WORKING PAPER Promoting Labour Standards through the European Neighbourhood Policy

This study examines the effectiveness of the EU’s efforts to improve labour standards among its neighbouring countries through its market integration-based regulatory governance approach. We consider the European Neighbourhood Policy as a critical case for assessing the prospects and limitations of the EU’s broader agenda to promote non-trade related policy objectives through its trade policy. Using a combination of regression and comparative analysis (in Moldova and Morocco), we show that upgrading in conditionality, assistance and dialogue can contribute to improvements in labour standards, albeit with significant shortcomings. In the absence of greater EU commitment towards strengthening domestic institutions, engaging labour organizations and adopting a problem-solving approach, EU-led improvements on the books will not result in changes on the ground. Even formal changes will be restricted to areas that are compatible with other major EU policy objectives, such as market integration, political and human rights. In this context, improvements at the implementation level have been facilitated to some extent by supplementary initiatives by labour organizations, albeit within the constraints of weak state capacity.


Introduction*
Over the past years, both the European Union (EU) and the United States (US) have increasingly sought to promote labor standards through unilateral, bilateral and/or regional trade integration, whereby market access power is used as a leverage to promote labor standards in third countries.Yet, studies exploring their divergent approaches and implications in various national settings often lead to contradictory results regarding effectiveness.While some find a significant positive association between the inclusion of labour provisions in trade policy, on one hand, and labour standards, on the other (Hafner-Burton 2005, Greenhill et al 2009, Kim 2012, Postnikov & Bastiaens 2014), others highlight major limitations in the design and implementation of such efforts, that ultimately undermine their effectiveness (Garcia & Masselot 2015, Giumelli & von Rozendaal 2016, Marx et al 2016).There is also disagreement regarding the most relevant tools and mechanisms through which labour standards in third countries can be improved through transnational regulatory governance.While some associate positive outcomes with the use of sanctions and incentives that influence the interests of relevant actors (Hafner-Burton 2005, Kim 2012), others emphasize the potential of more cooperative approaches like dialogue and assistance that shape preferences and facilitate policy learning (Oehri 2015, Postnikov & Bastiaens 2014).We also have relatively limited understanding of how domestic factors mediate the effect of labour provisions in trade agreements (Postnikov & Bastiaens 2014).There is thus need for more systematic research on the implementation and effectiveness of transnational efforts to promote labour standards through trade policy (Campling et al 2016, Harrison et al, 2019).
In this article, we focus on the EU's efforts to govern labour standards in developing countries through its trade integration-based governance approach.We thus ask: can the EU promote labour standards through its trade policy and if yes, through which processes and under what conditions?We address these questions through an empirical analysis of the implementation and effectiveness of the European Neighborhood Policy (ENP).Launched in 2004 to promote economic prosperity, social development and political stability among neighbouring countries (NCs), labour standards were considered instrumental for several of these ambitions (EC 2004c) and have been pursued by the EU ever since (Gstöhl, 2009), albeit with some changes over time (EC 2012c(EC , 2015c)).While the ENP is based on the legal and institutional foundations of bilateral trade agreements, it has been accompanied by a set of additional implementation strategies and tools that -borrowed from the EU's successful enlargement experience (Kelley 2006), have been associated with superior outcomes (Bruszt & McDermott 2012, 2016).These include: an upgrading in economic conditionality to incentivize harmonization with international and/or EU standards, deepening of political dialogue to set medium-term benchmarks and monitor progress, and expansion of assistance to facilitate reforms.Notwithstanding significant shortcomings (Börzel & Hüllen 2015, Casier 2011, Gstöhl 2009), the ENP still constitutes one of the EU's most ambitious foreign policies whereby a series of non-trade related policy objectives (NTPOs) are promoted through market integration (Schumacher 2017).We therefore consider the ENP to constitute a "critical case" to study the EU's efforts to promote labour standards through market integration, that can shed light on the prospects and limitations of the EU's trade-based regulatory governance approach.Its implementation across sixteen national contexts 1 , furthermore, provides an ideal empirical setting to understand the mediating role of domestic factors such as the strength of state capacity and civil society.
Our study advances the debate on the prospects and limitations of the EU's transnational labour governance approach through a set of important findings.These are based on a combination of regression analysis of the influence of the ENP on labour standards among NCs in general, and comparative study of its implications in the specific cases of Moldova and Morocco.Firstly, we show that notwithstanding a lack of prioritization of labour issues, the ENP did bring about a gradual improvement in labour standards of moderate strength among NCs.Secondly, we demonstrate that progress took place primarily at the formal level and in areas that were more compatible with political/human rights and market integration agendas of the EU.Third, we attribute this outcome to an implementation approach that relied primarily on the use of political dialogue to encourage reforms and address lack thereof, that favoured changes at the more "reportable" formal level.In the absence of adequate assistance and engagement of labour organizations to build domestic capacity to conform with EU expectations, however, this approach failed to ensure that workers benefit from greater market integration with the EU and the reforms that accompany it.Fourth, we find that improvements at the implementation level have been facilitated to some extent by the supplementary interventions of domestic and international labour organizations, but continued to be constrained by the persistence of weak state institutions.While the extent of ENP-led progress has been contingent on existing levels of state capacity among NCs -with superior outcomes in contexts with better administrative institutions; we do not find strong complementarity between ENP labour governance and strength of civil society (including labour organizations), in improvements of labour standards in NCs.The latter contributed to improvements in labour standards largely independently, rather than through direct interactions with EU efforts.
The structure of the paper is as follows.Upon a brief overview of the literature on the promotion of labour standards through trade agreements -the key institutional framework to set the rules of greater trade integration, we present an analytical framework that considers the potential of the ENP to improve domestic capacity to conform with EU expectations.The third and fourth parts consist of the regression and comparative case study analysis respectively.The conclusion brings together key findings from the two main empirical parts and interprets them in the context of ongoing debates regarding the EU's ability to promote NTPOs through its trade policies.

Promoting Labour Standards though Trade Agreements
The "normalization" of labor provisions in US and EU trade agreements (Van der Putte & Orbie 2016) has been accompanied by an expanding literature that analyzes the scope, implementation and impact of such arrangements, often in a comparative perspective (Grynberg & Qalo 2006, Oehri 2015, Postnikov & Bastiaens 2014).These studies generate some essential insights regarding the governance of labour standards through trade policy, but also have some important limitations.Studies reveal, for example, that while US trade policy emphasizes the enforcement of national labour regulation, including minimum wage, the EU has promoted primarily harmonization with core International Labour Organization (ILO) standards (Grynberg & Qalo 2006, Piore & Schrank 2016).The social ambitions of the EU, however, have intensified over time, often extending to harmonization with EU regulations in employment policy, and/or to the ILO's more encompassing decent work agenda (Young & Peterson 2006, Orbie & Barbarinde 2008).Notwithstanding such variation in scope, labour standards are often conceptualized and assessed as a homogeneous set in studies of implementation and effectiveness (Kim 2012, Postnikov & Bastiaens 2014, Lechner 2016), with only a few exceptions that focus on collective labour rights in particular and distinguish between reforms at the formal and practical levels (Greenhill et al 2009, Marx et al 2016).
Most of the literature examines the implementation strategies pursued to generate improvements.These are often characterized as "hard" and "soft" (Hafner-Burton 2005) or "hierarchical" and "network" approaches (Oehri 2015) to distinguish between coercive or incentivizing, versus cooperative approaches.The US is generally considered an adherent to the "hard" or "hierarchical" model that is more legally binding and can use sanctions to address non-conformance (Grynberg and Qalo 2006, Tsogas 2000, Oehri 2015).The EU, on other hand, is associated with the "soft" or "network" approach that relies on a combination of material incentives and cooperative tools, such as dialogue and assistance, to promote improvements (Bastiens & Postnikov 2014, Van der Putte & Orbie 2016).More recently, Oehri (2014) finds that while the EU and US labour governance approaches might differ in design, in practice, they both rely predominantly on the use of the "network" approach -assistance and dialogue -to address concerns.This distinction between "hard" vs "soft" approaches, however, is overly simplistic and sometimes confusing.Labour provisions in EU trade agreements, for example, have become more binding in character due in part to new monitoring and arbitration measures (Van der Putte and Orbie 2016).The "soft" attribute has moreover been used in association with different mechanisms of influence, including incentivization (Van der Putte and Orbie 2016), persuasion (Hafner-Burton 2005) and policy learning (Postnikov & Bastiaens 2014).
Less is known about the relative effectiveness of the divergent implementation approaches.Studies often regard 'hard' measures -that include arrangements to reward and/or punish member' behaviors -to render superior results, over those that do not (Hafner-Burton 2005, et al 2019, Greenhill et al 2009).More recent studies suggest that the difference is merely in the timing of their effectiveness as the US triggers reforms before negotiations start (Kim 2012), while the EU facilitates improvements only later on, as the implementation of its consultative approach enables changes in domestic preferences among state officials and policy learning by civil society organizations (CSOs) (Postnikov & Bastiaens 2014).The direct engagement of CSOs in the implementation of trade agreements, however, is relatively recent (Van der Putte & Orbie 2016) and empirical evidence regarding their effectiveness is still largely lacking (Campling et al 2016).Qualitative studies of such arrangements reveal significant problems in implementation due to the weaknesses of labour organizations and prioritization of other concerns, that undermine effectiveness (Marx et al 2016, Orbie & Van der Putte 2016, Harrison et al 2019).These studies thus often lead to somewhat contradictory results and have not considered the role of domestic factors in a more systematic manner.
The next section discusses the ways in which examining the implementation and effectiveness of the ENP to promote labour standards can address some of the shortcomings identified in the literature.

ENP as Domestic Capacity Oriented Regulatory Governance?
Rather than a simplistic distinction of alternative approaches, the emphasis in our study is on the potential of trade agreements and their accompanying measures to improve the capacity of domestic state and non-state actors to respond to external requirements in the context of greater market integration.This is consistent with other studies from the broader transnational governance literature that understand the persistence of poor conformity in developing countries not merely as a function of the material interests and political preferences of their governments, but also in terms of the ability of the state, companies and labour organizations to improve standards when economic competition intensifies (Bruszt & McDermott 2012, 2016, Locke 2013, Schrank 2013, Piore & Schrank 2016).Improvements in the domestic capacity of trade partners can thus be achieved through various types of measures; and the use of dialogue and assistance is no guarantee for improved ability of domestic actors to respond to market incentives.
To better understand the conditions under which trade-based governance approaches can bring about improvements in labour standards, we rely on a framework developed by Bruszt and McDermott (2016) to assess and compare the potential of regional integration regimes to bring about regulatory convergence in general.According to this framework, positive outcomes are more likely when: 1) the scope of regulation extends beyond legal reforms to changes at the institutional level; 2) investments are made in the institutional capacity of various stakeholders to implement reforms and to benefit from it; 3) a variety of public and private actors are engaged in the implementation process, that can mobilize resources and expertise to address problems linked to the management of the developmental externalities of reforms, and 4) problem-solving approach is pursued whereby the sources of non-conformity are identified and addressed.Transnational governance approaches Timea Pál, Franklin Maduko and László Bruszt that have these attributes, are more likely to lead to greater conformity with EU expectations than those that do not.An example in case is the EU's eastern enlargement approach that enabled broader institutional developments and greater regulatory convergence across several policy areas, than other transnational regulatory regimes (Bruszt andMcDermott 2012, 2016).
Given the significant influence of the eastern enlargement on the ENP (Kelley 2006), it is reasonable to expect that the tools and strategies that accompanied the ENP since its launch in 2004 increased the potential of the EU to pursue a domestic-capacity oriented approach with the NCs, relative to the Association and Cooperation agreements in the past.In what follows, we discuss some of the key changes that resulted from the ENP in terms of the key features underlined by Bruszt and McDermott.Firstly, the ENP resulted in a broadening and deepening of the scope of EU governance with respect to the NCs.The ENP entailed the possibility of deeper market integration -in the form of Deep and Comprehensive Free Trade Agreements (DCFTA) -based on gradual adaptation of national regulations to EU requirements in both trade and non-trade areas, including reforms at the institutional level (EC 2004c).To set more precise benchmarks, ENP Action Plans were negotiated between the EU and individual NCs, that established reform agendas for the next three to four years.In the realm of labour provisions, these reinforced commitments towards greater approximation with ILO standards and, in most cases, gradual harmonization with EU regulations.The ENP Action Plans extended to institutional reforms as well, through the inclusion of commitments regarding strengthening social dialogue and state regulatory institutions to facilitate implementation of regulatory reforms.
Secondly, the ENP also resulted in a significant expansion of EU assistance to improve the ability of NCs to achieve these reforms.The amount of financial assistance, delivered primarily through the European Neighborhood and Partnership Instrument (ENPI) from 2007 until 2013 and the European Neighbourhood Instrument (ENI) from 2014 until 2020 was significantly greater than the funding provided through the regional MEDA and TACIS programs in the past 2 .Additionally, NCs were offered access to the EU's SIGMA (Support for Improvement in Governance and Management), TAIEX (Technical Assistance and Information Exchange Instrument) and Twinning programs that had been successfully used to improve institutional capacity among candidate countries.
In the ENP framework, bilateral dialogue between the EU and NCs intensified, to guide progress and address potential problems.The primary institutional platform remained the Cooperation/ Association Councils and Committees established by the bilateral agreements, but were now informed by regular monitoring of progress to meet Action Plan commitments3.Committee meetings became more frequent and new sub-committees -dedicated to social and/or human rights issueswere established.The engagement of domestic non-state actors was initially to take place through social dialogue at the national level.More recently, civil society forums (CSFs) were established at the regional and bilateral levels to facilitate a more direct involvement of CSOs in the implementation process.These include the CSFs of the Union for the Mediterranean and Eastern Partnership -the regional initiatives of the ENP established in 2008 and 2009 to respond to demands for greater regional differentiation; as well as those that accompanied the DCFTA with Georgia, Moldova and the Ukraine.
Given the upgrading in its potential to improve domestic institutional capacity among NCs, our main expectation (H1) is that the ENP will facilitate additional improvements in labour standards, beyond any positive effect that trade agreements with labour provisions might have had.
The EU, however, pursues multiple and sometimes conflicting goals (Börzel & van Hüllen 2014) that might influence the extent to which its transformative potential is used to pursue improvements in labour standards relative to other competing goals.
The ENP's social integration objectives were very explicit during the initial stages of the ENP, as part of initial ambitions to promote the European Social Model beyond the enlargement (Gstöhl 2009).The emphasis over time, however, has shifted towards the promotion of social inclusiveness and sustainable development (EC 2012c(EC , 2015c)).The reforms of the ENP in 2011 and 2015 most likely reinforced this tendency of declining intensity of the ENP's social ambitions, as they influenced the EU policy priorities in relation to its southern and eastern NCs.In 2011, responding to developments in Arab countries and weaknesses in promoting political reforms in the past, the EU strengthened its focus on promoting deep and sustainable democracy, and introduced the "more for more" principle, according to which the EU forms stronger partnership with those NCs that embrace more democratic reforms (EC 2012c).In 2015, a series of security crisis that affected some eastern and southern countries4 then facilitated a shift towards a more differentiated, realist and pragmatic approach, whereby security concerns are prioritized (EC 2015c).The use of the EU's domestic capacity building approach to promote labour standards is therefore expected to have diminished over time, except for the countries that signed a DCFTA with the EU in 2014.
Variation in the use of the EU's transformative potential, however, might extend beyond differentiation between social and other objectives in general, to the level of individual labour standards.Inclusion of labour provisions in trade policies have often been treated as a homogenous category, but labour standards span across a whole range of issues that often have different economic, political and social dimensions.A common distinction, for example, is between fundamental rights, such as freedom of association and collective bargaining, elimination of forced and child labour, non-discrimination; and working condition standards regarding minimum wage, working hours, occupational health and safety (OHS) (Barrientos et al 2011).Child and forced labour, for example, constitute important human right violations that were more strictly enforced by the EU in the past to protect legitimacy of trade relations (Tsogas 2000).Freedom of association and collective bargaining is more closely connected to political rights, but could conflict with the interest of export-oriented companies to maintain a relatively unorganized and cheap labour force in the NCs.Working conditions standards can also conflict with trade interests, especially as several NCs became important locations for the outsourcing and offshoring activities of major multinational enterprises (MNEs) that export to the EU.
The second assumption of our study (H2) is thus that: Improvements in labour standards will vary over time and across specific area, in function of other EU policy priorities.
Finally, the implementation and effectiveness of the EU's transnational labour governance approach is expected to differ across developing countries based on the strength of state institutions and labour organizations.These factors are generally recognized as essential determinants of national labour regulation (Berliner et al 2015) and have been shown to play an essential role in developing countries' ability to respond to external requirements as well (Schrank 2013, Piore & Schrank 2016, Locke et al 2013).Weak state capacity is expected not only to impede the enforcement of higher standards and fail to ensure that potential benefits are reached and distributed, but to also constrain the ability of the EU to strengthen domestic institutions.The state capacity building efforts of the ENP are thus expected to generate more improvements in contexts where the national labour administration already has some reasonable organizational capacity to benefit from policy guidance, training and technical cooperation, and/or greater engagement with social partners.Similarly, consistent with recent findings regarding the important role played by labour organizations in the implementation of private governance efforts (Amengual & Chirot 2016), superior outcomes are expected in countries where these are more organized and representative, as they influence the outcomes of the EU to promote engagements and problem-solving strategies.
The third expectation (H3) is therefore that the effectiveness of the ENP will be greater in contexts with stronger state institutions (H3a) and/or more organized labour groups (H3b).
In the following two main sections, we use a combination of regression and comparative country case studies to test the empirical validity of these hypotheses.

Assessing the Effect of the ENP on Labour Standards
We first study the prospects of the EU's trade policy to promote labour standards in third countries by assessing the association between ENP and labour standards, beyond any positive influence that the inclusion of labour provisions in trade agreements might have had.The latter refers to all free and preferential trade agreements of the EU with developing countries that include labour provisions.But the ENP goes beyond this baseline, through the deployment of greater market integration benefits, provision of additional assistance and intensification of political dialogue.

Data and Empirical Model
We test our main hypothesis (H1) using an unbalanced panel data of 46 developing countries that had TAs with labour provisions in force with the EU, spanning from 1991 to 2014.These countries include the twelve NCs that became part of the ENP since 2004.We test hypothesis H1 using two empirical models.The first is a fixed effect regression model where we regressed an ordinal index of workers' rights (ranging from 0-2) on a dummy that captures whether countries have EU TAs in force with labour provisions and another dummy for the periods for which a subset of countries that are integrated in the ENP, while controlling for a set of political and economic variables that might influence labour standards.While linear regression offers ease of interpretability of results, it does not take in account the ceiling and floor restrictions on models that include ordinal variables (Winship & Mare, 1984) and may result to biased estimates.Motivated by this, our second empirical model is an ordered probit model, a generalization of probit models for the case where the dependent variable is ordinal.This model offers less biased estimates and ensures that model predicted outcomes lie within the ceiling and floor of the ordinal dependent variable.
The empirical specification for the fixed effect regression is of the form below: Where: y it is workers' rights in country i at time t, obtained from CIRI human rights dataset, that measures adherence to a set of internationally recognized labour standards including freedom of association and collective bargaining, prohibition on the use of forced and child labour, and acceptable working conditions with respect to minimum wage, hours of work, occupational safety and health etc (Cingranelli et al 2014).A score of 0 implies that workers' rights were very restricted; 1 indicates that workers' rights were mildly restricted; and 2 implies that workers' rights were fully protected in the year in question.
TA it is dichotomous independent variable that takes the value of 1 if country i in time t has EU TAs with labor provisions in force and 0 if otherwise.We identify all EU TAs with labor provisions from the dataset provided in Postnikov & Bastiaens (2014).This dataset was constructed from analyzing the text of each EU TA to ascertain which TAs contains labour standard provisions.Since this dataset was constructed for periods before 2010, we extend the dataset to 2014 by analyzing contents of more recent TAs and checking if previous TAs that were not in force as at 2010 are in force (See table A1 in the appendix).
ENP it is a dummy variable which takes the value 1 if country i is an ENP country in time t and 0 if otherwise.The coefficients of TA it and ENP it respectively are our main coefficient of interest5 .The former tells us whether labor provisions in EU trade agreements helped in improving workers' rights or not, while the latter looks at the influence of the ENP on the same outcome.
We include a set of control variables (Controls it ) that capture both economic and political conditions.Specifically, we include total trade in GDP ratio and GDP growth rate, obtained from the WorldBank database, that control for the influence of trade pressure/dependence and economic development on labour standards.The political control variables include the number of veto players, CSO freedom index, both taken from the Quality of Government (QoG) dataset, and state capacity index (Berliner et al 2015).Veto players is a dummy where 1 indicates more political constraint and thus less feasibility of policy change, while higher values of CSO freedom indicate stronger civil society.We also include a dummy variable that captures whether a country has a PTA with the US, to isolate the effect of the EU from other international governance frameworks.The ordered probit regression is of the same form as equation ( 1) above.The only difference is that we have one independent variable -EU TA and ENP -which lies between 0 and 2 instead of the dummy variables defined in (1) such that a value of 0 corresponds to "No TA with labour provisions", 1 denotes "EU TA with labour provisions in force" and 2 corresponds to "EU TA with labor provisions and ENP".Having our independent variable in this scale will enable the computation of the predicted probabilities of workers' rights for each level of our independent variable.
We estimate both the fixed-effects and ordered probit models with robust standard errors, and country and year fixed effects in all specifications.The year fixed effect controls for any observable or unobservable variable common across all countries in our dataset in each year which may have led to improvements or deterioration in labour rights, such as the global economic and political climate (e.g.financial crises).The country fixed effects account for both observable and unobservable timeinvariant country idiosyncratic factors that might be correlated with labour rights and ENP, such as religion, colonial heritage etc.Similar to Kim (2012), we lag the economic variables (total trade in GDP ratio and GDP growth).We present the estimation results for the fixed-effect and ordered probit models in Table 1, and the predicted probability that workers' rights are fully respected (based on the ordered-probit model) in Figure 1.

Interpretation of Results
Overall, our results indicate that both EU TAs with labour provisions and ENP has a positive and statistically significant effect on workers' rights.The highest effect is observed for countries that have been included in the ENP since 2004.It is worth re-emphasizing that the ENP countries already had EU TA with labour provisions in force prior to the ENP, so the co-efficient of ENP () isolates the additional effects of the ENP from the effects of EU TAs with labour provisions.For the case where our main independent variable lies between 0 and 2, we find that the relationship between this independent variable and workers' rights is positive and statistically significant; however, as the dependent variable is an ordinal one, we interpret the results in terms of the predicted probabilities.Figure ( 1), which corresponds to the estimates in column (4) of Table 1, indicates that the predicted probability that a country fully protects workers' rights is 0.043 when no EU TA is enforced, 0.093 after the PTA is enforced, and 0.176 after ENP.These results suggest that the ENP has been effective in the advancement of workers' rights in these countries.The control variables have the expected signs in both models.The coefficients of CSO freedom are positive and statistically significant in both models, while that of state capacity are also positive in both models, but significant in only the ordered probit model.Both Veto and GDP growth are positive but insignificant.We find that trade to GDP ratio is negative and statistically significant, consistent with findings by Greenhill et al (2009) that expansion of trade exerts a downward pressure on collective labour rights and practices.Finally, US PTAs is negatively associated with workers' rights, consistent with Kim (2012) -that countries engage in ex ante due diligence and improve labour rights before entering negotiations or signing TAs with the US.
Lack of more disaggregated data on workers' rights hinders our ability to identify the level and dimensions of labour standards that were most affected by the EU's regulatory governance instruments.We address this limitation and shed more light on the mechanisms of causality through the qualitative analysis of labour rights in the domestic contexts of Morocco and Moldova in the next section.
To evaluate changes over time in the effectiveness of ENP to exert an influence on labour standards, we test whether the 2011 ENP reform resulted in any major changes in the ability of the EU to promote labour standards among NCs.While this reform of the ENP, with a shift towards the "more for more" principle, could have intensified the positive influence, the increasing prioritization of political reforms in the aftermath of the Arab Spring, might have also undermined the use of ENP tools to promote labour standards more specifically.We test the validity of these distinct possibilities in two ways.Initially, we created 3 dummy variables.The first dummy variable takes the value 1 if there is an EU TA with labour provisions and 0 if otherwise.The second dummy variable takes a value 1 for the ENP countries between 2004 to 2010 and zero if otherwise, and the third dummy variable takes the value 1 from 2011 to 2014 for the ENP countries and 0 if otherwise.We estimate a fixed-effect regression specification similar to equation (1) and report the results in Column (1) of Table A2 in the appendix.Then, we estimate the ordered probit model and a fixed effect regression where we redefine the scale of the independent variable to range from 0 to 3.Where 0 indicates absence of both EU TA with labour provisions and ENP, 1 denotes EU TA with labour provisions, 2 denotes ENP between 2004 to 2010, and 3 denotes ENP between 2011 to 2014.We present the results in column (3) -for ordered probit model-, and column (2)-fixed effect regression-of Table A2 in the appendix.Clearly, the results suggest that the ENP was most effective in promoting workers' Timea Pál, Franklin Maduko and László Bruszt rights between 2004-2010 and we observed no statistically significant improvements in workers' rights during the period 2011-2014 for the ENP countries.
The third hypothesis (H3) states that the effectiveness of the ENP will be greater in contexts with stronger state institutions (H3a) and/or more organized labour groups (H3b).We test this hypothesis by interacting ENP with state capacity and CSO freedom in a similar specification as equation ( 1), and present the estimates in Table 2 below.Across all specifications, the results show that ENP was more effective in improving workers' rights in countries with stronger state capacity as shown in column 1 and 3 of Table 2.This finding suggests that stronger state capacity complements ENP efforts at improving workers' right.We do not find any significant relationship between the interaction of ENP and CSO freedom, which suggests that the effectiveness of the ENP is not further enhanced by the presence of stronger civil society organizations.Rather, civil societies free from external influence is positively associated with workers' rights, irrespective of whether the country is part of the ENP or not.

Robustness Checks
To see if our results are sensitive to slight modifications of the empirical specification and datasets, we conducted a series of robustness checks 6 .First, we estimate the relationship between EU TA and ENP on changes in workers' protections controlling for the initial level of labour standards, and the results were consistent with our main results.Second, we use alternative measures of CSO freedom, from Mosley and Uno (2007), and state capacity, from CIRI (2014) and our results remains unchanged.Finally, we ran the analysis with additional control variables, in the spirit of Postnikov and Bastiaens (2014) -log GDP and log GDP per capita -and the results were also not affected by these changes.

Some Quantitative Insights into Mechanisms of Influence
Finally, we also attempt to shed some light on the mechanisms underpinning the positive influence of the ENP on labour standards among NCs.While this is addressed primarily by the qualitative case analysis presented in the next section that relies on interviews conducted with relevant stakeholders in two of the NCs (Moldova and Morocco); here, we attempt to assess the relevance of ENP financial assistance to motivate improvements in labour standards among NCs.To do so, we estimated a linear probability model for only the NCs between 2004-2014 7 , where the dependent variable is a dummy that takes a value 1 if there is an improvement in workers' rights 8 , and 0 if otherwise.The results -presented in Table 3 -suggest a positive association between financial donations and improvement in workers' rights in the NCs, although the coefficient of log of financial assistance in columns (1) of Table 3 are significant only at 10 percent, and becomes statistically insignificant in columns ( 2) and ( 3), where we control for some economic and political variables similar to our mail specification.Overall, this finding suggests a weak association between these incentives and labour standard improvements.

ENP-Led Improvements in Moldova and Morocco
The comparative analysis of the implementation and implications of ENP labour governance in Moldova and Morocco allows us to shed more light on the main mechanisms driving (or hindering) improvements (H1), the scope of progress achieved (H2) and the mediating role of domestic context (H3).

Comparative Method and Data
Moldova and Morocco constitute ideal cases for such a comparative analysis.Notwithstanding significant differences across several economic, political and social dimensions, the two countries posed labour concerns at the outset of the ENP in 2004.Both countries ratified most core ILO conventions 9 , except for ILO convention on freedom of association and collective bargaining by Morocco, but their national labour codes had remaining caveats, including in the areas of freedom of association, elimination of child labour, non-discrimination and Occupational Health and Safety (OHS) (EC 2004a,b;USDS 2004a,b).Biggest concerns with respect to labor standards, however, pertained to the implementation of their national regulations, with government interference in union activity, child labour, informal work, gender inequality, high accident rates and payments below subsistence level prevalent in both countries.
The ENP governance approach deployed in the two countries to address these concerns was relatively similar between 2004 and 2014, relying on a combination of incentives, assistance and dialogue to facilitate harmonization with international and EU labour standards.Differences in their national contexts -particularly in state capacity and strength of labour organizations, provide an ideal opportunity to compare how these measures have facilitated (or hindered) improvements in labour standards across NCs, and in what way did domestic factors intervene in this process.While the main representative of workers' interests in Moldova over the past decades has been the unified National Trade Union Confederation of Moldova (CNSM) that included as members approximately half of the formal workforce in the mid 2000s; in Morocco, workers were represented by five trade unions10 with different political affiliations and a total density of mere 6 percent (USDS 2004a,b).State capacity was weak in both countries, but particularly low in Morocco (Cingranelli et at al 2014).Since 2016, the implementation of the ENP in the two countries diverged considerably, as the DCFTA (signed in 2014) came into force with Moldova and political cooperation with Morocco was interrupted by a trade conflict related to the Western Sahara.This provides an additional opportunity to assess the relevance of the ENP for the evolution of labour standards among NCs (H1&2) by comparing the extent of labour reforms in the two national settings over time.In the analysis, we rely on data from both secondary and primary sources.The former includes the initial ENP country assessments and subsequent progress reports prepared by the European Commission (from 2008 until 2019 for Moldova and from 2007 until 2015 for Morocco), complemented by labour reports by major international and national organizations such as the ILO, International Trade Union Confederation (ITUC) and the US State Department.The primary data refers to approximately 37 semi-structured interviews we conducted over the course of 2019 in Brussels (EU 1-6), Chisinau (MD 1-17), Rabat and Casablanca (MA 1-14) where we met with representatives of the European Commission and Council of the EU (in Brussels); state institutions -ministries of labour and economy, state labour inspectorates; employers' associations and trade unions, NGOs and domestic businesses, ILO country offices and EU country (in Moldova and Morocco).

ENP-Led Improvements
The implementation of the ENP facilitated the emergence of a series of incremental changes in labour standards of moderate strength in both countries between 2004 and 2015, but has then resulted in divergent trajectories since 2015, as summarized in Table 4. Overall, ENP-led improvements have been more prominent at the formal level and in the realm of fundamental labour rights, except for freedom of association and collective bargaining; with changes in working condition regulation limited to OHS.

Laws and Regulations
• Ratified all (Moldova) and most (Morocco) relevant international conventions There have also been some improvements at the implementation level.A series of governmental policies and programs were put in place to promote gender equality and elimination of child labour, albeit often without sufficient state funding (EC 2006(EC /2007(EC -2015a,b),b).The ENP has also contributed to an increase in the institutional capacity of the State Labour Inspectorates (SLI) of both Moldova and Morocco, with positive on the enforcement of national labour regulation, at least in the formal sector (USDS 2017a,b; MD12; MA10).The SLIs, for example, experienced improvements in the management of human resources and career development, organization of inspection services and visits, registry of work accidents, and started to place greater emphasis on prevention of noncompliances in their activities (MD5&6, MA6&7).A part of these improvements in implementation, while motivated by the ENP, have been achieved with the additional support of labour organizations and important deficiencies in implementation have nonetheless persisted, as we will explain later.
The relevance of the ENP for labour standard among NCs (H1) is also reflected by the different trajectory of reforms in the two countries over the past years.Legal and regulatory changes have been very substantial in the case of Moldova since it signed the DCFTA with the EU in 2014, while there were no major reforms in Morocco.With the DCFTA including direct references to EU Directives to harmonize with (Annex IV) and the Association Agenda setting specific timeline for their transposition, by 2019, Moldova adopted 7 out of 8 labor law acts, 5 out of 6 non-discrimination directives and 19 out of 25 OSH acts included in the DCFTA.These legislative changes, however, were accompanied by a set of administrative reforms to reduce regulatory burdens on companies, that led to a significant decline in the capacity of the SLI to ensure adequate implementation (EC 2019, MD 5&6), that will also be discussed in greater detail in the last sub-section.

Scope and Mechanism of Influence
These ENP-led improvements -more prominent in the realm of (most) fundamental labour rights and at the formal level -were the result of an implementation approach that has been selective in scope and did not place sufficient emphasis on improving domestic institutional capacity to improve workers' protections and their working conditions.
ENP efforts to promote labour standards have varied across individual issue areas and over time, consistent with initial expectations that the ENP regulatory governance approach might vary in accordance with other major policy objectives pursued by the EU (H2).The predominant focus of the ENP has been on the promotion of fundamental labour rights and of social dialogue that overlapped with the EU's ambitions to promote human and political rights, as originally expected.This can be best illustrated through the frequency of reporting on specific labour standards in the ENP progress reports of all NCs, shown in Table 5.As we can see in this table, the extent of progress achieved by NCs was evaluated most consistently in the areas of non-discrimination and social dialogue, followed by freedom of association and child labour.Working condition standards such as low payments and long overtime, that have facilitated the emergence of some NCs as relatively cheaper manufacturing sites for the offshoring and outsourcing activities of MNEs, however, have been largely overlooked.The only exception in this regard has been OHS that, given its overlap with product standards, acquired greater importance over time, particularly as bilateral relations advanced with some countries and the possibility of negotiation of a DCFTA increased.The table also confirms a declining trend in reporting on labour standards after the 2011 reforms, as the ENP shifted its focus towards promoting political reforms, that is consistent with the diminished positive association between ENP and worker protection identified also in the regression analysis.This tendency intensified further as the EU came to prioritize security concerns with most of its NCs (EU 6).Another indication of a selective ENP governance approach is that economic incentives were used only for policy areas prioritized by the EU, where some leverage was needed to address lack of progress.We did not find a strong direct link between the use of economic benefits and labour reforms in Moldova and Morocco.Both countries benefitted from reductions in trade tariffs and became eligible for DCFTA negotiations in the early 2010s, but their efforts to harmonize with external labour standards played no role in this (EU3,5; MA1).The promise of greater trade privileges might have played a role initially in encouraging reforms.Moldova, for example, became part of the GSP+ due to its ratification of international and human rights conventions.While the EC has also relied on financial assistance and mobility arrangements as alternative incentives and sanctions 11 , these have been used primarily to promote reforms in the areas of rule of law and market regulation, that were important for improving business and investment climate in their national contexts (EU3, 4&5).
According to some EU officials, NCs have realized relatively quickly that they will retain eligibility for financial assistance even if they don't make sufficient progress in areas of lower-priority for the EU (EU3).These findings explain the weak association between economic assistance and labour standards that was found in the regression analysis.
The EU has influenced labour reforms in Moldova and Morocco primarily through inter-state dialogue to guide reforms and address problems of non-conformance, without any significant engagement of labour organizations.This political dialogue has taken place primarily in the framework of the Association/Cooperation Councils and Committees where reform commitments were agreed upon, progress was monitored, and non-conformances discussed by top government representatives and state bureaucrats.In the framework of these engagements at the inter-governmental level, EU officials used a combination of political pressure, persuasion and policy guidance to address lack of progress.Although labour standards were rarely, if ever, flagged as a priority in the annual Council meetings, EU representatives would reiterate the importance of respecting core rights and achieving progress in harmonization with EU legislation in all areas, to ensure the advancement of their bilateral relations (EU4, MA6).More concrete recommendations and requests to address concerns raised by regular reports were formulated in Committee meetings.EU representatives, for example, requested the hiring of 100 additional state labour inspectors by the Moroccan government in 2010 to improve regulatory enforcement (MA6).These deliberations also became an important platform to familiarize NC government officials with EU standards, provide expertise on their transposition into national legislation and offer advice on policy measures to implement them (EU4, MA6, MD10&11).In the case of Morocco, for example, Association Committee meetings provided guidance for policy measures to promote non-discrimination.(MA6).Government officials from Moldova also praised these interactions for their valuable contributions to policy learning and development across all areas covered by the DCFTA (MD10&11).These meetings were also useful at increasing sensibility towards issues such as gender equality as both human and labour rights, and at raising awareness for the importance of OSH for both economic and social reasons (MA6).This problem-solving approach through inter-state deliberations, however, had two important 11 The amount of financial assistance offered to Moldova was reduced after concerns related to investigations of a major bank fraud.
limitations.Firstly, due in part to the primary focus of the assessment reports on major legislative and regulatory changes to evaluate progress, this has encouraged a narrow focus on addressing problems at the de jure level.This is consistent with other findings in the literature that transnational labour governance often limits progress to the more "reportable" reforms to demonstrate conformance, often in the absence or even deterioration of greater political and economic efforts to ensure implementation (Greenhill et al 2009, Peksen andBlanton 2017).Secondly, when NC governments remained reluctant, as for example, in the case of labour union protections in the case of Morocco, this implementation approach has remained largely ineffective to bring about change.The Moroccan government has, to this day, continued to refrain from ratifying ILO convention 87.
ENP assistance to build institutional capacity has been largely inadequate and insufficient to facilitate implementation of reforms and ensure that workers benefit from greater market integration.The assistance provided through the EN(P)I focused predominantly on improving state administrative capacity, with approximately 8 and 15 percent of the ENPI budget attributed to this end for Moldova and Morocco, respectively, between 2007and 2013(EC 2014).This has also benefitted to some extent the state labour administration from both Moldova and Morocco.The Ministry of Health, Labour and Social Protection (MHLSP) and State Labour Inspectorate (SLI) of Moldova both had civil servants that participated in Twinning programs on repeated occasions, primarily with the involvement of their counterparts from Romania and other Eastern European member states (MD5&6).State labour inspectors from Morocco also benefited from training with civil servants from France and Belgium (MA6&7).These trainings resulted, for example, in a growing emphasis on prevention in labour inspection services and greater attention conferred to risk assessments in organization of visits (MD5&6, MA6&7).In the OSH area, the European Agency for Safety and Health at Work (EU-OSHA) has cooperated with the SLI from Moldova and the National Institute of Living Conditions at Work (NILC) from Morocco over the past years.
ENP financial assistance was also considered too low to facilitate such ambitious reforms as the ones NCs committed to in the ENP Action Plans, and improving labour regulatory institutions has not been prioritized in the distribution process (EU3-6; MD 10, 11&12).The scarcity of assistance became especially evident as Moldova embarked on a more ambitious path of regulatory convergence since the DCFTA.Technical support to accompany the deepening of trade relations included, for example, an EU High Level Adviser in the field of Employment and Social Policies deployed to Chisinau to assist with regulatory harmonization in the employment and social policy area.At the Ministry of Labour (now MHSLP), however, there were too few legal experts working and the transposition of EU Directives often had to be done in a hasty manner to meet tight deadlines (MD 10&11).EU assistance to facilitate implementation on ground has been even more modest and failed to reverse a major backsliding in state enforcement capacity, as a set of administrative reforms adopted to reduce regulatory burdens on companies resulted in the loss of the SLI's ability to conduct announced visits and monitor OSH (MD5&6).This calls in question claims that more legally binding rules are generally more beneficial for improving labour standards (eg.Kim 2005).

Relevance of Domestic Factors
ENP efforts to engage domestic non-state actors have thus far been too limited to include labour organizations to address the persistence of low standards.The primary emphasis in the past was to encourage the involvement of domestic stakeholders primarily through social dialogue across NCs.The ENP has indeed contributed to some strengthening of tripartite social dialogue at the national levels in both countries, but the ability of labour unions to influence key policies remained limited (EC 2006(EC /2007(EC -2015a,b),b).Greater economic integration intensified their governments' concerns with economic competitiveness, often at the detriment of worker protection, particularly as union density has been declining and the influence of export-oriented businesses has grown (MA12&13, MD9&11, USDS 2015a,b).In Morocco, trade unions would often withdraw from social dialogue to protest the lack of consideration of their concerns; in Moldova, businesses would often use alternative and sometimes less transparent channels to influence government decisions12 (USDS 2016, MA 12, ITUC 2018).Union representatives were generally not aware of the existence of CSF forums and, with a long tradition to mobilize at the national level, they did not recognize these regional and transnational forums as an opportunity to leverage their influence (MD17, MA11).This is consistent with findings by Harrison et al (2019) that trade unions from Moldova have not yet been able to acquire a strong voice and participate in these forums (p.8).
In the absence of greater and more direct engagement of labour organizations, their influence on ENP labour governance has been limited to indirectly supplementing the weak institution building efforts of the EU to improve implementation of reforms.Labour unions from Moldova have indeed played a more active role in this regard than in Morocco.In Moldova, CNSM established its own labour inspectorate in 2012 to monitor enforcement of labour laws regarding freedom of association, labour relations and OSH (USDS, 2015a).While they lacked the authority to impose sanctions or coerce non-compliant companies to change their behaviours, they often engaged in joint problem-solving to address areas of concerns, and invested in measuring equipment to increase their credibility and leverage in the OSH area (MD10&11).One of the Moroccan trade unions, also launched a national observatory of employee's rights in 2010, but its subsequent activities have remained very modest and is no longer in existence.
In addition to the domestic unions, the ILO has also played an important role in reinforcing ENP labour governance efforts by mobilizing its expertise and relations with domestic actors to improve implementation.The ILO, for example, provided training for the SLI as well as trade unions and industry associations, and played a particularly important role in combatting child labour and building OHS expertise among domestic actors (MA11, MD8&9).Such interventions were at times directly connected to NC governments' efforts to meet ENP commitments, as for example, the Decent Work programme in Moldova that sought to facilitate governmental efforts to meet ENP Action Plan commitments (MD5&6).Most times, however, the connection has been limited to the overlap in the goals pursued by the ILO and the EU.In parallel with these efforts, however, the rising influence of large multinational companies in Moldova and of export processing zones in Morocco in labour intensive-manufacturing sectors often sought to reduce labour regulation in general, and undermined improvements in union rights and collective bargaining in particular (MD12-14;MA10&11, ITUC 2015, 2018).In addition to such deregulatory pressures and anti-union resistance in the context of the expansion of market integration with the EU that sometimes reinforced the low-cost based economic competitiveness strategy of national governments and export-oriented businesses (Maduko et al 2021), the persistence of weak state capacity also constrained more significant improvements at the implementation level.Some EU support notwithstanding, the material and human resources at the disposal of state administration responsible for protecting workers have remained very low in both countries, especially as their economies became increasingly internationalized and regulations grew more complex (MD11&12, 5&6).Labor inspectors from both countries, for example, continued to lack adequate administrative support and access to such essential materials as vehicles of transportation and computers.Due primarily to budgetary constraints, the number of labour inspectors has in fact declined over the past years in both countries, from 121 inspectors in 2002 to 109 in 2015 in Moldova and from 496 inspectors in 2004 to 398 in 2018 in Morocco, with a corresponding decline in the ratio of inspectors per workers (MD11&12, MA7).While enforcement of legislation in the formal sector improved, the state has not been able to formalize a significant part of the workforce that continued to operate in the informal sector, where working conditions are generally much lower and workers continue to lack legal protections.The share of informal employment in Moldova remained very high  The limitations imposed by these structural constraints were further aggravated in Moldova by the recent decline in the enforcement capacity of the Ministry of Labour and the SLI.Regulatory reforms adopted in 2017, based on the law on State Control of Business adopted at the recommendation of the World Bank to improve economic competitiveness in the context of greater integration with the world economy, resulted in some significant changes regarding the authority and organization of the SLI (MD 5,6,11&12).These included a reduced authority of the SLI to conduct unannounced visitsas they required a greater emphasis on review of documentation and limited impromptu inspections to cases where suspicious discoveries are made.But they also brought about a shift in authority over OSH to ten regional agencies responsible for conducting consolidated audits, that relied on civil servants without adequate trainings.The Ministry of Labour also lost its entire staff from its Department on Labour Relations and social Protections (EC, 2017a).Monitoring and enforcement of OSH regulation declined considerably in the country over the past years (MD 6&17).
In Morocco, the labour inspectorate system has some of the features of the Franco-Iberian labour enforcement model that was associated with positive developmental incomes in other developing economies (Schrank 2013, Piore andSchrank 2016), as labour inspectors are granted greater discretion to remediate problems of non-compliance (MD7).In the absence of greater emphasis by the EU on the enforcement of domestic regulation and/or more substantial ENP assistance, however, the SLI has largely failed to become an engine towards high-road of economic development whereby worker protection is facilitated alongside industrial upgrading.In the context of limited resources, the obligation for Moroccan inspectors to serve as mediators in disputes has in fact been at the detriment of their ability to conduct inspections (MA6&7, USDSb 2014).Companies would, moreover, often refrain from participating in the arbitration process to address the source of non-conformance (MA7).
These findings thus reveal that ENP-led improvements will be limited in contexts characterized by weak state institutions and labour organizations.Given the predominant focus of EU assistance on state institutions among NCs, the extent of progress is dependent on existing state administrative capacity.In the absence of greater engagement of labour organizations in the ENP labour governance approach, the strength of labour organizations is at most supplementary to EU efforts to improve labour practices.

Conclusions
In this paper, we have shown that upgrading the transnational governance potential of the EU did result in some improvements in labour standards among the NCs.We furthermore shed light on the processes through which these came about, revealing the relevance of a combination of political pressure, persuasion, policy guidance at the interstate level; and to a lesser degree, of state institution-building assistance.While the extent of reforms might have been less in the absence of economic incentives, we did not find a strong direct link between their use and labour reforms.The DCFTA accentuated regulatory reforms in Moldova, but failed to prevent a significant deterioration in state capacity.These findings cast doubt on the validity of expectations that more stringent measures are sufficient or even necessary for better outcomes (Hafner-Burton 2005, Lechner 2016), however, it supports the notion that better outcomes could be attained if such measures are complemented with domestic capacity-building assistance.
In addition to these prospects, we also found important limitations in the EU's transnational labour governance approach.Firstly, improvements in labour standards were contingent on their coherence with other EU priorities, limiting ENP-led reforms to some fundamental labour rights and failing to encourage progress in working conditions, except for OHS.Secondly, while setting benchmarks for reforms and monitoring progress presented the potential for a problem-solving approach, the primary focus on legal reforms limited progress to this 'reportable' level.Thirdly, EU efforts to strengthen institutions, moreover, have focused too narrowly on state administration and been insufficient to improve implementation.Finally, the EU has failed to engage domestic labour organizations in its problem-solving and institution-building efforts, discouraging also possible complementarities with the initiatives of labour organizations.These findings resonate with other studies that find that external labour governance efforts favor improvement at the formal level, often in the absence of or even deterioration in implementation (Greenhill et al 2009, Peksen andBlanton 2017).They fail, however, to provide robust empirical support to past assumptions that the EU approach facilitates improvements through policy learning by CSOs (Bastiens and Postnikov 2014).
The main policy implication of our research is that greater emphasis should be placed on improving the capacity of domestic state and non-state actors to participate in EU-led efforts to improve labour standards.Two areas of further research could be instrumental to the emergence of the EU as a smart power.The first pertains to a better understanding of how market-based private governance initiatives have affected labour standards in the NCs in the context of greater economic integration, exploring also the possibility of an indirect influence of the ENP on working conditions.Future studies could also examine how the EU could incentivize and support businesses from trade partners to opt for high-road of development, as for example through investments in skill training, improving labour standards in the process.

9
These were the conventions on Freedom of Association and Protection of the Right to Organise (No. 87); Right to Organise and Collective Bargaining (No. 98); Forced Labour (No. 29); Abolition of Forced Labour (No. 105); Minimum Age (No. 138); Worst Forms of Child Labour (No. 182); Equal Remuneration (No. 100); Discrimination in Employment and Occupation (No. 111)

-
estimated somewhere around 40 percent of the workforce in Moldova and as high as 60 percent in Morocco (USDS 2014/5a,b).

Table 3 : The Determinants of Workers' Rights Among NCs (LPM)
6 The estimates of the sensitivity checks are available upon request.7Sinceweareinterested in the underlying mechanisms for improvement in workers' rights for the ENP, we conduct this analysis for only NCs between2004-2014.8Thisdummy is constructed such that it takes a value 1 if there is an improvement in workers' rights or if there is no deterioration in workers' rights when workers' right is 2; and takes a value 0 if otherwise.

Table A2 : Effects of EU PTA and ENP on Workers' Rights (FE and Ordered Probit)
Timea Pál, Franklin Maduko and László Bruszt