A configurational approach to job quality analysis: forms of inequalities at work in Europe

Transformations in the structure and content of jobs have led to the development of numerous studies on job quality, the definition of which varies according to the discipline (Williams et al.2020) and according to national contexts (Gallie 2007b; Erhel and Guergoat-Larivière 2010). International institutions such as the OECD, the European Commission, Eurofound (2017), and the European Trade Union Institute (ETUI) also use the expression of job quality to make comparisons between the 28 EU countries (Leschke and Watt 2014). The academic literature focuses more on the individual and institutional determinants of job quality and its effects on well-being.

Some studies favour a subjective approach to job quality, based on the judgements of the workers themselves, which amounts to measuring job satisfaction (Clark 2005; Pichler and Wallace 2009). This subjective approach has been criticised on the grounds that it remains dependent on employees’ perceptions and their lack of awareness of the risks that their job may have on their health (Felstead et al.2019; De Bustillo et al.2011). To avoid this pitfall, objective approaches to job quality have relied on indicators of job quality rather than individuals’ evaluations (Gallie 2007a; De Bustillo et al.2011; Williams et al.2020). However, paying attention to the objective truth of work should not lead us to overlook its subjective truth (Bourdieu 1996). In order to obtain workers’ consent to work, employers try to conceal the objective exploitation that it entails, by highlighting the symbolic profits that workers can draw from it (Burawoy and Holdt 2012). Subjective assessments can therefore contradict the objective position, especially when workers declare themselves to be very satisfied with a job that is poorly paid or associated with poor working conditions (Baudelot and Gollac 2003). To account for this, we propose a configurational approach that makes it possible to measure the subjective assessments of job quality associated with a given objective position in social space (Bourdieu 1984), as well as the interactions between these two dimensions.

Most studies agree that job quality is multidimensional and includes criteria related to employment conditions and the nature of work (Warhurst et al.2022). Another point of agreement is that job quality has an impact on workers’ well-being (De Bustillo et al.2011; Felstead et al.2019), and can therefore have consequences outside work. Two main methods are used to measure job quality according to these criteria (De Bustillo et al.2011, 2022; Hauff and Kirchner 2022). The first consists in constructing a single index based on a score or scale, which amounts to representing the various dimensions by a unidimensional measure. The second consists in using a composite index that compares the scores of each dimension in a dashboard, which makes it possible to account for the different dimensions, but not for the relationships that exist between variables. To account for the multiple dimensions of job quality and the interactions between them, we use a different method of quantification which makes it possible to adopt a relational perspective (Emirbayer 1997), so as to analyse the ‘interaction effects’ among these dimensions.

The criteria that define job quality are not independent of one another, and their interactions themselves can produce effects. Our objective is not only to study job quality through the interactions between variables, but to construct the system of relations between variables and to position workers in a social space as described by Bourdieu (1984), that is, a hierarchical space structured by the unequal distribution of cultural and economic capital. This space of positions can be matched with several configurations of job quality, which are situated in a field within which both individuals and social groups are distributed. We therefore propose to break with the binary conception of job quality and with the opposition between objective and subjective definitions. In our configurational approach, the same variable does not always have the same effect on job quality: each effect depends on the job quality configuration in which the variable is inserted. For example, high investment in work can generate happiness if it is associated with good working conditions and some recognition, or unhappiness if it is associated with poor working conditions and low income. Based on the French case, Baudelot and Gollac (2003) show that there can be different configurations of unhappiness at work: one such configuration corresponds to the most dominated workers, who have poorly paid and uninteresting jobs, while another, distinct configuration corresponds to workers who suffer from too much pressure or a lack of autonomy, even though they are well paid and/or consider their work to be interesting and socially useful. Our configurational approach to job quality involves a form of representation by clusters, which makes it possible to account for the different possible combinations between variables (Coutrot and Perez 2022; Holman 2013; Valeyre et al.2009).

It thus remains to identify which variables explain the different configurations of job quality. Several micro-level variables have strong effects on the perception of job quality. Occupational position is undoubtedly one of the most significant such variables, with a strong class gradient in job quality (Green et al.2013; Gallie 2022). Moreover, managers and professionals, often in well-paid and secure jobs, are more likely than others to feel happy at work (Pichler and Wallace 2009; Arjona Perez et al.2010). The gender gap should also be mentioned, even if inequalities in the working environment between men and women are low in Europe (Green et al.2013). Jobs held by women often involve less autonomy and more monotony than those held by men (Gallie 2022), and these inequalities seem to be more pronounced at the top of the social space than at the bottom (Smith et al.2008). Age is another determining factor in job quality: young people enjoy less socioeconomic security (Erhel and Guergoat-Larivière 2010) and less autonomy at work (Esser and Olsen 2011) than older people. Employment status also matters: self-employment involves higher intrinsic job quality and job satisfaction, although self-employed people work harder and with less job security (Meager 2015), while being employed on a temporary rather than a permanent contract reduces job quality (Gallie 2019).

At the macro-level, job quality can also be influenced by national differences, for several reasons. First, the proportion of good quality jobs is higher in more affluent countries (Green et al.2013), even if job structures are becoming more polarised, with an increase in both skilled and unskilled jobs (Goos et al.2009). Second, national institutional structures can affect the level of job quality, through the regulation or liberalisation of labour markets (Hult and Svallfors 2002; Gallie 2007b), forms of work organisation (Valeyre et al.2009), the power of trade unions, and the political orientation of governments (Gallie 2007a). Using countries as macro-variables allows us to combine these several dimensions (Green et al.2013; Erhel and Guergoat-Larivière 2010).

To produce a complete picture of job quality, we propose to add a meso-level analysis that takes into account the type of sector (public or private) and the presence of unions in the company. It is known that the fact of working in the public sector has implications for relations between colleagues (Corby 2000) and provides a greater feeling of social usefulness compared with workers in the private sector (Hugrée et al.2015). The increasing similarity of working and employment conditions between the public and private sectors raises the question of whether this variable influences the perception of job quality. It has also already been shown that trade union density at the national level favourably influences job quality in European countries (Gallie 2007a; Leschke and Watt 2008), especially with regard to the autonomy of employees (Esser and Olsen 2011), notably because trade unions have an impact on the regulatory action of the state (Bosch and Weinkopf 2017). Nevertheless, given the decline in union membership, the question arises of whether unions still influence the employment relationship (Edwards 1990) in companies.

In summary, we propose to examine two research questions: how to account for the interactions between the subjective and objective dimensions that define job quality, and how to combine the relations between micro-, meso-, and macro-level variables in a configurational approach.¹

We use the European Working Conditions Survey (EWCS), which is a sample survey funded by Eurofound on the working conditions of Europeans. It has been repeated every five years since 1990, and covers a representative sample of around 1,500 working people, who are interviewed face-to-face in each country. It provides an overview of trends in working conditions and quality of employment, covering issues such as employment status, work organisation, work-life balance, and health and safety.

Using the 2015 survey, we chose to focus on 33,811 employed and self-employed workers from 29 European countries: Austria, Belgium, Bulgaria, Croatia, Cyprus, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, the Netherlands, Norway, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, Switzerland, and the United Kingdom.

In order to measure job quality, we rely on five dimensions that are widely accepted in the literature (Warhurst et al.2022): pay and other rewards; intrinsic characteristics of work, terms of employment, health and safety, and work-life balance (including work intensity). Drawing on the findings of other studies (Baudelot and Gollac 2003; Olsen et al.2010; Coutrot and Perez 2022), we measured their effects by focusing on variables that correspond to the expression of a judgement, an evaluation, or a perception. We can thus construct a typology using 12 variables²:

In order to understand the role of social structures in explaining the inequalities associated with this job quality typology, we perform a logistic regression for each cluster. The individual variables selected are sociodemographic (gender, age), education level, income level (quintile), occupational categories (European Socio-Economic Groups), and employment conditions (type of contract). The meso-level is related to the company context: the fact of working in either the public or private sector, and the size and presence of a trade union in the company. The macro-level is related to groups of countries, according to the typology suggested by Gallie (2018). We distinguish between North West (liberal model), Nordic (social democratic model), Continental (including France) and Mediterranean countries. The Eastern countries are subdivided into three, ‘reflecting their different historical and trade ties’.

In order to measure the relations between the different dimensions of job quality, we chose to carry out an unsupervised exploratory classification using K-means clustering, which makes it possible to construct a typology of the configurations of well-being (or not) at work. This method has two main benefits. First, the computation time of the K-means algorithm is adapted to the size of the distance matrix produced. Second, our use of clusters does not require knowledge of the hierarchy of the different class levels. The resulting typology is made up of five clusters, on the basis that the contribution of inter-class inertia as we can see in supplementary material, which presents the intra- and inter-cluster variances, but also because of the resulting size clusters and a concern to highlight the finesse of contrasting perception to work for clusters faced with the least favorable working conditions.

This cluster combines all the characteristics of well-being at work (Table 1). For each of the twelve variables selected, the respondents consistently score above the average of positive responses. They enjoy individual fulfilment, social recognition, and especially satisfaction with their remuneration, all of which guarantee them a feeling of happiness at work. Overall, they are fully committed to their work and derive great satisfaction from it. This cluster has similarities that which Holman (2013) calls ‘Active’ jobs, which corresponds to workers with high levels of job resource.

This cluster brings together workers who consider themselves to be well paid for their work, recognised, and able to escape stress and other forms of health hazards. On the other hand, they do not experience personal commitment (they ‘do not put their ideas into practice’) and are more often in the situation of not ‘learning new things’ at work. The fact that they are well integrated into work groups enables them to distance themselves from certain forms of strain and not to suffer too much from them. This cluster also has similarities with the ‘passive-independent’ cluster in Holman’s (2013) typology: low workload, standard working hours, relative job security, and low personal commitment.

This cluster contains workers who, for the most part, say that they are fulfilled and feel secure in their jobs, but who suffer from a lack of recognition. This feeling of being poorly recognised is expressed in two ways: first, with regard to the hierarchy, which results in the perception that they are poorly paid, and second, with regard to their colleagues, which is reflected in poorer relationships within work groups or even isolation in their work (they are less likely than the mean to get on well with their colleagues).

This cluster brings together workers who combine two major negative dimensions in their relationship to work. First, they feel that they are poorly paid and not recognised for their work. Second, they say that they do not experience any form of professional fulfilment: they very rarely claim that they can put their ideas into practice or that they are in a position to learn new things, and they do not express the feeling of work well done or of doing useful work. They therefore consistently feel devalued and are more likely than the mean to fear losing their job within six months. In this cluster, the level of meaning associated with work is low.

The members of this cluster are positive about their work and value it: they often consider it useful, take satisfaction in work done well, and enjoy learning new things and getting along with their colleagues. However, they claim to suffer from forms of insecurity: difficulties in balancing their professional and private lives, a feeling that their physical safety is at risk, emotional problems, and stress. As a result, and in view of their investment in their work, they consider themselves poorly paid and they often suffer from a lack of recognition. We again find some similarities with a category identified by Holman (2013), which he terms (‘high strain jobs’). In this category, workers report high personal commitment and face high workloads but have few resources to deal with them. Their relationship to job quality is ambivalent because they appreciate the sense of doing useful work but they suffer from working conditions that create severe difficulties in their job and in their personal lives.

The determinants of each cluster are identified through logistic regression models incorporating micro-, meso-, and macro-level variables (Table 2).

The members of the happy workers cluster hold high social positions: they are more often managers, professionals, and small entrepreneurs, highly educated and belonging to the highest income quintiles. The chances of belonging to this cluster are higher for those working in small companies, where they often enjoy a high degree of autonomy. In contrast, public sector workers are 1.19 times less likely than private sector workers to belong to this cluster. At the macro-level, citizens from the Nordic countries are almost twice as likely as those from Continental Europe to belong to this cluster, while those from Mediterranean and Eastern countries are 1.2 times less likely to belong to this cluster. This result confirms the specific position of the Nordic countries compared to that of the Continental countries (Gallie 2007a, 2007b).

Membership of the depreciated workers cluster follows opposite principles to those of the happy workers cluster. Less skilled workers are more likely to belong to this cluster than managers and professionals are. This is also the only cluster where employment status has a significant effect, since employees in temporary jobs are more likely to feel depreciated. This confirms the fact that precarity not only degrades the content of jobs (Gallie 2019), but also more broadly worsens the subjective relationship to work (Aleksynska 2018). Employees of large companies are more likely (OR = 1.27) to belong to this cluster (compared to employees of medium-sized companies), and public sector employees are less likely (OR = 1.12) than private sector employees. At the macro-level, citizens from Mediterranean and Eastern European countries are more likely to belong to this cluster than those from Continental and Nordic countries: these regions of Europe are characterised by greater job insecurity (higher unemployment or temporary employment rates) and weaker trade unions (especially in the Eastern European countries). The higher likelihood for citizens of North-Western countries (United Kingdom, Ireland) is explained by the scale of the development of low-wage occupations in these countries (Oesch and Piccitto 2019) and by high job insecurity (for example, the high proportion of part-time or temporary jobs). In these countries, less skilled workers are clearly distinguished from other occupational groups by a poor job quality.

Membership of the distant and pacified workers cluster is likely to be found among women, young people, clerks, and skilled service workers. They occupy an intermediate position in the social space and are distinguished by the routine nature of their jobs. This confirms that women and younger people are likely to have less autonomy and more monotony in their work. At the meso-level, the presence of a union in the company increases the chances of belonging to this cluster. In this case, we can guess that the union presence helps to maintain a collective distance from the work, and to avoid suffering or feelings of devaluation. National affiliation is not very significant for this cluster, even though citizens of Continental countries (especially in Germany),³ where office workers are numerous, are more likely to belong to this cluster than those of other groups of countries, with the exception of the South-Eastern countries.

The cluster workers in need of recognition gathers together occupational categories that are high in the social space (small entrepreneurs, managers, and professionals) but only averagely remunerated. Small entrepreneurs are more likely to belong to this cluster (OR = 2.7) than clerks and skilled service workers are. This result nuances those studies that emphasise the good job quality of the self-employed, and confirms that they benefit from a higher intrinsic job quality (Meager 2015), since this cluster is characterised by a feeling of satisfaction with work done well and its usefulness. Public sector employees are also more likely to belong to this cluster, and especially highly qualified public sector employees who feel poorly paid and not sufficiently recognised for their work. The correlation with national affiliations is weak for this cluster.

The cluster of insecure suffering workers more often concerns workers occupying low and intermediate positions in the social space: industrial and agricultural employees, but also small entrepreneurs. They are more likely to be middle-income workers with low educational qualifications. All other things being equal, public sector employees are more likely to belong to this cluster than those in the private sector. Finally, a surprising result is that the presence of a trade union in the workplace does not prevent people from belonging to this cluster of insecure suffering workers. The macro-level determinants of this cluster reveal more geographical differentiation than those of the first two clusters: compared to citizens of Continental countries, those from South-Eastern countries, but also from the Nordic countries, are less likely to belong to it; on the other hand, those from Mediterranean countries (particularly Greece and Spain, Table 3) are more likely (OR = 1.15) to belong to it. Among affluent countries, France and Sweden stand out, as they have the same proportion of workers in this cluster as Greece and Spain. For this cluster the differences between countries are less pronounced than for the other clusters. The variables of poor quality of working conditions (low autonomy, intense workload, physical hardship) seem to have a more significant effect than the micro and macro variables.

Like many studies that measure job quality (De Bustillo et al.2011; Leschke and Watt 2014; Antón et al.2015), we find an opposition between two poles: on the one hand, workers who benefit from all the positive dimensions of job quality and, on the other hand, workers who suffer from a large number of negative aspects related to their working conditions. Our configurational approach nevertheless makes it possible to nuance and enrich studies on job quality in three ways.

The first contribution of our approach is to show that the opposition between good jobs and bad jobs is not entirely binary and one-dimensional. The clustering method refines our knowledge of intermediate situations that do not appear in studies that use either a single index or dashboards of indicators. We can thus highlight the situation of workers who have good working conditions but who feel poorly paid and not sufficiently recognised, or of those who do not feel personally fulfilled in their work. Thanks to the nonhierarchical clustering method, we can distinguish two types of configurations where job quality is negatively assessed. One is characterised by workers who suffer because they have a high workload and because they are highly involved in their work, which they consider useful, without benefiting from sufficient resources, especially organisational autonomy and social support. This is the situation of job strain identified by Theorell etal. (1990), which is associated with psychological suffering, health problems, or ethical conflict (Coutrot and Perez 2022). The second is made up of employees who feel devalued because they lack social recognition and individual fulfilment owing to the low quality and lack of intellectual stimulation in their jobs. We also observe two dimensions in the perception of job quality: first, the accumulation of positive dimensions versus the accumulation of negative dimensions, and second, over-integration and over-investment at work versus individual relegation and symbolic devaluation. The recognition of these different configurations of job quality arises from taking account of a combination of factors, and from a methodology that is more complex than traditional measures of job quality.

The second contribution of our approach is to highlight forms of inequality that do not clearly appear in previous studies. At the macro-level, we find an opposition between, on the one hand, the Nordic countries characterised by good job quality - to a lesser extent in Sweden (Melldahl, 2022) - and, on the other hand, the Southern (Mediterranean) and Eastern countries, where job quality is lower (Antón et al.2015; Leschke and Watt 2014). The Continental and North-Western countries are in an intermediate position between these two poles (De Bustillo et al.2011; Gallie 2019). These differential effects of country groups can be explained by the opposition between more affluent countries with a high volume of skilled employment and less affluent countries with a lower volume of skilled employment (Green et al.2013). Our study confirms another result previously highlighted by studies based on dashboards of indicators: the Nordic countries are characterised by a high level of intrinsic characteristics (Smith et al.2008; Green et al.2013), and the Eastern countries by a low level of intrinsic characteristics, in particular in relation to autonomy at work (Esser and Olsen 2011). The clustering method nuances this geography of job quality in Europe by looking not only at good jobs, but also at the different forms. The cluster of insecure suffering workers shows that the Mediterranean countries, especially Greece and Spain, two countries hit by the economic and financial crisis, tend to be distinguished from the Northern and Eastern countries by the larger number of jobs that are relatively qualified and considered as socially useful, but which are carried out under degraded working conditions (high intensity, lack of autonomy, physical hardship, and with encroachment on personal life). However, membership of this cluster is found among workers from all countries, and this result confirms the advantages of our methodology: this job quality cluster is more related to individuals’ positions and work contexts than to national differences.

In terms of micro-level features, the configurational approach also confirms some known results regarding inequalities at work. The most highly qualified workers and occupational categories at the top of the social space are more frequently in a situation of experiencing well-being at work (De Bustillo et al.2011); manual workers and low-skilled workers, as well as the most precarious, do not benefit from a good job quality (Gallie 2019, 2022). However, the configurational approach also highlights inequalities which are not very visible when using the more usual methods, and which constitute internal boundaries within the major social groups. When the perception of job quality in Europe is analysed as a whole, we observe that the ‘bad’ jobs are not concentrated only among the least skilled jobs, and that the ‘good’ jobs are not concentrated only among the most highly-skilled jobs (Askenazy 2021). In our clusters, distant pacified workers perform high-quality jobs, particularly in terms of psychological security and a feeling of safety, but do not consider their work as a means of personal development; workers in need of recognition are more often qualified employees but consider themselves poorly paid in relation to their commitment to work, which Bourdieu would call a ‘misery of position’ (Bourdieu 1999). These two clusters highlight the subjective effects of certain inequalities at work (such as lack of recognition), which depend little on the country where one works and more on the work context (union presence for the distant pacified workers cluster and belonging to the public sector for the workers in need of recognition). Our focus on job quality reveals a strong segmentation in the middle and lower parts of the European social space and makes it possible to renew the concept of class fractions as developed by Bourdieu (1984). This complements work that highlights the development of an internal divide among white-collar workers based on the increasing precarity of the most highly-skilled jobs (Häusermann et al.2015; Mustosmäki et al.2017). More broadly, this shows that inequalities in working conditions constitute one of the internal boundaries within social classes in Europe (Hugrée et al.2022).

The third contribution of our approach is to highlight the importance of meso-level variables, such as the fact of working in the public or private sector and the presence of a trade union in the company, which are often neglected in job quality studies. In most of the work that focuses on data prior to the 2008 crisis, belonging to the public sector appears to be a factor that increases the chances of having a good job quality (Holman and McClelland 2011; Crespo et al.2013). However, since the financial crisis in 2008, budgetary pressures on the public sector have dramatically increased (Herrmann 2017), leading to reforms and austerity measures, with negative effects on the quality of jobs in the public sector (Keune et al.2020). Using data recorded after 2008, our results show that working in the public sector decreases the chances of being fully happy at work and, conversely, increases the chances of experiencing suffering, insecurity, and a lack of recognition (particularly in terms of remuneration). First, this result is probably directly linked to the budgetary austerity policies that led, in the 2000s and even more so after 2008, to a deterioration in the working conditions of public sector employees, particularly in hospitals and the education sector (De Beer and Keune 2022). In this respect, it can be noted that working in Mediterranean countries that have been hit hard by austerity policies increases the chances of belonging to the cluster of insecure suffering workers. Moreover, some studies show that New Public Management policies have contributed to the deterioration of employment conditions in the public sector (Sauer et al.2022) and to the disempowerment of public employees and civil servants (Kearney and Hays 1998). Although the managerial turn in national public administration has tended to make working conditions more similar in the public and private sectors, a process of material and symbolic downgrading of public sector employees compared to those in the private sector can be observed (Hugrée et al.2015). It should be noted that working in the public sector increases the chances of being in the suffering insecure workers cluster, which is characterised by a strong attachment to work that is done well and socially useful. This result brings to mind the contradiction between the public sector ethos – that is, a set of dispositions that privilege the public good over private interests – and the organisational changes introduced by New Public Management-based reforms (Gleeson and Knights 2006).

Our approach questions the importance of another variable at the meso-level, that of the union presence within a company. Many studies have shown a decline in the power of unions to limit wage inequalities (for example, in the United Kingdom, Bryson and Green 2015), even though they can have positive and negative impacts on workers’ job quality (Simms 2022). In our study, the presence of a trade union in the company favours membership of the cluster of distant and pacified workers, which would confirm the idea that the existence of forms of collective solidarity continues to have an impact on well-being at work (Baudelot and Gollac 2003). On the other hand, while several studies have shown the link between the growth of poor quality work and the reduction of a trade union voice (Kalleberg 2011), our study shows that the presence of a trade union within a company does not increase the chances of occupying very good quality jobs, and furthermore, does not protect against certain forms of suffering at work. It seems as if a trade union presence in companies no longer constitutes a protection against policies of intensification and increasing pressure on workers. This result can be understood as a reflection of the constraints resulting from the focus on financial objectives, which are now imposed on employers (Thompson 2003), independently of the balance of power and collective solidarity that may exist within the company.

The configurational approach to subjective relationships to work reveals a classic opposition between jobs perceived as being of good quality and others of poor quality, but it also reveals differentiated registers of job quality, which depend on micro-, meso- and macro-level variables. For example, working in the public or private sector, or in a company with a trade union presence are important criteria, but these meso-level variables can take on different meanings depending on how they are combined with other variables.

The configurational approach developed in this article invites us to consider the internal boundaries of social groups and the inequalities that may emerge from certain work situations, rather than focusing solely on the quantification of good and bad jobs. This perspective could be extended by diversifying the indicators of job quality, particularly those relating to the content of work and psychosocial risks, which are important dimensions of inequalities at work. In addition, we could deepen our knowledge of intermediate clusters by testing their stability over time, using several surveys to track the evolution of certain variables over several years.

We thank Milan-Bouchet Valat, Cédric Hugrée, and Philippe Askenazy for their valuable comments. We would also like to thank the reviewers for their insightful comments on the earlier versions of this article. We thank Sam Ferguson for the translation of the article.

No potential conflict of interest was reported by the author(s).