leave the EU was, in part, driven by campaigns that promoted negative feelings towards immigration (Goodwin and Heath, Reference Goodwin and Heath2016; Goodwin and Milazzo, Reference Goodwin and Milazzo2017; Hobolt, Reference Hobolt2016). Immigration was a central component of the debate during the referendum to leave the EU, with dissatisfaction with the EU being associated with anti-immigrant sentiment (Ford and Goodwin, Reference Ford and Goodwin2017). Data from the British Election Study shows that over half of Leave voters were “intensely opposed” to immigration, with 65% viewing immigration as “bad for the economy”, and 72% perceiving immigration as “undermining culture” (Goodwin and Milazzo, Reference Goodwin and Milazzo2017).

In relation to international students specifically, less research has been conducted on their experiences of and reactions to anti-immigrant sentiment. However, Mazzarol (Reference Mazzarol1998) and Mazzarol and Soutar (Reference Mazzarol and Soutar2002) highlight the risk of discrimination and lack of safety as two key social costs that prospective students consider when deciding where to study abroad. There have been several calls to investigate international student responses to more hostile policy environments. In the context of the U.S. Laws and Ammigan (Reference Laws and Ammigan2020) draws attention to the reduction in international student numbers following the election of Donald Trump in 2016. Similarly, Rose-Redwood and Rose-Redwood (Reference Rose-Redwood and Rose-Redwood2017) express the need for researchers to consider the impacts of anti-immigration policy environments on international students. Bartram (Reference Bartram2018) refers to the current political climate as a “landmark political and cultural shift with potentially significant implications and consequences for international students.” Despite these calls, there has been a lack of empirical work on how international students have been affected by anti-immigrant sentiment and hostile policy environments.

Although quantitative evidence that EU students were deterred by anti-immigrant sentiment is limited, recent results from surveys indicate that there has been an increased sense of a hostile environment since the referendum in 2016. Of those prospective students from the EU who were deterred from studying in the UK as a result of Brexit, the second biggest reason (after financial vulnerability) was that the “UK is less welcoming to international students like me” – with 57% of prospective EU students responding in this way (QS, 2018). Furthermore, in the 2022 iteration of the survey, prospective students from Western Europe and to a greater extent Eastern Europe were less likely to see the UK as “extremely” or “somewhat” safe in comparison to their counterparts in Latin America, North America, or Africa and the Middle East (QS, 2022). These results all point to a sense of vulnerability being experienced by prospective EU students that exists beyond financial insecurity.

2.4. COVID-19: The Impact on International Student Mobility

While the impacts of Brexit and the end of free movement came into force from August 1, 2021 for international students, earlier restrictions were enacted as measures of response to control the spread of COVID-19. The COVID-19 pandemic led to significant disruption to the global economy and movement of people between countries and made it more difficult for international students to travel abroad to study. The scale of disruption caused by the pandemic became clear in the Spring of 2020. Within the UK context, lockdown restrictions legally came into force at the end of March 2020. However, the pandemic and its responses progressed at different rates in different parts of the world, and other countries began their restrictions earlier. UK lockdown persisted on and off across 2020 and into early 2021, while restrictions remained in place into 2022.

The 2020 closure of university and college campuses had a significant impact on higher education providers, with many courses being delivered online (Hubble and Bolton, Reference Hubble and Bolton2020). While universities did attempt to mitigate the impact of COVID-19 on students there was a significant fear that overseas student numbers would fall as a result of the pandemic (European Commission, 2020). Concern arose that restrictions would lower applications for the 2020/21 academic year and that students would instead wait until the pandemic was over.

Surveys conducted into the deterrents for international students from OECD and EU countries to any destination as a result of the pandemic by the European Commission (2020) identified several key concerns for students from member states. These concerns included the ability to return to their country of origin due to travel restrictions, with 84% of prospective students being concerned about restricted travel options. Additionally, there was a concern that national administrations and universities would not be in operation, and as such student visas, residence permits, and admissions procedures would be affected. Another key concern was the ability of international students to support themselves financially due to the lack of student jobs.

Studies on the impact of the COVID-19 pandemic restrictions on international students are still in their infancy. Although research from González-Leonardo et al. (Reference González-Leonardo, Rowe, Potančoková and Goujon2024) shows how there was a decline in migration into high-income countries in Europe as a result of the pandemic, less is known about international students. However, looking at applications to UK universities during the pandemic, Di Pietro (Reference Di Pietro2023) found that the COVID-19 pandemic period was associated with a reduction in the number of overseas applications (not accepted applications) to UK universities of between 11% and 14% for those applying in the Summer and Autumn of 2020. Furthermore, the author finds that these declines were heterogeneous across different countries of origin, with greater declines seen in affluent countries. Other research takes a more global perspective, with Yang et al. (Reference Yang, Shen and Xu2022) pointing to evidence of a contraction in the global market for international student mobility as well as changes in the geographic patterns of international student mobility.

While literature points to a potential decline in international student mobility as a result of the pandemic, data of registered students released by HESA (2023) determines that there was little difference in the 2020/21 enrolments compared to a normal year. However, there are no studies with such recent application data as our study. Furthermore, analysis of the impact of COVID-19 has often overlooked the impact of Brexit, and vice versa, despite the two events being temporally linked (Figure 1).

Figure 1. Timeline for COVID and Brexit events.

2.5. Heterogeneity in the System: Differences Between UK Higher Education Institutions and Subjects

There are several studies that identify the importance of an institution’s reputation in the attracting of international students. The reputation of the institution is often considered an important “pull” factor for international students when choosing where to study Mazzarol (Reference Mazzarol1998). The reputation of the academic institution allows potential students to choose between the options available and provide some knowledge about the institution (Mazzarol and Soutar, Reference Mazzarol and Soutar2002; Beine et al., Reference Beine, Noël and Ragot2014). Given the importance of reputation, when exogenous shocks and policy changes occur, it may be the case that certain institutions are privileged over others. In these cases, universities that do not have as good of a reputation may suffer to a greater extent.

The UK system is marked by heterogeneity, and some differences exist within it that may be impacted variably by declines in student numbers. UK institutions vary depending on their research activity, perceived quality of teaching and learning, and the socio-economic mix of their students (Boliver, Reference Boliver2015). Croxford and Raffe (Reference Croxford and Raffe2015) suggest that an institutional hierarchy is deeply embedded within the UK HE system. These differentiations are best represented by the Russell Group, which was established in 1994 and represented the largest aggregation of research-intensive institutions in the country (Singleton, Reference Singleton2010). The Russell Group represents 24 leading research-intensive universities and began meeting in 1994 after the creation of the “new” universities. As of 2019/20, the Russell Group produces 68% of the UK’s world-leading research and teaches a quarter of all undergraduate students and a third of all post-graduate students (Russell Group, 2020).

Beyond institutions, there are differences across subject areas. From an economic perspective, different subject areas have different impacts on graduate earnings. According to reports from the Institute of Fiscal Studies (2018), the returns on early-career earnings are much greater for students who study subjects such as Medicine or Economics compared to subjects like Creative Arts and Agriculture. These returns are reflected in the popularity of different courses for international students, with UCAS (2022) reporting that the courses that attracted the most international students were Business and Management, Engineering, Law, and Computing. However, Creative Arts and Design was still a popular choice for international students, suggesting that economic returns are not the only driver of course choice for international students. Interesting, when considering the post-referendum period, work by Amuedo Dorantes and Romiti (Reference Amuedo-Dorantes and Romiti2021) identified a greater slowing in the growth of applications from EU students for STEM subjects compared to non-STEM subjects after the 2016 referendum despite the expected resilience of STEM subjects. All considered, it is likely that different subjects will see different effects from the impact of Brexit in terms of the reduction in student numbers. It is therefore important to consider these when understanding changes to external shocks and policy changes.

4.1. Data

Our data are sourced from the UK Universities and Colleges Admissions Service (UCAS) to capture successful undergraduate applications to UK HEIs, which we use as a proxy for origin-destination undergraduate student migration flowsFootnote ³ In the UK, UCAS is the main route of applications for students who want to study an undergraduate degree at a UK HEI and is the largest channel for internationally mobile students entering the UK (UCAS, 2020). We have access to count data indicating the number of students by country of origin who were accepted to study a particular subject in a particular institution for a given year using the UCAS applications service between the 2012/13 and 2022/23 academic years. These data accounted for 95% of all EU entrants and 60% of all non-EU entrants in 2019. Students from the Republic of Ireland were excluded from the analysis as they are not required to pay higher tuition fees and can still move freely to the UK and within the EU.Footnote ⁴

In addition, contextual data on stringency measures were obtained from the Oxford COVID-19 Government Response Tracker (Mathieu et al., Reference Mathieu, Ritchie, Rodés-Guirao, Appel, Giattino, Hasell, Macdonald, Dattani, Beltekian, Ortiz-Ospina and Roser2020). These data provide an index that measures the strictness of government policies related to COVID-19 in individual countries. A score of 100 represents the strictest response level, and 0 is the lowest level of stringency. The index was calculated for each day during the pandemic and is built of nine metrics: school closures, workplace closures, cancellation of public events, restrictions on public gatherings, closures of public transport, stay-at-home requirements, restrictions on international movements, and international travel controls. For our study, we took the mean value of the total stringency index for the month of May for both the country of origin and the UK in 2020, 2021 and 2022. The month of May was chosen as this is the final month students can accept their offers via UCAS. A range of other measures, in line with previous research by Neville et al. (Reference Neville, Rowe and Singleton2024), were gathered from ancillary data sources, which are shown in Table 1. These measures were integrated alongside the UCAS data in order to create a unique database of origin-destination flows augmented by characteristics of both the origin country and the UK for each year between 2012 and 2022.

Table 1. Data and sources

4.2. Difference-in-Differences (DID)

To assess the impact of Brexit on restrictions, we employ a DID methodology. DID is a quasi-experimental methodological approach that allows researchers to compare the characteristics of different sub-groups of the population, a control group and a treatment group, before and after an intervention or policy change, that is, a treatment. DID allows researchers to mimic an experimental research design using observational study data (Angrist and Pischke, Reference Angrist and Pischke2009). The aim of the method is to calculate the impact of the treatment on an outcome by comparing the average change over time in the outcome variable for the treatment group to the average change over time for the control group. In the case of this research, we aimed to estimate the impact of Brexit on EU student mobility. We compute this by comparing the numbers of EU student applications before and after the end of free movement against non-EU student applications who were not directly affected by Brexit.

For the DID analysis, we formally defined our treatment and control groups and before and after variables. A binary “treatment” indicator was created to differentiate between EU (treated) and non-EU (untreated) countries. We create two binary time-related variables to indicate whether the year was 2021 or 2022. The interaction between these variables was then used to assess the impact of Brexit on international student mobility. Table 2 shows the mean count of acceptances for each indicator.

Table 2. Distribution of control and treatment groups

A key consideration for the implementation of DID is its equal trends assumption. This assumption states that no time-varying differences exist between the treatment and the control group and the trend of the treatment and the control are assumed to be the same in the absence of the treatment (Bharadwaj, Reference Bharadwaj2010). In the case of our research, we assumed that the trend for non-EU international students and EU international students would have continued at the same rate if the change in free movement had not occurred and therefore they are suitably comparable. In addition, we ran a placebo test whereby we pretended the intervention occurred in the 2014/15 academic year to assess whether there were any notable changes in the interaction term. We chose 2014/15 as it is after the increase in fees in 2012/13 but before the Brexit referendum in 2016/17. We also include a DID estimator for the referendum in 2016 in order to control for the impact of the referendum before the tangible changes in policy began in 2021.We obtained a near-zero coefficient on both counts, which indicates common pre-trends across groups and suggests a common path of untreated potential outcomes if the treatment had not occurred.

The next stage was to split the data into Russell Group and non-Russell Group and different subject groupings. By splitting the data in these ways and running the analysis on each, we can uncover how the impacts of COVID-19 and Brexit manifest across different institutions and subject types. We aggregated subjects into five broader subject groupings using the coding scheme called JACSFootnote ⁵. The five subject groupings are: Arts and Languages; Business and Administration; Health and Life Sciences; Physical Sciences and Engineering; and Social Sciences and Humanities.

4.3. Modelling

To assess the impacts of Brexit on EU student applications when controlling for pandemic restrictions, we perform our DID in a hierarchical regression modeling framework. Specifically, we use a generalized mixed model framework as this is a flexible framework to incorporate a combination of random and fixed effect parameters as predictor variables and accommodate non-continuous responses such as counts (Rowe et al., Reference Rowe, Lovelace and Dennett2024). The annual count of accepted applications represents our dependent variable. Given that the distribution of the accepted applications count data is right-skewed and over-dispersed, a negative binomial regression model (NBRM) was implemented (see Rowe (Reference Rowe2021)).

Our models include a random intercept for origin country and random slopes for the DID estimator. The random intercept for origin country accounts for variation in the outcome variable attributable to the country, capturing differences between countries that might influence the outcome but are not explicitly modeled. The random slope allows the effect of Brexit to vary across different origin countries, capturing variations in the change in outcome before and after Brexit in each country. Before entering our model, independent continuous variables are standardized by subtracting their mean and dividing by their standard deviation. We then extract the random effects for DID estimator for each of our models, allowing us to plot the varying severity of the impact of Brexit across countries in the EU. The general equation for our model formalized in Eq. (1):

(4.1) $$ {Y}_{it}=f\left({a}_i+{d}_i+{p}_{it}+{p}_i+{p}_{jt}+\mathrm{eu}+\mathrm{post}+\mathrm{post}\times \mathrm{eu}+{\mathrm{string}}_it+{\mathrm{string}}_jt\right)+{\varepsilon}_{it}, $$

(4.2) $$ {\zeta}_{it}\sim N\left(\zeta, {\sigma}_{\zeta}\right). $$

$ {Y}_{it} $ captures the volume of flows between the country of origin and the UK $ (j) $ in a given year $ (t) $ . $ {\alpha}_i $ is a random intercept that varies by origin country $ i $ ; $ {d}_i $ relates to the distance between the origin and destination; $ {p}_{it} $ refers to origin-destination factors in a given year such as the size of the origin country population in the UK, origin country population size, origin country GDP per capita, rank of origin country institutions, the size of trade flows between the origin country and the UK, size of the origin country population at student age, and origin country unemployment. $ {p}_i $ relates to factors at the origin that do not change over time, such as colonial relationships and common language. $ {p}_{jt} $ relates to destination factors at a given year, including UK population, UK GDP per capita, and UK unemployment. eu is a binary indicator that shows whether a country is a member of the EU or not and post refers to a binary variable that identifies whether the time period is after the intervention. Post × eu is the interaction term (or DID indicator) of the effect of being a member of the EU in the post-intervention time period. $ {\mathrm{string}}_{it} $ and $ {\mathrm{string}}_{jt} $ refer to the severity of COVID-19 stringency measures in a given year.