Introduction

Although higher education (HE) reform recognises the primary role of employability in the mission of universities and other institutions, the concept of employability has been the subject of deep debate. The term employability combines the words employment and the ability to become employed (^{Vanhercke et al., 2014}); it is at the core of the debate on the relationship between HE and the world of work (see ^{Teichler, 2007}, for a survey of main issues) and consequently on graduates’ transition into the labour market (^{Alves, Alves and Chaves, 2012}; ^{Chaves and Morais, 2016}). However, it is up to the individual to become employable; this not only involves acquiring the necessary skills, but also being able to stand out in the labour market (^{Torres, 2019}).

Employability is a complex concept that encompasses multiple perspectives related to the supply of and demand for skills, and yet research usually focuses on the possessive approach, namely the skills and other attributes that make it easier to find a job (^{Holmes, 2013}). As this approach is influenced by the human capital argument, attention is placed largely on job requirements other than qualifications or professional experience (^{Andrews and Higson, 2008}; ^{Branine, 2008}). However, the lack of a clear definition of employability skills means that HE institutions inevitably find it difficult to provide their students with these skills.

Job advertisements are an important source of information and have been widely used to identify the employability skills graduates are expected to have (^{Bennett, 2002}; ^{Ahmed, 2005}; ^{Choi and Rasmussen, 2009}; ^{McArthur et al., 2017}), namely skills for low and medium level occupations (^{Mýtna Kureková et al., 2016}), and those for all levels of occupation (^{Brown and Souto-Otero, 2020}). While comparative studies remain scarce, some authors have compared different countries (^{Mýtna Kureková et al., 2016}; ^{Kennan et al., 2006}). Comparative research can enhance our understanding of the dynamics of the labour market where the job vacancies are posted. The information provided by job ads goes beyond the skills and attributes required by employers; they not only shed light on countries’ specificities but also on the employers’ attitudes towards the qualifications and abilities, aspects that need to be brought to the debate.

Our research uses jobs ads to compare the skill sets in various European countries and to explore the link between the requirements and labour market characteristics in each geography. For example, what are employers looking for when they hire graduates? Why do they inform (or not) job requirements? How do economic and labour market characteristics interact with job offers? Available literature has focused almost entirely on the skill set required in the labour market but has failed to examine this link even though it could provide HE and employment policy makers with useful information. We assume that job ads do more than simply publicise job vacancies. In fact, they reveal hidden information on the graduate labour market.

In an attempt to answer the above questions, we explore 2,812 online job advertisements targeting graduates (bachelor) and post-graduates (master), collected from European Union websites in 2019 and early 2020. More specifically, we aim to compare the job offers in Europe in terms of countries (n = 21), occupations and a set of country characteristics. We apply a fuzzy cluster analysis to identify types of skills and explore the association with types of occupation, as used in ^{Cedefop (2011}). Subsequently, variables such as field of education, industry, and country help provide a better characterisation of each skill cluster.

Literature review

Beyond the skill requirements in job ads

Job requirements must also be situated in the economic context. This section provides an overview of some of the literature on the underlying information contained in job ads to identify the skill set employers require. This can further the understanding of the impact of HE expansion, a subject that divides researchers between those who believe HE returns have declined (^{Mok, Wen and Dale, 2016}) and those who criticise the catastrophist theses (^{Chaves, Morais and Nunes, 2009}).

Some studies have addressed the types of occupation assigned to graduates. ^{Trow (1973}) was the first to warn of the lack of skilled jobs, arguing that the probability of graduates working in traditionally non-graduate sectors increases with the growing supply. This has led some social scientists to examine the association between skill requirements and occupational structure (^{Brown and Souto-Otero, 2020}), while others have provided taxonomies of graduate occupations (e.g.^{Elias and Purcell 2004}). The latter authors have discriminated four types of graduate job based on a comparison between two cohorts of graduates: traditional, modern, new, niche; they also identified non-graduate jobs. ^{Figueiredo et al. (2017}) used Elias and Purcell’s proposal to identify three types of job: traditional, new, and latent. The traditional jobs have a very high percentage of graduates in both cohorts; the new are similar to modern jobs and are made up of jobs where most of the younger workers have a higher education diploma. Finally, the latent jobs correspond to non-graduate jobs. An additional taxonomy used in the ^{Cedefop (2011}) study on job polarisation classifies the occupations in four categories, namely high-skilled, skilled non-manual, skilled manual, and elementary.

Alongside the discussion on graduate and non-graduate occupations, some research attempts to explain how these jobs are assigned. ^{Green and Henseke (2017}) refer to relative demand to highlight the gap between the availability of graduate jobs and the supply of graduates. They show that underemployment has registered an upward trend across Europe despite variations between countries. It is higher in Greece, Italy, and Ireland because of the disparities between the supply of graduates and graduate jobs, while Sweden has the highest proportion of graduates working in non-graduate jobs in Nordic countries. Other researchers provide evidence on the allocation of graduates to managerial, white collar and scientific and technical jobs in Portugal that require a degree and show there is a mismatch in only a small proportion of non-graduate jobs (^{Chaves, Morais and Nunes, 2009}). Other research on the Portuguese case indicates the mismatch is predominately in new and latent graduate jobs, and that fewer graduates have been employed in traditional jobs since HE expansion (^{Figueiredo et al., 2017}).

The literature also discusses overeducation (^{McGuinness, 2006}), credentials inflation (^{Collins 1979}; ^{Tomlinson and Watermeyer, 2022}), and crowding out (^{Klein, 2015}) to illustrate the impact of the oversupply of graduates in the labour market. The term credentials inflation refers to the increase in educational requirements for jobs as a result of the growing number of advanced education degrees conferred (^{Collins, 1979}). ^{Tomlinson and Watermeyer (2022}) admit that these credentials become a positional good, but the benefits of education will be restricted to some graduates. In such circumstances, some graduates tend to accept any available job, i.e., the lack of skilled jobs in some labour markets means graduates are assigned to non-graduate jobs (^{Teichler, 2000}). It seems that the overeducation / underemployment following the massification of HE is largely explained by a mismatch between the pace at which the supply of graduate increases and the availability of graduate jobs.

^{Marques, Suleman and Costa (2022}) bring new insights to this discussion, showing that overqualification is common in countries with low weight high-tech manufacturing and services sectors, as in the case of Southern European countries and Austria. This is because the high level of HE expansion occurred within a production regime that continued to be based on low technologies and domestic demand rather than on high-tech sectors (^{Hall, 2018}). The European Innovation Scorecard (^{Hollanders, Es-Sadki and Merkelbach, 2019}) classifies the European countries in four clusters based on the performance of the innovation system: the innovation leaders (Denmark, Finland, the Netherlands, Sweden, and Switzerland); strong innovators (Austria, Belgium, Estonia, France, Germany, Ireland, Luxembourg, Norway, and the United Kingdom); moderate innovators (Croatia, Cyprus, Czechia, Greece, Hungary, Italy, Latvia, Lithuania, Malta, Poland, Portugal, Slovakia, Slovenia, and Spain); and modest innovators (Bulgaria and Romania).

Based on the reported literature, we submit the following hypotheses to empirical testing:

H1: Employers demand a broad set of attributes that include not only soft and hard skills but also personal traits that might influence job performance because this is what makes ready-to-work employees. H2: The massification of HE and the increasing supply of graduates contributed to changes in hiring criteria. H3: Countries vary in terms of their sectoral specialisation, and this leads to different concentrations of industries and services with a high (low) level of skills. We expect job ads to vary according to country specific features.

Data and methodology

Empirical evidence

Skill sets and other attributes

Table 1 gives a qualitative overview of the skill set categories we can expect to find in each of the five fuzzy clusters. One is a null cluster, i.e., no specific skill is required, which we labelled as undifferentiated. We found two clusters of foreign languages, one specifying English and other languages, and the other comprising different languages; we grouped the latter into other foreign languages. Finally, the other two clusters detail a considerable number of skills. According to the weight of each skill, we labelled one as soft skills (cluster II), and the other specific skills (cluster III). It is clear from the clusters that a knowledge of languages does not appear as relevant. Most probably, employers consider it to be a key prerequisite for skilled jobs.

Table 1 Skills and abilities demanded by employers 

Soft skills comprise requirements such as interpersonal skills, quality awareness, communication, planning and organisation, ability to work under stress, innovation, responsibility, personal qualities, attitudes towards work, orientation towards goals, analytical abilities, and flexibility and adaptability. Here, employers are searching for candidates with a broad skill set for both highly skilled and elementary occupations; they are offering a permanent and full-time job and expect candidates to be available to travel for work. This skill set prevails in financial, public administration and construction sectors.

The specific skills cluster comprises technical skills, IT, interpersonal skills, problem solving, communication, orientation towards goals, and analytical abilities. Additionally, employers require a master’s degree and professional experience (4-10 years) to work in highly skilled occupations in ICT and financial sectors.

However, the most striking finding is in the undifferentiated cluster. No skill prevails in this cluster, and it can therefore be referred to as a null cluster. Employers are looking for graduates for skilled non-manual and skilled manual occupations, offering them temporary and part-time jobs in public administration, the education system and construction. No experience is required, but ethics-related attributes are relevant.

The skill set varies according to the type of occupation. For high-skilled occupations, employers tend to specify both soft and specific skills. The lack of information on required skills is most frequently found for skilled non-manual occupations, which clusters clerical jobs, personal services, and sales workers. Next, we examine how countries’ specificities impact the way employers express (or not) the required skills.

Country specificities and skill sets

The weights of countries in each fuzzy cluster prompt questions about employability skills and the extent to which ads are merely a vehicle for disseminating job offers (figure 1). It seems clear that labour markets in each country demand a different set of skills. For example, whereas job ads in Romania only refer to requiring a foreign language, English is generally specified in Malta. Figure 1 shows that the undifferentiated cluster prevails in a set of countries, namely Spain, Croatia, Greece, Ireland, Portugal, Slovakia, and Sweden. These findings raise questions about the country specificities, i.e., context-related factors, that might affect job offers.

Figure 1 Weights of clusters in each country 

As we are unable to access employers’ motivations, or the reasons behind the job offers and required skills, we chose to look at some country characteristics that shed light on what drives the demand for skills. Bearing in mind the supply and demand arguments reported in the literature, we include a set of economic and labour market characteristics that help circumvent the lack of information contained in job ads in each country. The Eurostat provides data on youth unemployment, level of job vacancies and the proportion of employment in high-intensive technology manufacturing (HITM) and high-tech knowledge intensive services (HTKIS). The goal here is to examine the probable association between these variables and the weight of the undifferentiated cluster in each country.

We started by standardising all variables to obtain a single metric to compare countries. Figure 2 displays the countries with the combination of variables referring to the particularities of the employment characteristics. A set of countries shows a higher proportion of undifferentiated job ads (³ 0.50) and a low level of employment in HITM and HTKIS. This is particularly the case of Southern European countries (Spain, Greece, Portugal, and Italy). At the other extreme, we found Austria, Malta, Netherlands, Switzerland, Belgium, Luxemburg, and Estonia⁴ with a low proportion of undifferentiated job ads and a high level of jobs in HTKIS and HITM. The outcome for liberal economies (Ireland and UK) is puzzling as employers are less likely to specify skill requirements even though a significant share of jobs is in qualified sectors. We also see this in Sweden and Finland for HTKIS.

Figure 2 The employment in high-tech manufacturing and services and undifferentiated cluster 

Another economic characteristic that deserves scrutiny is the innovation path. While most of these countries are moderate innovators, others are modest (Romania) or strong innovators (Ireland and UK). We therefore explore the link between the innovation pattern and the lack of information in the job ads. The leader and strong innovation countries are more likely to specify skill requirements than the moderate innovation countries when hiring graduates.

Next, we analyse the labour market outcomes which make employers express skill requirements more (less). They include the job vacancy index and youth unemployment, which we associate with undifferentiated skill clusters.

We see that SE countries tend to form a cluster, with a high proportion of undifferentiated ads, a low level of job offers and a high level of youth unemployment. Ireland also falls into this cluster. As noted earlier for HITM and HTKIS, the cluster with Austria, Malta, Netherlands, Switzerland, Belgium, Luxemburg, Estonia, and Romania remains consistent.

Discussion

Job advertisements are used in the literature to learn about employability skills (^{Beblavý et al., 2017}). Access to these data has been eased through their availability in websites, which has led to much research on skill requirements that informs HE policy makers and institutions. However, our findings raise an important question: should jobs ads be seen merely as a means of publicising job offers? We believe that data on the countries where these offers are publicised can further our knowledge of the graduate labour market. Moreover, our research focuses on the ads that lack detailed information on skills and tries to identify potential explanations.

The literature underlines the lack of clarity (^{Kennan et al., 2006}) in the terms used to describe the skills and other attributes in job ads, as well as the enormous number of abilities (230); for simplification, we aggregated some of these into broad categories. The data contained in the job ads provided a picture of the attributes graduates must have in the sampled countries; we found employers demand a broad set of attributes that include soft and hard skills, and personal traits (hypothesis 1). The two clusters that specified these attributes showed that employers prefer “ready-to-work” candidates (^{Brown and Souto-Otero, 2020}). The cluster labelled soft skills includes interpersonal and communication abilities, in addition to aptitude to work autonomously, to plan and organise own work, to deal with stress, positive work attitudes and goal orientation, which symbolise work readiness attributes. All these behavioural traits help reduce monitoring costs, which means employees voluntarily comply with their employment contracts that are generally incomplete (^{Bowles and Gintis, 2000}). Employability skills therefore comprise abilities that reduce employers’ monitoring and training costs and influence employees’ commitment to high job performance.

The same rationale is found in the cluster designated specific skills. Employers demand job-oriented skills together with cognitive abilities (analytical thinking) and soft skills (interpersonal and communication). There is also an underlying preference for person-oriented abilities, such as personal qualities and positive work attitudes (^{Branine, 2008}). In sum, employers target job applicants with a broad set of skills (hypothesis 1), but some do not specify the skills and abilities in their job ads (^{Feldman, Bearden and Hardesty, 2006}).

Our empirical analysis pointed to one cluster with no significant skill (undifferentiated) and two with only foreign language. Arguments presented in the literature help explain this outcome. For example, employers are probably trying to create a pool of graduate candidates and fail to understand that credentials are poor proxies for occupational performance (^{Brown and Souto-Otero, 2020}). However, they may also continue to believe that HE credentials signal productive and valuable skills for their business (^{Tomlinson and Anderson, 2021}) and therefore a sufficient reason to hire. On the other hand, they may be unable to describe the specific attributes (^{Kennan et al., 2006}) or prefer not to specify them to keep their competitors in the dark (^{Napierala and Kvetan, 2023}).

Our findings illustrate that some particularities of labour markets and productive regimes are associated with undifferentiated job ads. The data showed that both Southern European (SE) countries and liberal markets have a combination of high youth unemployment, a low level of job vacancies and the highest levels of undifferentiated job ads (figure 3; table 2). In addition to these labour market conditions, low employment in innovative activities seems to drive the assignment of graduates to non-graduate jobs. In fact, the estimates from Tobit regression point to the allocation of graduates in skilled non-manual occupations without any specific requirement, and even to elementary jobs. Employers in these labour markets, where job offers are scarce and youth unemployment is high, have probably raised hiring criteria due to the increase in the supply of graduates (^{Thurow, 1976}; ^{Green and Henseke, 2017}), creating credential inflation (^{Collins, 1979}) (hypothesis 2).

Figure 3 The labour market characteristics of countries and undifferentiated job ads 

Table 2 Estimates from Tobit regression: undifferentiated cluster and country characteristics 

*p< 0.05; **p< 0.01; ***p< 0.001.

Though less evident and puzzling, the link between undifferentiated job ads and employment in HITM and HITS provides further insights. Employers provide detailed information on skill requirements when qualified employment in manufacturing is high, but not for high-tech jobs in the service sector. Could this be because it is difficult to specify the kind of tasks required in service-related activities? This is more visible in the UK, Sweden, and Finland, where there is a high proportion of HTKIS jobs. However, Ireland also has a large share of employment in HITM, casting doubts on this thesis. It should be noted that underemployment grew most in Ireland (Anglo) and Sweden (Nordic) as well as in SE countries (^{Green and Henseke, 2017}) (hypothesis 3).

Growing underemployment is likely to intensify because on one hand young people strive to acquire credentials to be well positioned in the labour queue (^{Thurow, 1976}; ^{Tomlinson and Watermeyer, 2022}) and, on the other, the skilled jobs generated by economies do not grow at the same pace as the supply of graduates. As a result, there is a greater risk of wasting skilled and scarce resources (^{Marques, Suleman and Costa, 2022}) that could be used by countries to become more competitive and ensure well-being. This also raises doubts about HE’s responsibility for the employability of graduates.

The HE reform is based on the belief that the supply of graduates is the solution to skill problems and provides unprivileged social groups with the opportunity to access HE. European countries’ HE target and associated benchmarking fail to take into account demand and the varying skills required in the labour market. Our study suggests that the economic system has been unable to generate enough jobs to guarantee that available skills are used and ultimately that individuals get a fair return on investments in HE. The job ads revealed that the growing supply of graduates in some countries may simply be affecting the hiring criteria for HE graduates since no explicit skill is required. Therefore, young people invest in HE to obtain credentials as a response to these increasing requirements. In addition, our study shows that there is a mismatch between the supply of graduates and the pace of innovation in the productive system. While some graduates access better jobs in qualified sectors, others must accept whatever offer is available even if it is a non-graduate job. This mismatch fosters segmentation and perpetuates social inequalities among holders of HE diplomas, notably between women and men (^{Alves and Morais, 2021}).

Concluding remarks

Our study sought to go beyond the simple identification of graduates’ attributes described in job ads. The findings revealed that we can glean a better understanding of job ads if we consider certain factors such as the employers’ ability and/or willingness to specify skill requirements; the productive regime, which influences the level of employment in high-tech activities; a country’s level of innovation; and the type of occupations assigned to graduates. Other drivers that deserve scrutiny include the influence of economic conditions on the hiring criteria and search for skills. Future research should include longitudinal data on job offers to explore whether recession and expansion stages affect employers’ hiring criteria and consequently graduates’ employability.

We showed that it is necessary to delve into the demand side when discussing employability. Our aim was to demonstrate that unless the productive system is transformed, skills are wasted and the potential benefits of HE reduced. Policy makers should thus encourage employers to upskill jobs and not limit themselves simply to raising the skill requirements, which leads to credential inflation. In this situation, young people just participate in HE to obtain the credentials they need to get to the front of the labour queue.

However, this picture is incomplete if we overlook the social inequalities in the access to HE and acquisition of skills, as well as in benefiting from these attributes. The fate of young people before and after graduation is influenced by social class, family background, race, gender, and cultural matching, as well as by the type of HE institution that confers their degree.

On the other hand, although HE intends to provide a solution to skill problems, its benefits depend on how organisations take advantage of the supply of skills. In some cases, these investments can be counterproductive: graduates crowd out non-graduates from less demanding jobs and, consequently, expose non-graduates to unemployment or professional exclusion. Only a combination of education, economic, labour market and social policies can produce a win-win solution.

The data from job ads and country specific features allowed us to unveil outcomes that are usually less explored in the literature. However, our sample is relatively small. Further research should explore the potential of artificial intelligence to collect and classify ads and resubmit these research hypotheses to empirical tests with big data on both European and non-European countries.

Funding

This work was financed by Portuguese funds through FCT - Foundation for Science and Technology in the framework of the project no. 030016, “BRIGHET- Bringing together Higher Education, Training, and Job Quality”. Reference: PTDC/SOC-SOC/30016/2017.

Acknowledgements

We grateful acknowledge the comments from reviewers and participants of 6th International Conference on Higher Education Advances (HEAd’20), Universitat Politecnica de València, Spain, 2020.