The Impact of Acquiring Unrestricted Work Authorization on Romanian and Bulgarian Migrants in the United Kingdom

Abstract

When Romania and Bulgaria (the so-called A2 countries) joined the European Union in 2007, the United Kingdom imposed temporary restrictions on the employment and welfare entitlements of A2 citizens that lasted until January 1, 2014. This article analyzes the impact of the removal of these restrictions on the labor market outcomes and use of welfare benefits of A2 migrants. Applying difference-in-difference analysis to data from the UK’s Labour Force Survey, the results suggest that acquiring unrestricted work authorization had a significant negative impact on the incidence of self-employment among A2 migrants but there are no discernible effects on other labor market outcomes or on their receipt of a range of welfare benefits. The article offers potential explanations for these results.

Keywords

immigrants legal status labor market outcomes United Kingdom European Union

On January 1, 2007, Romania and Bulgaria (the so-called A2 countries) became members of the European Union (EU), a development that granted all Romanian and Bulgarian citizens the right to travel and move without restrictions to the United Kingdom (UK) and other EU member states. Acquiring EU status did not, however, automatically imply unrestricted rights for A2 citizens to work across the EU. The EU Accession Treaties allowed the 25 existing member states to impose transitional controls that limited access to the labor market and welfare benefits for Romanian and Bulgarian nationals for a maximum of seven years. Along with many other EU member states, the UK imposed these transitional controls for the full seven years. The UK differed from much of the EU in that it had not imposed similar transitional restrictions on nationals of 10 new member states—including the so-called A8 countries in central and eastern Europe—that acceded to the EU in 2004.¹

In essence, the imposition of transitional controls on A2 nationals meant that, from 2007 to 2013, Romanians and Bulgarians could work as self-employed but were obliged to obtain work authorization (in the form of an accession worker card, a type of work permit) if they wanted to take up jobs as employees in the UK. Accession worker cards were issued for employment in only a limited set of occupations that were open to A2 nationals (and any other non-EU workers) before accession and that continued to be open after accession. Furthermore, A2 migrants’ entitlements to unemployment and other forms of social assistance were restricted in various ways depending on specific status (e.g., whether self-employed or holding an accession worker card). On January 1, 2014, seven years after the A2 countries joined the EU, all these restrictions were lifted. As a result, Romanians and Bulgarians acquired unrestricted work authorization and wider access to the welfare state in the UK and all other EU member states.

EU enlargement in 2007 led to considerable increases in the number of A2 citizens in the UK, especially from Romania, which has a much larger population than Bulgaria (20 million and 7.2 million, respectively). The number of A2 nationals (aged 16+) in the UK increased tenfold over the decade since accession, from approximately 25,000 in late 2006 to 180,000 in late 2013 (i.e., just before employment restrictions were lifted), and 260,000 (including 190,000 Romanians) in 2015 (see Figure A.1 in the Appendix). The increase was quite gradual between 2006 and late 2011, with a sharper rise in 2012 and another after the second half of 2014.

The rise in Romanian and Bulgarian migrants living in the UK, especially following the lifting of employment restrictions in January 2014, has led to heated public debates—and considerable interest in new research—about the impacts of A2 immigration on the UK’s labor market and welfare state. Public concern has focused particularly on the effects of A2 migrants—and before that migrants from the previous EU accession countries—on the possible labor market outcomes of low-wage workers already in the UK and on A2 nationals’ access to “in-work tax credits” and social housing benefits, which are means-tested, non-contributory welfare benefits available to all British citizens and non-transitional EU workers on low incomes in the UK. Analysis of media reports on Romanians and Bulgarians in the UK press in 2013 reveals a considerable increase in media coverage of A2 immigration as well as a focus on the expected scale and effects of A2 migrants on poverty, welfare, and crime in the UK (Migration Observatory 2014). These public perceptions and concerns about Eastern European immigration played an important role in the UK’s referendum vote in June 2016 to leave the European Union (“Brexit”).²

This article focuses on the impact of the removal of employment restrictions on A2 nationals who were already living and mostly also working in the UK before January 1, 2014. More specifically, we analyze the impact of the change in the legal work status that A2 nationals experienced (from “EU national with restricted work rights” to “EU national with full and unrestricted work authorization” in the UK) on their labor market outcomes and receipt of welfare benefits. To address this, we consider the lifting of all work restrictions on January 1, 2014, as a quasi natural experiment that changed the legal work status of A2 migrants but did not affect the work status of other central and eastern European migrants (from the A8 countries) in the UK, who were never subject to transitional controls and who serve as the control group in our analysis.

Using pooled cross-section data from the Labour Force Survey (LFS), we employ difference-in-difference estimators to assess whether and how the change in A2 migrants’ legal work status affected their labor market outcomes and their use of benefits. We consider a large range of labor market outcomes (labor market participation; hours worked; earnings; the nature of the job such as incidence of self-employment, temporary work, and manual work) and welfare benefit claims (unemployment benefits, housing benefits, in-work tax credits, child benefits, and income support). Our article adds to the research literatures on the effects of migrants on labor markets and the welfare state and on the role of immigration and employment status as determinants of migrants’ outcomes in the labor market and their use of welfare benefits.

Gaining Full Work Authorization: What Effects Can We Expect for A2 Nationals in the UK?

Types of Immigration Status and Associated Rights to Work and Welfare before 2014

A2 citizens who were already in the UK in late 2013 had one of the five types of legal immigration and employment status shown in Table 1.

Table 1.

Overview of Possible Types of Immigration and Legal Work Status of A2 Migrants in the UK before 2014

		Immigration status
		Legally resident	Illegally resident
Work status	Not working	S1	S4
	Working legally	S2	×
	Working illegally	S3	S5

Numerous ways existed for A2 citizens to legally reside in the UK before 2014. For example, under EU rules they could have been legal residents as students or “family members of an EEA national (including A2 citizens) with a right to reside in the UK.”³ They could also have been legally resident if they were employed on accession worker cards, self-employed, or self-sufficient. The latter two categories were explicitly exempted from the temporary employment restrictions facilitated by the EU Accession Treaties. A self-sufficient person is defined as someone who “has sufficient resources for themselves and their family members [so as] not to become a burden on the social assistance system of the host Member State, and has comprehensive sickness insurance cover” (Gower and Hawkins 2013: 3). Some Romanians and Bulgarians may also have been legally resident based solely on UK laws, for example, as long-term immigrants with “indefinite leave to remain” (i.e., permanent residence) or UK citizenship, or as spouses of British citizens or non-EU immigrants with permanent residence status. A2 migrants who did not meet any of these conditions would, in all likelihood, have had no legal right to reside in the UK in 2013 (for more-detailed discussion, see Ryan 2008 and Gower and Hawkins 2013).

A2 nationals who were legally resident in late 2013 could have been not working (S1), working legally (S2), or working illegally (S3). A2 migrants working legally could have included, for example, students working part-time (up to the legal maximum of 20 hours a week), dependents (most of whom enjoy unrestricted work rights in the UK), self-employed persons, and migrants with accession worker cards that authorized employment in the UK. Authorization of employment as a worker (“employee”) was limited to A2 migrants who met the criteria for admission under the UK’s existing work permit schemes for non-EU workers (in the form they existed on January 1, 2007).⁴ Employment under most of these schemes was restricted to a specific employer named on the work permit. Non-compliant A2 workers (and their employers) were liable for prosecution. Under the transitional controls, before 2014 any A2 migrant with an accession worker card who worked in the UK legally and without interruption for a period of 12 months became exempt from the authorization requirement and thus acquired unrestricted employment rights.

The combination of “legal residence and illegal work” (status S3 in Table 1) is a potentially important and under-researched status that Ruhs and Anderson (2010) defined and analyzed as “semi-compliance.” In the context of A2 nationals in the UK before 2014, this status could involve a wide range of people and situations including, for example, a student working for more hours than legally allowed, an accession worker card holder who has been in the UK for less than 12 months and is working for an employer not specified on the accession worker card, and “false self-employment” (a key issue we discuss in more detail below).

The multitude of potential statuses of A2 migrants who were in the UK in late 2013, means that the removal of employment restrictions on January 1, 2014, could have had varying effects on different people. For example, for those residing illegally⁵ in 2013, removal of the employment restriction implied a legalization of their residence and work status. Those residing legally but working illegally in 2013 realized a legalization of employment status. And for A2 citizens legally working as self-employed persons or as authorized workers (i.e., with accession worker cards) for less than 12 months, the removal of employment restrictions on January 1, 2014, meant that they were free to legally work for any employer and carry out any type of work in the UK.

The types of immigration and work statuses of A2 migrants before 2014 were also associated with varying restrictions on access to welfare benefits (see Gower and Hawkins 2013 and Kennedy 2015). For example, A2 migrants who were employed as authorized workers could claim in-work benefits (including tax credits and housing benefits) but they were not eligible for means-tested out-of-work benefits during the first 12 months of their authorized employment. After 12 months of legal and uninterrupted employment, authorized A2 workers acquired the same access to welfare benefits possessed by other EEA workers. Self-employed A2 nationals had access to means-tested in-work benefits, but they could not claim out-of-work benefits if they stopped working. A2 migrants residing and/or working illegally in the UK during 2007 to 2013 had no access to welfare benefits.

On January 1, 2014, all of these restrictions on A2 migrants’ access to welfare benefits in the UK were lifted. Since 2014, A2 migrants who qualify as “workers” have enjoyed the same access to the UK’s welfare state as do other EEA workers.⁶ This access is somewhat more restricted than that for UK nationals living in the UK. In late 2013, the government announced a tightening of access to benefits for all EEA nationals coming to the UK. Under the new rules (most of which became effective in early 2014), newly arrived job seekers from the EEA have no access to housing benefits and can claim out-of-work benefits only after three months of proven job search. The new measures also included a strengthening of the habitual residence test (including stricter interviews) for EEA migrants claiming means-tested benefits (for more detail, see Kennedy 2015).

Research on Immigration/Work Status, Labor Market Outcomes, and Use of Welfare Benefits

The research literature on the effects of immigration status on migrants’ labor market outcomes and use of welfare benefits has focused on the impacts of two types of status: “illegal residence” and “legal but temporary residence and employment.” We briefly review the potential mechanisms of impact and empirical findings of the existing research below.

Theoretical Considerations: Illegal Status

Migrants who reside illegally in the UK (hereafter, illegally resident migrants) are subject to removal from the UK, and their employers are subject to fines. The risk of deportation can put illegally resident migrants in a vulnerable position in the host country (De Genova 2002). Some employers may offer illegally resident migrants lower wages and inferior employment conditions, either because they take advantage of migrant’s deportability or simply to account for the increased risk associated with employing migrants without legal residence rights. Taylor (1992) argued that cost-minimizing employers will allocate illegally resident migrants to jobs for which the expected cost of apprehension is lowest, and that such jobs are likely to be lower-skilled jobs. Employers may also take advantage of the potentially lower elasticity of labor supply of illegally resident migrants, which makes them less likely than other workers to quit their jobs in response to a reduction in wages (Hotchkiss and Quispe-Agnoli 2013). If employers enjoy monopsonistic power in the labor market (see, for example, Manning 2011), migrants without legal status may receive lower wages.

Deportability may also affect migrants’ labor market outcomes through mechanisms that are not directly related to employer discrimination. Illegal residence status may alter migrants’ behavior in the labor market in various ways (Kossoudji and Cobb-Clark 2002). Migrants without the right to reside may, for example, have lower reservation wages than workers with the right to legal residence. The fear of being deported could also discourage some migrants from investing in the development of host-country specific human capital (Chiswick 1984). At the same time, the risk of deportation could increase illegally resident migrants’ work effort compared to workers with legal status (Stark 2007). Illegal residence status could also influence the types of social networks migrants access, which, in turn, could affect their access to well-paying jobs (Massey 1987). A more general point is that illegal status usually constrains migrants’ choice of employment in the host country and thereby prevents migrants from maximizing the returns to their human capital (Calavita 1992).

Illegally resident migrants often have very limited access to welfare benefits. In many countries, including the UK, access to welfare benefits requires proof of legal residence (in addition to other eligibility criteria). Illegally resident migrants who use fraudulent residence permits or who use other people’s residence permits may be able to circumvent this barrier. The effect of illegal residence status on migrants’ access to benefits depends, therefore, on the national context, including for example, the ease of availability of fake residence permits and the efficacy of enforcement against such practices (see Papademetriou, O’Neil, and Jachimowicz 2004).

Theoretical Considerations: Temporary Residence Permits

Although not at constant risk of removal, migrants employed on legal temporary work permits may also experience worse labor market outcomes than permanent legal residents, because of their immigration status. Temporary work permits (including the accession work cards issued to newly arrived A2 nationals for authorized employment in the UK during 2007 to 2013) typically restrict migrants’ employment to the sector and employer specified on the work permit. Changing employers requires an application for a new work permit by the new employer. By restricting migrants’ choice of employment in the labor market, this requirement reduces workers’ bargaining power and may make it difficult to leave jobs with adverse employment conditions (Lowell and Avato 2007). Whether and to what degree migrants employed on legal temporary permits have a stronger or weaker bargaining position vis-à-vis their employers is an empirical question. Apgar (2015) suggested that illegally employed migrants may have some advantages over migrants who are legally employed on temporary work permits, because they may have greater job mobility and opportunities for improvement in occupational standing.

With regard to impacts on use of welfare services, most types of legal temporary residence status are associated with restricted access to the welfare state. There are, however, considerable cross-country variations in temporary migrants’ access to welfare benefits (see Ruhs 2013), as well as differences in the welfare rights associated with types of temporary status within countries (e.g., high- and low-skilled work permit holders and self-employed persons). Regardless of the specific type of restrictions, however, a change in status from temporary resident to permanent resident is typically associated with increased legal access to welfare benefits. In most countries, including the UK, the rights of migrants with permanent residence status are very similar to those of citizens (with the important exception of the right to vote in national elections).

Empirical Findings

In the UK and elsewhere, empirical analysis of the relationship between legal migration status and migrants’ labor market outcomes and receipt of welfare benefits has been hampered by a lack of data. Labor force surveys and other large-scale surveys usually do not record respondents’ official immigration status. Moreover, migrants residing illegally may be less likely to participate in government surveys. As a result, empirical research has focused on responses to policy shifts such as legalization programs. Most of this research has investigated the impacts of changes in immigration status on migrants’ employment outcomes rather than their use of welfare benefits.

Much of the empirical research relevant to this article has been carried out in the United States (US), especially in the aftermath of the 1986 Immigration Reform and Control Act (IRCA). US research on the relationships between immigration status and labor market outcomes may be relevant to the UK because both countries are liberal market economies (Hall and Soskice 2001) whose labor markets share a number of institutional and other characteristics (see, for example, Gautié and Schmitt 2010).

The IRCA granted amnesty (including eventual permanent residence status) to undocumented immigrants—approximately 1.7 million people outside agriculture⁷—who could prove continuous residence in the US since 1982. To help assess the impact of IRCA, a random sample of legalized migrants was surveyed in 1989 and in a follow-up survey in 1992. A number of studies used the data taken from these legalization processes to explore the impact of legal status on migrants’ labor market outcomes. Borjas and Tienda (1993) found that legal immigrants earned more than did undocumented immigrants from the same geographic region. Because of data limitations, the authors could not assess how much of this gap was attributable to differences in socioeconomic characteristics rather than other factors such as discrimination based on lack of legal status. Rivera-Batiz (1999) found that illegal status adversely affects migrants’ earnings even when controlling for migrants’ individual characteristics. This finding contradicted results from earlier (pre-IRCA) studies, many of which concluded that the wage differential between legal and illegal immigrants could be mostly accounted for by differences between the characteristics and human capital of the two groups (see, for example, Bailey 1985 and Massey 1987). Comparing legalized migrants’ earnings before and after legalization based on data from the U.S. Department of Labor’s Legalized Population Survey (LPS), Rivera-Batiz (1999) further concluded that legalization generated significant wage growth for migrants.

Kossoudji and Cobb-Clark (2000) used LPS data to analyze the occupational concentration and mobility of Mexican migrants legalized under IRCA. They found that legalization changed the mobility patterns of the legalized population, thereby creating new opportunities to move up the occupational ladder. In a later article that used LPS data on legalized migrants as well as data from the National Longitudinal Survey of Youth on a comparison group of Latino men, Kossoudji and Cobb-Clark (2002) found that IRCA had positive earnings effects for legalized migrants. Kossoudji and Cobb-Clark suggested that much of the wage growth following legalization could be attributed to increased returns to human capital (see also, Tienda and Singer 1995). Focusing on the agricultural labor market, Pena (2010) found that legal status had a positive but relatively small effect on the earnings of migrant workers.

Recent research suggests that the wage effects of legalization may have changed considerably over time and could now be much smaller than during the immediate aftermath of the IRCA in the late 1980s and early 1990s. Using data from the New Immigrant Survey in the US, the Lofstrom, Hill, and Hayes (2013) analysis is quite similar—in terms of research questions and the type of data available—to the one conducted in our study. They analyzed the changes in wages and occupational mobility of migrants who obtained legal permanent resident status in 2003 and who had previously worked in the US either without authorization or with authorization as temporary work/residence permit holders. Lofstrom et al. (2013) found very limited or no impacts of legalization on labor market outcomes, especially for low-skilled migrants. Any wage increases and upward occupational mobility attributable to legalization were limited to highly skilled migrants. Lofstrom and colleagues argued that the lack of a large effect of legalization on wages can potentially be explained by the widespread use of false employment authorization documentation, which, arguably, has undermined the threat and effectiveness of employer sanctions, especially in lower-skilled occupations.⁸

One problem that surrounds analysis of the outcome of legalization programs is determining to what extent any observed changes are due to a causal effect of legal status or rather to the differences in the process of self-selection into legal status. Fasani (2015) summarized the latest evidence for the US and Europe of the effect of various legalization programs with an emphasis on establishing the causal effect of such policies. The causal studies in Fasani’s overview typically appeared to find limited effects of legalization on earnings, employment, and occupational quality.

Drawing on longitudinal survey data on legal and illegal immigrants in Italy spanning an amnesty in 2002, Fasani (2015) also provided new evidence on the relationships between different types of immigration status (namely, illegal status, legal temporary status, and legal permanent status) and migrants’ labor market outcomes. Fasani found that migrants with a more stable residence status in the host country tended to have better labor market outcomes. He also found, though, that the causal effects of the 2002 legalization on migrants’ employment, earnings, and occupational quality were small. Fasani suggested this may be explained in part by the study’s focus on short-run effects (two years after legalization) as well as the design of the amnesty, which required amnesty applicants to be in employment at the time of applying for legal status (in this regard, it differs from legalization under IRCA in the US).

In the UK, only one study has examined the impacts of changing immigration status on the labor market outcomes of migrants. Using a research design similar to that employed in the current study, Ruhs (2017) analyzed the impact of gaining EU status with immediate full work rights on the earnings of A8 migrants, that is, East European workers whose countries joined the EU on May 1, 2004, and who were already working in the UK before that date—legally or illegally. The results of this exploratory analysis suggested a significant positive impact of acquiring EU status on earnings and that, in part, this effect was brought about by A8 workers gaining the right to freely change jobs after EU enlargement.

Ruhs (2017) focused on earnings only using a relatively small survey sample (not LFS). By contrast, our current analysis represents the first study of the impacts of acquiring unrestricted work authorization on a wide range of labor market outcomes of East European migrants in the UK, and the first analysis of the effects for Romanians and Bulgarians. It is also the first study that uses large-scale data from the LFS to conduct a plausibly causal analysis of the impacts of changing legal work and immigration status on labor market outcomes and welfare use of migrants in the UK.

Existing studies on the links between changing immigration status and migrants’ use of welfare benefits in high-income countries are limited. A relatively long-standing body of research exists on migrants’ participation in the welfare system, with considerable cross-country variations in empirical findings (for a review, see Kerr and Kerr 2011). Largely because of limitations in the available data, however, this literature usually employs broad distinctions between “migrants” and “natives” (and sometimes also between migrants from different countries) without investigating the specific effects of (changing) immigration status on migrants’ receipt of welfare benefits. For example, Borjas and Trejo’s (1991) analysis of immigrant participation in the US welfare system, based on census data from 1970 and 1980, found that the longer an immigrant household had been in the US, the more likely it was to receive welfare. Although the analysis cannot directly observe immigration status in the data, Borjas and Trejo investigated changes in immigrants’ welfare use after five years of residence in the US to proxy the potential effects of the acquisition of permanent residence status or citizenship. They concluded that an assimilation into welfare over time cannot be fully explained by immigrants’ increasing benefit eligibility over time (e.g., as they transfer from temporary to permanent residence status).

Hansen and Lofstrom (2003) used longitudinal data to study welfare participation in Sweden in 1990 to 1996, distinguishing between migrants (foreign-born people) and natives, and also between refugees and “nonrefugee immigrants.” They found that immigrants used welfare to a greater extent than did natives but they (especially refugees) assimilated out of welfare over time (without, however, reaching parity with natives, even after 20 years).

In the UK, research on migrants’ use of welfare has been similarly constrained by a lack of data that would allow an analysis of the effects of different types of immigration status, and of changing status over time. Drinkwater and Robinson (2013) used data from the LFS to examine the determinants of welfare participation of immigrants in the UK during the period 2004 to 2009. Distinguishing between seven immigrant groups based on area of origin (including A8 countries as a distinct group), Drinkwater and Robinson’s analysis explored the effects of a range of personal characteristics on welfare participation. Because of data limitations, however, they were unable to control for the immigration status and the associated welfare rights and restrictions of the migrants in their sample. Drinkwater and Robinson found that five out of the seven immigrant groups analyzed (including A8 nationals) were significantly less likely to claim benefits than were the UK-born. However, their analysis also showed that welfare benefit claims varied considerably across both immigrant groups and types of welfare benefits, which makes it difficult to generalize about welfare participation of immigrants. For example, A8 migrants were significantly less likely than the UK-born to claim unemployment-related benefits, income support, and sickness benefits, but significantly more likely to claim housing benefits and tax credits.

In their review of welfare assimilation studies, Kerr and Kerr (2011) argued that a limitation of most existing research is the lack of separation between welfare eligibility and welfare usage. This shortcoming is a result of the inability of most existing studies to identify immigration status and the associated restrictions on access to welfare of the migrants in the data. We address this limitation by focusing on the impacts of a change in legal status and welfare eligibility of a specific group of migrants—A2 nationals—in the UK.

A2 Nationals Gain Unrestricted Employment Rights in the UK

How might we expect the removal of work restrictions on January 1, 2014, to affect A2 migrants’ labor market outcomes and use of welfare benefits in the UK? One could argue that the lifting of work restrictions is likely to have positive effects on the labor market outcomes for Romanians and Bulgarians in the UK. For example, it is likely to increase their labor market participation and probability of working as an employee (rather than in self-employment), and it can be expected to have positive effects on job/occupational mobility and earnings. As regards welfare benefits, we may expect an increase in the take-up of those benefits for which A2 citizens became eligible after January 1. The change in benefit eligibility varied across A2 migrants with different types of immigration and work status in late 2013.

Conversely, gaining unrestricted work rights may not have had large effects on existing A2 migrants’ labor market outcomes and use of welfare benefits in the UK, for at least two reasons. First, although it was certainly possible for A2 migrants who were already in the UK in 2013 to be illegally resident and illegally working, in practice the effects of these types of illegality (including the threat of deportation) were, arguably, relatively minor. The reason is that although A2 migrants did not yet have full and unrestricted work authorization in 2013, their status as EU nationals meant that deportation, and indeed many other enforcement actions against their illegal employment, were likely not priority targets of the UK’s enforcement agencies.

Second, the exemption of the self-employed from the work restrictions imposed on A2 migrants before January 2014 meant that Romanians and Bulgarians did have the opportunity to legally work and access a wide range of benefits in the UK as long as they arranged and presented their work as self-employment. The UK has had a long-standing problem with false (or bogus) self-employment, especially (but not only) in the construction sector, where the share of self-employed in the workforce is around 40% (see Table A.2 in the Appendix). False self-employment implies a person effectively works as an employee but is formally registered (for tax and/or immigration purposes) as self-employed. The line between employment and self-employment is often unclear and frequently contested in the courts.⁹ Trade unions have long argued for stricter regulation of self-employment in the UK. For example, until April 2014 it was possible for employment agencies to engage workers as self-employed persons (REC 2014). It is possible, therefore, that A2 migrants used self-employment as a way of legally accessing and working in the labor market before January 2014, across a wide range of occupations. As suggested earlier, the removal of employment restrictions on January 1, 2014, did not involve a large change in self-employed A2 migrants’ legal access to Britain’s welfare state.

There are also potential factors that might lead to a decline in A2 nationals’ use of welfare benefits after the removal of employment restrictions in January 2014. For example, if the acquisition of unrestricted work rights leads to A2 nationals moving into better and higher-paid occupations in the UK, they may no longer be eligible for certain means-tested benefits.

So, potentially countervailing considerations and effects exist, which makes it impossible to formulate a clear expectation or hypothesis about the impacts of changing status on A2 migrants based on theoretical consideration alone. As a result, the presence, direction, and magnitude of the impact of gaining the right to unrestricted employment on A2 migrants’ labor market outcomes and use of welfare benefits in the UK is an open question for empirical research.

Data and Descriptives

Labour Force Survey

The analysis uses data from the Labour Force Survey (LFS). The LFS is a household survey carried out in the UK on a quarterly basis. Approximately 60,000 households and the individuals therein are surveyed every quarter. All LFS respondents are interviewed in five consecutive waves, with data on earnings collected in wave 1 and wave 5 only. The LFS does not record legal immigration status but data on country of birth and nationality are available—along with a rich set of variables relating to individual characteristics and labor market outcomes.

Given the difference-in-difference (DiD) approach of this study, we use data on working-age A2 migrants¹⁰ (the treatment group) and A8 migrants (the control group) from quarterly surveys in the four quarters of 2013 (“before”) and 2014 (“after”). These pooled cross-sectional data make use of all observations of A2 and A8 migrants in these eight quarters.¹¹ This sample includes approximately 900 A2 and 4,700 A8 migrants interviewed in 2013 and similar numbers in 2014.¹² To facilitate robustness checks, including “placebo” DiD, we make use of a larger selection of the LFS data, ranging from the first quarter of 2007 (when Romania and Bulgaria joined the EU) to the fourth quarter of 2015.

The LFS also provides an individual identifier variable that allows the matching of individuals who live at the same address across quarters, thus effectively creating a panel data set. For some 20% of migrants in our data set, specifically those interviewed for the first time in any of the quarters of 2013 and who remained in the same residence, we have information relating to time before and after January 1, 2014. The panel data set is much smaller, consisting of only 73 A2 migrants and 500 A8 migrants who were interviewed in both 2013 and 2014. For this reason we confine our estimates from the panel data set to the Appendix.

Descriptives

Table 2 above provides an overview of the pre-treatment characteristics (i.e., based on 2013 data) of the sample of A2 migrants and A8 migrants used in our analysis. To facilitate broader comparisons, we also include the characteristics of all other immigrants and UK-born individuals of working age in the LFS in that year. Table A.1 in the Appendix summarizes data on these variables spanning 2007 to 2015.

Table 2.

Characteristics of A2 Migrants, A8 Migrants, and Others, 2013

Characteristics	A2	A8	Other immigrants	UK-born
Population, age 16–64 (in thousands)	180	880	5,260	33,980
Employment
Employees (%)	31.4*	67.2	56.4	63.1
Self-employed (%)	42.7*	11.9	9.8	9.3
Unemployed (%)	6.1	6.2	6.7	5.9
Occupation
Managers, professionals, associate professionals (%)	18.1*	14.6	49.2	44.4
Administrative, skilled manual, caring (%)	39.7*	30.0	25.5	24.2
Sales, processing, elementary (%)	42.2*	55.5	25.3	24.2
Sector
Manufacturing (%)	4.5*	21.4	7.7	10.0
Construction (%)	30.3*	9.0	4.3	7.6
(of which % self-employed)	84.7*	61.6	40.3	38.0
Administration (%)	13.1*	7.7	5.3	4.5
(of which % self-employed)	75.1*	18.6	14.2	20.3
Health (%)	10.2	6.9	17.1	13.3
Other sectors (%)	41.9*	55.0	65.6	64.6
Benefits
UI (%)	0.5*	2.6	3.1	3.4
Sickness benefits (%)	0.4*	1.3	4.2	7.0
Child benefits (%)	18.6*	28.2	21.3	18.5
Tax credits (%)	11.1*	20.3	13.6	11.5
Housing benefit (%)	7.3*	10.2	10.3	8.4
Demographics
Female (%)	50.4	53.2	51.4	49.9
Age (years)	33.4*	32.7	39.1	39.9
Graduates (%)	37.0	36.1	45.1	21.8
Live in London (%)	54.7*	21.8	39.3	9.1
Students (%)	1.3	1.6	1.7	3.0

Source: Labour Force Survey, 2013; also see Appendix Table A.1.

Notes: Sample sizes A2 = 921, A8 = 4,963, Other immigrants = 30,223, UK-born = 212,217.

Denotes significantly different A2 compared to A8 at 5% level.

During 2013–2014, the UK economy was beginning to emerge from the Great Recession. As a result, aggregate employment grew by 700,000 and self-employment by some 300,000 over the year, raising the national share of self-employed in the working population by 0.6 points. As shown in Table A.1 and Figure A.1 in the Appendix, employment among A2 and A8 nationals was also rising during this period.

The unemployment shares of A2 migrants, A8 migrants, and other working-age individuals were similar throughout the period from 2007 to 2013. By contrast, considerable differences existed across groups in terms of employment and especially self-employment rates before 2014. In 2013, the employment rate of A2 migrants was lower than that of A8 migrants but higher than that of other individuals. This difference held true for most years from 2007 to 2013. The difference between self-employment rates, however, is striking: 43% of all working-age A2 migrants were self-employed in 2013 compared to 12% among A8 migrants and 9% among others. A2 migrants in the UK were considerably more likely to be self-employed than other people in the UK, even within occupation and sector.¹³ So, for example, a relatively large share of A2 migrants were employed in construction, a sector in which self-employment is considerably higher than elsewhere. This sectoral concentration is not enough, however, to explain the very high self-employment rates of A2 migrants in 2013. A2 migrants also had high rates of self-employment in various other occupations and sectors that were typically associated with much less self-employment. In the administration sector, 75% of A2 workers were self-employed compared to 20% of UK-born workers. Within occupations, 66% of A2 workers in elementary occupations were self-employed compared to 7% of others employed in elementary occupations. Similarly, 77% of A2 workers in skilled manual occupations were self-employed compared to 33% of other skilled manual workers. These findings suggest that self-employment may have been a strategic outcome between A2 migrants and employers that enabled them to work legally in the UK before the removal of work restrictions in January 2014. Table A.2 gives the changes in the self-employment share within each sector for A2, A8, and others over time. The table shows that in 2013 A2 migrants were more likely to be self-employed than other persons in all of the industries listed. It is also clear from Table A.2 (and from Figure A.1) that sharp falls occurred in the self-employment share for A2 workers after 2013 that were not observed among other groups of workers.

LFS data on use of welfare benefits in 2013 suggest considerable differences across A2 migrants and A8 migrants, as shown in Table 2. Specifically, A2 migrants’ use of child benefits, tax credits, and housing benefits in 2013 was considerably lower than that of A8 migrants and very similar to that of other individuals of working age. Only 11% of A2 migrants in the UK received means-tested tax credits in 2013, although almost half worked in “elementary” and “processing” occupations for which earnings are relatively low.

What do we learn about the legal immigration and work status of the A2 workers in our sample? Although the LFS does not include specific questions about these issues, the data allow three basic observations. First, just under half (43%) of A2 nationals in the LFS sample for 2013 reported to be in self-employment, which allowed them to reside and work legally in the UK. Second, a considerable share (31%) of A2 nationals reported working as employees. As previously discussed, three major avenues existed for A2 nationals to legally work as employees in the UK in 2013, namely, as students legally working part-time, as dependents with permission to work, or as accession card holders. As shown in Table 2, only 1% of our A2 sample were students in 2013. According to administrative data from the UK’s Home Office (2014), 17,300 accession worker cards were issued to A2 nationals in the UK from 2007 to 2014. This figure represents cumulative annual inflows, so it does not capture the stock of accession worker card holders in 2013 (since the total of 17,300 ignores outflows from 2007 to 2013, for example, of the A2 migrants admitted to the UK under temporary arrangements such as the Sector-Based Scheme). Given that an estimated 180,000 A2 nationals of working age were in the UK in 2013, it is unlikely that many of the A2 migrants in our LFS sample would have been legally employed on accession worker cards in 2013.¹⁴ The UK’s provisional Long-Term International Migration Estimates (LTIM), which are based on international passenger survey data, suggest that more than 60% of A2 nationals entering the UK from 2007 to 2013 did so for work-related reasons, about 25% for study, and the rest for other reasons including “joining/accompanying a family member.”¹⁵ Considered together, these data suggest that a considerable number of A2 nationals in 2013 may have been working illegally, or at least in semi-compliance with immigration rules (i.e., combining legal residence with illegal work).

Estimation Methods: Difference-in-Difference

Our statistical analysis of the impacts of acquiring unrestricted work rights on the labor market outcomes and use of welfare benefits of A2 migrants in the UK employs the difference-in-difference (DiD) approach. DiD estimation requires a treatment group of individuals affected by the policy and a comparison group of individuals unaffected by the policy. In this analysis, the policy change is the granting of unrestricted work rights. The treatment group consists of A2 migrants in the UK both before and after January 1, 2014. As a comparison group, we use migrants from the A8 countries who were also in the UK before and after January 1, 2014. As mentioned earlier, A8 migrants have enjoyed unrestricted work rights in the UK since their countries joined the EU in May 2004.

The DiD estimator (δ) is defined as the average difference in an outcome (denoted by Y in Equation (1) below) in the treatment group (A2) before and after the treatment (t = 0, t = 1, respectively) minus the difference in average outcomes in a comparison group (A8) before and after the treatment.

\hat{δ} = ({\overline{Y}}_{1}^{A 2} - {\overline{Y}}_{0}^{A 2}) - ({\overline{Y}}_{1}^{A 8} - {\overline{Y}}_{0}^{A 8})

We conduct DiD estimation using both pooled cross-sections and panel data. Given pooled cross-sectional data taken from before (t = 0) and after (t = 1) the treatment, the DiD can be estimated using simple regression as:

Y_{i t} = β_{0} + β_{1} A 2 + β_{2} A f t e r + δ (A 2 * A f t e r) + X_{i t} Β + u_{i t}

where i denotes the respondents and t denotes the time period; A2 = 1 if the respondent is an A2 national (i.e., in the treatment group), and 0 otherwise; After = 1 if the observation is in the second time period (i.e., t = 1), and 0 otherwise; A2*After = 1 if the respondent is an A2 national and the observation occurs in the second time period, and 0 otherwise; $${X_{it}}{\rm B}$ are individual-level variables and time-varying controls, and u is a normally distributed error term.¹⁶

Since most of the estimations use the sample populations of working-age people, the list of plausibly exogenous controls is somewhat restricted (since the controls have to apply to both working and non-working individuals and be exogenous with respect to the rules that determine the outcome variables).¹⁷ The set of controls used in all regressions includes dummy variables for age, gender, educational attainment, region of residence, and years living in the UK, along with seasonal dummies.

The validity of the DiD exercise rests on the parallel trend assumption, that is, that there were no before-treatment trends in the outcome variables that might otherwise contribute to a significant effect on the estimated interaction term. Table A.3 gives estimates of these before-treatment trends (the interaction of year dummies with the A2 variable for every year relative to a default base year of 2007—the year A2 migrants were allowed to enter the UK with transitional restrictions) for several of the outcome variables of interest.¹⁸ As can be seen, the interactions with respect to 2007 for A2 self-employment are not significant for all years up to 2013. This outcome provides additional evidence in support of the general DiD identification and estimation strategy with regard to self-employment. This finding also applies to the majority of other outcome variables we use, that is, the pre-treatment trends do not appear to be significant. There are, however, exceptions (specifically receipt of unemployment benefits, incidence of social housing, labor force participation, and manual working) for which the common trends assumption appears to be violated. Closer inspection reveals that labor force participation and manual working appear to have common trends beginning in 2008 (so the accession year of 2007 appears to matter for some other outcomes—as might be expected). Unemployment benefits and social housing, however, do not appear to have common trends and so the credibility of the estimation identification strategy is questionable for these two outcomes. As discussed below, we also carry out a range of robustness checks, including placebo DiD analysis for the years 2007 to 2013, a period with no change in legal work status for Romanians and Bulgarians in the UK.

Results and Discussion

Results

Our results suggest that the removal of work restrictions on January 1, 2014, had a negative impact on the incidence of self-employment among A2 migrants in the UK. Table 3 reports the results of DiD analysis using the pooled cross-sectional data. The coefficient of interest—the DiD estimator—is After*A2. We present estimates from seven models: without any controls (model 1); with controls, including age, education, gender, years in the country, quarter of observation, and region (model 2); only for those individuals who have been resident in the UK for more than one year, that is, excluding migrants arriving after 2013 (model 3); widening the control and treatment windows to two years, that is, using 2014 and 2015 as the period after the treatment (model 4); and “employed only” with and without industry and occupation controls (models 5 to 7).

Table 3.

Difference-in-Difference Estimates: A2 Compared to A8: Self-Employment

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	No controls	With controls	With controls reside >1 year	With controls reside >1 year 2012–2015	Employed only reside >1 year 2013–2014	Employed only reside >1 year + industry, occupation 2013–2014	Employed only reside >1 year+ industry, occupation 2012–2015
A2	0.333**	0.251**	0.253***	0.232**	0.322**	0.238**	0.236**
	(0.028)	(0.028)	(0.029)	(0.023)	(0.033)	(0.030)	(0.023)
After	−0.014	−0.022**	−0.022**	−0.006*	−0.031**	−0.027**	−0.013**
	(0.009)	(0.008)	(0.008)	(0.007)	(0.010)	(0.009)	(0.007)
After*A2	−0.101**	−0.082**	−0.079**	−0.113***	−0.098**	−0.068**	−0.122**
	(0.033)	(0.032)	(0.033)	(0.028)	(0.038)	(0.034)	(0.026)
Constant	0.113**	0.173	0.175	−0.078	0.170**	0.044**	0.003
	(0.007)	(0.056)	(0.057)	(0.149)	(0.070)	(0.022)	(0.070)
Observations	11,841	11,841	11,625	22,299	9,544	9,544	19.652
R-squared	0.081	0.174	0.178	0.154	0.209	0.367	0.354

Notes: Samples are pooled cross-sections. Robust standard errors in parentheses clustered at individual level. All regressions include controls for age (9 dummy variables), gender, region (18 dummy variables), education (3 dummy variables), years living in the UK, and quarter (3 dummy variables). In addition, column (7) includes industry (16) and occupation (10) dummy variables.

p < 0.05.

The DiD estimator in all these models is highly statistically significant and negative. The magnitude of the estimated impact of the removal of work restrictions on self-employment of A2 migrants is similar across the models, ranging from −8 to −12 percentage points. These differences in estimated effects are not statistically significant from each other. The point estimates, however, are always larger for the wider window, suggesting the impact of the change is ongoing (see also Figure A.1). Columns (5) and (7) in Table 3 restrict the sample to those in employment (employees and self-employed) and add both sector and occupation (1 digit) dummies. It is clear from these regressions that the difference-in-difference effect for A2 workers remains, suggesting that the relative (and absolute) shift away from self-employment was within sectors for a given occupation. We are unable to establish from our data whether this is because people switched employers to change self-employment status or changed self-employment with the same employer.¹⁹

Placebo analysis of the DiD model with controls (as in model 3 in Table 3) for the years 2007–2008 to 2012–2013 do not show any systematic effects, which again supports our claim that the treatment effect we have identified for 2013–2014 is genuine (see Figure A.2). Our finding of a negative effect on A2 self-employment is further strengthened by the fact that the average self-employment rate for all workers in the UK rose between 2013 and 2014, which suggests that the decline in the self-employment rates of A2 nationals cannot be explained by a change in broader structural features of the labor market (such as changes in supply of and demand for goods and output markets that involve self-employed workers).

Our results do indicate that A2 migrants remain more likely to be self-employed than other East European migrants even after the change in legal work status, but that the differential narrowed significantly after January 1, 2014. In Table 3, combining the main effect of being an A2 national (“A2”) with the interaction effect of being an A2 national after the change in legal work status (“A2*After”) still gives a positive and relatively large A2 self-employment propensity. Again, it is important to see this in the context of a fall in absolute numbers of A2 self-employment over this period, when self-employment numbers and self-employment as a share of total employment were both rising nationally.²⁰

We next investigate the impact of removing work restrictions on a wide range of other labor market outcomes of A2 migrants in the UK using the same procedure used for self-employment. The outcomes include labor force participation, propensities to hold temporary jobs or manual jobs, earnings, and hours worked. In contrast to the results for self-employment, we do not find statistically significant effects of the removal of work restrictions on any of these outcomes, as shown in Table 4.

Table 4.

Difference-in-Difference Estimates: A2 Compared to A8, 2013–2014: Other Dimensions (residence > 1 year)

	Labor force participation	Temporary job	Manual job	Log hourly wage	Hours worked > 0
A2	−0.029	−0.013	−0.060**	0.100	0.365
	(0.021)	(0.014)	(0.028)	(0.060)	(0.910)
After	0.006	0.003	0.029**	−0.004	0.175
	(0.009)	(0.006)	(0.012)	(0.015)	(0.280)
After*A2	−0.010	−0.001	−0.034	0.063	0.028
	(0.025)	(0.015)	(0.033)	(0.073)	(1.022)
Constant	1.013**	0.151**	0.958**	2.098	45.073**
	(0.073)	(0.049)	(0.098)	(0.084)	(2.549)
Observations	11,625	11,625	11.625	2,076	9,390
R-squared	0.089	0.026	0.121	0.219	0.187

Notes: Estimates include full set of controls. Robust standard errors in parentheses clustered at individual level. Wage data available for only 40% of employees in each sample.

p < 0.05.

Table 5 repeats the exercise to look for any effects of the removal of work restrictions on A2 migrants’ use of welfare benefits. Again, we do not identify any significant DiD estimates. Row (1) indicates that A2 nationals are less likely than A8 migrants to access welfare benefits, but no evidence supports that this changed after the removal of work restrictions in January 2014.²¹

Table 5.

Difference-in-Difference Estimates: A2 Compared to A8, 2013–2014: Welfare Benefits (residence > 1 year)

	Unemployment benefit	Income support	Sickness benefit	Child benefit	Tax credits	Housing benefit	Social housing
A2	−0.021**	−0.005	−0.004	−0.078**	−0.097**	−0.057**	−0.067**
	(0.005)	(0.003)	(0.005)	(0.024)	(0.020)	(0.018)	(0.014)
After	−0.004	−0.005**	−0.003	−0.014	−0.019	−0.015	−0.005
	(0.004)	(0.002)	(0.003)	(0.012)	(0.011)	(0.008)	(0.009)
After*A2	0.010	0.006	0.002	0.031	0.025	0.003	0.002
	(0.006)	(0.005)	(0.005)	(0.028)	(0.025)	(0.019)	(0.016)
Constant	0.031	−0.012	0.034	−0.526**	−0.075	0.059	0.455**
	(0.035)	(0.099)	(0.021)	(0.085)	(0.079)	(0.071)	(0.098)
Observations	11,625	11,625	11,625	11,625	11,625	11,625	11,625
R-squared	0.029	0.015	0.039	0.182	0.066	0.031	0.114

Notes: Estimates include same set of controls as in Table 3, column (3). Robust standard errors in parentheses clustered at individual level.

p < 0.05.

Discussion

Our empirical analysis leads to two key findings. First, acquiring unrestricted employment rights on January 1, 2014, had a significant negative impact on the self-employment rates of A2 migrants in the UK. Second, no evidence supports that the removal of employment restrictions had a significant impact on a range of other labor market outcomes and use of welfare benefits of A2 migrants in the UK. These results, especially the absence of significant effects on a range of labor market outcomes and use of welfare benefits, may at first seem counterintuitive. Our findings certainly go against the grain of public debate and concerns regarding the time the employment restrictions on A2 citizens were lifted.

How can we explain these results? An important part of the answer may be found in the link between our two key findings. The negative impact of the removal of employment restrictions on self-employment among A2 migrants who were already in the UK before 2014 suggests that a considerable number of Bulgarian and Romanian citizens used self-employment status as a way of working legally in the UK while employment restrictions were in place (i.e., during 2007–2013). That the removal of employment restrictions encouraged a switch to employment as an “employee” in 2014 suggests that self-employment in 2013 was a way, for workers and firms, to comply with existing immigration rules and employment restrictions rather than necessarily a requirement or reflection of the work or job done in the UK. The possibility of legal work through self-employment gave A2 migrants considerable freedom in the UK labor market before 2014. Critically, this freedom included the right to choose, accept, and end different types of work in the UK. Although there were rules in place that defined and distinguished between the nature of the work performed by employees and self-employed persons, in practice there was very little effective enforcement of these rules (HM Revenue & Customs 2013). Furthermore, as previously mentioned, until April 2014 it was legally possible for employment agencies to hire workers as self-employed persons. And so A2 migrants who were self-employed before January 1, 2014, may have found themselves in similar situations and employment relations (with similar bargaining power vis-à-vis employers) as A2 migrants who worked as employees after the removal of employment restrictions. Consequently, gaining the formal right to unrestricted employment in the UK had little impact on the labor market outcomes of self-employed A2 citizens in the UK.

This explanation of our zero-effects results of acquiring full work rights—through (partly false) self-employment—is similar to the reasons given by Lofstrom et al. (2013) to explain their finding that legalization did not lead to any noticeable wage gains for previously undocumented migrants in the US. Lofstrom et al. argued that the lack of a large effect of legalization on wages can potentially be explained by the widespread use of fraudulent employment authorization documentation, such as false Social Security numbers, which, arguably, undermined the threat and effectiveness of employer sanctions. In our case, it was the use and abuse of self-employment status rather than fake documentation that may have generated a similar effect.

As discussed earlier, although our data do not allow us to identify individual immigration status, it is likely that a large share of A2 migrants who were not self-employed but still working in the UK in 2013 did so illegally, or at least in semi-compliance with immigration rules (i.e., combining legal residence with illegal work). If so, why did the switch from illegal working in 2013 to legal employment in 2014 not result in a positive impact on other labor market outcomes? At least part of the answer may be found in the low levels of enforcement against A2 migrants working illegally during 2007 to 2014.²² As EU nationals, A2 migrants in the UK were under limited threat of deportation, and enforcement against their employment was, arguably, not a priority of the UK’s immigration and enforcement agencies. Consequently, the risks of semi-compliance, that is, violation of legal restrictions of the right to work, as perceived by workers and employers were likely to have been quite limited. As a result, A2 migrants who were illegally employed before 2014 may not have been in a significantly weaker bargaining position vis-à-vis their employers than were workers legally employed after employment restrictions were lifted in January 2014.²³

Our finding of a lack of a discernible impact of the removal of work restrictions on A2 migrants’ use of welfare benefits in the UK has a number of potential explanations. One important reason is likely to be the relatively limited change in welfare eligibility for many (although clearly not all) A2 nationals in our sample. As discussed earlier, A2 migrants who were self-employed before 2014—who constitute a substantial share of A2 nationals in our sample—had access to similar elements of the UK’s welfare state as did A2 migrants after January 1, 2014, meaning that the removal of work restrictions did not significantly affect their legal entitlements (and hence their use) of welfare benefits. Moreover, the new restrictions on welfare access that all non-UK EU citizens faced after 2013 may have helped keep numbers of welfare claimants low for both A2 and A8 nationals.

Finally, how to explain our finding that A2 migrants remain more likely to be self-employed than other East European migrants even after acquiring full work rights in the UK? We cannot answer this question with certainty in this article. Eurostat (2011) data showed that the self-employment shares of employment in the A2 countries and A8 countries were very similar, at around 19%, which is higher than the numbers we observe for the UK. So, although it could be argued that migrants from central and eastern Europe have a greater propensity to work as self-employed, there is no evidence that A2 citizens’ propensity to work as self-employed is intrinsically higher than that of A8 nationals.²⁴ Our data suggest that A2 migrants move out of self-employment after a period of lag time following the removal of work restrictions in the UK. We consider this an important area for future research.

Conclusion

The enlargement of the European Union in 2004 (when the A8 countries joined) and in 2007 (when the A2 countries joined) generated heated policy debates about the effects of migrants from the new member states on the labor markets and welfare state of the existing EU countries, especially in the UK, which has experienced much more rapid growth of East European migrants since 2004 than have most other member states. A majority of the existing research on A8 and A2 migrants in the UK has focused on the effects of newcomers, that is, of new arrivals after May 2004 (when A8 workers acquired the unrestricted right to live and work in the UK), after January 2007 (when A2 citizens acquired the right to freely move but not work in the UK), and after January 2014 (when employment restrictions on A2 migrants were lifted). By contrast, this article contributes to research and our understanding of the effects of these changes in the legal immigration and employment status of A8 and A2 citizens on those East European migrants who were already in the UK before the changes occurred.

Specifically, we analyzed the impact of the removal of employment restrictions on A2 nationals who were already living and for the most part also working in the UK before January 1, 2014. Applying DiD analysis to data taken from the LFS, we found that the removal of employment restrictions had a significant negative impact on A2 migrants’ incidence of self-employment. At the same time, we could not identify any discernible effects on A2 migrants’ other labor market outcomes and use of welfare benefits in the UK. We argue that one reason acquiring full work rights may not have had a large impact is that considerable numbers of Bulgarian and Romanian citizens used self-employment status as a way of working legally in the UK before January 2014, including working in jobs typically done by employees. This finding is an example of how the nature of work is, at least to a degree, endogenous to the nature of controls. Low levels of enforcement against such practices meant that A2 citizens may have enjoyed considerable freedoms in the UK labor market—as well as formal rights to access welfare benefits—while employment restrictions were still in place. Limited enforcement against illegal working of A2 nationals (other than those falsely self-employed) may also explain why the removal of employment restrictions did not lead to a large change in the labor market outcomes of migrants who switched from illegal working in 2013—a practice that likely involved considerable numbers—to legal employment in 2014.

In addition to contributing to research on a seriously under-studied issue, our article raises at least two important issues for policy debates. First, we show that not all changes in the legal status of migrant workers will automatically lead to changes in labor market outcomes and/or use of welfare benefits. Our analysis suggests that the effects critically depend on the specific change in status involved (e.g., change in legal immigration and/or work status) as well as the national context, especially with regard to the enforcement of immigration and employment laws. This finding has important implications for the design of legalization programs around the world, highlighting the need for country-specific policies that account for the national institutional context and its role in shaping the effects of different types of legalization policies.

A second implication relates to policy debates about the design, activation, and operation of transitional employment restrictions for citizens from new EU member states. Transitional controls, which individual member states can choose to impose or not impose, have been common features of the accession treaties between existing member states and new countries joining the EU. Our analysis suggests that the role and effectiveness of these employment controls—in terms of regulating the actual employment, outcomes, and impacts of workers from new EU member states—can be critically influenced by the exemption of self-employed persons, the characteristics of the national labor market (e.g., the incidence of self-employment across sectors and occupations of the economy), and the degree of enforcement against false self-employment. Although the exemption of self-employed persons from employment restrictions affects all member states that opt to impose the transitional controls, the nature of the labor markets and degrees of enforcement vary across countries and over time. In other words, the effectiveness of transitional controls as tools that enable EU countries to tightly regulate the labor market outcomes and effects of migrants from new member states is always likely to be limited as well as variable across countries and over time.

Footnotes

Appendix

Table A.4.

Difference-in-Difference Estimates: A2 Compared to A8, 2013–2014, Panel: Self-Employment

	(1)	(2)	(3)	(4)
Variable	No controls	With controls	No controls residence > 1 year	With controls residence > 1 year
A2	0.344**	0.257**	0.344**	0.256**
	(0.060)	(0.061)	(0.059)	(0.061)
AFTER	0.039**	0.028*	0.040**	0.029**
	(0.011)	(0.012)	(0.012)	(0.012)
AFTER*A2	−0.025	−0.022	−0.026	−0.023
	(0.045)	(0.047)	(0.045)	(0.047)
Constant	0.085**	0.087	0.087**	0.092
	(0.015)	(0.141)	(0.013)	(0.142)
Observations	1,118	1,118	1,110	1,100
R-squared	0.102	0.220	0.100	0.219

Notes: Robust standard errors in parentheses clustered at individual level.

p < 0.05; *p < 0.1.

Acknowledgements

For their helpful comments, we thank Eliza Galos and Bernard Ryan.

The Labour Force Survey data used in this study were made available through the UK Data Service, produced by the Office for National Statistics (ONS) and supplied under special license by the UK Data Service. The data are Crown Copyright and reproduced with the permission of the controller of HMSO and Queen’s Printer for Scotland. The use of the data in this work does not imply the endorsement of ONS or the UK Data Archive in relation to the interpretation or analysis of the data. For information regarding the data and/or computer programs utilized for this study, please address correspondence to the authors at martin.ruhs@conted.ox.ac.uk and j.wadsworth@rhul.ac.uk.

Notes

References

Apgar

Lauren A.

2015. Authorized status, limited returns: The labour market outcomes of temporary Mexican workers. EPI Briefing Paper. Washington, DC: Economic Policy Institute.

Autor

David

. 2003. Outsourcing at will: The contribution of unjust dismissal doctrine to the growth of employment outsourcing. Journal of Labor Economics 21(1): 1–42.

Bailey

Thomas

. 1985. The influence of legal status on the labor market impact of immigration. International Migration Review 19(2): 220–38.

Borjas

George J.

Tienda

Marta

. 1993. The employment and wages of legalized immigrants. International Migration Review 27(4): 712–47.

Borjas

George J.

Trejo

Steven

. 1991. Immigrant participation in the welfare system. Industrial and Labor Relations Review 44(2): 195–211.

Calavita

Kitty

. 1992. Inside the State: The Bracero Program, Immigration and the INS. New York: Routledge.

Chiswick

Barry R.

1984. Illegal aliens in the United States labor market: An analysis of occupational attainment and earnings. International Migration Review 18: 714–32.

De Genova

Nicholas P.

2002. Migrant “illegality” and deportability in everyday life. Annual Review of Anthropology 31: 419–47.

Drinkwater

Stephen

Robinson

Catherine

. 2013. Welfare participation by immigrants in the UK. International Journal of Manpower 34(2): 100–112.

10.

Eurostat. 2011. Labour Market Statistics, 2011 edition. Luxembourg: Eurostat Pocketbooks. doi:10.2785/15405.

11.

Fasani

Francesco

. 2015. Understanding the role of immigrants’ legal status: Evidence from policy experiments. CESifo Economic Studies 61(3–4): 722–63.

12.

Jérôme

Gautié

Schmitt

John

(Eds.) 2010. Low-Wage Work in the Wealthy World. New York: Russell Sage Foundation.

13.

Gower

Melanie

Hawkins

Oliver

. 2013. Ending of transitional restrictions for Bulgarian and Romanian workers. Research Briefing, SN/HA/06606. London: House of Commons Library.

14.

HM Revenue & Customs. 2013. Onshore employment intermediaries: False self-employment. Consultation Document. London: HM Revenue & Customs.

15.

Hall

Peter

Soskice

David

. 2001. Varieties of Capitalism. Oxford: Oxford University Press.

16.

Hansen

Jorgen

Lofstrom

Magnus

. 2003. Immigrant assimilation and welfare participation: Do immigrants assimilate into or out of welfare? Journal of Human Resources 38(1): 74–98.

17.

Home Office. 2014. Immigration statistics, January to March 2014. London: Home Office.

18.

Hotchkiss

Julie L.

Quispe-Agnoli

Myriam

. 2013. The expected impact of state immigration legislation on labor market outcomes. Journal of Policy Analysis and Management 32(1): 34–59.

19.

Kennedy

Steven

. 2015. Measures to limit migrants’ access to benefits. House of Commons Library Briefing Paper No. 06889. London: House of Commons.

20.

Kerr

Sari Pekkala

Kerr

William R.

2011. Economic impacts of immigration: A survey. NBER Working Paper No. 16736. Cambridge, MA: National Bureau of Economic Research.

21.

Kossoudji

Sherrie A.

Cobb-Clark

Deborah

. 2000. IRCA’s impact on the occupational concentration and mobility of newly-legalized Mexican men. Journal of Population Economics 13(1): 81–98.

22.

Kossoudji

Sherrie A.

Cobb-Clark

Deborah

. 2002. Coming out of the shadows: Learning about legal status and wages from the legalized population. Journal of Labor Economics 20(3): 598–628.

23.

Lofstrom

Magnus

Hill

Laura

Hayes

Joseph

. 2013. Wage and mobility effects of legalization: Evidence from the New Immigrant Survey. Journal of Regional Science 53(1): 171–97.

24.

Lowell

B. Lindsay

Avato

Johanna

. 2007. The wages of skilled temporary migrants: Effects of visa pathways and job portability. Paper presented at a meeting of the Population Association of America, New York.

25.

Manning

Alan

. 2011. Imperfect competition in the labour market. In Ashenfelter

Orley

Card

David

(Eds.), Handbook of Labor Economics, Volume 4, Part B, pp. 973–1041. Oxford: Elsevier.

26.

Massey

Douglas S.

1987. Do undocumented migrants earn lower wages than legal immigrants? New evidence from Mexico. International Migration Review 21(2): 236–74.

27.

Migration Observatory. 2014. Bulgarians and Romanians in the British National Press. Accessed at http://www.migrationobservatory.ox.ac.uk/wp-content/uploads/2016/04/Report-Bulgarians_Romanians_Press_0.pdf (January 15, 2017).

28.

Orrenius

Pia M.

Zavodny

Madeline

. 2015. The impact of E-Verify mandates on labor market outcomes. Southern Economic Journal 81(4): 947–59.

29.

Papademetriou

Demetrios

O’Neil

Kevin

Jachimowicz

Maia

. 2004. Observations on regularization and the labor market performance of unauthorized and regularized immigrants. Paper prepared for the European Commission. Brussels: DG Employment and Social Affairs.

30.

Pena

Anita Alves

. 2010. Legalization and immigrants in US agriculture. BE Journal of Economic Analysis & Policy 10(1): Article 7.

31.

[REC] The Recruitment and Employment Confederation. 2014. Changes to tax legislation from 6 April 2014. Accessed at https://www.rec.uk.com/__data/assets/pdf_file/0004/140098/Tax-and-NICs-changes-6-April-2014-REC-briefing-note-1-28-03-14.pdf (February 2017).

32.

Rivera-Batiz

Franciso L.

1999. Undocumented workers in the labor market: An analysis of the earnings of legal and illegal Mexican immigrants in the United States. Journal of Population Economics 12: 91–116.

33.

Ruhs

Martin

. 2017. The impact of acquiring EU status on the earnings of East European migrants in the UK: Evidence from a quasi-natural experiment. British Journal of Industrial Relations (Early View). doi:10.1111/bjir.12223.

34.

Ruhs

Martin

. 2013. The Price of Rights: Regulating International Labor Migration. Princeton, NJ: Princeton University Press.

35.

Ruhs

Martin

Anderson

Bridget

. 2010. Semi-compliance and illegality in migrant labour markets: An analysis of migrants, employers and the state in the UK. Population, Space and Place 16(3): 195–215.

36.

Ryan

Bernard

. 2008. The accession (immigration and workers authorisation) regulations 2006. Industrial Law Journal 37(1): 75–88.

37.

Stark

Oded

. 2007. Work effort, moderation in expulsion, and illegal migration. Review of Development Economics 11(4): 585–90.

38.

J. Edward

Taylor

. 1992. Earnings and mobility of legal and illegal immigrant workers in agriculture. American Journal of Agricultural Economics 74(4): 889–96.

39.

Tienda

Marta

Singer

Audrey

. 1995. Wage mobility of undocumented workers in the United States. International Migration Review 29(1): 112–38.