The Role of Domestic Factors in the EU’s Governance of Labour Standards through Trade

This paper analyses the role of domestic factors in the ability of the EU to promote NTPOs through formal trade policies and market integration, as two distinct governance approaches. Focusing on labour standards, we examine how economic and political factors in developing economies influence the ability of key stakeholders – governments, businesses and labour organizations – to undermine or reinforce EU governance efforts. We use a combination of regression and comparative case analysis to assess associations and understand processes on ground. The quantitative analysis shows that improvements in labour standards are closely associated with 1) labour standards in key export markets, 2) technological sophistication of exports, 3) state capacity and 4) freedom of CSOs. Comparative analysis of how differences in state strategies and capacity, key export destinations and strength of labour organizations across Moldova and Morocco set in place divergent pathways of labour standard improvements allows us to shed more light on the potentials and limitations of EU governance efforts.


Introduction *
How and under what conditions can the European Union (EU) promote non-trade related policy objectives (NTPOs) such as labour standards among its lesser developed trade partners? Is this merely a matter of using the right type of conditionality and assistance in the specific field? Or does it involve a more encompassing approach that require adjusting trade and non-trade related policy objectives to diverse local conditions? We address these questions through an analysis of the extent and manners in which domestic factors in developing countries condition the ability of the EU to promote labour standards through trade 1.
Over the past years, EU efforts to promote labour standards among its trade partners have intensified. The inclusion of labour provisions in EU trade agreements has become increasingly common, as demonstrated by the Trade and Sustainable Development (TSD) chapter that has become an integral part of EU free trade agreements (FTAs) (Harrison et al 2019). Expansion of trade and investments, especially since the emergence of global value chains (GVCs) in the 1990s, have also led to an intensification of market-based mechanisms for diffusion of labour standards, illustrated by the increasing prevalence of private transnational regulation (Locke, 2013). Studies of the promotion of labour standards through formal trade policy or actual trade integration have so far focused primarily on the external features of international public and market-based private governance. A key debate, for example, has been around the relative effectiveness of the use of economic incentives and sanctions, assistance and dialogue, and/or engagement of labour organizations by public or private external governance actors (Hafner-Burton 2005, Kim 2012, Postnikov and Bastiaens 2014, Oehri 2015, Amengual and Distelhorst 2020, Pike 2020. More recently, however, an increasing number of studies acknowledge the role of political and economic factors from developing countries in the implementation of these external governance approaches. These point towards the relevance of exposure of domestic firms to global competition, institutional capacities and industrial strategies of governments and business organizations, independence and influence of labour organizations, and the labour standards of key export destinations (Greenhill et al 2009, Bruszt and McDermott 2014, Amengual and Chirot 2016, Bair et al 2020, Short et al 2020.
Building on these studies, we set up an integrated analytical framework to examine the role of the interests and capacity of major stakeholders from developing countriesstate, exporting firms and labour organizationsto facilitate or hinder implementation of inter-national and market-based governance approaches from the EU. We then test the explanatory power of this framework through a combination of quantitative and qualitative analysis. By combining large scale regression analysis to identify and assess the relative strength of associations between labour standards and domestic factors, on one hand, and comparative studies of processes of change in the domestic contexts of Moldova and Morocco, on the other, we shed more light on the ways in which key actors from developing countries respond to EU labour governance based on the institutions and structures in which they are embedded in. The main contribution of our study is thus to expand our understanding of the relative importance and of the mediating role of domestic factors identified in past studies. We also further research in this area by highlighting the relevance of sectoral structure of trade integration and the differentiated ability of individual EU member states to trigger changes in developing economies.
Our quantitative study shows the role played by economic mechanism in domestic responses to external regulatory norms. Export to countries with more demanding labor standards has a positive effect, as does exporting in sectors with larger complexity and requiring higher skills. We also find * This project was supported by European Union Horizon 2020 research and innovation grant 770680 (RESPECT). 1 Labour standards, in our analysis, refer to both fundamental labour rights, such as freedom of assembly and collective bargaining, elimination of forced and child labour, non-discrimination; and working condition regulations, such as minimum wage, overtime restrictions, occupational health and safety. Robert Schuman Centre for Advanced Studies Working Papers support for the positive effects of strengthening civil society organizations (CSOs) and the upgrading of state capacity. Our case studies help us to establish links between these factors and action on the ground by key economic actors: the state, business and labor. We can identify a vicious circle of the low road of integration to regional markets with export going primarily to countries with lower labor standards from sectors of lower technological complexity requiring primarily unskilled labor. In this setting, incumbents base their export strategy of maintaining their competitiveness on the basis of cheap labor. State capacity to design policies of industrial upgrading are weak and the state apparatus is open to the pressure of domestic business to resist improvements in labor standards that could increase labor costs. External assistance to unions or nongovernmental organizations (NGOs), or state administration have very little positive effects. We can also point, however, to a potentially more positive virtuous circle among these factors and actors. Export to countries with more demanding labor standards and with goods of higher complexity might alter the incentives of the state and business, both the domestic and of the external firms in the production chain. If the state has stronger administrative capacity and it can design and implement upgrading programs, the probability of the coming about of a developmental alliance supporting improvements in labor standards among domestic actors also increases.
The structure of the paper is as follows. We first review some of the insights regarding the role of domestic factors emerging from the literatures on the promotion of labour standards through trade. We then present an analytical framework that highlights how the state, business and labour organizations from developing countries are expected to facilitate (or hinder) the implementation of EU governance mechanisms. The following two sections consist of the regression and comparative analysis respectively, that serve for drawing some general conclusions at the end of the paper.

Promoting Labour Standards through Trade
A growing body of studies examines the link between labour standards and trade, focusing primarily on how political and private actors from major advanced economies can promote labour standards in developing economies. This field of study is largely divided between the literature that examines the promotion of labour standards through the inclusion of labour provisions in formal trade agreements and deployment of accompanying instruments to ensure implementation (eg. Hafner-Burton 2005, Kim 2012, Postnikov and Bastiaens 2014, Oehri 2015; and those that analyze the diffusion of labour standards through market integration, particularly through GVCs and the accompanying private governance initiatives (Greenhill et al 2009, Distelhorst et al 2015, Short et al 2020. Over the past years, an increasing number of studies from these literatures have shifted attention from the exclusive focus on international or transnational aspects of these governance approaches to greater consideration of the role of domestic factors across the different developing countries where these external arrangements have been implemented. In the literature on labour provisions in trade agreements, studies that focus on the implementation of TSD chapters in the EU's FTA agreements across various national settings reveal the relevance of domestic interests and political institutions. In their paper on the potential of the EU-Asia FTAs to promote social norms, Garcia and Masselot (2018) note the lack of commitment or outright resistance among some trade partner governments to EU governance of labour standards through international agreements. In an examination of the EU-Colombia agreement, Marx, Lein and Brando (2018), highlight the significant influence exerted by the Ministry of Trade in the implementation process, as well as the lack of capacity of NGOs and of trade unions to promote workers' interests in the civil society mechanisms of the agreement. Focusing on the case of Peru, Orbie and Puttie (2016) raise concerns regarding the ability of trade unions and state labour inspectorates to ensure adequate implementation of reforms adopted to meet EU requirements. In comparative study of the operationalization of the TSD framework across the Caribbean, South Korea and Moldova, Harrison and colleagues (2019) reveal additional constraints set by a lack of prioritization of labour standards by CSOs and antagonistic statesociety relations. More research is needed to understand the mediating role of domestic interests and institutions across developing countries in general (Postnikov and Bastiaens 2014).
In the literature on market-based governance of labour standards, a group of empirical studies emphasize the role of incentives set by social sustainability expectations of public and private entities from major exports markets in raising standards among trade partners. Greenhill, Mosley and Prakash (2009) show that exporting to countries with rigurous collective labour rights leads to more stringent laws in the countries of origin. Cao, Greenhill and Prakash (2013) also find a similar association between importing and exporting country standards in the realm of human rights. Malesky and Mosley (2020) show that when opportunities to export to the US arise, Vietnamese firms are more willing to invest in improving their own labour practices, including skills and better working conditions. In a contrary sense, Adolph, Quince and Prakash (2017) find that as African countries expand their exports towards China, they start to reflect the latter's lower labour standards. Other studies have placed greater emphasis on the relevance of political institutions and interests. In a study of the implications of expansion of trade for collective labour rights in Costa Rica, for example, Mosley (2008) underlines the importance of trade unions and their ties to left-leaning political parties. Schrank (2009Schrank ( , 2013 argues that policymakers and bureaucrats from the Dominican Republic were able to meet external demands from the US to improve labour standards by reinforcing their state labour inspectorate. The literature on private transnational regulation also calls attention to the important role of developing country governments and labour organizations that may hinder or support alternative arrangements (eg. Distelhorst et al 2015;Locke et al 2013). In their work on the apparel sector in Bangladesh, for example, Bair, Anner and Blasi document the ways in which the state and suppliers, concerned about rising operational costs and continued cost pressures by lead firms, undermine an ambitious private initiative to address health and safety concerns in the aftermath of the Rana Plaza tragedy. Amengual and Chirot (2016) also find that improvements in labour standards in the garment sector in Indonesia through the Better Work programme are largely contingent on the support of the Ministry of Manpower and Transmigration. They attribute state support to political mobilization by trade unions, thus reiterating the important role that labour organizations can play even in developing countries, where freedom of association and independence of trade unions is often limited. Short, Toffel and Huggill (2020) also find that firms are more likely to improve practices in response to private governance initiatives in national contexts with denser NGO networks and relatively free press, that pose greater risk for harmful practices to be exposed.
These studies provide some important insights into which economic and political factors might be relevant for the ability of the EU to promote labour standards in developing economies. Our contribution is to integrate these perspectives under a common framework that allows us to assess their relative importance in a more systematic manner. We also extend analysis to the sectoral composition of trade to better capture the influence of economic pressures on the abilities of domestic actors to respond to EU labor standard requirements. Our policy proposals are built on combining the exploration of the different domestic factors that shape the incentives of key players, with the insights of the literature dealing with the way various economic integration strategies can shape the incentives and the capacities of domestic actors McDermott 2012, Bruszt andLangbein 2020). According to the latter, trade policies that include assistance that alter the capacities of domestic actors to profit from increased trade can also change their expectations about the costs and benefits of improved labour standards.

Linking Domestic Factors to EU Labour Governance Approaches
This section focuses on our hypotheses regarding the role of domestic factors from developing countries in the effectiveness of the EU to promote labour standards either through inter-national or market-based governance. Informed by past studies, we focus on the role of the state, businesses and labour organizations that as interested parties affected by changes in labour standards, might further or block change promoted by EU actors. While international trade policies, on one hand, and market integration -particularly through GVCs, on the other, obviously overlap, they correspond to different labour governance approaches and have different policy implications. Notwithstanding some efforts to engage social partners and CSOs, EU governance through international trade agreements is more formal and Robert Schuman Centre for Advanced Studies Working Papers political in character that focuses predominantly on the governments of trade partners. Market-based governance of labour standards, on other hand, operates primarily at the firm and sector level, although it can also affect governments in function of their national relevance. As such, these two alternative mechanisms set different opportunities and constraints for incumbents from developing countries to interfere with their implementation and mediate their impact.

State-Led Process of Change
Governments of developing countries are generally expected to respond primarily to formal requirements stipulated at the international level. The significance of state efforts to conform with such expectations is often presumed to vary primarily in function of the magnitude of the incentives and assistance provided (eg. Schimmelfenning and Scholtz 2008). Neighbouring countries, for example, were offered the possibility of full market integration and access to substantial assistance as part of the European Neighbourhood Policy (ENP), that exceed the benefits offered to most other developing country trade partners. The extent of reforms adopted by developing country governments is, nonetheless, expected to vary across national contexts in function of other factors as well. Even within the ENP, governments of individual countries have sought different levels of approximation with the EU based largely on geopolitical considerations and economic development strategies (eg. Börzel and Lebanidze 2015). Governments that actively pursue an export-oriented economic development will most likely exert greater efforts to meet the requirements of both public and private actors from the EU to ensure market access. This process is expected to affect labour reforms primarily at the formal level, as legal reforms are generally easier to adopt and can serve as visible trademarks for governmental efforts to meet EU requirements.
The implementation of these reforms, however, will be contingent on the economic competitiveness strategy and strength of state capacity, that influence the political will and ability of the government to enforce reforms. As we have seen in the case of the Dominican Republic, improvements in labour practices were possible due to the existence of a state labour inspectorate that combined worker protection with economic efficiency (Schrank 2009(Schrank , 2013). This in turn was possible due to a public administration system that relied on meritocratic recruitment, higher wages and fostering of an organizational culture (Piore and Schrank 2008). The capacity of the state to facilitate industrial upgrading towards higher value added activities, where cost pressures are generally lower, can also facilitate improvements in labour standards Lee 2016, Marques andEberlein 2020). In the absence of such capacities, however, economic pressures exerted through international markets, particularly through some GVCs, can greatly constrain domestic inclination to raise labour standards on the ground. In the case of Bangladesh, dependence of the national economy on textile and apparel exports has fuelled government resistance to domestic and external pressures to improve labour standards (Bair et al 2020). This is because suppliers from low-skill labour intensive sectors like the garment sector are often exposed to particularly strong cost and competition pressures (eg. Anner 2020). Such pressures are generally attenuated in the more technologically complex sectors like electronics and automotive, where the ratios of mid to high-skilled production is generally higher and prospects for workers are better (Barrientos, Gereffi and Rossi 2011). Governments of developing countries are thus also often exposed to market-based economic and social pressures, with influence on the implementation of national labour regulation.
Hypothesis 1: The extent and type of labour standard improvements facilitated by the state will be contingent on a combination of a) overall interest in greater economic integration with the EU, b) state administrative capacity for regulatory enforcement and/or industrial policy, and c) the national economy's dependence on low-skill labour intensive exports.

Business-Driven Processes of Change
Exporting businesses are expected to respond predominantly to market-based incentives and pressures, with differences across developing economies in function of the key destinations and sectoral composition of exports.
Firms seeking to export to large markets with strict standardsparticularly the EU -are at the core of studies on upward regulatory convergence (Vogel andKagan 2002, Bradford 2012). While the California or Brussels effect generally work for product standards that can be easily monitored, the emergence of private regulation overcomes some of the disadvantages that process standards like labour standards face in this regard (Greenhill et al 2009). CSOs and lead firms from developed economies with high standards can monitor and coerce exporting firms and their governments to improve labour standards (Peterson et al 2018, Amengual andDistelhorst 2020). Consumers and firms from developed economies are also often willing to pay more for products manufactured in ethical conditions (Hainmueller andHiscox 2015, Distelhorst andLocke 2018). This is echoed by findings that exporting to advanced economies can also entail higher markups (Malesky and Mosley 2018).
While the EU is often understood as a homogeneous entity in these regards, thus activating the Brussels effect (Bradford 2012), we highlight the important differences that exist across individual member states both in terms of stringency of labour regulations and economic prosperity. Notwithstanding continuous efforts towards convergence, such divisions are still particularly important between old and new member states (Meardi 2013). New member states might not meet a minimum threshold that is often necessary to trigger changes in practices that are as culturally and institutionally embedded as those related to fundamental labour rights (Cao et al 2013). We therefore consider exports to individual member states rather than the EU at large, and expect to find differences in firm-level adjustments in developing countries that export primarily to old relative to those that sell predominantly to new EU member states.
The effectiveness of market-based mechanisms is also expected to vary across sectors due to differences in economic pressures exerted through the more and less technologically sophisticated GVCs. The success of private regulation is often hindered by significan cost and competition pressures Kuruvilla 2020, Bair et al 2020). The intensity of such pressures tends to vary across sectors due to differences in technological and skill complexity, organizational structure and governance of GVCs (Barrientos et al 2012;Gereffi et al 2005). In the textile and apparel sector, for example, where low-skill labour intensive production is more prevalent and competition is more global, lead firms tend to pay increasingly low prices to suppliers while also imposing short lead times and high order volatility (Anner 2020). The intensity of such pressures might be mitigated to some extent by financial and technical assistance from the state and/or business associations to facilitate industrial upgrading (Gereffi and Lee 2016). Economic pressures on supplier firms are presumably weaker in technologically more sophisticated GVCs, like automotive, where production often requires more skilled workers, major lead firms rely on regional suppliers and relations are more durable (Sturgeon, van Biesebroek and Gereffi 2008).
Social and economic pressures on exporting firms might also affect state regulation, and as such, influence EU governance at the international level as well, especially when businneses are politically influential. The EU has, after all, actively supported the engagement of social partners and CSOs in the implementation of trade agreements through institutionalized dialogue (eg. Ashraf and van Seters 2020). Export-oriented producers might support labour reforms at the formal level to assuage the EU and of pressure groups from the EU that their labour standard requirements are met. At the implementation level, however, the influence of export-oriented businesses on regulatory policies in developing countries is generally in the opposite direction, especially when the role of low skill labour intensive sector is significant. Robert Schuman Centre for Advanced Studies Working Papers Hypothesis 2: Export-oriented firms are more inclined to improve practices when a) exporting technologically more sophisticated products and/or b) to end markets with higher labour standards, especially when they benefit from state or business association assistance.

Labour-Driven Improvements
Labour organizationstrade unions or NGOs involved in labour issuescan facilitate improvements in labour standards through either inter-national or market-based mechanisms. Their influence is expected to be contingent on their organizational strength and political influence. The former refers primarily to the independence and representativeness of trade unions. The latter can be increased by the social dialogue and new civil society mechanisms instituted by the EU, but also through connections to political parties.
At the international level, trade unions and NGOs can provide essential information to the EU about major labour concerns and contribute to possible solutions through its civil society arrangements. But as these mechanisms are still relatively new, we expect labour organizations to exert their influence primarily through the pressures on their national governments that can reinforce EU requirements to improve worker protections. Their contribution, empowered by the social dialogue institutions at the national level, could be particularly essential to promote greater protection with respect to working conditions, complementing the primary focus of EU labour provision on fundamental labour rights. Considering the significant gap that often exist between formal regulations and practices on ground, labour organizations could exert greater pressure on the state or complement state capacity to improve enforcement.
Labour organizations can also facilitate the effectiveness of private regulatory initiatives by providing essential information and engaging in international campaigns. Traditionally, their direct involvement in the regulatory efforts of lead firms in global value chains have been relatively limited. Labour organizations have, however, facilitated changes through contestation of existing practices that lead firms were then forced to address (Amengual and Chirot 2016) and have sometimes used private arrangements to substitute for weak state enforcement . Finally, labour organizations could also contribute to improvements by facilitating industrial upgrading or ensuring that the gains are shared with workers in export-oriented firms and/or sectors (Gereffi and Lee 2016).
Hypothesis 3: Labour organizations, when organizationally strong and influential, can facilitate improvements in labour standards through both governance mechanisms.

Regression Analysis
In this section, we examine the explanatory power of some of the hypotheses. We do so through a regression analysis of the association between labour standards and several domestic variables across developing countries. Our key variables of interest are state capacity (H1b), a measure of labour standards in key export markets (H2a), and CSO freedom (H3). Textile and apparel and technologically more sophisticated exports are proxies for economic constraints on state and business-led improvements in labour standards (H1c&H2b).

Data and Econometric Model
We estimate the relationship between labour standards and our key variables of interest using an unbalanced panel of 45 countries over the period 1990-2014. These include developing economies from Africa, Eastern Europe, Central Asia, the Middle East, Latin America, and the Caribbean that have trade agreements in force with the EU. Among these countries, twelve have become part of the ENP since 7 2004 2. Countries with labor provision in their EU trade agreement are identified by looking at the specific details of the agreements (see Bruszt et. al 2020 for detail).
As our dependent variable is an ordinal variable, we estimate an ordered probit regression model, as specified in equation (1).
Where y it is the worker's right index of country i in time t. This variable was taken from the CIRI database (Cingranelli et al 2014) and ranges from 0 to 2. It measures mainly the extent workers enjoy internationally recognized labor rights and minimum standards regarding working conditions. A score of 0 reflects that workers' rights were severely restricted; a score of 1 indicates that workers' rights were somewhat restricted; and a score of 2 indicates that workers' rights were fully protected during the year in question.
BTC it is the bilateral trade context (BTC) index proposed in Greenhill et al. (2009) that is the weighted average of worker's rights among a country's export partners. Specifically, for each country and in each year in our data, we construct the share of exports to each of the country's export partners, and we use these shares as weights in taking averages of the worker's rights for the country's export partners. The measurement equation takes the form below: Where i denotes a country in our data, j is the export partner of country i, and t is year and N is the number of countries that country i exports to. Expijt is the total exports from country i to country j in year t. is the total export of country i in year t, and measures the worker's rights in the export destination. The export data used in this index was taken from the World Integrated Trade Solution (WITS) database. Clearly, countries with larger export share to destinations with stronger worker's rights, will have high values of the BTC index, and vice versa. This index is part of the variables that helps us test for hypothesis 2. Specifically, it enables us to test whether a country's worker's rights depends, amongst other things, on the worker's rights in its main export destinations.
is the ratio of net export of textile/apparels products in total exports in a country in period t, and is the ratio of net export of technology-intensive products in total exports in period t 3. Technology intensive classification is based on the work by Lall (2000). The product-export-destination data comes from the UN Comtrade database. These variables proxy for the cost and competitiveness pressures exerted through GVCs on the ability of governments and domestic businesses to conform with EU expectations to raise labour standards, that are relevant for both hypothesis 1 and 2. We thus expect a negative relationship between and a positive relationship between with workers' rights, respectively.
is an index of civil society freedom which was taken from the variety of democracy (VDEM) database (Coppedge et al 2019). Higher values indicates higher level of civil society freedom, while lower values imply higher level of civil society repression. The variable proxies for the strength of labour organizations and we expect a positive relationship between CSO freedom and labor standards (H3). 2 These are: Algeria, Armenia, Azerbaijan, Egypt, Georgia, Israel, Jordan, Lebanon, Moldova, Morocco, Tunisia and Ukraine. 3 We measure these variables using the equations = and = net ℎ Robert Schuman Centre for Advanced Studies Working Papers is a variable which represents the level of judicial independence in each country. This is used as a proxy for state capacity (H1b) in this paper to capture the ability of the state to enforce labour reforms adopted to meet EU requirements. This variable was taken from the CIRI database and it ranges from 0 to 2, where higher values represents a more independent Judiciary. We also expect a positive relationship between this variable and labor standards.
ENP&TAit is a control variable which ranges from 0 to 2 and captures the intensity of international labour governance efforts by the EU. It takes the value 0 if country i does not have any EU trade agreements with labor provisions in force in period t, 1 if it has an EU trade agreement (TA) with labor provisions in force in period t, and 2 if the country joined the ENP. Since labor provisions in EU trade agreements improves worker's rights , controlling for this variable isolates the effect of domestic factors on labor standard from the effect attributed to EU trade agreements with labor provisions.
We also include a number of other control variables such as FDI inflows, political constraint, GDP (aggregate and per capita) and GDP growth rate. The FDI inflows, is the amount of new inflows of foreign direct investments to GDP ratio in a country each year. This controls for the effect of foreign direct investments on labor standards and helps to isolate the effect of foreign owned production from technology-intensity of exports or export destinations. The political constraint variable, taken from the Quality of Government (QoG) database, measures the ease of changing government policy given that one political actor has strong preferences for such policy change. The GDP serve as measures of national income and income growth and captures the fact that different levels of economic prosperity can have different effect on labour standards.

Results
We present our ordered probit model results in Table (1) and the marginal effects in Table (2). These results confirm the existence of statistically significant associations between labour standards, on one hand, and our key domestic variables, on the other, in the expected directions.
The estimates in Table (1) indicate a positive relationship between BTC and workers' rights across all specifications. This implies that higher labor standards among countries where a country exports larger proportion of its foreign sales is associated to an improvement in the country's labor standards, consistent with findings by Greenhill et al (2009).
We also find that a higher share of technology-intensive product in total exports is positively associated with higher values of worker's rights, while a greater share of textile and apparel in exports is associated with a decline in worker's rights. This is consistent with initial expectations that economic integration through low-skilled labour intensive sectors, like textile and apparel, will more likely hinder domestic efforts to meet EU labour standard requirements. Prospects are much more favourable in technologically more sophisticated sectors.
The freedom of civil society organization index (CSO freedom) is positively associated with workers' rights. This implies that policies that strengthen the freedom of the civil society will be instrumental in ensuring that labor standards are upheld. This positive influence can be exerted, at least in part, through the engagement of CSOs with the international or transnational labour governance approaches from the EU. Similarly, with respect to judicial independence, there is a statistically significant positive association between labour standards and state enforcement capacity in developing countries, that could also facilitate the implementation of EU's governance efforts to achieve more progress.
We control for a number of variables including the ENP & TA variable that is positively associated with our dependent variable, confirming that having an EU trade agreement with labor provisions or joining the ENP leads to better outcome in labour standards. This is consistent with previous findings in the literature (Postnikov andBastiaens 2014, Bruszt et al 2020). We conduct some robustness checks to test whether our results are sensitive to deviations from model specifications and variables redefinitions. First, we estimate a fixed effect regression model similar to equation (1). Since the dependent variable is ordinal, fixed-effect regression does not take into account the ceiling and floor restrictions on models that include ordinal variables (Winship and Mare 1984) and may result to biased estimates, however it offers an ease of interpretation. In the second sensitivity checks we use net exports of technology intensive, and textiles and apparels products to the EU, instead of the total net exports of these products. We find results from both sensitivity checks to be largely consistent to our baseline results.
Since our dependent variable is ordinal (ranging from 0 to 2), interpretation and quantification of the relationships identified in Table (1) are more adequate through the computation of marginal effects. Specifically, the marginal effects are the partial effects of our explanatory variables of interest on the probability of observing a value of worker's rights=0, worker's rights=1, and worker's rights=2. We present the marginal effects corresponding to column (4) of Table (1) in Table (2) below 4. These confirm that our independent variables of interest exert a substantial influence on the probability of demonstrating higher labour standards. 4 The results in Table (2) are expressed in terms of a one standard deviation from the mean of the domestic factors. Robert Schuman Centre for Advanced Studies Working Papers The estimates in column (1) of Table (2) imply that an increase by one standard deviations from the mean of the worker's right index of a country's key export destination, is associated with being 5.8% less likely to have a worker's rights score of 0, 4.6% more likely to be have a worker's score of 1, and 1.2% more likely to have a worker's rights score of 2. This suggests that labour standards in a country are very sensitive to labour standards of the country's trading partners.
In column (2) of Table (2), we find that a one standard deviation from the mean of the ratio of net export of high technology products to total exports will be associated with being 8.1% less likely to having a worker's rights score of 0; 6.4% of being more likely to have worker's right score of 1, and 1.7% of being more likely to have a worker's rights score of 2. In column (3), as expected, we find the opposite results. A one standard deviation from the mean of the net export of textiles and apparel products in total exports is associated with 6.9% of being more likely to have a worker's right score of 0; 5.5% less likely to have a workers' rights score of 1, and 1.4% less likely to have a workers' rights scores of 3. This suggest that policies which promotes production and exports of technology intensive products as against textile and apparel will increase the likelihood of stronger worker's rights.
In column (4), we find that a one standard deviation from the mean of CSO freedom is associated with being 8.3% less likely to having a worker's rights score of 0, 6.6% more likely to having a workers' rights score of 1 and 1.7% more likely to having a worker's rights score of 2. In column (5), we find that a one standard deviation from the mean of judicial independence is associated with being 7.2% less likely to have a workers rights of 0; 5.8% of being more likely to have a worker's rights of 1, and 1.5% of being more likely to have a workers' right of 2. Clearly, this implies a strong sensitivity of workers' rights to both CSO freedom and state capacity (proxied by Judicial Independence) within a country. We now explore whether these domestic factors exert part of their influence on labour standards through the increased effectiveness of the EU's international governance approach, captured in our study through the ENP & TA variable. To this end, we interact ENP&TA with judicial independence, CSO freedom and BTCas the variables of greatest policy relevanceto determine the magnitude of their positive influence on labour standards across countries without labour provisions in their EU TAs, with labour provisions in their TAs and ENP countries. By comparing the coefficients across these groups of countries, we can draw some conclusions on the extent to which their positive influence has been interconnected with the international labour governance approach of the EU. We present the results from these specifications in column (1) and column (2) of table (3). In column (1), we find that among countries in periods that they neither have EU TA (with labor provisions) or ENP agreements, judicial independence is strongly positively associated with labor standards. This pattern also exists in countries with EU TA, however, the magnitude of this relationship is weaker. Finally, the estimates show that, among ENP countries, judicial independence is uncorrelated with labor Robert Schuman Centre for Advanced Studies Working Papers standards. We find a similar pattern of results in column (2) of table (3). The particularly low coefficient levels among the ENP countries can be attributed in part to the fact that both judicial independence and CSO freedoms have been very low with very slight within country variation among the eastern and southern ENP countries. Overall, these results suggest that irrespective of joining an EU TA with labor provisions or ENP, policies that improves judicial independence and CSO freedom, can foster improvements in labor standards. It also suggests an absence of complementarity, at least during the 1990 until 2014 period, between CSOs and state capacity on the one hand, and EU international labour governance efforts on the other, to bring about improvements in labour standards. These might come about, however, as the EU makes more investments in raising the domestic capacities in these societies where state institutions and strength of CSOs have been traditionally low.
The positive association of increasing magntitude between ENP/TA and BTC, however, does reveal a synergistic effect between the EU's international labour governance approach and market-based governance of labour standards, as the positive influence on labour standards is the strongest when both the EU and private entities from key EU sets the most intense expectations regarding social sustainability among developing country trade partners.

Enabling and Constraining Factors in Moldova and Morocco
In this section, we use the cases of Moldova and Morocco to expand our understanding of the processes that underpin associations identified in the regression analysis and further examine the validity of the hypotheses. More specifically, we explore how integration into labour versus technology-intensive value chains influences the willingness and ability of governments and businesses to meet EU expectations regarding labour standards (H1&H2). We investigate the processes through which the stringency of labour standards in major export markets affects standards within their contexts (H2). Finally, we examine the role of state capacity and labour organizations in promoting labour standards through the EU's international or market-based governance approaches (H1&3). The primary source of data that we rely on consists of approximately 31 semi-structured interviews we conducted over the course of 2019 in Chisinau (MD1-17), Rabat and Casablanca (MA1-14) where we met with representatives of ministries of labour and economy, state labour inspectorates, employers' associations and private businesses, trade unions, ILO country offices and EU delegations. These are complemented with review of secondary reports prepared by various international, EU and national organizations.

Comparative Strategy
Moldova and Morocco constitute ideal cases for a comparative analysis of the role of domestic factors in the ability of the EU to promote labour standards among trade partners. Over the past decades, the two economies have become integrated into similar EU labour governance systems, at both international and market levels, but differ considerably across several political, economic and social features. Notwithstanding important limitations, both countries have experienced some incremental improvements in labour standards over the past decades .
At the international level, their relations with the EU have been governed through the ENP that reinforced commitments to international and EU labour standards stipulated in their formal agreements, and deployed additional incentives and assistance to promote reforms (EC 2004(EC , 2005(EC , 2013. Moldova and Morocco have also become increasingly integrated with the EU market, particularly in the textile and apparel, electronics and machinery sectors, with implications on their exposure to market pressures and private governance initiatives from the EU. In 2018, approximately 65 percent of their total exports were destined to the EU, representing about 16 percent of their national economies (WITS, 2020). Textile and clothing was of greater relevance in the past, but has continued to expand in absolute value and still represents around 15 percent of their total exports. Over the past years, both Moldova and Morocco made considerable progress in expanding exports of machinery and electronics products, reaching 20 percent of their total exports in 2018.
The national contexts in which these relatively similar EU labour governance approaches have been implemented, however, differ significantly. There are three major differences that are particularly relevant for this study. Firstly, while the Moldovan government has demonstrated great commitment towards approximation with the EU by signing a DCFTA in 2014, the Moroccan government has been more reluctant in this regard and suspended negotiations. Secondly, while the EU constitutes the most important export destination for both countries, a significant part of Moldovan exports are destined to Romania and other new member states, but the predominant part of Moroccan exports are captured by France and Spain, where worker protection are very high (CIRI 2014). A third important difference pertains to the strength of labour unions in the two countries. Workers and unions from Moldova benefit from more freedoms and protections than in Morocco (ITUC 2020). Trade union density has been much higher at approximately 50 percent in the mid 2000s in Moldova relative to a mere 6 percent in Morocco (USDS 2004a,b). Finally, while in Moldova workers are represented by the unified National Trade Union Confederation of Moldova (CNSM), in Morocco, there have been five nationally representative unions with different political inclinations.
Based on the analytical framework presented in this paper, these differences are expected to have implications on the ability of the EU to promote labour standards in Moldova and Morocco, respectively. While the DCFTA, exporting to EU member states with higher labour standards, and stronger labour unions are presumed to facilitate improvements, including in the textile and apparel sector, their lack thereof is expected to hinder conformance with EU expectations. The remaining parts of this section analyses how disparities in these regards across Moldova and Morocco have led to the emergence of divergent pathways of improvements in the two contexts, each with its own advantages and disadvantages.

National Reforms with Weak Implementation in Moldova
In Moldova, the state has undertaken ambitious reforms, with trade union support, to demonstrate conformance with DCFTA requirements, but the low cost-competition strategy pursued by the state and businesses have undermined enforcement. Business-led improvements in actual practices have been connected primarily to the expansion of technology intensive sector rather than incentives set by key export markets.
Since the launch of the ENP in 2004, the Moldovan government has made continuous efforts to meet EU labour standard expectations, resulting in reforms of fundamental labour rights and working condition regulations . This tendency has intensified over the past years, as negotiations for a DCFTA were launched in 2012 and convergence with EU regulations were prioritized. The agreement signed in 2014 included specific references to EU directives to guide further approximation efforts (Annex III). By 2019, Moldova transposed into its national legislation 7 out of 8 labor law acts, 5 out of 6 non-discrimination directives and 19 out of 25 Occupational Safety and Health (OSH) acts (M10&11). These reforms did entail increased protections for workers, especially in areas such as non-discrimination and OSH. Implementation, however, has been undermined by a combination of weak political commitment to strengthen labour regulation that might diminish the competitiveness of the export-oriented economy, and lack of adequate state capacity to ensure enforcement even in areas that were relatively compatible with this strategy. The state labour administrative systemthe Ministry of Health, Labour and Social Protection (MHLSP) and State Labour Inspectorate (SLI)have benefitted from some assistance from the EU as part of the ENP, but the material and administrative resources remained inadequately low and posed increasing challenges as the economy and regulations grew more complex (MD5&6,10&11). The number of inspectors in fact declined from 121 in 2002 to 109 in 2015 (USDSa 2003, MD6). Even in the OSH area that is more compatible with EU product standards, the SLI and business associations lacked the capacity to ensure adequate implementation (MD5&6,16).
In the context of greater trade integration with the EU and Commonwealth of Independent States (CIS) countries, national regulation in Moldova has been further hindered by reliance on a competition strategy based on low costs and regulations. The investment promotion strategy, formulated by the Ministry of Economy and state agency Invest Moldova, continued to emphasize the availability of relatively cheap and well educated labour force (MD2,316). The main strategic planthe National Development Strategy "Moldova 2020" -adopted by the government in 2012 has prioritized the reduction of regulatory burden on businesses to boost entrepreneurship and competitiveness (MD2&5&6). This strategy led to a further decline in state labour administrative capacity as regulatory reforms adopted in 2017 to achieve this goal resulted in the loss of the authority of the SLI to conduct unannounced visits and enforce OSH regulations (MD8&10). The number of labour inspectors declined to a mere 59 by 2018 (MD5&6). The MHLSP also lost its entire staff from its department on labour relations and social protections (MD10).
This competition strategy based on low labour costs and regulation has been related to the position of Moldovan producers in European value chains and the significant economic pressures they have been exposed to. Export-oriented manufacturing from Moldova has benefitted from a relocation of more cost sensitive parts of production from Romania as the latter experienced a rise in wages and became a member of the EU. With Romania and other Eastern European economies, capturing the predominant part (54 percent) of Moldovan exports over the years (WITS 2020), the incentives for domestic businesses to meet external social sustainability expectations have been relatively weak. Pressures to retain low production costs, have moreover continued to be very strong from both western and eastern European firms (MD14&15). State support to move from lower to higher value added activities has been limited thus far. The state did launch some programs, as for example the Industrial Cluster Strategy, but these generally lacked both the necessary coordination capacity and financial resources for successful implementation (MD2). The state has nevertheless facilitated the expansion of the machinery and electronics sectors, through fiscal incentives and other benefits provided to foreign investors, especially in export processing zones (EPZs). While a large part of activities still consist of relatively low skill operations in this sector, working conditions, particularly the level of wages, are generally superior than in the rest of the economy (MD11,13&16). The textile and apparel sectors has also experienced some industrial upgrading, but the predominantapproximately 80 percent -of production still takes place in Cut and Make or Lohnarbeit regime, where domestic value added is limited and pressures on retaining low labour costs are significant (MD7&14&15).
In this context, the export oriented business community has used its political influence to reinforce the low cost and regulation strategy of the state. This influence was in part exerted through the main tripartite institution National Commission for Consultation and Collective Bargaining (NCCCB) where, notwithstanding EU intentions to promote social inclusion, the objective of economic competitiveness has often been prioritized over worker protection. While businesses have largely supported state-led efforts towards approximation with EU norms at the formal level, they would sometimes use these as opportunities to introduce more flexible provisions. In the case of Council Directive 98/59/EC on collective redundancies, for example, reasons for dismissing workers were expanded, and the obligation to consult trade unions before doing so was abolished (MD12, ITUC 2018). But businesses would also exert their influence through alternative and sometimes even less transparent channels (USDS 2016a, ITUC 2018. At the revision of the national labour code of Moldova in 2017, the government favoured a draft proposal prepared by the Economic Council to the Prime Minister -a consultation platform supported by EU development agencies, where business representation has been more prominent, over the one prepared in the more balanced tripartite NCCCB (MD12&14). There is also a sense that associations representing foreign company actors have used their wide access to the government to push for decreasing labour inspections over the past years (ITUC 2015(ITUC , 2018. The national trade union confederation CNSM did facilitate some improvements in labour standards through its support of state efforts to meet EU labour standard expectations. The CNSM has been very active in the framework of the NCCCB to represent worker interests in the harmonization process, but its influence on final decisions has been mixed due to the prioritization of economic competitiveness by the state and businesses (MD11-13). At the implementation level, noteworthy developments included the establishment of its own labour inspectorate and the conclusion of a cooperation agreement with the SLI in 2012 to improve implementation of national regulation. CNSM monitoring prioritized the areas of union rights, as concerns regarding MNEs resistance to unionization were prevalent, but also in the realm of labour relations and OSH. While the unions lacked the authority to impose sanctions or coerce changes, they often engaged in joint problem solving to address areas of concerns (MD10&11). They also invested in measuring equipment to increase the credibility and leverage in the OSH area (MD12). These efforts, however, have been insufficient to compensate fully for weak and declining state capacity.

Market-Driven Upgrading of Labour Standards in Morocco
In Morocco, on other hand, improvements have been connected primarily to the market-based governance mechanism of major EU lead firms, facilitated by domestic business associations and state initiatives, without a major overhaul of national regulation.
The government of Morocco has also adopted some reforms to ensure harmonization with international labour standards that it committed to in ENP Action plans (CE 2005(CE , 2013. With negotiations over the DCFTA paused, however, regulatory convergence with EU labour directives has been more modest. The state labour administrative system has also benefitted from some assistance from the EU through the ENP. But as in Moldova, state capacity to ensure enforcement of labour reforms continued to remain low, with the state labour inspectorate affected by a lack sufficient human, financial and technical resources (USDSb 2019). The number of labour inspectors also declined from 496 inspectors in 2004 to 398 in 2018 (USDSb 2004(USDSb , 2018. This has been particularly problematic in Morocco as inspectors are also required to serve as mediators in labour disputes. Of greater importance for working conditions in the case of Morocco, have been state policies directed towards speeding up the development pace of industrial sectors, as for example the Industrial Acceleration Plan (IAP) (MD 3&5). The plan attached great significance to labour issues as well, including improvements in skills and transition from the informal to formal sector, with the aim of better responding to Moroccan and foreign enterprises' requirements in this regard (MA3-5).
With Moroccan companies exporting primarily to France and Spain (51 percent of total exports in 2018) where labour regulations are more stringent and consumer expectations regarding social sustainability more prominent, lead firms played an important role in encouraging changes in domestic practices. Two of the major EU lead companies that purchase a significant part of Moroccan production -Spanish firm Inditex that buys approximately 50 percent of textile and apparel products, and French company Renault Group that produces 80 percent of vehicles, constitute some of the largest lead companies within their sectors that have international framework (IFAs) in place with global unions, suggesting a more serious commitment towards promoting labour standards among their suppliers. Social sustainability requirements of Inditex and other major EU companies, for example, incentivized the Moroccan Textile and Garment Industry Association (AMITH) to launch its own social governance initiative known as Fibre Citoyenne in 2003 (MA13). The latter consisted of an industry code of conduct and related label to be issued to factories that passed a social audit. Instrumental to its development was the IFA concluded between Inditex and the International Texile, Garment and Leather Worker Federation (ITGLWF) in 2007, now part of IndustriALL, through which Inditex pledged to source only from suppliers with this certification (Miller, 2011). Similarly, in the automotive sector, the IFA concluded between the Renault Group and international trade union IndustriAll facilitated cooperation among international and French labour unions with Moroccan unions, leading to some progress in collective bargaining over wages (MA10). The importance of lead firm governance measures in this sector is revealed by the fact that, according to AMITH representatives, inspections of labour standards in the Moroccan textile and apparel sector are performed primarily by private auditors (MA13).
In addition to their labour practices and requirements, major lead companies from Spain and France in these sectors were at times also instrumental at promoting social upgrading through other aspects of their business practices. In the case of the textile and apparel sector, for example, buyer companies have increasingly demanded full package production over time and provided some technical assistance to major suppliers to achieve it (MA13). Changes in domestic business practices to meet these expectations have also been greatly facilitated by state industrial policies, most notably the IAP, in collaboration with AMITH. The IAP and AMITH played an important role in upgrading technical expertise in the Moroccan textile and apparel sector, including building design capabilities (MA13). The Ministry of Industry has played an important role in this upgrading process through the provision of financial assistance through programs such as the Locomotive program set up to promote cluster collaboration among SMEs. The assistance also entailed significant investments in training programs to upgrade skill, through organizations such as the Casa Moda Academy, founded through a public-private partnership with AMITH; the Ecole Supérieure des Industries du Textile et Habillement (ESITH) that also entails a close collaboration with the business association, as well as the technical labs of AMITH. By 2019, approximately 30 percent of production constituted full package production and an increasing number of Moroccan companies have become involved in fashion design (MA13). Industrial upgrading in the sector has been associated with improvements in working conditions for the more regular and skilled employees in the sectors, but had some negative implications on the more vulnerable groups of irregular and less skilled employees (Rossi, 2013).
Similarly, in the automotive sector, the state and business association have played an important role in expanding and improving the supplier base for lead companies, particularly the Directorate of Automotive Industries of the Ministry of Industry, tasked with developing and implementing IAP programs to support investment and training, and sectoral business association AMICA, that acts as an interface between ministerial departments and members (MA4). A manifestation of their cooperative support includes the establishment and operation of the Automotive Industry Training Institute, where essential training has been provided at four centers, that falls under the authority of the Ministry of Industry but three centers are managed directly by AMICA and one by Renault (MA5). As in Moldova, working conditions in the machinery manufacturing sector are generally superior than in other industries, especially with respect to the level of wages (MA8-11).
The role of trade unions in influencing the effectiveness of EU labour governance efforts has been more limited in Morocco than in Moldova. In the context of lower representation, organizational and political fragmentation, their influence on policy outcomes through tripartite consultation has been more modest. Trade unions would often withdraw from tripartite social dialogue to protest the lack of adequate consideration of their concerns (MA10&11). At the implementation level, one of the Moroccan trade unions, also launched their own monitoring efforts in 2010, by establishing a national observatory of employee's rights in 2010, with the task of collecting data on cases of infringements of trade union freedoms and labor laws, but its subsequent activities have remained very modest and is no longer in existence (CE 2011, MA12). According to trade unions, their ability to affect changes has also been hindered by restrictions on their entry to export processing zones (EPZ) where a significant part of export-oriented manufacturing takes place (ITUC, MA9&10). The ability of Moroccan unions to compensate for weak state enforcement capacity has therefore been even more limited than in Moldova.

Conclusions
In this study we considered two approaches through which the EU can promote NTPOs such as labour standards among its trade partners: international and market-based governance. Going beyond a predominant focus on the features of these external governance approaches, we used a combination of qualitative and quantitative analysis to understand how domestic factors across developing countries condition their effectiveness. The analyses underpin the importance that EU efforts to promote NTPOs through trade need to rely on greater consideration of how trade policies and integration influence the willingness and ability of key stakeholdersthe state, businesses and labour organizationsto respond to normative requirements that often accompany them.
Economic factors, in the form of EU member state destination and sectoral composition of exports, condition the incentives and ability of governments and businesses to improve labour standards. A shift away from textile and apparel to more technologically sophisticated sectors generally entails an improvement in working conditions. Continued reliance on textile and apparel products, however, retains competition and cost pressures, that undermine the effectiveness of EU labour governance efforts. Exports to economically more advanced EU member states with high labour standards accentuate the effectiveness of market-based governance, especially when business associations are strong, as they increase the incentives and in some cases capacity of domestic businessess to improve workers' skills and working conditions. Political factorsstate strategies and capacity, independence and strength of labour organizationsare also influential, with potential for positive implications for both EU governance approaches. In addition to the enforcement capacity of state labour administration, we highlight the relevance of developmental capacity, that can facilitate improvements in working conditions through industrial upgrading. While the freedom and strength of labour organizations is associated with more positive outcomes, and illustrate how they can compensate for weak state enforcement capacity, we find that their limited political influence and business resistance to unionization still set limitations to their ability to reinforce EU labour governance efforts. Finally, we find that improvements are more likely when EU governance through formal policies is reinforced by market-based labour governance.
From the point of view of workers' protection, our findings confirm that international and marketbased labour governance mechanisms have their distinct advantages and disadvantages. While marketbased governance mechanisms can facilitate improvements in working conditions, these are often limited to a segment of workers (the most skilled with regular contracts) in export-oriented sectors. Inter-national labour governance triggers reforms at the national levels and often extends to fundamental labour rights, but these often manifest primarily at the formal level.
At a broader policy level, our findings support the call for treating NTPOs as parts of externally supported developmental programs that aim to improve the conditions of domestic gains from increased trade. Such an integrated approach requires linking the enforcement of external trade and non-trade related rules to improving the capacity of domestic public and private actors to benefit from increased trade. Isolated, or fragmented policies that target CSO, states' administrative capacity or tripartite institutions alone have little effect if they are separated from assistance that can increase domestic benefits from trade. The latter requires improving the capacities of domestic public and private actors to design and implement policies that can improve the composition of export and lead to industrial upgrading. EU efforts to strengthen labour standards across all member states and mitigate economic pressures by lead firms on their suppliers could also facilitate more socially sustainable practices among developing countries.