Policy Research Working Paper 11161 Environmental Stringency and Firms’ Participation in Global Value Chains Evidence for MENA Countries Nada Hazem Chahir Zaki Middle East and North Africa Region Office of the Chief Economist June 2025 Policy Research Working Paper 11161 Abstract The Middle East and North Africa (MENA) region stands countries and thus lends support to the Porter Hypothesis. among the most vulnerable areas to the impacts of climate The paper also shows that these regulations increase the change. At the same time, with lax environmental regula- effect of spending on research and development on GVC. tions, this region’s integration into Global Value Chains Yet, the results are less conclusive for the role of environ- (GVC) is modest. Thus, this paper aims to examine the mental treaties. These results remain robust when a mixed effect of environmental stringency on GVC participation multilevel approach is used, and when large exporters, who in MENA countries. To do so, using the World Bank might lobby to affect policy choices, are dropped from the Enterprise Surveys, this paper analyzes how environmental analysis. In addition, at the sectoral level, national regula- regulations and treaties affect both the extensive and the tions are associated with higher GVC participation in the intensive margins of GVCs. The main results show that food sector in the MENA region and lower participation in national environmental regulations increase the likelihood the plastics one. Finally, regulatory stringency increases the of integrating into GVCs when it is measured using both the probability of GVC participation for both SMEs and large simple and the strict definitions. This result highlights the firms, with the effect generally stronger for SMEs. role of such regulations in attracting GVCs in developing This paper is a product of the Office of the Chief Economist, Middle East and North Africa Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://www.worldbank.org/prwp. The authors may be contacted at nada.hazem@feps.edu.eg and chahir.zaki@feps.edu.eg. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. Produced by the Research Support Team Environmental Stringency and Firms’ Participation in Global Value Chains: Evidence for MENA countries 1 Nada Hazem 2 Chahir Zaki 3 JEL classification: F12, Q56. Keywords: Global value chains; Firms; Environmental stringency UNIT: Chief Economist’s Office of the Middle East and North Africa Region TTL: Ernest John Sergenti 1 The authors gratefully acknowledge the financial and analytical support from the Office of the Chief Economist for the Middle East and North Africa (MNACE) under the Decarbonization and Diversification Research Programs (TTLs: Ernest John Sergenti and Hoda Assem). The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent. The authors are grateful to Giorgia Giovannetti and two anonymous referees for constructive feedback. 2 PhD student, University of Paris 1 Pantheon Sorbonne and Faculty of Economics and Political Science – Cairo University. Email: nada.hazem@feps.edu.eg 3 Chaired Professor of Economics, University of Orléans, Laboratoire d’Economie d’Orléans and Economic Research Forum (corresponding author). Email: chahir.zaki@univ-orleans.fr and chahir.zaki@feps.edu.eg 1. Introduction The Middle East and North Africa (MENA) region 4 stands among the most vulnerable areas to the impacts of climate change. The region’s ecosystem is increasingly facing a serious number of problems such as droughts, heatwaves and rising sea levels (Lionello et al., 2014). Being highly-dependent on climate-sensitive agriculture and having a significant share of population residing in coastal areas, the MENA region is threatened by severe repercussions in terms of food security, human health and livelihoods (Ezzeldin, Adshead, and Smith, 2023). Accordingly, a significant number of MENA countries have begun to pursue measures to mitigate climate change and shift towards a low-carbon, resilient economy. Nearly every country in the region has submitted a roadmap outlining its contributions to the global framework established by the Paris Agreement, aimed at fostering a climate-resilient future (Sieghart and Betre, 2018). At the same time, the region’s integration into Global Value Chains (GVCs) remains modest. Based on Ayadi et al. (2020), the MENA region possesses an untapped potential to make substantial contributions to GVCs given its favorable geographical location, proximity to main markets, resource endowments, and abundant human capital. They also mention that integrating GVCs could play a vital role in stimulating the region's exports, enhancing employment opportunities, as well as boosting firms’ growth and productivity. Accordingly, improving MENA countries’ insertion into the global networks of production is crucial for strengthening economic resilience and fostering competitiveness in the region. Indeed, transitioning towards more sustainable and diversified economies could create both opportunities and challenges for the MENA region’s integration into GVCs. On the one hand, more diversification and lower reliance on fossil fuels could increase the region’s ability to integrate into more sophisticated and higher-value-added activities within the GVC (Fayek and Zaki, 2023). Besides, adherence to international environmental standards and certifications is highly attractive to green investments and companies seeking to align their position with the leading international norms (Wang et al., 2021). This becomes even more important as several export destinations start imposing restrictions on the carbon content of exports. This applies, for instance, to the European Union (EU) that recently imposed the Carbon Border Adjustment Mechanism (Fontagné and Schubert, 2023). The latter aims at imposing a fair price on the carbon emitted during the production of polluting goods such as chemicals, cement, steel, etc., which are mainly goods where several developing countries (including MENA) have a comparative advantage at. Hence, while the EU is an important trade partner for the MENA region, imposing stringent environmental regulations will oblige firms to innovate and upgrade their export structure to abide by such new standards, in line with the Porter Hypothesis (PH). At the same time, compliance costs to strict environmental regulations may be significantly high that they hinder domestic producers from integrating GVCs, by decreasing their competitiveness (Yu and Choi, 2024). Against this background, considering the importance of fostering GVC integration while simultaneously upholding environmental commitments, this paper examines how more stringent environmental regulations affect firms’ GVC participation in the MENA region. Thus, the contribution of the proposed research is threefold. First, the study aims to examine the validity of the PH in the context of different countries, with a special focus on the MENA 4 In this paper, we employ a slightly different definition of the MENA region than that used by the World Bank Group. 2 region compared to the rest of the world. To the best of our knowledge, this is the first study to shed light on this intricate relationship in the MENA region. Second, the paper uses two proxies for environmental stringency at the external and the internal levels, namely domestic environmental regulations and environmental treaties, relying on a comprehensive database for environmental law and policy. Third, the study adopts a firm-level analysis, drawing upon different measures of GVC participation previously developed by Dovis and Zaki (2020) and Urata and Baek (2020). This novel approach allows for a deeper understanding of the dynamics between environmental regulations and GVC integration by acknowledging the heterogeneous behavior of firms and the variations in their responses to environmental stringency. Our main results show that national environmental regulations increase the likelihood of integrating into GVCs when it is measured using both the simple and the strict definitions. This result highlights the role of such regulations in attracting GVC in developing countries and thus lends support to the Porter Hypothesis. This finding is also confirmed when we introduce research and development spending, whose impact on GVC is amplified when interacted with environmental regulations. Yet, the results are less conclusive for the role of environmental treaties which might not be fully enforced or vague. The results remain qualitatively similar when we use a mixed multilevel approach and when we drop top exporters. In addition, at the sectoral level, national regulations are associated with higher GVCs in the food sector and lower in the plastics one. Finally, both SMEs and large firms benefit from environmental stringency, with SMEs particularly gaining when they hold foreign certification and have foreign ownership. The remainder of this paper is organized as follows. Section 2 reviews the literature and presents the theoretical framework. Section 3 introduces the data and presents some stylized facts on environmental stringency and GVC participation in MENA. Model specification and estimation methodology are presented in Section 4. Section 5 discusses the main empirical results. Section 6 concludes and provides some policy recommendations. 2. Literature Review The effect of environmental regulations on international trade has been heavily addressed in the academic literature within the context of two main hypotheses: the Pollution Haven Hypothesis (PHH) and the Porter Hypothesis (PH). According to the PHH, environmental regulations result in higher production costs for domestic producers, negatively impacting their international competitiveness and hence their exports. In response to this situation, companies may choose to relocate pollution-intensive industries to countries with less strict environmental regulations, where lower compliance costs enable them to gain a competitive advantage and become pollution havens (Copeland and Taylor, 1994). The empirical literature on the validity of the PHH was inconclusive. Some studies failed to provide evidence for the PHH (Van Beers and Vanden Bergh, 1997; Harris Konya, and Matyas, 2002; Cole and Elliott, 2003). Other studies provided support to the PHH by detecting a negative effect of environmental stringency on exports and a positive effect on imports of dirty industries (Wilson, Otsuki, and Sewadeh, 2002; Ederington and Minier, 2003; Ederington, Levinson, and Minier, 2005; Jug and Mirza, 2005; Levinson and Taylor, 2008; Kellenberg, 2009; De Santis, 2012; Dou and Han, 2019, and Nunez-Rocha et al., 2024). Conversely, the Porter hypothesis (PH) asserts that well-designed environmental regulations can improve the competitiveness of domestic producers by stimulating their innovative capacities and encouraging them to invest in technologies that are both greener and more 3 efficient. In other words, the hypothesis posits that gains from environmental policies’ enforcement can compensate and may even exceed the required compliance costs (Porter, 1991; Porter and van der Linde, 1995). In fact, a positive effect of environmental regulations on European exports was detected by Costantini and Mazzanti (2012) and Tsurumi, Managi, and Hibiki (2015). Similar results were found in the context of developing countries by Wang, Zhang, and Zeng (2016), Ramzy and Zaki (2018), and Qiang et al. (2022). All of the aforementioned studies primarily focused on analyzing the impact of environmental regulations on gross trade. A new strand of literature has shifted towards examining the validity of the PHH and the PH using data on trade in value added. This new approach accounts for the growing fragmentation of production across various countries and the rise of GVCs. For instance, Duan, Ji, and Yu (2021) show that the larger the difference in environmental stringency between exporting and importing countries, the more pollution-intensive the value- added embodied in exports. In addition, Huang (2020) detected a U-shaped relationship between environmental regulation intensity and the domestic value-added rate of industrial Chinese exports, indicating that the negative cost effect dominates at first before being surpassed by the innovation effect. In the same context, Wang et al. (2021) show that environmental regulations have upgraded the GVC position, measured by the difference between forward and backward participation 5, of China's industrial sector. The main channel through which the PH can affect trade in gross or value added is increased R&D investment that constitutes an important mediation channel of this effect (Dangelico, 2016; Doran and Ryan, 2016; Liu et al., 2022; and Zhao et al., 2022). Likewise, Yuan et al. (2023) find that environmental stringency increases the upstreamness, distance between intermediates and final products, of Chinese manufacturing enterprises in GVC as it allowed them to engage in higher value-added activities through enhanced innovation and research capabilities. Several reasons explain why environmental regulations matter for R&D (Clausen et al., 2022). First, based on the endogenous growth theory (Romer, 1990), policy interventions are needed to boost R&D that is a source of innovation. Environmental regulations can be considered as a policy tool aiming at protecting a public good, which is the environment. Second, regulations create pressure that leads the firm to reduce the costs and the compliance requirements, and thus, firms will have to innovate. Third, and a consequence of the first two points, R&D helps optimizing factors of production and reducing information asymmetry, which improves the firms’ performance, and thus can increase its participation into GVCs. Finally, from a GVC and trade openness perspective, the local and global competition will create more incentives to innovate. At the firm level, more recent studies focus on the nexus between environment friendly measures and GVC integration, confirming the PH hypothesis. For instance, Fayek and Zaki (2023) and Paschoaleto and Martínez-Zarzoso (2024) show that investing in green investments affect the likelihood of participation into GVC, not its intensity. In other words, it affects GVCs measured at the extensive not the intensive margin, pointing out that they are perceived as a fixed cost. Conversely, Siewers et al. (2024) argue that firms that are part of a GVC are more likely to invest in green measures as they are generally more productive. 5 GVC forward participation refers to the country’s domestic value added embodied in exports of third countries, whereas GVC backward participation refers to the foreign value added embodied in the country’s exports. 4 Against this background, we contribute to this literature by examining the impact of environmental stringency proxied by national regulations and international treaties on GVC participation, measured at the firm level. 3. Data and Stylized Facts To measure environmental stringency, we rely on the Ecolex dataset that includes information on treaties, international soft-law and other non-binding policy, national legislation, and judicial decisions related to the environment. The study focuses on the adopted environmental regulations and the signed environmental treaties by different countries. Data on environmental regulations are available between 1935 and 2021, whereas data on environmental treaties are available till 2020. We focus in this paper on the total number of laws and treaties for each country 6. As for firm participation in GVCs, firm-level pooled data from the World Bank Enterprise Surveys (WBES) are used. These surveys encompass a wide array of topics concerning the business climate and environment, gathering data from 168,057 firms across 146 developing economies and emerging markets including the MENA region. The surveys primarily focus on the manufacturing and services sectors and were conducted over a span of time ranging from 2006 to 2021. The objective of this section is to understand the patterns of both environmental stringency and GVC integration, with a special focus on the MENA region. 2.1. Environmental stringency in the MENA region There are notable differences in the number of environmental regulations and treaties among different regions. For instance, Europe and Central Asia (ECA) leads in both categories, with an average of 642 environmental regulations per country and 78 distinct environmental treaties adopted between 2006 and 2021, reflecting a robust regulatory framework and a strong commitment to international cooperation. Latin America and Caribbean (LAC) follows closely, with an average of 623 regulations and 45 treaties. As for the MENA region, it exhibits low counts compared to the rest of the regions in terms of environmental regulations. However, it performs better at the level of environmental treaties, coming in third place after ECA and LAC, with 41 environmental treaties (Figure 1). This can help understand why environmental treaties might not be very effective in the MENA region. Indeed, as they are not complemented by enforceable national laws that guarantee the implementation of international agreements, the latter will not play a role in improving the environmental performance of MENA countries. 6 The main topics covered in the database are laws and treated related to agriculture and rural development, air and atmosphere, cultivated plants, energy, environment, fisheries, food and nutrition, forestry, general, land and soil, legal questions, livestock, mineral resources, sea, waste and hazardous substances, water, wild species and ecosystems. 5 Figure 1. Number of environmental regulations and treaties by region (2006-2021) (a) Environmental regulations (b) Environmental treaties ECA 642 ECA 78 LAC 623 LAC 45 NA 409 MENA 41 EAP 169 EAP 31 SSA 156 SSA 28 MENA 150 NA 20 SA 57 SA 12 0 200 400 600 800 0 20 40 60 80 100 Notes: Panel (a) reports the average number of environmental regulations per country, calculated by dividing the total number of regulations in each region by the number of countries in that region. Panel (b) reports the total number of distinct environmental treaties signed between 2006 and 2021. Each treaty is counted only once per region, regardless of how many countries within the region are signatories, to avoid double-counting. ECA: Europe and Central Asia; LAC: Latin America and Caribbean; SSA: Sub-Saharan Africa; EAP: East Asia and Pacific; MENA: Middle East & North Africa; NA: North America; SA: South Asia. Source: Authors’ own elaboration using Ecolex. When looking at the time evolution of the number of environmental regulations and treaties in the MENA region, there is a clear upward trend in both categories over the 2006-2021 period (as shown in Figure 2). The number of environmental regulations steadily rises from 252 in 2006 to 3,149 in 2021, indicating a gradual expansion in the environmental regulatory frameworks over time. This goes in line with the trend observed in other world regions, which also experienced a consistent rise in the number of environmental regulations over the same period, reflecting a global shift towards stronger environmental governance and compliance (see Figure A1 in appendix). Yet, despite this, it is important to recall that the number of regulations represents 14% of that in Latin America and 45% of that in Sub-Saharan Africa. This was confirmed by Abou-Ali (2012) who argues that, while several MENA countries have established a ministry (or other institutions) to deal with environmental issues, the latter are still neglected and are not streamlined in economic policies or public priorities. This is also in line with Omojolaibi and Nathaniel (2020) who show that environmental regulations in MENA did not reach the optimal level where it can exert a positive impact on environmental outcomes. The count of environmental treaties in MENA exhibits consistent growth, climbing from 18 in 2006 to 40 in 2015, then experiencing almost stagnation for the rest of the period. This pattern was also observed in other regions, despite the different starting points (see Figure A2 in the Appendix). For instance, in 2006, Europe and Central Asia started with a number of treaties that is higher than the number achieved by some regions by the end of the period such as North America and South Asia. However, other world regions, including MENA, are starting to catch up by demonstrating a commitment to signing more environmental treaties over the years. 6 Figure 2. Evolution of the number of adopted environmental regulations and treaties in the MENA region (2006-2021) 3500 40 40 40 40 40 41 45 38 39 39 36 37 40 3000 35 32 35 2500 29 30 2000 25 18 1500 20 15 1000 10 500 5 0 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 Environmental regulations Environmental treaties Notes: Figure shows the cumulative sum (stock) of environmental regulations and treaties adopted by all MENA countries each year. Source: Authors’ own elaboration using Ecolex. To examine the differences in environmental stringency among MENA countries, Figure 3 depicts the number of environmental regulations and treaties adopted by each country in the region. Algeria emerges as the country with the highest number of environmental regulations in the region at 663 regulations, followed closely by Malta (612 regulations), then Morocco (466 regulations). Tunisia and Egypt also demonstrate significant efforts, with 408 and 243 regulations, respectively. However, some countries exhibit lower counts, such as Saudi Arabia, Israel, Yemen, and Iraq, which have fewer than ten regulations each, as shown in Figure 3a. This is not surprising given that most of these economies are oil rich and have a comparative advantage in oil industries. As for the distribution of environmental treaties across countries in the MENA region, it also varies significantly (Figure 3b). Syria leads with 17 signed environmental treaties, while Egypt and Iran come in second place with 9 treaties each. On the other hand, countries like Kuwait and Palestine have only signed one environmental treaty between 2006 and 2021. In terms of the regulations and treaties content, the MENA region has a diverse range of environmental regulations and treaties, covering various aspects of the environment. The most common environmental regulations fall under the categories of cultivated plants (590) and food & nutrition (537), showcasing a significant focus on food security in the region. Whereas, environmental treaties are particularly focused on categories such as environment genetics (11), energy (8), and agricultural and rural development (8). 7 This analysis shows that, while the MENA region is comparable to other emerging and developing regions in terms of environmental treaties, it performs worse in terms of national legislation. Thus, from a de jure perspective, it is lagging behind. In addition, from a de facto perspective, the deficiency of its institutions adds another layer of complication to the enforcement of such laws or regulations. 7 For the content of environmental treaties, we re-conducted the analysis after excluding Syria, and there was no significant change in the sub-categories order. 7 Figure 3. Number of environmental regulations and treaties, by MENA country (2006-2021) (a) Environmental regulations (b) Environmental treaties Algeria 663 Syria 17 Malta 612 Egypt 9 Morocco 466 Iran 9 Tunisia 408 Morocco 8 Egypt 243 UAE 6 Lebanon 180 Libya 6 Oman 104 Malta 5 UAE 92 Tunisia 5 Djibouti 92 Jordan 5 Bahrain 66 Yemen 5 Jordan 48 Djibouti 4 Syria 42 Saudi Arabia 4 Iran 31 Israel 4 Kuwait 27 Algeria 3 Qatar 18 Lebanon 3 Palestine 17 Bahrain 3 Libya 13 Oman 2 Saudi Arabia 9 Qatar 2 Israel 8 Iraq 2 Yemen 7 Kuwait 1 Iraq 3 Palestine 1 0 200 400 600 800 0 5 10 15 20 Source: Authors’ own elaboration using Ecolex. Figure 4. Scope of environmental regulations and treaties in MENA (2006-2021) (a) Environmental regulations (b) Environmental treaties Cultivated plants Environment genetics Food & nutrition Energy Livestock Environment genetics Agri & rural development Fisheries Water Agri & rural development Sea Water Waste & hazardous substances Cultivated plants Land & soil Wild species & ecosystems Wild species & ecosystems Livestock Sea Energy Forestry Air & atmosphere Fisheries Forestry Air & atmosphere Mineral resources General Waste & hazardous substances 0 200 400 600 800 0 2 4 6 8 10 12 Notes: One environmental regulation/treaty can be classified under more than one category. Source: Authors’ own elaboration using Ecolex. 8 2.2. Firms’ GVC participation in the MENA region In order to identify firms participating in GVCs, the paper follows the definitions outlined by Dovis and Zaki (2020). The broadest definition, termed GVC1, encompasses firms engaged simultaneously in both exporting (either directly or indirectly) and importing activities. Moving towards stricter definitions, GVC2 includes firms that are exporters and importers while also possessing an internationally recognized quality certification. Similarly, GVC3 incorporates firms that are two-way traders and have a share of their capital owned by a foreign entity. The most stringent definition, GVC4, combines all four criteria (exporting, importing, international certification, and foreign ownership), representing the highest level of involvement in GVCs. All these variables are binary where they take the value of 1 if the firm is part of a GVC and zero otherwise. Later in the paper, the focus will be mainly accorded to the basic and most advanced definitions of GVC participation (GVC1 and GVC4) for two main reasons. First, they show the two extremes (lowest and highest) of a firm’s integration into a GVC. Second, they help understand the main differences related to the impact of environmental stringency on the depth of GVC. Table 1 displays the share of firms that could be classified as integrated into a GVC within each MENA country 8, according to the abovementioned definitions 9. The table shows how the portion of firms participating in GVCs decreases as the stringency of the definition increases. Given that GVC4 represents the most stringent and comprehensive definition, it consistently applies to a very small share of firms within the country. At the country level, Tunisia exhibits the highest GVC participation across all MENA countries. Around 34.4% of Tunisian firms engage in two-way trading activities (GVC1), whereas 4.7%, in addition to being two-way traders, hold an international certification and have foreign capital investment (GVC4). Jordan, Lebanon, and Israel also demonstrate significant participation. Conversely, countries like Iraq and Yemen show comparatively lower participation, suggesting potential challenges (as they are conflict-affected countries) in accessing and integrating into GVCs. Table 1. Share of firms integrated into GVCs, by MENA country MENA country GVC1 GVC4 Djibouti 13.2 0.0 Egypt 6.0 0.4 Iraq 2.3 0.1 Israel 29.1 3.5 Jordan 27.5 1.2 Lebanon 29.0 0.6 Malta 21.6 3.1 Morocco 23.8 0.9 Palestine 24.3 0.0 Tunisia 34.4 4.7 Yemen 4.0 0.3 Average MENA 19.6 1.4 Notes: Figures in the table represent the percentage of firms engaging in GVCs within each country. Sampling weights are applied. Source: Authors’ own elaboration using WBES. 8 See Table A1 in Appendix for details on the available survey years for MENA countries. 9 See Table A2 in Appendix for GVC2 and GVC3 results. 9 When distinguishing between firms of different sizes, Table 2 shows that large firms have a higher potential to be integrated into GVCs compared to small and medium enterprises (SMEs), which is consistent with a large body of the literature (Antras and Chor, 2022). Large firms often benefit from economies of scale, which allow them to produce goods and services at lower costs. This efficiency enables them to compete more effectively in global markets and make them more attractive to foreign investment. At the same time, large firms are more likely to have the capacity and resources to meet the required compliance standards for obtaining international certification. This is why, a higher share of firms within the large category are identified as integrated into a GVC at both the basic and strict definitions. Table 2. Share of firms integrated into GVCs in MENA countries, by firm size GVC1 GVC4 MENA country SMEs Large SMEs Large Djibouti 13.2 0.0 0.0 0.0 Egypt 4.9 24.6 0.2 3.5 Iraq 2.3 0.0 0.1 0.0 Israel 26.7 59.5 2.3 18.3 Jordan 24.8 50.8 1.0 3.5 Lebanon 28.0 45.5 0.5 2.3 Malta 19.0 52.0 2.4 11.1 Morocco 21.0 47.0 0.5 3.7 Palestine 24.3 30.3 0.0 3.2 Tunisia 27.5 75.7 3.5 12.1 Yemen 2.2 47.3 0.0 8.1 Average MENA 17.6 39.3 1.0 6.0 Notes: Figures in the table represent the percentage of firms engaging in GVCs within each size category in each country. Sampling weights are applied. Source: Authors’ own elaboration using WBES. To measure the depth of firms’ GVC participation (intensive margin), we follow Urata and Baek (2020). Thus, the paper constructs two indices based on the most lenient and the strictest definitions of GVC participation (GVC1 and GVC4). The first index captures the participation depth of two-way trading firms (GVC1) by multiplying the share of exports in total sales by the share of imported intermediates in total inputs, as follows: = ( )*( ) The second index is a score that measures the participation depth of GVC4 firms. To construct the index, we rely on a principal component analysis (PCA) that accounts for the share of exported products, the share of intermediate inputs, and the share of foreign ownership 10. Figure 5 displays the extent to which GVC1 firms engage in GVCs, on average, in each of the MENA countries. Djibouti seems to have the highest average depth of GVC integration in the region, despite showing moderate figures in GVC participation at the extensive margin. Malta and Tunisia follow closely, demonstrating strong GVC participation at both the intensive and extensive margins. On the other hand, Iraq remains at the lower end in terms of both the 10 The fourth dimension, international certification, is a dummy that already takes 1 for all GVC4 firms. 10 likelihood and the depth of GVC participation. The previous analysis confirms that there is some heterogeneity across MENA countries at both the intensive and the extensive margins of GVC. Regarding the diversified economies of the MENA region, the best one is Tunisia, followed by Jordan, Morocco then Egypt. This confirms that, despite their potential, their investment climate, trade policy and institutional setting do not help trading firms, especially in the case of Egypt (see Aboushady and Zaki (2019) for Egypt and for Ruckteschler et al. (2022) for Morocco), which affects their integration into GVCs. Figure 5. Average depth of firms’ GVC integration of MENA countries, basic definition 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Djibouti Malta Tunisia Palestine Jordan Israel Morocco Yemen Lebanon Egypt Iraq Notes: Displayed figures measure GVC participation at the intensive margin (on average) in each MENA country. The intensive margin is calculated by multiplying the share of exports in total sales by the share of imports in total inputs. Sampling weights are applied. Source: Authors’ own elaboration using WBES. After examining these different patterns for both GVC and environmental regulations, the next section presents the methodology we use to examine the nexus between these two variables. 3. Methodology The impact of environmental stringency on firms’ integration into GVCs is examined following two steps. First, a pooled Ordinary Least Squares (OLS) model is estimated as follows: = + + + + + + where GVC is measured using several dimensions at both the intensive and the extensive margins. The latter is measured following Dovis and Zaki (2020) through different dummy variables that measure export status, import status, international certification, and type of ownership. The first definition (GVC1) is the most lenient as it encloses firms that are exporters and importers at the same time. The second definition (GVC4) is stricter as it combines the four dimensions together: firms that are simultaneously exporters and importers, that also have an international certification and a foreign ownership of its capital. At the intensive margin level (Urata and Baek, 2020), GVC measures the participation depth through an index that includes the share of exported products, the share of intermediate inputs, and the share of foreign ownership. The subscripts i, j, c, t, and g denote firm, sector, country, year, and city respectively. is a vector that includes firms’ characteristics that are expected to affect GVC participation such as firms’ age, size and share of government ownership. is the difference between the year in which the most recent survey is released and the year in which the establishment began operation. is the share of government ownership that is likely to reduce the integration of a firm into GVCs (unless this makes the firm politically connected). is a categorical variable that takes the value 1 for small firms, 2 for 11 medium firms and 3 for large ones. Large firms get advantage of their size and engage in economies of scale, which, in turn, allows them to enjoy lower costs of production and might be more likely to integrate into GVCs. is the error term. Envijct measuring environmental stringency is introduced through two variables: the number of domestic environmental regulations and the number of environmental treaties signed by each country. Given that we pool data for different countries and years, we include year, country, city, and sector fixed effects ( , , and ) to control for unobservables. is the disturbance term. All the regressions are run for the whole sample and then environmental stringency variables are interacted with the MENA dummy. Several empirical remarks are worth mentioning. First, our estimations are run using a pooled Ordinary Least Squares (OLS) estimation method (when the dependent variable is continuous, as in the intensive margin) and a Linear Probability Model (when the dependent variable is binary, as in the extensive margin). Yet, given that we observe the GVC intensity variable only when the probability of participating into a GVC is equal to 1, GVC firms might not be randomly selected and the data is, thus, truncated. This is why we proceed with a Heckman selection model where the firm’s productivity is used as a selection variable that explains the first stage (GVC participation), not the GVC intensity (second stage) in line with the Melitz (2003)’s model. A similar approach was also followed by Karam and Zaki (2020). Moreover, we empirically test our choice of selection variable and show that productivity is a significant determinant of a firm’s likelihood to participate in a GVC but has no significant effect on the intensity of GVC participation. Finally, as we merge a macro variable (environmental stringency) with firm-level data, we cluster errors by country and year. We extend the analysis in several ways. First, to explicitly test the Porter Hypothesis, we introduce a variable measuring R&D and interact it with environmental regulations. We expect that the interaction term should be positive. Second, we introduce a measure of the quality of institutions as a proxy for the implementation of such environmental regulations. Indeed, the higher the quality of institutions, the more likely both treaties and regulations will be implemented. The third extension pertains to examining how firm size and sectoral characteristics matter. This is why we examine the heterogeneous impact of environmental stringency on small and medium (SME) vs. large enterprises as the latter are more likely to have the relevant resources to abide by stringent regulations. We also examine how the results change by sector as some economic activities (such as chemicals, plastics and non-metals) might be more polluting than others (such as food or textiles). Finally, as a robustness check, we proceed in two ways. We first run a multilevel analysis. The final sample includes variables defined at two different levels: firm-level variables and the newly added environment indicators that capture country characteristics. In this situation, performing a regression analysis, ignoring the hierarchical structure of data, e.g. simply adding the environment indicators in OLS estimations, produces biased estimates (Burstein et al., 1978; Aitkin and Longford, 1986). Specifically, Moulton (1990) shows that including aggregate variables in micro-level OLS estimations results in downward biased standard errors. However, clustering alone may not solve the issue (Cheah, 2009). To avoid bias, we employ a (linear) multi-level model or mixed effects model (Snijders, 2011; Rabe-Hesketh and Skrondal, 2010; Searle et al., 1992). Our specification includes two different levels: we allow firm participation to depend on firm characteristics (first level), as well as on country-level characteristics, namely environmental stringency (second level). A similar approach has been used, for instance, by Giovannetti et al. (2013) and Ayadi et al. (2024) to investigate how firm- 12 level characteristics and context factors (defined at the province or macro level) affect the the firms’ international behavior. In addition, one can argue that there is a reverse causality between GVC integration and environmental legislation as large firms (that are more likely to integrate into GVC) might lobby for or against environmental regulations (Grey, 2018), which can affect both treaties and regulations. Even if firms in developing countries are less likely to have the power to lobby and influence environmental regulations, we run our baseline regressions by dropping the top 10% of exporting firms (using the share of exports in total sales) which can affect the government decision. 4. Empirical results 4.1. Baseline regressions Our baseline results are presented in Tables 3 and 4 for both the extensive and the intensive margins of GVCs, respectively. In the analysis, we focus on two definitions of GVC: the simple one (two-way traders) and the strict one (two-way traders, foreign ownership, and international certification). For the extensive margin of GVC, first, our controls are in line with the literature as medium and large firms are more likely to be part of GVCs, relative to small ones that are not able to bear the sunk cost of exporting. This result holds for both of the two definitions of GVC at the extensive margin level. In addition, both age and government ownership are not statistically significant. The literature on the impact of age on GVC is not conclusive given that old firms have a long history and might be more experienced to integrate into a GVC. In contrast, new firms might be able to innovate and thus integrate into GVCs. On the other hand, the impact of government ownership is not surprising as, in general, private firms might be more productive and thus more able to integrate into GVCs, especially in developing countries. This is in line with the results of Szarzec et al. (2021) who argue that state-owned enterprises are not positively nor negatively associated with economic growth given that their impact is conditional on the quality of institutions. While developing countries have more deficient institutions, it is not surprising that governmental firms do not improve the integration in to GVCs. As per our variables of interest, Table 3 shows that national environmental regulations increase the likelihood of integrating into GVCs when it is measured using both the simple and the strict definitions. This result highlights the role of such regulations in attracting GVC in developing countries and thus lends support to the Porter Hypothesis, according to which strict environmental regulations can increase efficiency and innovations, leading to a higher competitiveness (Porter, 1991 and Takalo et al., 2021). As per environmental treaties, while they deter GVCs measured by the simple definition, they have an insignificant impact on GVCs in its strict version. These results do not change when each variable is introduced individually or together. The results on environmental treaties might be counterintuitive. Yet, two potential explanations help understand this finding. First, if environmental treaties are vague, countries might have more incentives to sign them while, de facto, these agreements might not be effective in reducing CO2 emissions (Kassab and Zaki, 2020). Second, if environmental treaties are not backed and supported by enforceable national legislations, they will not be 13 implemented 11. Indeed, the role of national legislations is to set the general schemes of protecting the environment and empower the related institutions to oversee and implement the relevant policies. In the absence of such rules, international agreements will not be sufficient. Table 3. Environmental stringency and GVC participation at the extensive margin A. All regions GVC1 GVC4 (1) (2) (3) (4) (5) (6) Ln (Age) 0.009 0.015 0.015 -0.000 0.001 0.001 (0.011) (0.012) (0.012) (0.002) (0.002) (0.002) Medium 0.084*** 0.075*** 0.075*** 0.014*** 0.013*** 0.013*** (0.010) (0.009) (0.009) (0.004) (0.004) (0.004) Large 0.248*** 0.240*** 0.240*** 0.086*** 0.078*** 0.078*** (0.023) (0.025) (0.025) (0.012) (0.012) (0.012) Ln (Gov. own.) -0.003 -0.001 -0.001 0.000 0.000 0.000 (0.013) (0.014) (0.014) (0.004) (0.005) (0.005) Ln (Env. reg.) 0.102*** 0.106*** 0.021** 0.021** (0.034) (0.032) (0.010) (0.010) Ln (Env. treaties) -0.075** -0.084** -0.000 -0.002 (0.038) (0.041) (0.008) (0.008) Constant -0.544** 0.240** -0.375* -0.123* 0.003 -0.120* (0.212) (0.099) (0.226) (0.064) (0.020) (0.069) Observations 111,782 107,025 107,025 111,782 107,025 107,025 B. Interaction with MENA GVC1 GVC4 (1) (2) (3) (4) (5) (6) Ln (Age) 0.009 0.015 0.015 -0.000 0.001 0.001 (0.011) (0.012) (0.012) (0.002) (0.002) (0.002) Medium 0.084*** 0.075*** 0.075*** 0.014*** 0.013*** 0.013*** (0.010) (0.009) (0.009) (0.004) (0.004) (0.004) Large 0.248*** 0.240*** 0.240*** 0.086*** 0.078*** 0.078*** (0.023) (0.025) (0.025) (0.012) (0.012) (0.012) Ln (Gov. own.) -0.003 -0.002 -0.001 0.000 0.000 0.000 (0.013) (0.014) (0.014) (0.004) (0.005) (0.005) Ln (Env. reg.) 0.109*** 0.115*** 0.024** 0.025** (0.036) (0.034) (0.011) (0.011) Ln (Env. reg.) x MENA 0.291 0.629*** 0.120*** 0.051* (0.185) (0.146) (0.040) (0.027) Ln (Env. treaties) -0.078** -0.089** -0.003 -0.005 (0.039) (0.042) (0.008) (0.009) Ln (Env. treaties) x MENA 0.053 -0.124** 0.051*** 0.049*** (0.094) (0.054) (0.018) (0.011) Constant -0.699*** 0.235** -0.687*** -0.187** -0.002 -0.168** (0.262) (0.098) (0.263) (0.073) (0.019) (0.075) Observations 111,782 107,025 107,025 111,782 107,025 107,025 Notes: Robust standard errors clustered at the country-year level in parentheses. *** p<0.01, ** p<0.05, *p<0.1. Country, city, year and sector fixed effects are included. When we interact environmental stringency variables with the MENA dummy (see Table 3b) 12, the baseline results do not change as environmental regulations are positive and environmental treaties are negative. Yet, their interaction with the MENA region gives interesting results. 11 While the enforceability, the content and the coverage differ between agreements, the dataset we rely on does not allow us to take this into account. This point can be subject to further investigation. 12 It is important to note that as we include country dummies, the MENA dummy is dropped from the regression. 14 While the marginal effect of national legislations for the MENA region is positive and statistically significant, that of environmental treaties is negative in the case of the simple definition of GVC (column 3). For the strict definition (column 6), all the results remain qualitatively the same with the exception of the interaction of treaties with the MENA that becomes positive and statistically significant at 1%. Thus, to be deeply integrated in GVCs, international treaties and national legislations might complement each other. This is due to the fact that, with the proliferation of environmental standards in different countries (such the Carbon Border Adjustment Mechanism) and of environmental provisions in trade agreements, developing economies might be obliged to adopt cleaner production techniques to integrate GVCs. This will be particularly the case for the MENA region as the EU is one of its major trade partners. Indeed, Maliszewska et al. (2023) show that, in the MENA region, the most exposed products to CBAM are fertilizers from Egypt, cement from Tunisia, Saudi Arabia, Qatar, and Kuwait; and iron and steel from Morocco. Clearly, if these countries would like to maintain their market shares in the EU or integrate in GVCs, they will have to upgrade their production techniques. At the intensive margin level, Table 4a and b show that, while the control variables exhibit the similar effects as in the extensive margin, most of the environmental stringency variables are statistically insignificant for both the simple and the strict definitions of GVCs, whether they are interacted with the MENA dummy or not 13 (when they are introduced simultaneously in columns 3 and 6). These results point out to what extent such environmental regulations can be perceived as a fixed cost to be part of GVC. Indeed, while they affect the extensive margin, they do not exert any effect on the intensive one. This finding is in line with the New-New Trade literature, as investments in cleaner production technologies to abide by these regulations are associated to a fixed cost, where only more productive firms can bear it (Girma et al., 2008 and Delmas and Pekovic, 2012). Such costs include the reshaping of the production organization; training employees to adapt to these new technologies; reducing waste and pollution; and adopting new business models (Hart and Ahuja, 1996 and Fayek and Zaki, 2023). However, the results of the intensive margin should be analyzed with caution as the number of observations drops by more than 70%, potentially leading to a selection bias. In fact, as mentioned before, the intensive margin variable is observed only when the extensive margin variable is equal to one. This is why we control for the selection bias in our dataset by running a two-stage Heckman selection model (see Table 5) with productivity being the selection variable 14 (Melitz, 2003). Hence, we compute a firm’s productivity by taking the ratio of total sales to the number of employees. Our Heckman results show that productivity is positively associated with the probability of integrating into a GVC, while the remaining control variables hold the same sign. This applies to both the simple and strict definitions, as shown in Table 5a. Interestingly, firm age becomes significant once we control for the selection bias (columns 2 and 4), indicating that older firms are more likely to integrate into GVCs. Government ownership also becomes positive and significant, suggesting that it may facilitate GVC participation. However, government ownership has a negligible or even slightly negative effect on GVC intensity. This is similar to other studies arguing that state-owned firms or firms with 13 Only the national legislation variable is positive and significant at 10% when it is introduced individually in both the simple and the strict definition. 14 We show in Table A3 how productivity is only a significant determinant of a firm’s likelihood to participate in a GVC and has no significant effect on the intensity of GVC participation. 15 partial government ownership may more easily integrate into GVCs (as they face lower entry barriers), but they struggle to compete internationally at deeper levels of integration. As per our variables of interest, our results confirm the previous findings where national laws have a positive impact on the GVC integration (at both the intensive and the extensive margin levels) measured by the simple definition. Yet, for the strict definition, both of the regulations and the treaties are positively associated to GVC participation at the extensive margin level, confirming their necessity to be part of deeper value chains. It is important to recall that the difference between the strict and the simple definition of GVC pertains to foreign ownership and international certification that might be more sensitive to such environmental treaties. 15 When our variables of interest are interacted with the MENA dummy, Table 5b shows that similar results are obtained, where the interaction with MENA is not significant. This shows that the general effect of regulations or treaties is neither amplified nor reduced for the MENA region. However, for the strict definition of GVC, the interaction between environmental treaties and MENA is negative. In a nutshell, our findings show that the impact of national regulations on GVC participation, particularly the extensive margin, is positive and rather robust for the whole sample, when the selection bias is controlled for. A 1% increase in environmental regulations is associated with a 3.6% increase in the likelihood of GVC participation under the simple definition and a 0.13% increase under the strict definition (columns 2 and 4 of Table 5a) 16. Yet, the results of environmental treaties are less robust and less conclusive. We now extend this baseline analysis by examining some potential mechanisms and the heterogeneity observed at the firm size and sectoral levels. 15 Regressions were also run for GVC2 and GVC3 definitions (see Table A4 in the appendix). 16 To obtain the marginal effects, the selection equation of the Heckman model was re-estimated using a standalone Probit model, and Stata’s margins command was used to compute the average marginal effect of environmental regulation stringency. 16 Table 4. Environmental stringency and GVC participation at the intensive margin A. All regions Simple definition Strict definition (1) (2) (3) (4) (5) (6) Ln (Age) -0.023*** -0.029** -0.029** -0.040 -0.223*** -0.222*** (0.008) (0.011) (0.011) (0.082) (0.041) (0.041) Medium 0.011 0.034*** 0.034*** -0.047 -0.025 -0.024 (0.015) (0.010) (0.010) (0.095) (0.155) (0.155) Large 0.046** 0.071*** 0.071*** 0.062 0.220 0.218 (0.019) (0.014) (0.014) (0.148) (0.136) (0.135) Ln (Gov. own.) -0.005 -0.006 -0.006 0.067 0.078 0.078 (0.010) (0.010) (0.010) (0.082) (0.075) (0.075) Ln (Env. reg.) 0.058* 0.053* 0.325* 0.262 (0.030) (0.029) (0.190) (0.195) Ln (Env. treaties) 0.032 0.021 -0.142 -0.180 (0.029) (0.028) (0.245) (0.239) Constant -0.090 0.200*** -0.080 -1.933 0.757 -0.609 (0.181) (0.072) (0.166) (1.177) (0.563) (1.137) Observations 26,770 25,044 25,044 3,771 3,558 3,558 B. Interaction with MENA Simple definition Strict definition (1) (2) (3) (4) (5) (6) Ln (Age) -0.023*** -0.029** -0.029** -0.040 -0.223*** -0.222*** (0.008) (0.011) (0.011) (0.082) (0.041) (0.041) Medium 0.011 0.034*** 0.034*** -0.048 -0.026 -0.025 (0.015) (0.010) (0.010) (0.095) (0.155) (0.155) Large 0.046** 0.072*** 0.071*** 0.061 0.219 0.217 (0.019) (0.014) (0.014) (0.148) (0.135) (0.135) Ln (Gov. own.) -0.005 -0.006 -0.006 0.067 0.077 0.078 (0.010) (0.010) (0.010) (0.082) (0.075) (0.075) Ln (Env. reg.) 0.058* 0.050 0.351* 0.297 (0.030) (0.030) (0.188) (0.193) Ln (Env. reg.) x MENA -0.015 -0.014 2.681*** -5.198 (0.098) (0.087) (0.706) (10.547) Ln (Env. treaties) 0.041 0.028 -0.176 -0.224 (0.031) (0.031) (0.244) (0.238) Ln (Env. treaties) x MENA -0.093* -0.051 2.299*** 6.796 (0.048) (0.038) (0.619) (8.742) Constant -0.084 0.192*** -0.062 -2.543** 0.607 -0.226 (0.192) (0.072) (0.172) (1.181) (0.551) (1.850) Observations 26,770 25,044 25,044 3,771 3,558 3,558 Notes: Robust standard errors clustered at the country-year level in parentheses. *** p<0.01, ** p<0.05, *p<0.1. Country, city, year and sector fixed effects are included. 17 Table 5. Heckman model results A. All regions Simple definition Strict definition (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Productivity) 0.083*** 0.032*** (0.004) (0.003) Ln (Age) -0.042*** 0.041*** -0.150*** 0.060*** (0.003) (0.008) (0.030) (0.013) Medium 0.135*** 0.545*** 0.578*** 0.634*** (0.009) (0.013) (0.219) (0.031) Large 0.315*** 1.262*** 1.270*** 1.391*** (0.016) (0.014) (0.443) (0.030) Ln (Gov. own.) -0.007* 0.023** 0.018 0.026* (0.004) (0.011) (0.030) (0.015) Ln (Env. reg.) 0.028** 0.153*** 0.135 0.020*** (0.011) (0.030) (0.095) (0.007) Ln (Env. treaties) -0.064*** -0.189*** -0.094 0.086*** (0.016) (0.039) (0.139) (0.025) Lambda 0.240*** 0.732** (0.017) (0.365) Constant -0.390*** -3.912*** -3.264*** -3.266*** (0.103) (0.196) (1.230) (0.083) Observations 23,248 97,125 3,457 97,120 B. Interaction with MENA Simple definition Strict definition (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Productivity) 0.083*** 0.029*** (0.004) (0.003) Ln (Age) -0.042*** 0.041*** -0.150*** 0.055*** (0.003) (0.008) (0.030) (0.013) Medium 0.135*** 0.545*** 0.624*** 0.637*** (0.009) (0.013) (0.235) (0.031) Large 0.316*** 1.262*** 1.370*** 1.400*** (0.016) (0.014) (0.479) (0.031) Ln (Gov. own.) -0.007* 0.023** 0.020 0.027* (0.004) (0.011) (0.030) (0.015) Ln (Env. reg.) 0.026** 0.145*** 0.148 0.019** (0.012) (0.030) (0.095) (0.008) Ln (Env. reg.) x MENA -0.122 -0.113 -1.021 0.061 (0.078) (0.213) (0.888) (0.045) Ln (Env. treaties) -0.064*** -0.186*** -0.095 0.086*** (0.016) (0.040) (0.139) (0.025) Ln (Env. treaties) x MENA 0.021 -0.119 1.988 -0.239** (0.081) (0.215) (1.542) (0.100) Lambda 0.242*** 0.811** (0.018) (0.393) Constant -0.393*** -3.903*** -3.522*** -3.220*** (0.104) (0.196) (1.307) (0.084) Observations 23,248 97,125 3,457 97,120 Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *p<0.1. Country, year and sector fixed effects are included. 18 4.2. Extensions As was mentioned before, we extend the analysis in several ways. First, to test the validity of the Porter Hypothesis, we introduce a variable that equals 1 if the establishment invested in R&D in the last fiscal year and interact it with environmental regulations. Innovation is indispensable as firms will have to abide by the new regulations, either domestically or internationally, and therefore innovate (Lanoie et al., 2008). Table 6 shows that environmental regulations and R&D independently positively impact GVC participation. Their interaction is also positive and statistically significant at both the intensive and extensive levels, when using the simple definition of GVC participation. This points out how the positive impact of environmental regulations on both the likelihood and intensity of GVC participation is reinforced when firms spend on R&D. Clearly, R&D helps optimize factors of production, reduce information asymmetry, and lower the cost of compliance (Rubashkina et al., 2015). This would in turn increase resource efficiency, and thus can increase the firm’s participation in GVC. As for environmental treaties, they interestingly stimulate GVC participation intensity when firms invest in R&D (Table 6, column 3). The result for international treaties tends to be less significant when both regulations and treaties are included in the same regression (column 6). In contrast, when using the strict definition of GVC participation, the interaction between R&D and treaties is more significant and displays a higher coefficient for both the likelihood and intensity (Table A5). 17 As mentioned earlier, under the strict definition, firms have foreign ownership and hold international certifications, which might be more sensitive to international treaties. It is important to note that, when it comes to the MENA region, its interaction with environmental treaties and national legislations turns to be insignificant when they two variables are introduced simultaneously (column 5 and 6). Thus, there is not additional marginal effect for the MENA, compared to the positive effect implied by environmental regulations, R&D and their interaction. Second, we examine whether the quality of institutions matter in promoting GVCs. The literature on the determinants of firms’ participation in GVCs shows indeed that the quality of institutions (among other macro determinants) matter for GVCs (Urata and Baek, 2020; Alhassan et al. 2021; and Fernandes et al. 2022). Better institutions promote property rights, reduces uncertainty and the cost of investment, which can boost the firms’ productivity and thus its integration into GVCs. Table 7 shows that government effectiveness positively affects the extensive margin of GVC for the simple definition and both the intensive and extensive margin for the strict definition. Clearly, deeper and higher levels of GVC involving complex products will be more sensitive to institutions. In addition, the better the quality of institutions, the more likely environmental regulations will be implemented. This result is similar in MENA regressions. 17 We also ran separate regressions using R&D as an outcome variable. We found that higher environmental regulations are significantly associated with higher likelihood of spending on R&D. On the other hand, environmental treaties showed no significant effect (see Table A6). 19 Table 6. R&D channel – Heckman model results using GVC simple definition A. All regions (1) (2) (3) (4) (5) (6) Intensive Selection Intensive Selection Intensive Selection Ln (Env. reg.) 0.029* 0.250*** 0.034* 0.249*** (0.018) (0.050) (0.018) (0.050) R&D -0.006 0.301*** 0.007 0.436*** -0.015 0.337*** (0.016) (0.047) (0.017) (0.046) (0.020) (0.055) Ln (Env. reg.) x R&D 0.010*** 0.034*** 0.007** 0.036*** (0.003) (0.008) (0.003) (0.010) Ln (Env. treaties) 0.032 -0.012 0.036 0.011 (0.028) (0.079) (0.028) (0.079) Ln (Env. treaties) x R&D 0.022*** 0.017 0.014* -0.026 (0.007) (0.021) (0.008) (0.024) Lambda 0.182*** 0.209*** 0.207*** (0.016) (0.018) (0.018) Constant -0.272** -3.742*** -0.293** -3.438*** -0.320** -3.679*** (0.126) (0.292) (0.128) (0.288) (0.130) (0.293) Observations 20,664 82,094 18,976 77,564 18,976 77,564 B. Interaction with MENA (1) (2) (3) (4) (5) (6) Intensive Selection Intensive Selection Intensive Selection Ln (Env. reg.) 0.026 0.249*** 0.029 0.239*** (0.018) (0.051) (0.019) (0.052) R&D -0.006 0.301*** 0.007 0.436*** -0.015 0.337*** (0.016) (0.047) (0.017) (0.046) (0.020) (0.055) Ln (Env. reg.) x R&D 0.010*** 0.034*** 0.007** 0.037*** (0.003) (0.008) (0.003) (0.010) Ln (Env. reg.) x MENA -0.128** -0.021 -0.105 0.098 (0.061) (0.181) (0.080) (0.227) Ln (Env. treaties) 0.040 0.014 0.038 0.026 (0.029) (0.080) (0.029) (0.081) Ln (Env. treaties) x R&D 0.022*** 0.017 0.014* -0.026 (0.007) (0.021) (0.008) (0.024) Ln (Env. treaties) x MENA -0.126** -0.345** -0.036 -0.187 (0.062) (0.173) (0.081) (0.225) Lambda 0.183*** 0.210*** 0.208*** (0.016) (0.018) (0.018) Constant -0.279** -3.741*** -0.296** -3.437*** -0.324** -3.666*** (0.126) (0.292) (0.128) (0.288) (0.130) (0.293) Observations 20,664 82,094 18,976 77,564 18,976 77,564 Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and country, year and sector fixed effects are included. See Table A5 in the appendix for the results of the strict definition. 20 Table 7. Institutions, Environmental stringency and GVC – Heckman model results A. All regions Simple definition Strict definition (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Env. reg.) 0.031*** 0.157*** 0.112 -0.007 (0.012) (0.030) (0.095) (0.008) Ln (Env. treaties) -0.064*** -0.165*** -0.102 0.008 (0.016) (0.041) (0.140) (0.027) Gov. effectiveness -0.001 0.134*** 0.552*** 0.377*** (0.018) (0.047) (0.185) (0.015) Lambda 0.238*** 0.635** (0.018) (0.279) Constant -0.386*** -3.713*** -2.429** -3.160*** (0.105) (0.207) (0.956) (0.086) Observations 22,821 95,034 3,390 95,029 B. Interaction with MENA Simple definition Strict definition (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Env. reg.) 0.026** 0.142*** 0.120 -0.013 (0.012) (0.031) (0.096) (0.008) Ln (Env. reg.) x MENA -0.054 0.244 -0.791 0.133*** (0.094) (0.265) (0.902) (0.042) Ln (Env. treaties) -0.065*** -0.164*** -0.107 0.013 (0.016) (0.041) (0.140) (0.027) Ln (Env. treaties) x MENA -0.174 -0.883** 1.607 -0.272*** (0.138) (0.395) (1.561) (0.095) Gov. effectiveness -0.007 0.122** 0.538*** 0.388*** (0.018) (0.047) (0.187) (0.016) Lambda 0.240*** 0.630** (0.018) (0.273) Constant -0.395*** -3.712*** -2.442*** -3.150*** (0.106) (0.207) (0.939) (0.087) Observations 22,821 95,034 3,390 95,029 Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and country, year, and sector fixed effects are included. 21 The third extension pertains to examining how firm size matters. This is why we examine the heterogeneous impact of environmental stringency on small and medium (SME) vs. large enterprises. To do so, we split the sample into two sub-samples one for SMEs (defined using the World Bank definition) and one for large firms 18. Tables 8a and b show the results for the extensive margin with and without the interaction with the MENA dummy. Two important findings are worth to be highlighted. First, our previous results hold with national regulations having a positive impact (for the simple and strict definitions) and treaties a negative one (for the simple definition). Second, we find that regulatory stringency significantly increases the probability of GVC participation (extensive margin) for both SMEs and large firms, though the effect is generally stronger for SMEs. This aligns with Hoogendoorn et al. (2015), who highlight that although SMEs may face constraints in greening processes, external pressures— such as regulations—can push them toward more sustainable practices. However, Lynch- Wood and Williamson (2014) argue that while traditional regulations generally drive environmental behavior among SMEs, responses vary significantly across firms, as their capacities and orientations determine their receptiveness to regulation. Notably, the positive effect of environmental treaties on SMEs under the strict definition underscores the role of certification and foreign ownership in helping SMEs comply with environmental treaties. When the environmental variables are interacted with the MENA region, the marginal effect of national regulations is positive and that of treaties is negative for the strict definition in the case of SMEs. The other interactions are not statistically significant, pointing out that the aforementioned results do not differ for this region, compared to the global sample (see Table 8b). Environmental treaties are generally statistically insignificant at the intensive margin level. 18 See Table A7 in the Appendix for further results on SMEs where, instead of splitting the sample, we interact the environmental legislations with the firm size dummy. 22 Table 8. Environmental stringency and GVC – by firm size A. All regions Simple definition Strict definition SMEs Large SMEs Large (1) (2) (3) (4) (5) (6) (7) (8) Intensive Selection Intensive Selection Intensi Selection Intensive Selection m. m. ve m. m. Ln (Env. reg.) 0.022 0.153*** -0.039* 0.033 0.131 0.022** 0.118 0.021** (0.015) (0.035) (0.021) (0.060) (0.205) (0.011) (0.111) (0.010) Ln (Env. treaties) -0.037* -0.153*** -0.038 -0.152** -0.219 0.173*** -0.161 0.026 (0.021) (0.047) (0.027) (0.074) (0.308) (0.038) (0.161) (0.033) Lambda 0.086*** 0.388*** -0.333 1.816*** (0.020) (0.043) (0.528) (0.528) Constant 0.277* -4.131*** -0.349** -1.973*** -0.210 -3.355*** -3.862*** -1.736*** (0.144) (0.246) (0.150) (0.347) (1.780) (0.121) (1.183) (0.111) Observations 12,809 75,525 10,439 21,600 1,015 21,596 2,442 21,596 B. Interaction with MENA Simple definition Strict definition SMEs Large SMEs Large (1) (2) (1) (2) (1) (2) (1) (2) Intensive Selection Intensive Intensive Selection Intensive Selection Selection m. m. m. m. Ln (Env. reg.) 0.020 0.139*** -0.038* 0.045 0.129 0.015 0.144 0.024** (0.015) (0.036) (0.021) (0.062) (0.207) (0.011) (0.119) (0.010) Ln (Env. reg.) x MENA -0.198** -0.200 0.042 0.107 -1.863 0.248*** -0.925 -0.049 (0.101) (0.255) (0.132) (0.393) (1.690) (0.072) (1.067) (0.059) Ln (Env. treaties) -0.040* -0.145*** -0.038-0.155** -0.227 0.173*** -0.165 0.024 (0.021) (0.048) (0.027) (0.075) (0.308) (0.038) (0.170) (0.033) Ln (Env. treat.) x MENA 0.099 -0.234 0.016 0.286 3.078 -0.702*** 2.718 0.055 (0.090) (0.234) (0.183) (0.556) (2.979) (0.167) (2.021) (0.132) Lambda 0.088*** 0.385*** -0.328 1.953*** (0.020) (0.043) (0.568) (0.591) Constant 0.267* -4.115*** -0.346** -1.989*** -0.223 -3.274*** -4.149*** -1.716*** (0.144) (0.246) (0.149) (0.348) (1.871) (0.122) (1.301) (0.112) Observations 12,809 75,525 10,439 21,600 1,015 21,596 2,442 21,596 Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and country, year and sector fixed effects are included. Finally, our fourth extension pertains to the sectoral heterogeneity. Indeed, some economic activities (such as petroleum products, chemicals, plastics, and non-metals) might be relatively more polluting than others (see Figure 6). Bearing in mind that the MENA region is a large exporter of fuel (66% of the region’s merchandise exports) and manufactures compared to other regions, it is important to examine the heterogeneous impact of environmental regulations on different sectors. 23 Figure 6: Industries emissions Lime, Cement, Construction Materials 12.0 Fertilizers 3.5 Residual Petroleum Products 2.7 Inorganic Chemicals 2.3 Non-Ferrous Metals 1.6 Petroleum Products 1.2 Iron and Steel 1.1 Chemical Materials 1.0 Organic Chemicals 1.0 Paper, Paperboard 0.6 Pulp and Waste Paper 0.6 Metals Manufactures 0.2 Wood Manufactures 0.1 Veneers, Plywood 0.1 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 Source: Authors’ own elaboration using Classification Harris et al. (2002). Emissions intensity is from ESA, 2010 measured as metric tons of CO2 per $1000 in constant 2000 dollars. Table 9 shows that for the simple definition measured at the extensive margin 19, the impact of environmental legislations is not significant for food, non-metals, and textile. Yet, they deter GVC in plastics, being a dirty sector but boost GVC in chemicals and other manufacturing sectors for the global sample. The negative impact of treaties is obtained for chemicals, food, and other manufacturing, with a stronger coefficient for chemicals, being among the most polluting sectors (measured by its CO2 emissions, see Figure 6). Yet, while the result of the food sector seems to be counterintuitive, it is important to note that this sector faces a lot of sanitary and phyto-sanitary measures that hinder its insertion into GVCs. Thus, environmental regulations might be capturing the effect of such measures. Finally, the textile sector, being a labor-intensive sector (and thus not highly polluting) is not affected neither by national legislation nor by treaties. Interestingly, Table 9b shows that, when environmental regulations are interacted with the MENA region, they are associated with a positive impact on non-polluting sectors such as food and other manufacturing in GVC. This result is of particular importance as the food sector (agriculture and processed food) has a potential to help the MENA region increase its integration in GVC given the comparative advantage of some countries (such as Egypt, Morocco, and Tunisia) in agriculture exports (AATM, 2023). Yet, as it was mentioned before, better compliance of agricultural products to international standards is a must to increase the competitiveness of these countries. 19 As the number of observations per industry is very low, we do not run the regression for the intensive margin. 24 Table 9. Environmental stringency and GVC participation at the extensive margin, by manufacturing sector (basic definition) A. All regions GVC1 (1) (2) (3) (4) (5) (6) Chemicals Food Non-metals Other Plastics Textiles Manuf. & rubber Ln (Env. reg.) 0.311*** -0.077 -0.275 0.096** -0.211* -0.223 (0.076) (0.098) (0.217) (0.041) (0.113) (0.137) Ln (Env. treaties) -0.227** -0.149*** -0.114*** -0.197 (0.098) (0.053) (0.038) (0.126) Constant -1.269** 0.802 1.557 -0.144 1.309 1.559** (0.492) (0.582) (1.226) (0.246) (0.767) (0.681) Observations 4,175 14,044 3,337 36,119 2,052 11,470 B. Interaction with MENA GVC1 (1) (2) (3) (4) (5) (6) Chemicals Food Non-metals Other Plastics Textiles Manuf. & rubber 20 Ln (Env. reg.) 0.311*** 0.053 -0.275 0.107** -0.211* -0.222 (0.076) (0.066) (0.217) (0.042) (0.113) (0.142) Ln (Env. treaties) -0.227** -0.150*** -0.120*** -0.198 (0.098) (0.052) (0.040) (0.127) Ln (Env. reg.) x MENA 9.242*** 1.250*** 0.067 (0.236) (0.168) (0.473) Ln (Env. treaties.) x MENA -5.800*** -0.265*** (0.357) (0.058) Constant -1.269** -8.176*** 1.557 -0.392 1.309 1.510* (0.492) (0.344) (1.226) (0.254) (0.767) (0.882) Observations 4,175 14,044 3,337 36,119 2,052 11,470 Notes: Some coefficients are dropped in the table because of collinearity, with low number of observations. Robust standard errors clustered at the country-year level in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and country, city and year fixed effects are included. When GVCs are measured in a strict way (see Table 10a and b), similar results are obtained for national legislations in the chemical (positive) and plastics and rubber (negative) sectors, with an insignificant impact of environmental treaties. When these variables are interacted with the MENA region, national legislation has the same positive effect on the food sector, confirming the previous results. It is important to note that in these regressions, we are examining the impact of all environmental legislation and treaties on each sector. However, a further investigation using sector-specific legislation is needed to better elucidate sectoral heterogeneities. 20 Given the limited variability of this variable for the plastic sector, the environmental treaties variable was dropped, especially that country and year fixed effects are included. 25 Table 10. Environmental stringency and GVC participation at the extensive margin, by manufacturing sector (strict definition) A. All regions GVC4 (1) (2) (3) (4) (5) (6) Chemicals Food Non-metals Plastics Textiles Other & rubber Manuf. Ln (Env. reg.) 0.536*** 0.017 0.124*** -1.182*** 0.011* 0.006 (0.101) (0.033) (0.035) (0.076) (0.006) (0.008) Ln (Env. treaties) -0.018 0.000 -0.021*** -0.014* (0.106) (0.019) (0.007) (0.008) Constant -2.941*** -0.081 -0.696*** 7.091*** -0.015 0.010 (0.575) (0.207) (0.190) (0.458) (0.037) (0.053) Observations 4,175 14,044 3,337 2,052 11,470 36,119 B. Interaction with MENA GVC4 (1) (2) (3) (4) (5) (6) Chemicals Food Non-metals Plastics Textiles Other & rubber Manuf. Ln (Env. reg.) 0.536*** 0.033 0.124*** -1.182*** 0.011* 0.007 (0.101) (0.037) (0.035) (0.076) (0.007) (0.008) Ln (Env. treaties) -0.018 0.000 -0.021*** -0.015* (0.106) (0.019) (0.007) (0.008) Ln (Env. reg.) x MENA 0.504*** 0.028 -0.093*** (0.110) (0.043) (0.032) Ln (Env. treaties.) x MENA -0.230 0.040*** (0.143) (0.009) Constant -2.941*** -0.669*** -0.696*** 7.091*** -0.035 0.025 (0.575) (0.186) (0.190) (0.458) (0.054) (0.056) Observations 4,175 14,044 3,337 2,052 11,470 36,119 Country FE Yes Yes Yes Yes Yes Yes City FE Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes Yes Yes Yes Notes: Some coefficients are dropped in the table because of collinearity, with low number of observations. Robust standard errors clustered at the country-year level in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and country, city and year fixed effects are included. 4.3. Robustness As our sample includes variables defined at two different levels: firm-level variables and environmental regulations indicators that capture country-year characteristics, we run a mixed effects multilevel analysis. Thus, we allow firm participation to depend on firm characteristics (first level), as well as on country-level characteristics, namely environmental legislations and treaties (second level). The advantage of this approach is that it helps us avoid the downward biased standard errors. Table 11a and b confirm once again our baseline results where national legislation matters more than environmental treaties, when they are introduced individually or together and whether they are interacted with the MENA region or not. It is important to note that these results hold for the simple definition of GVC, not the strict one where all the environmental regulations turn to be insignificant. Recall from the descriptive analysis that the number of GVC4 firms is quite low in our dataset, and even more in the MENA subsample 21. 21 As the sample size is small for the intensive margin and the results are not comparable, we opted for the extensive margin only in the mixed effects model. 26 Table 11. Mixed multilevel results A. All regions Simple definition Strict definition Ln (Env. reg.) 0.0140** 0.0153*** 0.00199 0.00155 (0.00577) (0.00579) (0.00139) (0.00136) Ln (Env. treaties) -0.000839 -0.0168 0.00605 0.00443 (0.0203) (0.0206) (0.00482) (0.00464) Constant 0.0180 0.0731** 0.0275 -0.00344 0.000104 -0.00454 (0.0393) (0.0368) (0.0405) (0.00884) (0.00795) (0.00938) Observations 111,787 107,030 107,030 111,787 107,030 107,030 B. Interaction with MENA Simple definition Strict definition Ln (Env. reg.) 0.0172*** 0.0170*** 0.00223 0.00171 (0.00609) (0.00596) (0.00156) (0.00151) Ln (Env. reg.) x MENA -0.0448*** 0.00676 -0.00534 -0.00256 (0.0171) (0.0337) (0.00329) (0.00517) Ln (Env. treaties) 0.0104 -0.00533 0.00707 0.00536 (0.0230) (0.0232) (0.00539) (0.00519) Ln (Env. treaties) x MENA -0.152** -0.191* -0.0156 -0.0114 (0.0601) (0.113) (0.0114) (0.0188) Constant 0.00420 0.0668* 0.0152 -0.00463 -0.000513 -0.00583 (0.0396) (0.0368) (0.0403) (0.00911) (0.00804) (0.00966) Observations 111,787 107,030 107,030 111,787 107,030 107,030 Notes: Robust standard errors clustered at the country-year level in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and sector and year fixed effects are included. Finally, to control for the reverse causality between GVC and environmental laws and treaties, we drop the top 10% of exporting firms (using share of exports in total sales). The rationale is that large firms (that are more likely to integrate into GVC) might lobby for or against environmental regulations, which affects the elaboration of new laws or the ratification of international environmental agreements. Tables 12 confirms the baseline results (as in Table 5). Indeed, there is a positive and robust effect of environmental regulations on GVC (measured with the simple and strict versions), especially at the extensive margin level. However, treaties have a negative effect in the case of the simple definition and an insignificant one in the case of the strict definition. 27 Table 12. Results after dropping the top 10% of exporting firms A. All regions Simple definition Strict definition (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Env. reg.) 0.018*** 0.141*** 0.018 0.045*** (0.007) (0.032) (0.073) (0.009) Ln (Env. treaties) -0.024** -0.122*** -0.118 0.041 (0.009) (0.044) (0.098) (0.031) Lambda 0.024** 0.121 (0.010) (0.205) Constant 0.097* -4.071*** -1.488* -3.889*** (0.058) (0.205) (0.780) (0.103) Observations 16,326 88,138 2,041 88,134 B. Interaction with MENA Simple definition Strict definition (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Env. reg.) 0.019*** 0.140*** 0.031 0.046*** (0.007) (0.033) (0.074) (0.009) Ln (Env. reg.) x MENA 0.044 0.353 0.516 -0.067 (0.053) (0.252) (0.877) (0.056) Ln (Env. treaties) -0.024** -0.115*** -0.114 0.039 (0.009) (0.044) (0.098) (0.031) Ln (Env. treaties) x MENA -0.002 -0.340 0.164 0.061 (0.050) (0.244) (1.306) (0.123) Lambda 0.023** 0.113 (0.010) (0.213) Constant 0.098* -4.069*** -1.488* -3.867*** (0.058) (0.205) (0.807) (0.104) Observations 16,326 88,138 2,041 88,134 Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and country, year, and sector fixed effects are included. We relied on the share of exports in total sales to drop the top 10 of exporting firms. 28 5. Conclusion and Policy Implications This paper investigates the effect of environmental stringency on GVC participation in MENA countries using the World Bank Enterprise Surveys. We therefore analyze how environmental regulations and treaties affect both the extensive and the intensive margins of GVCs. Our main results show that national environmental regulations increase the likelihood of integrating into GVCs when the latter is measured using both the simple and the strict definitions. This result highlights the role of such regulations in attracting GVC in developing countries and thus lends support to the Porter Hypothesis. Indeed, we show that these regulations increase the effect of spending on research and development on GVC. Yet, the results are less conclusive for the role of environmental treaties that might be either vague or not implemented. These results remain robust after we control for the selection bias we have in the GVC variable, when we use a mixed multilevel approach that controls for the different levels (macro and micro) and when drop the top 10% exporting firms that can affect government policies related to the environment (and can thus control for the reverse causality between GVC and environmental regulations). In addition, at the sectoral level, national legislations are associated to higher GVCs in the food sector and lower in the plastics one. Finally, when the firm size is taken into account, while large firms seem to benefit from such regulations given their agility to abide by stringent standards, SME can benefit when they have a foreign certification and a foreign ownership (part of their GVC measurement). This topic is of a particular importance as it addresses two important, and interrelated, challenges in the MENA region that are weak environmental regulations and low integration into GVC. Thus, from a policy perspective, several implications can be concluded from our analysis. First, by demonstrating the potential for environmental regulations to foster GVC integration, the paper offers insights for policymakers aiming to balance economic growth with environmental sustainability. Second, if MENA countries adopt stringent environmental measures, they might be able to attract FDI in less polluting sectors, leading to an increase in their GVC integration. This will also help them diversify their economies away from natural resources that are large emitters of CO2. Third, to help firms comply with such environmental regulations and treaties, research and development is a must. This is why governments in the MENA region should adopt more policies that foster green innovation to increase GVC integration. This includes budget allocation for R&D, tax subsidies for R&D, and intellectual property rights. Fourth, finding effective mechanisms to better implement and make national legislation more enforceable is indispensable. Indeed, our results show to what extent treaties are less effective in attracting GVCs given that they might be vague, without specific commitments nor objectives. This is why national legislations are necessary to address climate change issues. They might be complemented with either environmental treaties or environmental provisions in trade agreements. Fifth, SMEs that face several challenges in terms of GVC integration are still unable to adopt and abide by stringent environmental regulations. Thus, greening SMEs agenda in MENA countries and investing in clean infrastructure is thus indispensable (waste management infrastructure, loans to adopt greener production techniques, greener clusters with larger firms). Finally, from a sectoral perspective, these developments might affect their integration into traditional sectors that might be more polluting such as the oil sector, refineries, chemicals, and construction. 29 References [1.] Abou-Ali, H. (2012). Environment and regulations in MENA. Economic Research Forum, pages 1-14. [2.] Aboushady, N., & Zaki, C. (2019). Investment climate and trade margins in Egypt: Which factors do matter? Economics Bulletin, 39(4), 2275–2301. [3.] Aitkin, M., & Longford, N. (1986). Statistical modelling issues in school effectiveness studies. Journal of the Royal Statistical Society. Series A (General), 149(1), 1–43. https://doi.org/10.2307/2981882. [4.] Alhassan, A., Zoaka, J. D., & Ringim, S. H. (2021). Africa as headwaiter at the dining table of global value chains: Do institutions matter for her participation? African Development Review, 33(3), 560-576. [5.] Antràs, P., & Chor, D. (2021). Global value chains (NBER Working Paper No. 28549). National Bureau of Economic Research. https://ssrn.com/abstract=3804547 [6.] Ayadi, R., Giovannetti, G. and Marvasi, E., Zaki, C. (2024) “Trade Networks and the Productivity of MENA Firms in Global Value Chains”, Structural Change and Economic Dynamics, vol. 69, pages 10–23. [7.] Baghdadi, L., Ben Kheder, S., & Arouri, H. (2019). Impact of nontariff measures on firms in Tunisia. Review of Development Economics, 23(4), 1562-1579. [8.] Burstein, L., Linn, R. L., & Capell, F. J. (1978). Analyzing multilevel data in the presence of heterogeneous within-class regressions. Journal of Educational Statistics, 3(4), 347– 383. https://doi.org/10.2307/1164777. [9.] Cheah, B. C. (2009). Clustering standard errors or modeling multilevel data. University of Columbia, 2-4. [10.] Clausen, L. P. W., Nielsen, M. B., Oturai, N. B., Syberg, K., & Hansen, S. F. (2023). How environmental regulation can drive innovation: Lessons learned from a systematic review. Environmental Policy and Governance, 33(4), 364-373. [11.] Cole, M. A., & Elliott, R. J. (2003). Do environmental regulations influence trade patterns? Testing old and new trade theories. World Economy, 26(8), 1163-1186. [12.] Copeland, R., & Taylor, M. (1994). Trade, Growth, and the Environment. Journal of Economic Literature, 42(1), 7–71. [13.] Costantini, V., & Mazzanti, M. (2012). On the green and innovative side of trade competitiveness? The impact of environmental policies and innovation on EU exports. Research policy, 41(1), 132-153. [14.] Dangelico RM (2016) Green product innovation: where we are and where we are going. Business Strategy and the Environment, 25(8):560–576. [15.] De Santis, R. (2012). Impact of Environmental Regulations on Trade in the Main EU Countries: Conflict or Synergy? The World Economy, 35(7), 799–815. [16.] Delmas, M. A., & Pekovic, S. (2012). Environmental standards and labor productivity: Understanding the mechanisms that sustain sustainability. Journal of Organizational Behavior, 34(2), 230–252. [17.] Doran, J., & Ryan, G. (2016). The importance of the diverse drivers and types of environmental innovation for firm performance. Business Strategy and the Environment, 25(2), 102-119. 30 [18.] Dou, J., & Han, X. (2019). How does the industry mobility affect pollution industry transfer in China: Empirical test on Pollution Haven Hypothesis and Porter Hypothesis. Journal of cleaner production, 217, 105-115. [19.] Dovis, M., & Zaki, C. (2020). Global value chains and local business environments: Which factors really matter in developing countries?. Review of Industrial Organization, 57, 481-513. [20.] Duan, Y., Ji, T., Lu, Y., & Wang, S. (2021). Environmental regulations and international trade: A quantitative economic analysis of world pollution emissions. Journal of Public Economics, 203, 104521. [21.] Ederington, J., & Minier, J. (2003). Is Environmental Policy a Secondary Trade Barrier? An Empirical Analysis. Canadian Journal of Economics, 36(1), 137–154. [22.] Ederington, J., Levinson, A., & Minier, J. (2005). Footloose and Pollution-Free. The Review of Economics and Statistics, 87(1), 92–99. [23.] Ezzeldin, K., Adshead, D., & Smith, P. (2023). Adapting to a New Climate in the MENA Region: An assessment of physical risk management and climate adaptation finance in the MENA region. UN Environment Programme Finance Initiative. https://www.unepfi.org/themes/climate-change/adapting-to-a-new-climate-in-the-mena- region/ [24.] Fayek, M., & Zaki, C. (2023). Does "going green" promote global value chains integration? LEO Working Paper Series, DR 2023-21. https://doi.org/10.5281/zenodo.10390660 [25.] Fernandes, A. M., Kee, H. L., & Winkler, D. (2022). Determinants of global value chain participation: Cross-country evidence. The World Bank Economic Review, 36(2), 329- 360. [26.] Fontagné, L., & Schubert, K. (2023). The Economics of Border Carbon Adjustment: Rationale and Impacts of Compensating for Carbon at the Border. Annual Review of Economics, 15(1), 389-424. [27.] Giovannetti, G., Ricchiuti, G., & Velucchi, M. (2013). Location, internationalisation and performance of firms in Italy: A multi-level approach. Applied Economics, 45(18), 2665–2673. https://doi.org/10.1080/00036846.2012.665597. [28.] Girma, S., Hanley, A., & Tintelnot, F. (2008). Exporting and the environment: A new look with micro-data. Kiel Institute for the World Economy Working Paper, No. 1400. [29.] Grey, F. (2018). Corporate lobbying for environmental protection. Journal of Environmental Economics and Management, 90, 23-40. [30.] Harris, M., Konya, L., & Matyas, L. (2002). Modeling the Impact of Environmental Regulations on Bilateral Trade Flows: OECD, 1990–1996. World Economics, 25(3), 387–405. [31.] Hart, S. L., & Ahuja, G. (1996). Does it pay to be green? An empirical examination of the relationship between emission reduction and firm performance. Business Strategy and the Environment, 5(1), 30–37. [32.] Hoogendoorn, B., Guerra, D., & Van Der Zwan, P. (2015). What drives environmental practices of SMEs?. Small Business Economics, 44, 759-781. [33.] Huang, J. (2020). The Impact of Environmental Regulation on the Domestic Value Added Rate of China’s Industrial Exports—Empirical Test Based on Panel Data of 31 Prefecture Level Cities. American Journal of Industrial and Business Management, 10(8), 1437-1453. [34.] Jug, J., & Mirza, D. (2005). Environmental regulations in gravity equations: evidence from Europe. World Economy, 28(11), 1591-1615. [35.] Karam, F., & Zaki, C. (2020). On women participation and empowerment in international trade: Impact on trade margins in the MENA region. The Journal of International Trade & Economic Development, 30(3), 384–406. [36.] Kassab, M., & Zaki, C. (2020). Do environmental treaties reduce CO₂ emissions? Evidence from a dynamic panel approach. Environmental Economics and Policy Studies, 22(4), 519–547. [37.] Kellenberg, D. K. (2009). An empirical investigation of the pollution haven effect with strategic environment and trade policy. Journal of international economics, 78(2), 242- 255. [38.] Kowalski, P., López González, J., Ragoussis, A., & Ugarte, C. (2015). Participation of developing countries in global value chains: Implications for trade and trade-related policies (OECD Trade Policy Papers No. 179). OECD Publishing. https://doi.org/10.1787/5js33lfw0xxn-en [39.] Lanoie, P., Patry, M., & Lajeunesse, R. (2008). Environmental regulation and productivity: Testing the porter hypothesis. Journal of productivity analysis, 30, 121- 128. [40.] Levinson, A., & Taylor, S. (2008). Unmasking the Pollution Haven Effect. International Economic Review, 49(1), 223-254. [41.] Li, J., Ji, J., & Zhang, Y. (2019). Non-linear effects of environmental regulations on economic outcomes, Management of Environmental Quality, 30(2), 368-382. [42.] Liao, Z., & Liu, Y. (2021). What drives environmental innovation? A meta‐analysis. Business Strategy and the Environment, 30(4), 1852-1864. [43.] Lionello, P., Abrantes, F., Gacic, M., Planton, S., Trigo, R., & Ulbrich, U. (2014). The climate of the Mediterranean region: research progress and climate change impacts. Regional Environmental Change, 14, 1679-1684. [44.] Liu, H., Chen, L., & Shan, Y. (2022). Does environmental regulation affect global value chain position in service sectors? Evidence from 41 major economies. Frontiers in Environmental Science, 10, 1051015. [45.] Lynch-Wood, G., & Williamson, D. (2014). Understanding SME responses to environmental regulation. Journal of Environmental Planning and Management, 57(8), 1220-1239. [46.] Maliszewska, M., Chepeliev, M., Fischer, C., and Jung E. (2023) “Trade and Development Chart: CBAM’s impact on exports to the EU”, The Trade Post, The World Bank. [47.] Melitz, M. J. (2003). The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica, 71(6), 1695–1725. [48.] Moulton, B. R. (1990). An illustration of a pitfall in estimating the effects of aggregate variables on micro units. Review of Economics and Statistics, 72(2), 334–338. https://doi.org/10.2307/2109724. 32 [49.] Nunez Rocha, T., Martinez-Zarzoso, I., and Zaki, C. (2024) “What Type of Trade is Promoted by Environmental Provisions in Trade Agreements?”, Annals of Economics and Statistics, No.156 (December 2024), pages 207-236. [50.] Omojolaibi, J., & P Nathaniel, S. (2022). Assessing the potency of environmental regulation in maintaining environmental sustainability in MENA countries: An advanced panel data estimation. Journal of Public Affairs, 22(3), e2526. [51.] Paschoaleto, R. D. L., & Martínez-Zarzoso, I. (2024). Environmental regulations and firms' integration in global markets: Using a new environmental performance index. Empirica, 51(3), 829–876. [52.] Porter, M. (1991). America’s Green strategy. Scientific American, 264(4), 168. [53.] Porter, M. E., & Van der Linde, C. (1995). Toward a new conception of the environment-competitiveness relationship. Journal of economic perspectives, 9(4), 97- 118. [54.] Qiang, O., Tian-Tian, W., Ying, D., Zhu-Ping, L., & Jahanger, A. (2022). The impact of environmental regulations on export trade at provincial level in China: evidence from panel quantile regression. Environmental Science and Pollution Research, 29(16), 24098-24111. [55.] Rabe-Hesketh, S., & Skrondal, A. (2010). Multi-level Modelling (Vols. 1–4). Sage, London. [56.] Ramzy, M., & Zaki, C. (2018). Do environment regulations matter for EU-MENA trade? Applied Economics, 50(39), 4197-4221. [57.] Romer, P. M. (1990). Endogenous Technological Change. Journal of Political Economy, 98(5), S71–S102. [58.] Rubashkina, Y., Galeotti, M., & Verdolini, E. (2015). Environmental regulation and competitiveness: Empirical evidence on the Porter Hypothesis from European manufacturing sectors. Energy policy, 83, 288-300. [59.] Ruckteschler, C., Malik, A., & Eibl, F. (2022). Politics of trade protection in an autocracy: Evidence from an EU tariff liberalization in Morocco. European Journal of Political Economy, 71, 102063 [60.] Searle, S. R., Casella, G., & McColluch, C. E. (1992). Variance Components. Wiley, New York. [61.] Shu, C., Zhao, J., Yao, Q., & Zhou, K. Z. (2024). Green Innovation and Export Performance in Emerging Market Firms: A Legitimacy-based View. Management and Organization Review, 20(1), 85–110. [62.] Sieghart, L. C., & Betre, M. M. (2018). Climate change in MENA: challenges and opportunities for the world's most water stressed. MENA Knowledge and Learning Quick Notes Series, No. 164. Washington, DC: World Bank. [63.] Siewers, S., Martínez-Zarzoso, I., & Baghdadi, L. (2024). Global value chains and firms' environmental performance. World Development, 173, Article 106395. [64.] Snijders, T. A. (2011). Multilevel analysis (pp. 879-882). Springer Berlin Heidelberg. [65.] Szarzec, K., Dombi, Á., & Matuszak, P. (2021). State-owned enterprises and economic growth: Evidence from the post-Lehman period. Economic Modelling, 99, 105484. [66.] Takalo, S. K., Tooranloo, H. S., & Parizi, Z. S. (2021). Green innovation: A systematic literature review. Journal of Cleaner Production, 279, 122474. 33 [67.] Tsurumi, T., Managi, S., & Hibiki, A. (2015). Do Environmental Regulations Increase Bilateral Trade Flows? B.E. Journal of Economic Analysis and Policy, 15(4), 1549-1577. [68.] UNCTAD. (2013). World investment report 2013: Global value chains – Investment and trade for development. United Nations. https://unctad.org/system/files/official- document/wir2013_en.pdf [69.] Urata, S., & Baek, Y. (2020). The determinants of participation in global value chains: A cross-country, firm-level analysis. Asian Development Bank Institute Working Paper Series, No. 1116. [70.] Van Beers, C. & Van den Bergh, J. (1997). An Empirical Multi-country Analysis of the Impact of Environmental Regulations on Foreign Trade Flows. Kyklos, 50(1), 29-46. [71.] Wang, J. C., Jin, Z. D., Yang, M., & Naqvi, S. (2021). Does strict environmental regulation enhance the global value chains position of China's industrial sector?. Petroleum Science, 18(6), 1899-1909. [72.] Wang, Z., Zhang, B., & Zeng, H. (2016). The effect of environmental regulation on external trade: empirical evidences from Chinese economy. Journal of Cleaner Production, 114, 55–61. [73.] Wilson, J.S., Otsuki, T., & Sewadeh, M. (2002). Dirty Exports and Environmental Regulation: Do Standards Matter to Trade? Policy Research Working Paper, No. 2806. Washington, DC: World Bank. [74.] Yu, Z., & Choi, Y. (2024). Is the impact of environmental regulations on the global value chains biased towards technological effects or refuge effects?. Energy Strategy Reviews, 53, 101385. [75.] Yuan, K., Cui, J., Zhang, H., & Gao, X. (2023). Do cleaner production standards upgrade the global value chain position of manufacturing enterprises? Empirical evidence from China. Energy Economics, 128, 107185. [76.] Yusof, N. A., Tabassi, A. A., & Esa, M. (2020). Going beyond environmental regulations—The influence of firm size on the effect of green practices on corporate financial performance. Corporate social responsibility and Environmental Management, 27(1), 32-42. [77.] Zhao, L., Zhang, L., Sun, J., & He, P. (2022). Can public participation constraints promote green technological innovation of Chinese enterprises? The moderating role of government environmental regulatory enforcement. Technological Forecasting and Social Change, 174, 121198. 34 Appendix Figure A1. Evolution of environmental regulations by region (2006-2021) Source: Authors’ own elaboration using Ecolex. Figure A2. Evolution of environmental treaties by region (2006-2021) Source: Authors’ own elaboration using Ecolex. 35 Figure A.3: Exports share of merchandise exports (%) - 2021 100 4 5 4 4 90 13 19 80 25 70 22 60 68 71 46 50 83 40 66 37 30 9 20 13 8 23 10 4 12 13 5 9 0 4 East Asia & Europe & Latin America Middle East & South Asia Sub-Saharan Pacific Central Asia & Caribbean North Africa Africa Agricultural raw materials Food Fuel Manufactures Ores and metals Other Source: Authors’ own elaboration using the World Development Indicators online dataset. Table A1. List of available MENA countries and corresponding years in the sample MENA country Survey years Djibouti 2013 Egypt 2013, 2016, 2020 Iraq 2011 Israel 2013 Jordan 2013, 219 Lebanon 2013, 2019 Malta 2019 Morocco 2013, 2019 Palestine 2013, 2019 Tunisia 2013, 2020 Yemen 2013 Source: Authors’ own elaboration using WBES. 36 Table A2. Share of firms integrated into GVCs, by MENA country MENA country GVC2 GVC3 Djibouti 1.7 3.5 Egypt 2.3 0.7 Iraq 0.5 0.1 Israel 22.3 4.6 Jordan 12.4 2.4 Lebanon 7.0 0.8 Malta 8.3 3.9 Morocco 2.4 9.9 Palestine 1.8 0.5 Tunisia 12.4 10.1 Yemen 1.1 0.5 Average MENA 6.6 3.3 Notes: Figures in the table represent the percentage of firms engaging in GVCs within each country. Sampling weights are applied. Source: Authors’ own elaboration using WBES. Table A3. Choice of the selection variable Extensive margin Intensive margin Simple definition Strict definition Simple definition Strict definition Ln (Productivity) 0.024*** 0.005*** 0.003 0.009 (0.004) (0.001) (0.005) (0.024) Ln (Age) 0.007 0.001 -0.028** -0.257*** (0.010) (0.002) (0.011) (0.036) Medium 0.076*** 0.012*** 0.035*** -0.135 (0.010) (0.003) (0.012) (0.144) Large 0.236*** 0.077*** 0.070*** 0.155 (0.025) (0.013) (0.015) (0.136) Ln (Gov. own.) 0.000 0.000 -0.010 0.062 (0.013) (0.004) (0.009) (0.079) Ln (Env. reg.) 0.073*** 0.019*** 0.023 -0.011 (0.024) (0.006) (0.022) (0.156) Ln (Env. treaties) -0.111*** -0.008 0.005 0.020 (0.033) (0.008) (0.028) (0.186) Constant -0.384** -0.158*** 0.095 0.492 (0.165) (0.046) (0.159) (0.875) Observations 97,125 97,125 23,246 3,448 Notes: Robust standard errors clustered at the country-year level in parentheses. *** p<0.01, ** p<0.05, *p<0.1. Country, year and sector fixed effects are included. 37 Table A4. Heckman model results for GVC2 and GVC3 definitions A. All regions GVC2 GVC3 (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Env. reg.) -0.027* -0.005 0.134 0.019** (0.016) (0.039) (0.095) (0.008) Ln (Env. treaties) -0.030 -0.067 -0.095 0.085*** (0.024) (0.054) (0.139) (0.025) Lambda 0.102*** 0.715** (0.017) (0.356) Constant -0.058 -4.970*** -3.195*** -3.247*** (0.162) (0.282) (1.197) (0.084) Observations 11,105 97,125 3,457 94,630 B. Interaction with MENA GVC2 GVC3 (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Env. reg.) -0.028* 0.016 0.147 0.017** (0.016) (0.040) (0.095) (0.008) Ln (Env. reg.) x MENA -0.024 0.656** -1.013 0.076* (0.123) (0.301) (0.888) (0.045) Ln (Env. treaties) -0.029 -0.068 -0.096 0.086*** (0.024) (0.055) (0.139) (0.025) Ln (Env. treaties) x MENA -0.039 0.078 1.968 -0.278*** (0.138) (0.328) (1.543) (0.102) Lambda 0.103*** 0.791** (0.017) (0.384) Constant -0.060 -4.998*** -3.439*** -3.198*** (0.162) (0.282) (1.269) (0.084) Observations 11,105 97,125 3,457 94,630 Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and country, sector and year fixed effects are included. 38 Table A5. R&D channel – Heckman model results using GVC strict definition A. All regions (1) (2) (3) (4) (5) (6) Intensive Selection Intensive Selection Intensive Selection margin margin margin Ln (Env. reg.) 0.251 0.022** 0.226 0.011 (0.172) (0.010) (0.171) (0.011) R&D 0.049 0.266*** -0.534*** 0.111 -0.488** 0.089 (0.192) (0.077) (0.171) (0.082) (0.198) (0.094) Ln (Env. reg.) x R&D 0.050* 0.016 -0.011 0.008 (0.028) (0.013) (0.032) (0.016) Ln (Env. treaties) -0.072 0.151*** -0.069 0.138*** (0.267) (0.035) (0.268) (0.037) Ln (Env. treaties) x R&D 0.399*** 0.111*** 0.402*** 0.100** (0.082) (0.036) (0.087) (0.040) Lambda 0.879* 1.000** 0.989* (0.474) (0.499) (0.508) Constant -2.537 -3.250*** -1.705 -3.220*** -2.402 -3.251*** (1.616) (0.097) (1.622) (0.100) (1.737) (0.104) Observations 82,090 2,994 77,560 2,788 77,560 2,788 B. Interaction with MENA (1) (2) (3) (4) (5) (6) Intensive Selection Intensive Selection Intensive Selection margin margin margin Ln (Env. reg.) 0.252 0.025** 0.247 0.011 (0.172) (0.010) (0.173) (0.011) R&D 0.078 0.268*** -0.501*** 0.133 -0.441** 0.119 (0.199) (0.077) (0.179) (0.082) (0.207) (0.094) Ln (Env. reg.) x R&D 0.049* 0.013 -0.012 0.006 (0.028) (0.013) (0.033) (0.016) Ln (Env. reg.) x MENA -0.494 -0.034*** -1.022 0.086* (0.600) (0.007) (0.925) (0.048) Ln (Env. treaties) -0.052 0.158*** -0.055 0.151*** (0.269) (0.035) (0.270) (0.037) Ln (Env. treaties) x R&D 0.394*** 0.095*** 0.401*** 0.085** (0.081) (0.036) (0.087) (0.040) Ln (Env. treaties) x MENA -0.318 -0.077*** 1.120 -0.269** (1.014) (0.016) (1.587) (0.109) Lambda 1.004** 1.242*** 1.154** (0.506) (0.343) (0.548) Constant -2.875* -3.231*** -2.066 -3.203*** -2.956 -3.234*** (1.707) (0.097) (1.732) (0.100) (1.863) (0.104) Observations 82,090 2,994 77,560 2,788 77,560 2,788 Country FE No No No No No No Year FE Yes Yes Yes Yes Yes Yes Sector FE Yes Yes Yes Yes Yes Yes Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *p<0.1. All control variables and country, sector and year fixed effects are included. 39 Table A6. R&D as an outcome variable (1) (2) (3) R&D R&D R&D Ln (Age) -0.004 -0.001 -0.001 (0.004) (0.004) (0.004) Medium 0.066*** 0.057*** 0.057*** (0.012) (0.010) (0.010) Large 0.193*** 0.175*** 0.174*** (0.027) (0.026) (0.026) Ln (Gov. own.) -0.004 -0.009 -0.009 (0.022) (0.019) (0.019) Ln (Env. reg.) 0.057* 0.054* (0.032) (0.032) Ln (Env. treaties) 0.034 0.031 (0.056) (0.054) Constant -0.247 0.009 -0.317 (0.201) (0.141) (0.254) Observations 90,559 85,810 85,810 Notes: Robust standard errors clustered at the country-year level in parentheses. *** p<0.01, ** p<0.05, *p<0.1. Country, year and sector fixed effects are included. 40 Table A7. Alternative regression – Firm size A. All regions Simple definition Strict definition (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Productivity) 0.094*** 0.034*** (0.003) (0.003) Ln (Age) -0.034*** 0.078*** -0.129*** 0.083*** (0.003) (0.007) (0.034) (0.012) Ln (Gov. own.) -0.007* 0.027** 0.013 0.025 (0.004) (0.011) (0.029) (0.015) SME -0.343*** -1.432*** -1.467*** -1.272*** (0.022) (0.038) (0.402) (0.063) Ln (Env. reg.) 0.008 0.108*** 0.100 0.027*** (0.011) (0.030) (0.096) (0.009) Ln (Env. reg.) x SME 0.024*** 0.050*** 0.062** -0.013 (0.003) (0.008) (0.030) (0.013) Ln (Env. treaties) -0.059*** -0.258*** -0.107 0.019 (0.016) (0.042) (0.139) (0.028) Ln (Env. treaties) x SME 0.004 0.122*** 0.144* 0.182*** (0.006) (0.019) (0.085) (0.031) Lambda 0.212*** 0.729** (0.016) (0.342) Constant 0.053 -2.453*** -1.812** -1.859*** (0.096) (0.200) (0.793) (0.086) Observations 97,125 23,248 97,120 3,457 B. MENA region Simple definition Strict definition (1) (2) (3) (4) Intensive margin Selection Intensive margin Selection Ln (Productivity) 0.125*** 0.012 (0.012) (0.015) Ln (Age) -0.056*** 0.035 -0.188 -0.064 (0.009) (0.023) (0.365) (0.045) Ln (Gov. own.) -0.023* 0.044 -0.159 0.043 (0.014) (0.039) (0.298) (0.058) SME 0.017 -0.460* 1.514 -1.065** (0.087) (0.258) (5.805) (0.479) Ln (Env. reg.) -0.141 0.183 -0.427 -0.031 (0.123) (0.351) (1.695) (0.064) Ln (Env. reg.) x SME 0.046*** 0.127** -1.007 0.301*** (0.017) (0.051) (1.611) (0.096) Ln (Env. treaties) 0.288** 0.321 3.883 0.408* (0.146) (0.394) (4.967) (0.214) Ln (Env. treaties) x SME -0.197*** -0.519*** 2.654 -0.670** (0.054) (0.153) (3.714) (0.271) Lambda 0.169*** -2.938 (0.048) (5.859) Constant 0.430 -3.647** -1.812** -1.859*** (0.630) (1.764) (0.793) (0.086) Observations 11,124 2,198 11,123 241 Notes: Robust standard errors in parentheses. *** p<0.01, ** p<0.05, *p<0.1. Country, sector and year fixed effects are included. 41