1. Introduction

This article assesses the relationship between the working conditions of European employees and two major global trends: production fragmentation across borders as reflected in involvement in global value chains (GVCs); and technological progress driven by digital technologies that combine hardware (advanced robots and 3D printers), software (big data analytics, cloud computing and artificial intelligence (AI)) and connectivity (UNIDO 2019, xvi). In particular, the main goal of this article is to investigate whether digitally driven technological factors alter the core relationship between GVCs and different aspects of working conditions (wages and job quality) in developed countries.

There are many determinants of working conditions, including the growth of temporary employment (Aleksynska 2018), characteristics of employment structure within firms (Clark, D’Ambrosio and Zhu 2021), organization of work and the social relations embedded in the labour process (Briken et al. 2017; Gandini 2019; Harley 2018), the use (and abuse) of technologies (Salanova, Llorens and Ventura 2014; Badri, Boudreau-Trudel and Souissi 2018; Brynjolfsson, Mitchell and Rock 2018) and the complexity of value chains (Berliner et al. 2015; Bernhardt and Pollak 2016; Parteka, Wolszczak-Derlacz and Nikulin 2024). In this article, we focus on the interplay between the latter two forces: technological progress and GVCs. The proliferation of GVCs raises concerns about their impact on working conditions (e.g. Barrientos, Gereffi and Rossi 2011; Barrientos et al. 2016; Berliner et al. 2015; Bernhardt and Pollak 2016; Nikulin, Wolszczak-Derlacz and Parteka 2022), the protection of workers’ rights (Delautre, Echeverría Manrique and Fenwick 2021) and social relations at work (Reinecke et al. 2018). Some 70 per cent of global trade involves GVCs,¹ such that changes in business models due to cross-border production fragmentation cannot be neglected. At the same time, GVCs cannot be isolated from the exponential progress in digital technologies, including AI, which has been another powerful force in recent decades (Aghion, Jones and Jones 2019; Hernandez and Brown 2020; Lu and Zhou 2021). It has fuelled the intensification of cross-border production links and the “second unbundling” (Baldwin 2013) of, and interacting with, modern labour markets (Agrawal, Gans and Goldfarb 2019; Acemoglu and Restrepo 2018; Brynjolfsson and Mitchell 2017; Brynjolfsson, Mitchell and Rock 2018; Lane and Saint-Martin 2021; OECD 2023).

We focus on developed economies in Europe because the struggle to ensure conditions of decent work, addressed in international policy measures and strategies such as the United Nations 2030 Sustainable Development Goals, is not restricted to the developing world and problems such as hazardous conditions or child labour (Delautre, Echeverría Manrique and Fenwick 2021). The quality of working life is not only diversified across European industries and jobs (Eurofound 2020), depending on the sex, age, contractual status and occupation of workers (Eurofound 2021) but is also – at least in some respects – globally unsatisfactory. This problem is reflected in the European Union’s current 2030 Agenda, which “calls for opportunities for full employment and decent work for all” (Goal 8 – our own emphasis).² It was also previously reflected in the Treaty on the Functioning of the European Union, the Lisbon Strategy and the Europe 2020 Strategy, which all established improving labour rights as one of the main targets for European labour markets. Still, the 2021 European Working Conditions Telephone Survey (EWCS) (Eurofound 2022) reports that 49 per cent of European workers frequently work at high speed, while 19 per cent said that their job involved being in emotionally disturbing situations. Moreover, “the reality of the changing workplace” results in the growth of psychosocial risk or work intensity and the blurred boundaries between work and non-work life (Eurofound 2021). These problems could be exacerbated by the fact that in many European countries, employment shares have increased in occupations with greater exposure to modern digital technologies, including AI (Albanesi et al. 2023).

A reading of the related literature (see section 2) allows us to identify several specific research gaps to be addressed in our Europe-focused study. First, GVC-focused research has rarely dealt with such aspects of work as occupational safety and health, job satisfaction or job security (Budría and Baleix 2020; Geishecker 2012). Even when the social consequences of GVCs are analysed, they are mainly quantified in terms of their impact on wages or the risk of job displacement (Baumgarten, Geishecker and Görg 2013; Ebenstein et al. 2014; Geishecker, Görg and Munch 2010; Parteka and Wolszczak-Derlacz 2019 and 2020; Shen and Silva 2018; Hummels, Munch and Xiang 2018). These studies do not capture the full complexity of working conditions, including less quantifiable aspects such as the quality of physical working environments, social support and management quality, career development prospects, work–life balance and the impact of work intensity on health and well-being (Eurofound 2021). We investigate the multifaceted nature of working conditions, including the economic dimension of work and its effects on living conditions, by providing estimates using monetary information (wage data) and information on different aspects of job quality.³ Our key underlying assumption is that non-wage job dimensions affect employees in ways that are as important as wages in terms of well-being. Such an approach is in line with the concept of equivalent income used in the well-being literature (e.g. see Decancq, Fleurbaey and Schokkaert 2015; Fleurbaey 2015) or demand-control theory (Karasek and Theorell 1990), linking job demands and job strain with the mental and physical condition of workers. One may postulate that poor working conditions – not including wages – are compensated by higher salaries, but the empirical evidence for this is rather weak (e.g. see Bonhomme and Jolivet 2009; Fernández and Nordman 2009).

Second, studies focusing on job quality have thus far not assessed the role of production fragmentation in the context of the dynamically changing technological landscape. Social scientists seem to be several steps behind the actual advancements in the digital sphere, including the increasing role of AI technologies in the work context (Albanesi et al. 2023; OECD 2023). Digital technologies, including AI, can be quantified at the level of occupations using the task-based approach (Autor, Levy and Murnane 2003; Acemoglu and Restrepo 2018) that views occupations as a bundle of tasks differing in exposure to technology. We can now measure the AI exposure of tasks typical for a given occupation (Webb 2020) and of the abilities that those tasks require (Felten, Raj and Seamans 2018 and 2019) or the “suitability for machine learning” of specific work activities, tasks and occupations (Brynjolfsson, Mitchell and Rock 2018). Such measures have mainly been used to assess the impact of AI on employment patterns in the United States (Acemoglu et al. 2022) or in Europe (Albanesi et al. 2023). While there is rich evidence about the effects of information and communications technology or automation on workers performing routine tasks (e.g. Autor, Levy and Murnane 2003; Autor and Handel 2013; Autor and Dorn 2013; Frey and Osborne 2017; Goos, Manning and Salomons 2014; Marcolin, Miroudot and Squicciarini 2016; Spitz-Oener 2006; Acemoglu and Restrepo 2020), the first studies on worker-level exposure to AI have only just started to emerge. In the light of a study for the European Parliament (Deshpande et al. 2021), the current evidence and forecasts for the link between AI and labour markets are rather mixed. In terms of the impact on working standards, the study indicates that AI offers opportunities (reducing dangerous or unhealthy working conditions and improving accessibility to certain jobs) but also creates physical and psychosocial risks.

We contribute to the literature on working conditions and GVCs by analysing the role played by developments in AI alongside other features of technological progress (computerization and automation). Our econometric results show that, on average, involvement in GVCs correlates negatively with wages and (slightly) positively with some non-monetary aspects of job quality, especially once we consider the exposure of jobs to software or robots. As regards the exposure of jobs to AI, the effect of the intensification of GVC involvement is negative for wages but insignificant for job quality. We thus find that the influence of GVCs on wages and various facets of job quality may be diversified, such that non-wage aspects should be analysed in addition to the wages themselves. At the same time, we show that the role of digital technology in affecting the GVC–working conditions nexus is marginal or even insignificant. In other words, technological factors do not affect the core relationship between working conditions and GVCs in an economically meaningful way.

Following our review of literature related to the determinants of working conditions in section 2, section 3 documents some descriptive evidence on working conditions in Europe. In section 4, we present our key results, linking observed trends in job quality and wages with GVCs and the technological features of jobs. We discuss our conclusions in section 5.

2. Determinants of working conditions: A literature review

The literature on working conditions is extensive. As there is no unique definition of working conditions and/or job quality (Clark 2015; Steffgen, Sischka and Fernandez de Henestrosa 2020), multidimensional workers’ well-being may be analysed from different perspectives, implicating alternative methodological approaches. We can quantify the economic aspects of jobs using monetary indicators (such as wages), combined with information on working time or type of work (evening/night/shift/temporary) (e.g. Aleksynska 2018; Piasna 2018; Rossi 2013). However, other non-monetary features, reflected in such intrinsic aspects of work as autonomy, social utility, interpersonal relations or social context of work, constitute critical features of workers’ well-being (Cascales Mira 2021; Clark, D’Ambrosio and Zhu 2021; Gallie, Felstead and Green 2012). For instance, Clark, Kristensen and Westergård-Nielsen (2009) suggest that job satisfaction may be related to co-workers’ wages, where the association may be both positive, as higher co-workers’ wages may provide information about prospects (see also Javdani and Krauth 2020), but also negative if the worker’s wage is below the median wage (Card et al. 2012). Job-position gender diversity is related to higher worker well-being (Clark, D’Ambrosio and Zhu 2021). To make things more complicated, the perception of what constitutes a “good job” is highly subjective, differing between genders, for instance (Kaufman and White 2015). Job attribute preferences reflect the desire for specific work-related outcomes and may also be diversified across other workers’ characteristics, such as domestic circumstances, highest qualification held and occupation (Sutherland 2012).

The related economic literature has also attempted to evaluate the role of external global economic changes, such as production fragmentation across borders and technological progress. Vertical specialization, first quantified via offshoring indicators (Baumgarten, Geishecker and Görg 2013; Ebenstein et al. 2014; Egger, Kreickemeier and Wrona 2015) and recently by GVC measures based on global input–output data (Feenstra and Sasahara 2018; Parteka and Wolszczak-Derlacz 2019 and 2020), has been shown to have profound implications for labour markets. The literature is abundant, but many studies have assessed the phenomena in a purely economic way, based on information on wages (Baumgarten, Geishecker and Görg 2013; Ebenstein et al. 2014; Geishecker, Görg and Munch 2010; Parteka and Wolszczak-Derlacz 2019 and 2020; Shen and Silva 2018).⁴ The use of wages as an indicator of working conditions can be partially justified by the concept of social upgrading.⁵ This reflects the improvement in workers’ well-being resulting from involvement in global production (Milberg and Winkler 2011). However, the empirical literature on the broadly understood social consequences of trade and the proliferation of GVCs beyond wages gives contrasting results. Some empirical studies confirm a positive relationship, showing improvement in labour standards in companies that are more active in international trade (Nadvi et al. 2004; Bair and Gereffi 2001). Another strand of research finds that the link between economic and social upgrading is industry-specific (Bernhardt and Pollak 2016). Some authors claim that greater GVC involvement may not produce better pay or working conditions (Gimet, Guilhon and Roux 2015; Lee and Gereffi 2013; Lee, Gereffi and Lee 2016).

Moreover, the linkages between GVCs and non-monetary aspects of working conditions have mainly been analysed from the perspective of developing countries (Bair and Gereffi 2001; Barrientos et al. 2016; Kabeer and Mahmud 2004; Lee, Gereffi and Lee 2016; Rossi 2013). The question of how job quality varies among different groups of workers in developed countries is rarely raised in relation to GVCs and the existing evidence is mostly country- and industry-specific. For instance, Smith and Pickles (2015) find that in the Slovak clothing industry, wages and benefits in export-oriented companies may be higher, but employment stability is not. Lloyd and James (2008), in turn, report a positive impact of GVCs on the health and safety of workers employed in the UK food processing industry. Budría and Baleix (2020) investigate the effects of production fragmentation on individual job satisfaction and the perceived risk of job loss among German workers, finding that offshoring is negatively associated with job satisfaction. Studies on the production fragmentation–labour market nexus that offer a broader, cross-country perspective also tend to focus on wages as an indicator of labour conditions (Parteka and Wolszczak-Derlacz 2019 and 2020). Nikulin, Wolszczak-Derlacz and Parteka (2022) go one step further and examine how involvement in GVCs affects the wages, working hours and additional payments of workers from 24 European countries. Their results indicate a diversified effect of GVCs on working conditions, depending on the measure used: workers in sectors more deeply involved in GVCs have lower and less stable earnings, but they are also less likely to have to work overtime. The authors call for further research to consider diverse aspects of workers’ well-being.

Another strand of related research deals with the worker-level effects of technological progress. It is widely recognized that rapid technological changes are closely linked to labour market outcomes, including wages and employment (for a review, see Georgieff 2024; Goos 2018). An influential stream of literature addresses the displacement effect, typical for highly routine jobs prone to computerization and robotization (Frey and Osborne 2017), and the degree of substitution between robots (automation) and workers (Acemoglu and Restrepo 2018 and 2020). According to the skill-based technological change and routine-based technological change hypotheses (Acemoglu and Autor 2011; Autor et al. 2003; Goos, Manning and Salomons 2014), low-skilled workers constitute the most vulnerable group, while highly skilled workers may benefit from new technologies.⁶

Existing studies show that the implications of the newest technologies are complex and differ from a simple case of replacing workers with machines. Digital technologies have important effects on workers’ well-being, since they may put pressure on work–life balance or be the source of “technostress” (Tarafdar, Cooper and Stich 2019; Salanova, Llorens and Ventura 2014; Berg-Beckhoff, Nielsen and Ladekjær Larsen 2017). AI may impact human–machine interactions, resulting in new and changing work environments (Lane and Saint-Martin 2021), and affecting workers’ well-being (Nazareno and Schiff 2021). Generative AI can have effects on both job quantity and job quality (Gmyrek, Berg and Bescond 2023).

As regards the effects of technological progress on employment, the linkages depend on the suitability of job-related tasks for machine learning (Brynjolfsson and Mitchell 2017). The analysis conducted by Antón, Fernández-Macías and Winter-Ebmer (2020) reveals that robotization may be negatively associated with work intensity, whereas there is no linkage with the physical environment or skills and discretion dimensions of job quality. Turja et al. (2024) find that, on average, job satisfaction is lower in robotized workplaces than in non-robotized ones. In the case of AI technologies, the effects are not entirely related to the displacement effect, and for some workers these technologies may even be beneficial: in high-skill occupations, the role of AI solutions may be positive (Lane and Saint-Martin 2021, 23). Webb (2020) shows that, while exposure to robots and software is mainly typical for highly routine jobs, AI “performs tasks that involve detecting patterns, making judgments, and optimization” (Webb 2020, 3) – tasks typical of many high-skill occupations. Despite broad concerns about AI’s potential to substitute for labour, Felten, Raj and Seamans (2019) show that, while exposure to AI is not significantly related to employment growth, it is positively correlated with wage growth. The recent evidence for Europe provided by Albanesi et al. (2023) suggests that AI may be related to employment growth, especially for skilled workers, but the effect on wages is rather negligible. Similarly, Acemoglu et al. (2022) find a significant employment surge in AI-exposed establishments, but no detectable effect in either employment or wages at the occupation or sector level. The authors argue that the effect of AI is limited because “AI technologies are still in their infancy” (Acemoglu et al. 2022, S337) and only a small part of the US economy is involved in AI-related tasks. The study conducted for Germany reveals a positive effect on wages from AI and a small but positive effect on wages from robotization (Grimm and Gathmann 2022). The newest evidence on the effect of generative AI shows a possible positive surge in productivity growth (Brynjolfsson, Li and Raymond 2023; Calvino and Fontanelli 2023).

Summarizing the empirical evidence on technology–wage links, the net result of automation on wages is ambiguous because of the two opposite effects: substitution and productivity. The final result depends on routinization (or skill) level (e.g. Acemoglu and Restrepo 2020; van der Velde 2020). Similarly, the empirical evidence on the impact of AI on wages is also rather mixed (for a review, see Georgieff 2024).

Overall, working conditions are not easy to quantify, given their multidimensional and highly intangible nature. One approach is based on composite indicators of job quality (for an overview, see Cascales Mira 2021). There are many indicators for job quality in Europe, proposed by the ILO, Eurostat and Eurofound (for a review, see, among others, Cazes, Hijzen and Saint-Martin 2015). They rely both on aggregate indices and individual aspects of job quality. For instance, the Job Quality Index developed by the European Trade Union Institute (ETUI) (see Leschke, Watt and Finn 2008) and calculated for European countries includes such aspects as: (i) wages; (ii) non-standard forms of employment; (iii) working time and work–life balance; (iv) working conditions and job security; (v) skills and career development; and (vi) collective interest representation. Periodically published EWCS reports (Eurofound 2017, 2020, 2021 and 2022) aim to capture the complexity of job quality and rely on indices for physical environment, work intensity, working time quality, social environment, skills and discretion, prospects and earnings. The EWCS is the primary source of data employed in the next section.

3. Data and empirical approach

3.3. Model

Our key task is to estimate the intertwined relationship between different aspects of working conditions, the intensity of GVC involvement and the technological content of occupations. To this end, we run the following augmented Mincerian regression:

$\begin{array}{l}{\mathit{\text{JQ}}}_{\mathit{\text{iojsc}}}^k=\alpha \hspace{0.17em}+\hspace{0.17em}{\beta }_1\mathit{\text{Worke}}{r}_i\hspace{0.17em}+\hspace{0.17em}{\beta }_2\mathit{\text{Fir}}{m}_j\hspace{0.17em}+\hspace{0.17em}{\beta }_3\mathit{\text{Pro}}{d}_s\hspace{0.17em}+\hspace{0.17em}{\beta }_4{\mathit{\text{GVC}}}_{\mathit{\text{sc}}}\hspace{0.17em}+\hspace{0.17em}{\beta }_5\mathit{\text{Tec}}{h}_o\hspace{0.17em}\\ +\hspace{0.17em}{\beta }_6\mathit{\text{GV}}{C}_{\mathit{\text{sc}}}\hspace{0.17em}\times \hspace{0.17em}\mathit{\text{Tec}}{h}_o\hspace{0.17em}+\hspace{0.17em}{D}_c\hspace{0.17em}+\hspace{0.17em}{D}_s\hspace{0.17em}+\hspace{0.17em}{\epsilon }_{\mathit{\text{iojsc}}}\end{array}$        (1)

where i stands for worker, o occupation, j company, s sector of employment, c country and k the type of job quality (JQ) EWCS index. As an alternative to JQ, we employ the log of wage. In JQ estimates (relying on EWCS data), the set of individual characteristics (Worker_i) includes sex, age, educational attainment and skills (four types, based on occupation). In the case of wage regression, based on more detailed SES data, Worker_i characteristics include sex, age, educational attainment and type of employment (a full-time/part-time job binary variable). Firm_j stands for firm-related job characteristics: in models using the EWCS indices, we use type of contract (unlimited/temporary) and type of employment (part-time or full-time); in wage regression, we use length of service in the enterprise, type of employment (full-time/part-time) and form of economic and financial control (public/private)¹⁴. Sector productivity (Prod_s, expressed in logs) is equal to the ratio of value added to the total number of hours worked by employees. GVC_sc is a proxy for country sector-specific involvement in GVCs, while Tech_o refers to an occupation-level technology exposure measure. We also add the interaction between GVC and Tech, which takes into account the possibility that the effect of GVCs on working conditions depends on the type of technological exposure (GVC_sc × Tech_o). The marginal effect of GVCs on job quality is equal to:

$\frac{\delta \mathit{\text{JQ}}}{\delta \mathit{\text{GVC}}}={\beta }_4+{\beta }_6\mathit{\text{Tech}}$        (2)

This is similar for wages:

$\frac{\delta \text{ln}\left(\mathit{\text{wage}}\right)}{\delta \mathit{\text{GVC}}}={\beta }_4+{\beta }_6\mathit{\text{Tech}}$        (3)

and both can be represented graphically (see figures 1 and 2). Additionally, we include country and sector fixed effects: D_c should clear all country-specific characteristics such as labour market regulations, and D_s the remaining characteristics of sectors.

Figure 1

Predicted job quality EWCS indices due to changes in GVCS, at different levels of technological exposure of jobs (illustrating the results from tables 1, 2, 3)

Notes: The lines on the chart correspond to technological exposure level. Division of occupations into categories of low/medium/high exposure according to the index values (low = 10, medium = 40, high = 80).

Source: Our own calculation based on data from the EWCS and WIOD, and Webb (2020).

Figure 2

Predicted log wages due to the changes in GVC intensity at different levels of technological exposure of jobs (illustrating the results from table 4)

Notes: The lines on the chart correspond to the technological exposure. Division of occupations into low/medium/high software, robot and AI exposure according to the index values (low = 10, medium = 40, high = 80).

Source: Our own calculations based on job quality indices from the 2015 EWCS merged with the 2014 SES, the 2016 WIOD and technological exposure indicators from Webb (2020).

In the model using the job quality indices from the EWCS (in the range 0–1), equation (1) is estimated by fractional probit,¹⁵ while in the wage-based model the equation is estimated using weighted ordinary least squares (OLS). Weights are calculated on the basis of the SES grossing-up factor adjusted to the number of observations per country (to ensure that each country is equally represented in the sample). Regressions for both types of dependent variable are estimated with robust standard errors clustered at the country-sector level.

In the next section, we present the estimation results for the EWCS indices (section 4.1) and wages (section 4.2).

4. Results

5. Conclusions

A comprehensive view of the joint impact of the production internationalization and digital progress on job quality and social conditions at work is still lacking, including in the European context. The purpose of our analysis in this article was to shed new light on the differences in working conditions across Europe. In particular, we aimed to broaden the understanding of Europeans’ well-being at work by: (i) using a multidimensional approach to the quantification of working conditions (analysing wages and several aspects of job quality); and (ii) jointly assessing the role played by the dependence of European labour markets on GVCs and various types of digital technologies.

Concerning the first point, our analysis has allowed us to put the social dimension of working conditions into a typical, purely economic view of the impact of GVCs/technology on workers. As wages do not capture the full complexity of the work-related factors that determine workers’ well-being, our analysis includes such non-monetary dimensions of job quality as the quality of the physical and social environment at work, career development prospects and work intensity. Indeed, we find that working conditions tend to differ not only between European countries and across occupations but also with respect to particular aspects of working life. The comparison of cross-country averages is not informative because workers’ well-being depends on specific dimensions of job quality, as well as on sector, occupational and personal characteristics. A detailed cross-country microeconomic perspective is thus necessary.

Concerning the second point, we were particularly interested in examining how the two global phenomena of intensification of cross-border production links and rapid progress in digital technologies affect job quality (compared to wages). We have estimated several econometric models linking wages and six job quality indices from the EWCS with GVC intensity and the technological characteristics of workers’ occupations (and several additional controls). We have thus combined three important perspectives present in the related literature: (i) labour economics/sociological research on working conditions and decent work; (ii) international economic studies on production fragmentation; and (iii) the literature on the labour market effects of technological progress driven by digital solutions.

We find that on average (that is, when controlling for individual and firm characteristics), involvement in GVCs correlates negatively with wages and (slightly) positively with some non-monetary aspects of job quality, once we consider the exposure of jobs to software or robots. When we consider the exposure of jobs to AI, the effect of the intensification of GVC involvement is negative for wages, but insignificant for the job quality indices. We thus find that the influence of GVCs on wages and different types of job quality measures may vary. This calls for an analysis of non-wage aspects in addition to wages themselves. Importantly, we find that technological factors do not alter the core relationship between global production links and job quality and wages in an economically meaningful way.

Our approach addresses the problem of a proper quantification of worker-level effects of global production fragmentation. Real-world problems of work overload, technostress and bad social or physical work environments cannot be solved without identifying their sources. Our analysis indicates that the information contained in input–output tables, once matched with detailed micro-level data on job quality, provides a fuller view of how GVCs impact workers. We hope that our results, exploring heterogeneity across countries, sectors, occupations and workers, convincingly confirm the multidimensional nature of working conditions. They are relevant to the formulation of adequate and coordinated policy responses to worsening labour standards due to pressure from cross-border competition, such as through GVCs, and the challenges resulting from rapid digital progress. Fair and high-quality job standards are of the highest priority because health problems and worsening job performance caused by bad working conditions require policy intervention.

We deliver empirical evidence on the impact of new types of technology, contributing to the growing body of literature on linkages between digital technology and labour markets. On the one hand, a recent survey among workers and employers in Europe revealed relatively positive expectations regarding the impact of AI on the labour market and working conditions (Lane, Williams and Broecke 2023). On the other hand, and as our study confirms, empirical evidence on the effects of digital technology on wages is highly inconclusive.

Our study is based on pre-pandemic data. Accordingly, the next important question that needs to be addressed is how the COVID-19 pandemic and changes in the world of work in its aftermath have affected job quality. The impact is also likely to have been unevenly distributed across workers. During the pandemic, some, such as healthcare workers, worked on the front line, while others had to close their businesses, and many of those working at home faced increased stress and pressure on their work–life balance; all experienced day-to-day uncertainty. These effects have had direct and indirect impacts on working practices and conditions in the post-pandemic landscape. Future research describing the impact of the pandemic on workers’ well-being is thus needed.

Acknowledgments

This article draws upon the analysis presented in a working paper published by the ETUI (Nikulin, Parteka and Wolszczak-Derlacz 2022). Joanna Wolszczak-Derlacz acknowledges financial support for these studies from Gdańsk University of Technology under Americium International Career Development “Excellence Initiative – Research University” Grant No. DEC08/2023/IDUB/II.14/AMERICIUM. Aleksandra Parteka acknowledges financial support from the National Science Centre, Poland (2020/37/B/HS4/01302). Dagmara Nikulin acknowledges financial support from Gdańsk University of Technology under the Argentum Triggering Research “Excellence Initiative – Research University” Grant No. DEC-8/1/2023/IDUB/I3b/Ag. We are also grateful to ETUI members for their comments and advice.

Competing interests

The authors declare that they have no competing interests.

Appendix