States struggle to establish multilateral cooperation on migration – yet they include more and more migration provisions in preferential trade agreements (PTAs). This article sheds light on this phenomenon by introducing the Migration Provisions in Preferential Trade Agreements (MITA) dataset. Covering 797 agreements signed between 1960 and 2020, this dataset offers a fine-grained coding of three types of migration provisions: those that facilitate the international mobility of service providers and labor migrants, protect migrant rights, and control unauthorized migration. Against the backdrop of limping multilateralism, we examine PTAs’ migration policy content with regard to two key cooperation dilemmas: conflicts of interest within developed countries and between them and developing countries. Facilitating business and labor mobility might be a possible way around the first dilemma, commonly referred to as the ‘liberal paradox': the tension between economic demands for openness and political calls for closure. Nevertheless, this facilitation is largely limited to highly skilled migrants and agreements between developed economies. Provisions for migration control tend to be included in agreements between developed and developing countries, which signals that states use issue-linkages to address the second dilemma, i.e. interest asymmetries. Finally, provisions for migrant rights stand out because they do not deepen over time. Our findings suggest that while PTAs have become an increasingly common venue for migration governance, the issue-linkage between trade and migration cooperation perpetuates entrenched divisions in the international system. The MITA dataset will allow researchers and policymakers to track the evolution of the trade-migration nexus and systematically investigate the motives for and effects of various migration provisions in PTAs.
Author contributions to research design and conceptualization: S.L. 50%, P.L. 25%, P.H.-S. 25%; statistical analysis: S.L. 10%, P.L. 80%, P. H.-S. 10%; writing: S.L. 40%, P.L. 30%. P. H.-S. 30%.
The order of authors reflects the significance of the authors’ contributions.
The Funding information section was missing from this article and should have read ‘This research was supported by the nccr – on the move funded by the Swiss National Science Foundation grant 51NF40-182897’.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
1 Introduction
A burgeoning literature addresses the inclusion of non-trade issues, such as human rights, labor, environmental protection or security provisions, in preferential trade agreements (PTAs, i.a., Ariel & Haftel, 2021; Dür et al., 2014; Lechner, 2018; Milewicz et al., 2018; Morin et al., 2018; Raess et al., 2018). Nevertheless, the proliferation of PTA provisions facilitating labor mobility, protecting migrant rights and fighting irregular migration has remained understudied. This is surprising because PTAs’ migration policy content goes well beyond states' existing multilateral commitments. What is more, such provisions have recurrently spurred political controversy across the globe. India’s demands for greater concessions on immigration have been the major obstacle to the establishment of a post-Brexit trade agreement with the UK. On the day Rishi Sunak became the UK's prime minister, his trade minister Greg Hands sought to allay public concerns arguing that "in the area of trade, what we’re talking about is mode four arrangements. These aren’t immigration arrangements. These relate to business visas not for permanent settlement" (Wickham, 2022).1 Such controversies are not new. In 2003, the US Senate sharply criticized the inclusion of visa commitments in the PTAs the US concluded with Chile and Singapore. It adopted a Resolution stating that "trade agreements are not the appropriate vehicle for enacting immigration-related laws or modifying current immigration policy."2 This resolution put a de facto ban on the US Trade Representative's competence to include further migration commitments in trade agreements (Umberger, 2008). Migration provisions in trade agreements have also come under trade unions’ attack for disadvantaging domestic workers, for instance by abolishing the qualification requirements for certain professions (Ekman & Engblom, 2019: 182ff.), and for failing to protect the socio-economic rights of migrant workers. As the largest federation of unions in the US put it, such provisions would be "trading away migrant rights" (AFL-CIO, 2016, see also Chaisse & Meng, 2016; Panizzon, 2010). Finally, the third type of migration provisions included in PTAs, the commitment to fight unauthorized migration, also is far from being uncontroversial. The negotiation of detailed provisions for the readmission of migrants staying irregularly in the EU in the ‘post-Cotonou’ partnership agreement with the countries of the Organisation of African, Caribbean and Pacific States of 2021, are a case in point (Carbone, 2022).
Against this backdrop, a detailed stock-taking and a better understanding of the content and development of migration policy provisions in PTAs seems overdue. This article contributes to filling this gap in the trade and migration policy literatures by introducing the Dataset on Migration Provisions in Preferential Trade Agreements (MITA). Covering all migration provisions in 797 PTAs signed worldwide between 1960 and 2020, MITA constitutes the most comprehensive data collection on migration content in PTAs. Its 236 variables provide for a detailed measurement of provisions that regulate people’s mobility, protect migrant rights and foster migration control, and allow for large-N comparative analyses of PTA design in these three dimensions.
Advertisement
In this dataset-article, we substantiate MITA's analytical value by assessing PTAs’ migration policy content against two key cooperation dilemmas in the migration policy literature. The first dilemma concerns conflicts of interest within states, i.e., states' difficulty to reconcile economic demands for mobility with political calls for closure, also known as the 'liberal paradox' (Hollifield, 1992; see also Freeman, 1995; Hollifield & Foley, 2022; Ruhs, 2013). The second cooperation dilemma stems from conflicts of interest between states and the asymmetric structure of interdependence, which pits countries that mainly identify as a destination for immigration against countries that identify as countries of origin (Betts, 2001; Lahav & Lavenex, 2013; Sykes, 2013). This asymmetry, especially between the North and the South, means that states’ interests in international cooperation are rarely reciprocal (Adamson & Tsourapas, 2019; Hatton, 2007; Money & Lockhart, 2018; Peters, 2019). As a result, developed and developing countries have typically pursued diverging foreign policy objectives on migration and international codification has remained patchy.
The article first introduces the MITA dataset, its connections to existing datasets, its structure and its contents. We then discuss the main obstacles to international cooperation identified in the literatures on economic migration, migrant rights and migration control. These cooperation dilemmas become the starting point of our investigation of how far the distribution of migration provisions related to mobility, rights and control in PTAs reiterate or eventually transcend established divisions over migration cooperation between developed and developing countries.
2 Migration provisions in trade agreements – the MITA dataset
The Migration Provisions in Preferential Trade Agreements (MITA) dataset is a comprehensive source of data about the migration-related content of preferential trade agreements.3 The endeavor to establish this dataset fits into the broader trend of quantifying policies in the fields of migration (De Haas et al., 2014; Helbling et al., 2017) and trade, including non-trade issues such as environmental and labor standards or security provisions (Ariel & Haftel, 2021; Dür et al., 2014; Lechner, 2018; Milewicz et al., 2018; Morin et al., 2018; Raess et al., 2018). So far, these efforts have remained unconnected. The existing datasets on migration policy focus primarily on national policies and leave out relevant international provisions. At the same time, we lack a systematic attempt to study the growing nexus between trade and migration in PTAs. The datasets on non-trade issues in PTAs do not generally pay attention to migration content, as they usually concentrate on broader categories such as labor and social rights (Milewicz et al., 2018; Raess et al., 2018). The most detailed existing measurement of migration provisions (Pauwelyn et al., 2019) is part of the World Bank Deep Trade Agreements dataset, which includes provisions falling in the "Visa and asylum" category developed by the WTO.4 The latter covers 100 trade agreements and codes a selection of 30 variables that capture commitments "beyond what is covered under GATS mode 4" (idem: 228). The extended version of the DESTA database (Dür et al., 2014), which codes provisions on the temporary entry of business persons, is another source with partial coverage of migration provisions in PTAs. The nine variables in this area capture the general inclusion of business mobility in PTAs, but do not supply information about the categories of workers who benefit from facilitated mobility or the specific conditions under which they do so. Nor do they address provisions for migrant rights and migration control.
The MITA dataset goes beyond these pioneering studies in its coverage of both the number of PTAs and the scope and depth of their migration content. To construct our dataset, we build on the list of PTAs complied by the DESTA project (Dür et al., 2014), the WTO’s PTA database, as well as the websites of governments and intergovernmental organizations, which results in a total of 797 preferential trade agreements signed between 1960 and 2020.5 Each agreement constitutes one unit of observation. We cover bilateral, plurilateral and regional agreements, thus making the MITA dataset one of the broadest and most inclusive datasets on trade agreements. We understand migration in a broad sense as the movement of natural persons across international borders, irrespective of any specific duration of stay in the destination country. Accordingly, we define ‘migration provisions' in PTAs as rules and regulations that aim to govern international migration. We then classify migration provisions based on states’ cooperation objectives and distinguish between mobility provisions, rights provisions, and control provisions, which we conceptualize further below. The three types of provisions correspond to different dimensions of national migration policies and can be attributed to existing classifications in dedicated migration policy indices, such as the DEMIG Policy dataset, which differentiates between border control, legal entry, integration and exit (De Haas et al., 2014). Migration provisions in PTAs cover legal entry policies in the form of mobility provisions, integration policies in the form of rights provisions, and immigration deterrence and enforcement in the form of control provisions.
Advertisement
The dataset thus covers all migration provisions in PTAs including their modalities, which adds up to a total of 236 variables and allows us to measure the scope and depth of countries’ commitments. This fine-grained measurement provides a valuable data source for descriptive and analytical purposes, especially by making it possible to aggregate variables into higher-level indices to accommodate different research objectives.
The MITA dataset consists of four parts (see Fig. 1). The technical variables relate to the agreement itself, such as the signing parties, the year it was signed, and its type. These variables provide valuable context information and make mergers with additional data on the PTAs or the agreements' signing parties (e.g., DESTA database by Dür et al., 2014) possible. The remaining parts cover the three types of migration provisions, whereby the mobility section is partially organized around 11 categories of people (see below) and thus contains a number of sub-variables per category. We code PTAs’ migration-related content following a detailed codebook (see Appendix). The codebook was discussed with external experts and refined in several rounds of explorative coding. To ensure reliability, each agreement has been coded by two independent coders and disagreements between the two coders were subsequently discussed and jointly resolved with a referee to determine the final codes. The Kappa-value of 0.814 suggests a strong inter-coder reliability. We use a common dichotomous scale to code the presence or absence of a particular provision in each agreement.
Fig. 1
Structure of the MITA dataset
×
3 Mobility, migrant rights and migration control provisions in PTAs
Unlike other non-trade issues, such as human rights or environmental protection, migration has seen little codification in multilateral treaties. Although two UN compacts on migration and refugees concluded in 2018 emphasize the need for closer international cooperation, states have remained reluctant to concede national sovereignty over the admission of foreigners and to make binding commitments (Kainz & Betts, 2021; Lavenex, 2020). Existing international norms date back to the post-World War II period and concentrate on states' responsibilities towards refugees (the 1951 Geneva Convention and its 1967 Protocol) and migrant workers (the poorly ratified 1949 and 1975 Conventions of the International Labor Organization and the 1990 UN Convention on the Protection of the Rights of All Migrant Workers and Members of their Families, UNCMRW, Chetail, 2019). However, two core aspects of migration policy—the admission of economic migrants and the control over irregular migration—are widely considered "the last bastion of state sovereignty" (Dauvergne, 2014, p. 92). The only exception is a limited set of commitments that facilitate the mobility of “natural persons” moving for commercial purposes, which were established in the WTO’s 1995 General Agreement on Trade in Services (GATS) and are commonly referred to as "mode four" liberalization (Dawson, 2013; Mattoo & Carzaniga, 2003; Trachtman, 2009). International cooperation on migration has progressed more at the regional level (Geddes, 2021; Lavenex, 2018) and through various bilateral arrangements, including bilateral labor agreements (Peters, 2019), readmission agreements (Stutz & Trauner, 2022), more informal 'migration diplomacy' (Adamson & Tsourapas, 2019) and, as we show below, PTAs.
Provisions that facilitate labor mobility, protect migrant rights or foster migration control relate to PTA's general objectives to different extents. Labor mobility is inherent to market integration, and the GATS recognizes the economic value of facilitating the cross-border movement of persons who deliver services. The rights of migrant workers can be seen both as a corollary of labor rights that have been included as safeguards against the potential negative externalities of free trade on labor, and as a commitment to ensuring the social and economic rights of PTA partner countries’ citizens abroad. The third type of migration provisions in PTAs,—the parties’ commitment to fight irregular migration—cannot be related to improving their commercial relations, nor to addressing the externalities of the trade agreements, and therefore constitute the most unexpected migration policy content in PTAs.
These three types of migration provisions relate to distinct areas of states' immigration policies and invoke different logics and dilemmas of international cooperation. In the following subsections, we briefly present these provisions before investigating the patterns of their distribution in PTAs.
3.1 Mobility provisions
The category of ‘mobility provisions’ comprises all clauses that facilitate the cross-border movement of natural persons and is the most complex type of migration provisions in PTAs. The inclusion of such provisions in PTAs is rather surprising considering the hardening political climate regarding migration around the world. The fact that states have entered legally binding international commitments that facilitate these flows points to the existence of important economic interests. These interests are rooted in the turn to service- and knowledge-based economies, the ensuing global 'race for talent,’ the proliferation of multinational enterprises and, more generally, the intensification of transnational investment and business practices in advanced economies (Weinar & Klekowski 2020). Powerful associations of service industries, such as the European Services Forum and the US Coalition of Service Industries, have long been lobbying for easier business mobility, arguing that natural persons’ mobility is often inseparable from the delivery of a service or transnational investment (Drake & Nicolaïdis, 1992; Sapir, 1994). Eventually, the focus on highly skilled business migration linked to investment shaped countries’ commitments in the GATS, sidelining the demands for wider openings for less-highly skilled workers independent from foreign investment voiced by developing countries under India’s lead (Dawson, 2013; Lavenex & Jurje, 2021).
To capture all forms of mobility provisions in PTAs, we differentiate between general labor mobility, which grants individuals access to countries' labor markets, and service mobility, which grants them access for the purpose of service provision, similar to what the GATS has defined as ‘mode four’ (cf. Lavenex & Jurje, 2015; Mattoo & Carzaniga, 2003). In practical terms, both forms imply labor migration in the sense of entering the partner country for the purpose of work.
The coding scheme differentiates three categories of mobility commitments: the substantive categories of covered persons, the modalities of mobility, and the status of mobility provisions within a PTA. The main categories of persons that PTAs provide for are business visitors (BV), contractual service suppliers (CSS), independent professionals (IP), and intra-corporate transferees (ICT).6
These categories differ in their connection to commercial presence (i.e., foreign direct investment or ‘mode 3’ according to the GATS) and their level of skills (see Table 1). The categories linked to commercial presence are ICTs and BVs. ICTs move from one branch of a multinational company to another, while retaining their work contract in the country of origin. In most individual GATS commitments and PTAs, ICTs are defined as executives, managers and/or specialists, meaning employees who either hold high positions within a company or possess specific skills needed for the activity in the other country. The commitments for ICTs’ duration of stay can reach several years, whereas the commitments for other categories of persons are usually limited to one year or less. BVs are employees of a company who wish to do business in the other country. In most cases, their business visits are linked to the establishment of a company in the partner country. In contrast, IPs and CSSs are not linked to commercial presence and do not have predetermined levels of skills. IPs are self-employed individuals who move to provide a service in another country, while CSSs have employment contracts with a firm in their country of origin and are sent by their firm to deliver a service in the host country. In addition to these four main categories, PTAs also include commitments on other business people, who sometimes overlap with the ICT and CSS categories (executives, managers, specialists, investors, installers, business sellers, trainees, other business persons) and non-business people (such as cultural professionals, tourists, students and researchers).
Table 1
Main categories of service suppliers under GATS mode four
Source: Table adapted from Lavenex and Jurje (2015)
The second category of mobility commitments coded in MITA consists of modalities that regulate the mobility between the signing parties. They allow us to identify the extent to which PTAs regulate the admission and stay of the various categories of persons. We differentiate between modalities that constrain and modalities that facilitate the mobility of natural persons (Table 2).
Table 2
Modalities that regulate mobility
Facilitating mobility
Constraining mobility
Exception from economic needs test
Recognition of professional qualifications
Sector expansion
Spouses’ rights
Exemption from/limitation of quota
Right to domestic employment
Fee limitation
Facilitated application and entry procedure
Visa renewal mechanism
Information provision
Regulatory cooperation
Most-favored-nation clause
National treatment
Dispute settlement mechanism
Compliance commitments with GATS or other international agreements that facilitate mobility
Economic need tests
Skills, experience, and pre-employment requirements
Limitation of business activity/sector
Proportional employment restrictions
Quantitative limits
Limited duration of stay
Clause on national security or public health interest
Technology transfer requirements
Restrictions of permanent settlement in the country
Mobility restrictions in the country
Finally, the third category of commitments measures the status of the mobility provisions in an agreement by whether facilitating mobility is defined as an overall objective of the agreement and whether mobility provisions have their own dedicated chapter.
To assess the validity of our measurement, we compare the aggregate number of mobility provisions in our dataset to the number of commitments on business mobility in the DESTA database (Dür et al., 2014). We find a high correlation of r = 0.83 between the two datasets (see the Appendix for more details on the validity tests).
3.2 Rights provisions
The second type of migration provisions concerns the rights of labor migrants and refugees. This type of provision is most closely related to other non-trade issues in PTAs, as it also addresses the potential negative externalities of free trade, such as labor or social rights. Existing datasets on non-trade issues in PTAs either code broader categories, such as labor rights, human rights or economic and social rights (Milewicz et al., 2018; Lechner, 2018), or only focus on labor rights and exclude migrant rights (Raess et al., 2018). Provisions on migrant rights are distinct from those on labor rights: while the latter are meant to protect all workers within a country, the former aim to protect the rights of migrants who have moved from one PTA partner country to another or, in some rare cases, the socio-economic rights of refugees residing in PTA parties. Migrant rights provisions can be linked to mobility provisions in the same PTA or can apply to all migrants from a given country in the PTA’s partner country. Consequently, they can either be a corollary of the facilitation of mobility in PTAs or an instance of issue-linkage, where sending countries aim to strengthen their citizens’ rights via PTA negotiations. Even though migrant socio-economic rights are codified at the multilateral level in the ILO Conventions and 1990 UN Migrant Workers Convention (UNCRMW, Lonnroth, 1991; Pécoud, 2017), these conventions are poorly ratified. Wealthy and democratic immigrant-receiving countries recognize many of these rights in domestic law but have traditionally been reluctant to make binding international commitments on migrant rights that they perceive as a costly and/or symbolic constraint to their discretion over migration (Ruhs, 2012). In consequence, no country of the global North has ratified the UNCRMW.
In the MITA dataset, we capture three categories of provisions for migrant rights: substantive rights, the modalities of regulatory cooperation between the parties and the status of migrant rights in the PTAs.7 Substantive rights cover general anti-discrimination clauses and specific economic and social rights, including equal access to social security, the right to transfer social insurance capital or refugees’ access to the labor market. Regulatory cooperation commitments involve dialogue and coordination on migrant rights as well as compliance provisions (international conventions on migrant rights, such as ILO Conventions or the UNCRMW). As with the other types of provisions, the status of migrant rights provisions is measured by whether they appear in the agreements’ objectives and by whether there exists a dedicated chapter on the matter.
3.3 Control provisions
Provisions for migration control are provisions related to immigration enforcement and the prevention of unauthorized immigration. These provisions are different from mobility and rights provisions or other non-trade issues, because they do not refer to the liberalization of markets or the promotion of global public goods but to the enforcement of territorial borders. The capacity to curb irregular migration and to deport unauthorized migrants has become a top priority in Europe, Australia and North America (Money & Lockhart, 2018). Yet, reaching these objectives largely depends on the collaboration of sending and transit countries, which benefit little from controlling emigration or forced returns because such collaboration is costly, it reduces the benefits they derive from emigration (such as remittances) and creates challenges related to the reintegration or relocation of returning migrants. Therefore, the fundamental conflict of interest between countries of origin and transit, which tend to be developing countries, and high-income receiving countries is even greater in this respect (Ellermann, 2008, p. 171). This asymmetry of interests leaves few opportunities for mutual gains from cooperation (Axelrod & Keohane, 1985) and developing countries' collective power in multilateral institutions makes it unlikely for cooperation on migration control to materialize therein (Lahav & Lavenex, 2013, p.757f.). Instead, developed economies have intensified their efforts to enlist sending and transit countries in such forms of cooperation through capacity building, training, and political pressure, albeit with limited success (Kainz & Betts, 2021; Money & Lockhart, 2018). Against this backdrop, PTAs may constitute an attractive venue whereby wealthy liberal democracies can mobilize their economic leverage through strategic issue-linkage, i.e., demanding the inclusion of provisions on migration control in PTAs with developing countries in exchange for commercial concessions (Jurje & Lavenex, 2014). In this case, the inclusion of readmission and border-control measures in PTAs constitutes an instance of strategic venue-shopping and issue-linkage on the part of developed countries, which use PTAs to leverage their superior market power in order to extract concessions from the economically weaker developing countries.
We measure control provisions via three categories of commitments: substantive provisions, the modalities of regulatory cooperation between the signing parties, and the status of migration control in the PTA. Substantive provisions comprise commitments to fight irregular migration, commitments linking development aid to cooperation on migration control, as well as commitments on the readmission and re-integration of unauthorized migrants. The second category includes commitments on establishing a dialogue or regulatory cooperation on migration control (for both prevention and enforcement), as well as compliance provisions related to existing agreements on migration control. Finally, the status of migration control is measured by whether migration control is mentioned as an overarching objective of the agreement, whether there is a dedicated chapter on it and whether the provisions on migration control are subject to the dispute settlement mechanism established in the PTA.
4 Trends and patterns of migration regulation in PTAs
What can the MITA dataset tell us about PTAs’ role as a venue for migration policy cooperation along the three dimensions of mobility, rights and control? In the following, we first provide an overview of the evolution of migration provisions in conjunction with the general proliferation of non-trade issues in PTAs. We then shed light on which types of countries are more likely to sign PTAs with mobility, rights and control provisions, respectively. For that purpose, we build a measure for each of the three types of provisions based on whether such provisions are included in each agreement (dummy variable) and the number of such provisions in each agreement (continuous variable).8 The first measure captures whether an agreement covers mobility, migrant rights or migration control, whereas the second measure sheds additional light on the depth of these commitments. Instead of employing a substantive weighting of the different provisions, these frequency measures reveal the extent and the level of detail of migration-related content in a PTA. We only apply the continuous frequency measure to PTAs with at least one migration provision. To study the relationship between migration provisions in PTAs and overarching cooperation dilemmas, we classify the signing parties based on their level of development, which structures their position in the global migration dynamics: developed countries attract most of the world’s migrants, whereas developing countries send most of the world’s migrants.9 We choose two identification strategies: first, we distinguish OECD countries from non-OECD countries. As a robustness test, we use an alternative operationalization, which distinguishes high-income countries from non-high-income countries based on the World Bank's world development indicators.10 Accordingly, we group the agreements according to their signatories, distinguishing between PTAs between developed countries, PTAs between developing countries, and PTAs between developed and developing countries.11 While well established in the migration literature, this broad categorization between developed and developing countries hides rich variation within each category. Therefore, in a second step, we look at the countries with the highest numbers of PTAs with mobility, rights and control provisions, respectively, to further explore the context of migration governance through trade agreements. The discussion is based on a descriptive analysis of our two types of frequency measures, as well as pair-wise mean comparisons that use t-tests and regression models that assess how migration provisions vary across signatories' levels of development. Finally, we contextualize the proliferation of migration provisions against the more general trend of deeper trade agreements by calculating the correlation between migration provisions and PTAs’ overall depth, as well as with substantively related non-migration provisions, such as service trade liberalization for mobility and labour standards for migrant rights.
4.1 The evolution of migration content in trade agreements
The longitudinal perspective shows that migration provisions have become increasingly common in PTAs over time (Fig. 2a). Between 1960 and 2020, the share of new trade agreements with migration provisions has continuously increased from rare occurrences of around 20% between the 1960s and the 1980s to around 75% of all PTAs signed between 2010 and 2020. The three types of migration provisions have largely evolved in parallel, but mobility provisions started to proliferate earlier and, especially after the adoption of the GATS in 1995, their growth rate has been higher than that of rights and control provisions. By and large, this proliferation mirrors the more general trend of including non-trade issues in PTAs (Fig. 2b; Milewicz et al., 2018).
Fig. 2
Evolution of PTAs with migration provisions.
Note: The figures display the frequency of PTAs with at least one migration provision. The left plot shows the annual number of new signed PTAs that include migration provisions (bars) and their share of the total number of signed PTAs as a moving average over a decade (line). The right plot displays the cumulative number PTAs with the three types of migration provisions over time, as well as PTAs with non-trade-issues (corruption, environment, labor issues). Based on N = 797 agreements. Source: MITA & DESTA
×
The sum of migration provisions in PTAs, which indicates the deepening of such commitments, shows a similar pattern (Fig. 3). Mobility provisions stand out for becoming more numerous (or deeper) over time and the average sum of control provisions in PTAs has broadly stabilized, but provisions for migrant rights have become slightly shallower after 1990. Most recent PTAs with migration provisions include around 25 mobility provisions, but only around three to four rights or control provisions, respectively.
Fig. 3
Average depth of PTAs with migration provisions.
Note: The figures display the depth of migration provisions in PTAs with at least one such provision. The smoothed lines display the average number of mobility, rights and control provisions per PTA over time. Data points jittered to reduce overplotting.
Source: MITA
×
4.2 Facilitating mobility in PTAs
Mobility provisions come in two shapes: linked to trade in services (the so-called ‘mode four’ provisions) (N = 240 agreements, around 30% of all PTAs) and, more rarely, in the form of liberalization of labor mobility (N = 40 agreements, around 5% of all PTAs). The latter are mainly found in regional integration agreements.
The fact that states increasingly liberalize economic mobility via PTAs may point to a possible way around the 'liberal paradox,’ which pits economic demands for foreign labor against citizens' anti-immigration concerns. As the term suggests, such conflicts of interest are particularly strong in liberal democracies. The latter combine open, internationalized service economies competing for 'talent' with democratic institutions and constituencies opposed to the opening of more immigration channels. PTAs’ bilateral set-up allows for tailor-made deals in response to economic demands and factors in power differentials by allowing the economically stronger party to set the agreements’ conditions. PTAs’ trade context also offers several advantages over alternative avenues, such as bilateral labor agreements or unilateral reforms of immigration laws. As the quote by the British trade minister in the introduction to this article indicates, framing labor mobility as part of PTAs emphasizes migration’s commercial aspect and embeds people’s cross-border movements in a wider set of transactions that foster growth and prosperity (Hoffmeyer-Zlotnik, 2020; Lavenex & Jurje, 2015). Furthermore, the focus falls on temporary mobility and categories of workers that are generally excluded from contentious debates about the implications of immigration on wages, labor conditions or welfare, namely highly skilled managers, specialists and other business people. In sum, the mobility provisions found in PTAs have a de-politicizing effect, which singles these labor flows out from overarching immigration debates. This is evident in the relatively weak involvement migration ministries have in negotiating such commitments. Analyses of the Uruguay Round, which led up to the GATS, corroborate the leadership of countries’ trade ministries, which acted in concert with economic lobbies, over migration ministries. Using highly technical language, such as 'mode four' liberalization, to designate measures that facilitate the cross-border mobility of natural persons, trade talks circumvented the use of migration terms and thus largely avoided political debates (Drake & Nicolaïdis, 1992; Lavenex, 2007). In some cases, the sensitivity of migration commitments comes to bear later, as prominently shown by the case of the US Congress in the debates surrounding the ratification of the PTAs with Chile and Singapore, or, most recently, the UK-India trade talks (see above and Umberger, 2008).
While the inclusion of mobility provisions in PTAs may thus mitigate the first obstacle to international cooperation—the conflicts of interest that result from the 'liberal paradox' within developed economies—the same cannot be expected of the second impediment, which has to do with the asymmetry of the interdependence and the concomitant conflicts of interest between developed and developing countries. Highly skilled business migration is less pertinent to most developing countries, which do no not have large, export-oriented service economies, nor do they promote outward foreign investment through multinational companies. Apart from these economic factors, many developing countries may also be reluctant to ease the outward mobility of highly skilled workers, as this could exacerbate their risk of brain drain (Özden & Schiff, 2006). Instead, developing countries favor facilitating mobility for low-skilled workers, in order to export labor overflows and benefit from remittances (Peters, 2019). A prominent example is India, which, both in the GATS negotiations and again in its bilateral talks first with the EU and now with the UK, seeks to widen the scope of skill levels under 'mode four' commitments (Lavenex & Jurje, 2021). Nevertheless, this type of labor migration is in abundant supply and developed countries can unilaterally satisfice their needs through domestic policies or bilateral labor agreements (Peters, 2019; Ruhs, 2013). Indeed, our PTA data corroborates the preference in the GATS commitments for skilled and investment-related categories of persons, such as ICTs or BVs, which are in the interest of developed countries (Dawson, 2013), and lends support to a clearly more limited level of liberalization for CSSs or IPs (see Fig. 4). Apart from these four main categories, we find a number of other categories of lower frequency – mostly linked to investment and highly skilled (investors, business sellers), and, less frequently, not highly skilled persons (installers, trainees). These observations lead us to expect that mobility provisions concentrate in PTAs between developed countries and are less frequent in mixed agreements between developed and developing countries, as well as in agreements between developing countries.
Fig. 4
Frequency of mobility categories in PTAs.
Note: The bar plot displays the share of PTAs that facilitate the mobility of different categories of natural persons. These are business visitors (BVs), intra-corporate transferees (ICTs), independent professionals (IPs), contractual service suppliers (CSSs), investors (INVs), business sellers (BSs), installers (INSs) and trainees (TRs). Based on N = 797 agreements. Source: MITA
×
In a next step, we assess the pattern of mobility provisions across the signing parties’ level of development. The pattern lends tentative support to our expectations: While around 55% of agreements between OECD countries include mobility provisions, only every fourth agreement between non-OECD countries does so (Fig. 5a). Mixed agreements between an OECD and a non-OECD country have mobility provisions in about half of all cases. These differences across disparate levels of development are even more pronounced when we rely on the World Bank classification of countries’ levels of development (see Figure A4 in the Appendix). A similar pattern emerges when we examine the depth of mobility provisions (Fig. 5b). We find that the number of such provisions tends to be highest in agreements with OECD countries (there are no significant differences between OECD-only agreements and mixed agreements). PTAs signed between non-OECD countries have the lowest number of mobility provisions, a pattern also confirmed in analyses relying on the World Bank’s classification (see Figure A5 in the Appendix). Finally, we find a strong correlation between agreements that cover trade in services and the inclusion of mobility provisions (r = 0.72), as well as between the inclusion of mobility provisions and the overall depth of economic liberalization (r = 0.62).12 These results support the conclusion that mobility provisions are closely linked to countries’ trade liberalization agenda.
Fig. 5
Mobility provisions by parties' level of development.
Note: The bar plot displays the share of PTAs with mobility provisions (in dark grey) based on parties’ level of development. The box plots show the distribution of the absolute number of mobility provisions in PTAs that include at least one such provision. The dots represent single PTAs and are jittered to reduce over-plotting. The plot shows p-values for pairwise group mean comparisons based on t-tests. See Tables A1 and A2 in the Appendix for the regression models estimating the PTA type comparisons. Source: MITA
×
As expected, the facilitation of mobility mainly takes place between developed countries. Nevertheless, we find a substantial number of provisions in mixed agreements with non-OECD countries. We further explore this pattern by looking at the ten countries with the highest number of PTAs that include mobility provisions (Fig. 6). With more than 60 such agreements, the EU is the leading actor on this front. Other developed countries in the Top10 are Singapore, Korea, the United Kingdom and the United States with around a quarter of the EU’s PTAs – however, in comparison, these countries often have a higher share of PTAs that contain mobility provisions than the EU. Developing countries with a high absolute and relative number of their PTAs including mobility provisions tend to be the large economies in Latin America and China. Emerging economies share developed countries' interest in both inward and outward investment and in facilitating the mobility of business-people in multinational companies (Ekman & Engblom, 2019, p. 170). Indeed, they played an active role in the promotion of such provisions in the stalled Doha Round of the WTO and have sought far-reaching commitments in bilateral trade negotiations (Lavenex & Jurje, 2021). Our data confirms that the PTAs concluded by emerging economies with strong ties to world markets document a strong interest in facilitating trade-related mobility.
Fig. 6
Number of PTAs with mobility provisions by country or bloc.
Note: The bar plot displays the ten signing parties with the highest absolute number of PTAs that include mobility provisions. The relative share is displayed in percentages. Source: MITA
×
4.3 Promoting migrant rights through PTAs
Provisions for migrant rights are much rarer than overall mobility provisions, with about 10% of all PTAs in our dataset containing at least one such provision. Their development has also been less dynamic over time. Although the number of PTAs that include migrant rights has significantly increased since the 1990s, the average depth of these provisions has actually decreased over the same period (see Fig. 2 above). The most frequent provision is a general non-discrimination clause, followed by commitments to the transfer of social insurance and access to social benefits. Less than 4% of all PTAs include provisions for refugee protection (see Fig. 7).
Fig. 7
Frequency of rights provisions in PTAs.
Note: The bar plot displays the share of PTAs that include substantive provisions for migrant rights. DIS stands for non-discrimination of migrant workers, TRA for social insurance transfer, SOC for access to social security and REF for commitment to protect refugees. Based on N = 797 agreements. Source: MITA
×
In contrast to mobility commitments, migrant and refugee rights are not a corollary of commercial interests. Furthermore, they are subject to stronger multilateral codification in dedicated fora (the ILO, the UNCRMW and the 1951 Geneva Convention on Refugees). Multilateral cooperation on migrant rights has been driven by sending countries that seek to protect the rights of their citizens residing abroad. This is also reflected in the ratification patterns of the two ILO Conventions and the UNCRMW (Kainz & Betts, 2021; Lonnroth, 1991). One could therefore assume that the inclusion of commitments for migrant rights in PTAs is a demand from sending countries and that such commitments in mixed agreements result from issue-linkages, in which developing countries make trade concessions conditional on developed countries' consent to such clauses (Axelrod & Keohane, 1985). Yet, developing countries have little leverage to engage in such bargains. Moreover, developed democratic countries already guarantee a certain level of protection for migrants' economic and social rights as part of their commitment to human rights, even if they do not support further codification in international law (Chetail, 2019). The fact that developed countries sign on provisions for migrant rights in PTAs but not in multilateral treaties might thus be better explained by the possibility to only include select provisions and exclude more controversial issues, such as the rights of unauthorized migrants, which figure in the 1975 ILO Convention and the UNCRMW. This leads us to assume that, rather than strategic issue-linkages, the human rights character of migrant rights, the possibility to exclude controversial norms and the demand of emigration countries combine to explain the inclusion of such commitments in PTAs. We therefore expect that the inclusion of provisions for migrant rights shows less variation across development levels than that of mobility provisions.
Figure 8a shows that rights provisions are included almost exclusively in agreements that involve at least one OECD country—they appear in around 20% of all such cases. Rights provisions appear even slightly more often in agreements between two OECD countries than in mixed agreements. This pattern is also confirmed when we use the World Bank’s classification (see Figure A4 in the Appendix). Nevertheless, examining the depth of rights provisions per PTA paints a different picture (see Fig. 8b). There is a slight tendency for rights provisions to be more numerous in agreements that involve developing countries and most numerous in mixed agreements. Although migrants' socio-economic rights are included in different chapters, they have an affinity with labor standards. Therefore, we further check for the correlation between the two types of rights in PTAs.13 Our data reveals a moderate correlation of r = 0.40. Finally, as expected, rights provisions are less strongly associated with the depth of economic liberalization than mobility provisions (r = 0.34).
Fig. 8
Rights provisions by parties' level of development.
Note: The bar plot displays the share of PTAs with rights provisions (in dark grey) based on parties’ level of development. The box plots show the distribution of the absolute number of rights provisions in PTAs. The dots represent single PTAs and are jittered to reduce over-plotting. The plot shows p-values for pairwise group mean comparisons based on t-tests. See Tables A1 and A2 in the Appendix for the regression models that estimate the PTA type comparisons. Source: MITA
×
To further explore the existing cross-country variation, we look at the ten parties with the highest number of PTAs with rights provisions (see Fig. 9). The EU stands out for including migrant rights in around 50% of its PTAs; in this respect, it has an even more dominant position than in the case of mobility provisions. The EU is followed by the UK and Canada, two other major trade actors. While we also find Latin American countries (Peru, Chile and Colombia) in the Top10, European countries, Canada and – albeit less so—the US clearly have a higher share of PTAs with migrant rights. Finally, in contrast to the ranking on mobility provisions, we find no Asian countries on this list.
Fig. 9
Number of PTAs with rights provisions by country or bloc.
Note: The bar plot displays the ten signing parties with the highest absolute number of PTAs that include rights provisions. The relative share is displayed in percentages. Source: MITA
×
4.4 PTAs as a venue for migration control
Like provisions for migrant rights, migration control provisions are included in about 10% of all PTAs. Their proliferation has been less steep than those of mobility and rights provisions and their depth has stagnated after 2010 (see Figs. 2 and 3 above). Control provisions concentrate on cooperation in the fight against irregular migration and the readmission of nationals of the signatory parties staying irregularly in the other party (see Fig. 10).
Fig. 10
Frequency of control provisions in PTAs.
Note: The bar plot displays the share of PTAs that include substantive migration control provisions. These concern the fight against irregular migration (IRR), readmission commitments (REA), the link between development and migration control cooperation (DEV), and commitments concerning the re-integration of returned migrants and refugees (RST). Based on N = 797 agreements. Source: MITA
×
With their focus on law enforcement, provisions for migration control mainly bind wealthy migrant-receiving countries and less wealthy migrant-sending countries. Such provisions should therefore be most frequent in PTAs between developed and developing countries.
Figure 11 confirms that such provisions figure mainly in mixed agreements that involve both OECD and non-OECD countries (around 20% of agreements), followed by agreements between OECD countries (around 10%), and are practically absent in agreements that do not involve OECD countries. The differences in the pattern produced by using the World Bank classification are minor (see Figure A4 in the Appendix). The predominance of mixed agreements is even more pronounced for the depth of control provisions (see Fig. 11b). The World Bank classification yields the same pattern (see Figure A5 in the Appendix). This finding corroborates the expectation that control provisions target developing countries, which tend to be origin and transit countries for migrants who seek to enter developed countries without authorization.
Fig. 11
Control provisions by parties’s level of development.
Note: The bar plot displays the share of PTAs with control provisions (in dark grey) based on parties’ level of development. The box plots show the distribution of the number of control provisions in PTAs. The dots represent single PTAs and are jittered to reduce over-plotting. The plot shows p-values for pairwise group mean comparisons based on t-tests. See Tables A1 and A2 in the Appendix for the regression models estimating the PTA type comparisons. Source: MITA
×
Once again, the EU leads on this indicator, followed by the United Kingdom; both include control provisions in 39% of their PTAs (Fig. 12). Farther behind are countries that have signed PTAs with the EU and cooperate with the latter under its external migration policy, including the ACP states, Georgia, Moldova and Ukraine (Carrera et al., 2019). There are only five non-EU/UK agreements that include control provisions: the 2006 Japan-Philippines PTA, the 2010 South Korea-India PTA, and Peru’s PTAs with Costa Rica (2011), Guatemala (2011) and Panama (2012).
Fig. 12
Number of PTAs with control provisions by country or bloc.
Note: The bar plot displays the ten signing parties with the highest absolute number of PTAs that include control provisions. The relative share is displayed in percentages. Source: MITA
×
Since the provisions against irregular migration are the most weakly related to commercial exchanges out of all three types of migration provisions, we do not expect their presence to correlate strongly with the depth of economic liberalization in a given PTA. Indeed, we only find a moderate correlation between the inclusion of control provisions and the depth of economic liberalization in PTAs (r = 0.28). This might be the reason why other developed countries have not hitherto engaged in this issue-linkage, despite also being concerned about irregular migration. Rather, the EU's dominance echoes its reliance on PTAs as instruments of 'market power' in external relations (Damro, 2012) and highlights the political salience of the topic. Originally a domestic competence, immigration from third countries has come on the EU agenda in connection with the abolition of internal border controls codified in the 1985 Schengen Agreement. Consequently, the evolving common immigration policies have concentrated on ensuring strict entry requirements and high standards of control at the EU's external borders, including via cooperation with countries of transit and origin (Geddes et al., 2020). Considering that such cooperation is primarily in the interest of the EU, this trade-migration control linkage can be read as a move to leverage the EU's market power in order to incentivize third countries' collaboration (Jurje & Lavenex, 2014).14
5 Conclusions
With the Migration Provisions in Preferential Trade Agreements (MITA) dataset this article introduces a politically highly salient but hitherto much neglected 'trade and' issue. In contrast to other 'trade and' issues, such as environmental protection and human and labor rights, which all benefit from dedicated multilateral agreements, international migration stands out because international cooperation in the area is relatively rare. Against this background, and given the domestic sensitivity of the issue, the inclusion and proliferation of migration provisions in legally binding trade agreements is puzzling at least. Therefore, we introduce the dataset through the prism of the key cooperation dilemmas identified in the international migration literature: conflicts of interest within and between states.
The first constellation is captured by the notion of the 'liberal paradox' characteristic of wealthy democracies where economic demand for openness clashes with political calls for closure. The second constellation denotes the opposing interests of developing and developed states, which arise from the asymmetry of migration flows. Generally speaking, PTAs’ bi- or plurilateral, malleable set-up, the opportunities they offer for issue-linkages, and the de-politicizing effect of framing migration norms as commercial commitments present an attractive context for advancing some cooperation on international migration where it otherwise fails.
Regarding the 'liberal paradox,’ our findings corroborate the expectation that PTAs only partially help overcome states’ cooperation dilemmas. On the one hand, we see a steep rise in the number of PTAs that contain mobility provisions and a dynamic evolution in the number of such provisions per PTA. On the other hand, the commitments tend to focus on highly skilled migrants linked to investment, trade in services and multinational corporations – that is, categories of persons that primarily interest developed economies and are less politically contested. Consequently, mobility commitments concentrate in PTAs signed by developed countries and large emerging economies.
We find strong support for the argument that developed countries use PTAs as a venue to mitigate conflicts of interest with developing countries via issue-linkages in the realm of cooperation on migration control. However, this issue linkage is, with very few exceptions, limited to the EU’s PTAs. This corroborates the notion that trade connections constitute the primary tool of EU foreign policy and signals the EU's willingness to mobilize its market power in the pursuit of non-trade related migration policy goals (Hoffmeyer-Zlotnik et al., 2023).
Yet another pattern appears for migrant rights. Together with mobility, the latter represent the oldest type of migration provisions in PTAs. Their focus on non-discrimination and the portability of social security benefits mirrors the demands of migrant-sending countries codified in ILO/UN Conventions. Developing countries' interests are also reflected in the slightly deeper content of these provisions in PTAs signed between developing and developed countries. Yet, migrant rights also stand out for shallowing out, rather than deepening, over the last decades. This echoes the very low ratification of related multilateral conventions and herewith some of the main antagonisms in international migration governance.
Summing up, the MITA dataset shows that PTAs have become an established and expanding venue for international migration governance, especially as far as certain forms of temporary labor mobility and, more selectively, migration control and migrant rights are concerned. Nevertheless, rather than providing a panacea, the distribution of migration provisions across countries' development status mirrors long-standing divides anchored in the asymmetry of migration flows and levels of wealth.
To conclude, the dataset makes it possible to move the study of the trade-migration policy nexus much further than the glimpses we provide in this first introductory analysis. Possible applications include more fine-grained causal analyses of the driving forces behind the different types of migration provisions and their depth and scope. MITA can be used to systematically investigate why countries link trade and migration, what explains the inclusion and substance of migration provisions in PTAs, and what effects these provisions have on a series of economic and political outcomes, such as, for instance, migration legislation, migration flows or commercial exchanges. We hope that the dataset is going to be a valuable resource for scholars of migration and trade governance in comparative politics, political economy and international relations alike.
Acknowledgements
We thank Julia Gubler and Laura Mauricio for their excellent research assistance in the coding of migration provisions. We are grateful for the generous comments on earlier versions of the article that we received from Simon Hug, Thomas Sattler, Tanja Schweinberger, Stefanie Walter, and the four anonymous reviewers. The participants of the ECPR General Conference 2022, the 28th International Conference of Europeanists 2022 and a 2020 workshop at the University of Geneva further contributed valuable feedback. Furthermore, we thank Antonia Carzaniga, Elspeth Guild and Daniel Sormani for their expert inputs on the codebook. Any remaining errors are the responsibility of the authors.
Declarations
Conflict of interest
The authors whose names are listed immediately below certify that they have NO affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants; participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements), or non-financial interest (such as personal or professional relationships, affiliations, knowledge or beliefs) in the subject matter or materials discussed in this manuscript.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher's note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
"Mode four" refers to the four ways in which trade in services is liberalized under the WTO/GATS. Mode one refers to cross-border supply; mode two to consumption abroad, mode three to commercial presence, and mode four to the mobility of the natural person providing the service (Mattoo & Carzaniga, 2003).
This is based on the international categorization scheme, the list of WTO-X policy areas that includes “illegal immigration”, “visa and asylum” and “social matters” as migration-related regulations (Horn et al., 2010). The policy area "Visa and asylum" includes the exchange of information, drafting legislation and training (including the international movement of persons), "Illegal immigration" comprises re‐admission agreements and the prevention and control of illegal immigration. The area of "social matters" is migration-related as it includes the coordination of social security systems and non-discrimination in working conditions.
See the Appendix for the full list of agreements. The number of agreements covered by the MITA dataset is slightly lower than that of the DESTA dataset because we do not include country accessions, withdrawals and agreements in the state of negotiation. Instead of including such different ‘agreement events’, we choose to only have one dataset entry for each agreement. Similarly, we do not include agreements signed before 1960.
Nevertheless, it is important to note that there are no agreed definitions, even on the most common categories of natural persons, so agreements use a range of different definitions. Moreover, national immigration rules often allow entry under different schemes and definitions than those defined in PTAs – for example, while the PTA defines an ICT as a worker who retains his or her work contract with the sending company, in practice the same person can often also obtain a local work contract and enter the partner country following the national labor immigration rules.
The sub-variables for the mobility categories that measure the modalities of the mobility provisions are not included individually but merged into dummy variables for whether a modality is present or absent from the agreement (see also variable list in the Appendix).
We do not classify countries based on their status as migrant-receiving or –sending countries because no accurate data is available for the whole period of our analysis.
There are 58 agreements between OECD countries, 317 mixed agreements and 422 agreements between non-OECD countries. The World Bank classification includes a slightly larger number of countries into the group of developed countries, yielding 78 agreements between developed countries, 238 mixed agreements and 347 agreements between developing countries.
The analysis uses a variable that captures whether trade in services is covered by a PTA and an aggregated depth index from the DESTA database (Dür et al., 2014).
Labor standards are measured by a variable (nti_labour) from the DESTA database (Dür et al., 2014). The variable captures whether a PTA covers this area.
