1 Introduction

Investing in higher education is a key focus of both short-term and long-term national and international development agendas. It is closely tied to enhanced economic and social stability (Raileanu-Szeles & Tache, 2016). Education, in broad terms, is considered a form of human capital investment that contributes to economic growth (Molina & Rivadeneyra, 2021).

As a result, so far higher education has been analysed from different perspectives. For instance, the potential gender differences in overeducation (Boto-García & Escalonilla, 2022), the potential impact of education level on earnings (Card, 1994; Lemieux, 2006; Meghir & Palme, 1999), the flow of individuals from the education system to the labour market (Albert Verdú et al., 2003), the determinants of the unemployment duration after the university (Borra et al., 2009), the youth employment trajectories (Verd et al., 2019), the level of individual productivity based on signals such as educational attainment, work experience (Spence, 1973). In addition, the relationship between postgraduate education and the labour market has been examined (Morikawa, 2015).

Recently, increased participation of people in higher education and the big difficulty and uncertainty for graduates to get a job corresponding to their education level has increased researchers’ interest in analysing the transition of graduates from the university to the labour market. For example, Stewart (2007) and Heckman and Borjas, (1980) point out the importance of this transition and suggest that poor early career outcomes can adversely affect later outcomes. Therefore, in the study of this transition process, the possible educational mismatch can also be examined. The educational mismatch means the relation between the graduate's educational background or field of study and the competences required to do the job the worker has accepted. The lack of correspondence between education—employment and the oversupply of graduates diminishes the well-being of graduates because it promotes, among other aspects, a reduction in expected labour income associated with higher education (Duncan & Hoffman, 1981; Freeman, 1976; Iriondo Múgica, 2022).

There are usually two kinds of educational mismatch described in the literature: a horizontal or a vertical mismatch. According to the International Labour Organization, a horizontal mismatch occurs when a person's degree, abilities, or field are inappropriate for the position. In contrast, vertical mismatch occurs when the education or qualification level is either lower or higher than what is necessary (ILO, 2014).There are several ways to characterise a horizontal and vertical mismatch and their nuances. According to Pérez Navarro, (2021), Alam et al., (2020), Verhaest et al. (2017) and Wolbers (2003) horizontal mismatch generally refers to a mismatch between the worker's field of study and the requirements of their job (skills and field of knowledge). Contrarily, the vertical mismatch is usually categorized as either overqualification or underqualification and is related to the degree of qualification required for a position (Alam et al., 2020; Pérez Navarro, 2021). In general, educational mismatches occur when graduates work in different qualification-level jobs or in areas other than the ones they studied. In this study, we are going to refer to overqualification as vertical mismatch, and horizontal mismatch when the type of field of education is different from the one that is required.

The extensive body of research on overeducation has generated diverse methodologies for measuring both vertical and horizontal mismatch over the past five decades. The discourse delineated by Boto-García and Escalonilla (2022), Verhaest and Omey (2010), Verhaest and Omey (2006) introduces three distinct avenues for assessing overeducation: job analysis (JA), Individual self-assessment (ISA), and the Statistical approach (SA). Each approach showcases its unique merits and drawbacks. The JA method involves scrutinizing job requirements and comparing them with individuals' educational qualifications and skills, as highlighted by Hartog (2000). Nevertheless, the potential obsolescence of occupation-related information, especially concerning rapidly evolving and technologically inclined occupations, poses a challenge (McGuinness, 2006).

In contrast, while the ISA method emerges as the most prevalent approach, it remains vulnerable to subjectivity and potential overstatement of job requirements by respondents striving to elevate their job standing. Finally, the SA gauges overeducation by comparing an individual's education level with the average or most prevalent educational attainment within a specific occupation. However, a downside of this method is its inclination to reflect educational qualifications without distinguishing between the actual skills needed to perform a job and those merely necessary to secure it.

The loss of well-being due to education-employment mismatches prompts research on this field. The importance of studying this phenomenon has been highlighted by seminal works suggesting a macroeconomic and microeconomic perspective to analyse the relationship between mismatches and human capital. From a macroeconomic perspective, increasing investment in human capital is positive and has a beneficial effect on the economy by contributing to the economic growth of countries, productivity, and competitiveness. Meanwhile, from a microeconomic standpoint, investment in human capital may not necessarily increase the well-being of individuals when educational mismatches exist in economies. Primarily vertical mismatches result in a loss of human capital as they reduce educational returns, i.e., the increase in labour income associated with an additional year of schooling (Duncan & Hoffman, 1981; Freeman, 1976).

The study of vertical and horizontal educational mismatches is pivotal as it clarifies the alignment between education/training and the skills required in the labour market. It serves as an indicator of job creation concerning individuals' skill sets. Analysing educational-to-work mismatches among young individuals significantly enhances understanding of how graduates transition into the labour market. This analysis is crucial in evaluating the efficacy of higher education investments, assessing workforce readiness to meet market needs, and mitigating the underutilization of graduates' skills (Boto-García & Escalonilla, 2022). As a result, this mismatch may have a significant impact on each country's policies and education initiatives, as well as providing insight into the relevance and importance of education and educational levels.

The vertical and horizontal educational mismatches have been present in the majority of the economies. In the EU for instance, the average of people between 20 and 64 years old with overqualification was around 21% between 2009 and 2019 (Eustat, 2022). Our analysis is focused on Spain, as it presents the highest average percentage in the same period with around 36% of people in mismatch through overqualification. Nevertheless, this average percentage has been rising in the last decade in almost all European countries, principally in Slovakia, Greece and Malta, where an increase in overqualified workers between 2009 and 2019 was more than 10%. The study of the Spanish case is particularly relevant given the persistently high rates of youth unemployment in the country, which can be compared to similarly elevated rates in some Southern European countries, such as North Macedonia (35.6%), Greece (35.2%), Spain (32.5%), and Italy (29.3%) in 2019 (Eustat, 2021). Moreover, Spain shares common characteristics and challenges with other Southern European countries (Marcenaro-Gutierrez & Lopez-Agudo, 2021), suggesting that, while the external validity of the study may be limited, the Spanish case could provide valuable insights into educational-to-work mismatches in certain European countries (Green & Henseke, 2017).

The high percentages of mismatches and the young unemployment represent important topics of political agendas in many countries. It is reflected by the goals proposed in the Lisbon Strategy (2000–2010) of the EU, its successor, the Europe 2020 Strategy (2011–2020) and the United Nations’ 2030 Agenda for Sustainable Development. These political agendas focus on the quality of employment; particularly for the youth.

Therefore, this research aims to analyse the effect of a master’s degree on the probability of being in educational mismatches (vertical or horizontal) in the Spanish labour market. For this purpose, we use the University Graduates' Labour Market Insertion Survey 2019 that was carried out in Spain applying specific discrete choice models.

This study builds upon Pérez Navarro's (2021) research, focusing on the misalignment between education and job requirements in Spain's job market. To address this critical issue, we formulate the following research questions:

  1. 1.

    (RQ1) How does the probability of experiencing a vertical mismatch vary based on factors such as the attended university's public or private status, field of study, full-time employment status, type of contract, and parental educational level?

  2. 2.

    (RQ2) What role does the field of study and type of occupation play in preventing situations of horizontal mismatches?

  3. 3.

    (RQ3) Does holding a master's degree significantly reduce the likelihood of an individual being overqualified for their position, thus decreasing vertical mismatches?

  4. 4.

    (RQ4) Is the attainment of master's level studies among the most influential factors in mitigating horizontal mismatch?

To our knowledge, unlike previous studies focusing solely on the discrepancies between education and employment and their determinants, our research is the first to concurrently analyse horizontal and vertical mismatches among Spanish students, including those with a master's degree. Our findings regarding RQ1 and RQ3 indicate that attending a public university and not pursuing a master's degree increases the likelihood of a vertical mismatch. Conversely, studying scientific disciplines such as health, pure sciences, engineering, and architecture reduces this probability. Concerning RQ2 and RQ4, not having a master's degree and majoring in health sciences decrease the likelihood of a horizontal mismatch. Regarding horizontal mismatch, lower-level occupations, part-time jobs, and majors in humanities, arts, or pure sciences are more likely to experience this mismatch.

The research indicates that an oversupply of graduates exacerbates credential inflation, and enhancing university admission procedures could alleviate this concern. To tackle these mismatches, it is advisable to enhance collaboration between universities and the private sector, offer internships for master's degree students, and diversify educational offerings. Adjustments to educational programs should adhere to the rigorous standards set by Spain's National Agency for Quality Assessment and Accreditation (ANECA). Any institution or university planning to introduce a new master's program must adhere to ANECA's regulations and criteria.

This paper is organised as follows. Section 2 reviews the literature; Section 3 contains data. Section 4 introduces the results and discussion. Section 5 presents the conclusions, and finally, Appendix 1 describe the methods.

2 Literature review

The vertical and horizontal mismatches and their impacts have been analysed for a long time in many different countries. Similarly, a considerable amount of literature has been published linking education with the investment in human capital. As first studies on this topic can be considered Freeman (1976), Becker (1964) and Mincer (1974).

Focusing on the mismatch pioneer literature carried out in the US, Freeman (1976) analysed the relevance of the graduate’s cohort size in the transition to employment, the gap in salaries, and the successful job conditions. His results suggest that there is an important oversupply of college-educated labour, and point out that this is one of the reasons why having a university degree did not predict economic success or prevent a vertical mismatch in the US since the 1970s. Smith and Welch (1978) re-evaluated Freeman's (1976) findings, concluding that there was no overeducated American population, but there was a relevant increase in the population due to the baby boom, which could explain why university degrees did not increase the likelihood of better employment conditions.

A similar analysis of the US labour market regarding the vertical mismatch and its wage effects was made by Duncan and Hoffman (1981). They showed that a vertical mismatch produces an important reduction in wages, which was even higher for the black population. According to their results, the educational level increases the returns of education on wages, but its effect is lower than expected. A recent study by Shin and Bills (2021) analysed the factors that have influenced vertical and horizontal mismatch over time in the US. labour market. Their results suggest that even though individual’s characteristics are possible causes of vertical and horizontal educational mismatch, the occupational factors such as information supply and demand, company size or job sector, appear to contribute the most to explain these mismatches.

Educational mismatch and its effects have also been analysed in Asia. For instance, Liu et al. (2021) examined the factors that influence the overeducation and the wage effects in China. The findings suggest that individuals with a higher quality of higher education have less vertical mismatch and fewer reductions on wages caused by vertical mismatch. Moreover, they concluded that an academic curriculum with highly practical components could help to reduce the overeducation. In a similar study, Morikawa (2015) showed the following positive effects of postgraduate degrees on labour market outcomes in Japan. Firstly, postgraduate studies produce higher employment rates, mainly for women and older people. Secondly, the wage of university students increases by roughly 30%—40% with a postgraduate education.

The educational mismatches are also analysed for the EU countries. For example, McGuinness et al. (2018) synthesized and discussed different definitions of vertical and horizontal educational mismatch in applied literature. This study highlighted that in general, the vertical mismatch was higher for younger people (24 to 29 years old) in their first job. From the graduates’ point of view, the level of education in the first employment was higher than what was required for the real job development tasks.

In Spain, Corominas et al. (2010) evaluated the match between education and jobs for Catalans graduates. Their results showed that a good match was dependent on the field of study. For instance, the highest percentages of the mismatch were found in Social Sciences and Humanities. However, students' perception of the quality of education-job fit improved after three years of graduation as their promotion opportunities improved.

In another case study, Esteban and Vidal (2020) studied the vertical mismatch in Spain focusing on gender differences. They showed that having more than one university degree and having a good knowledge of technology reduces the job-education mismatch more for women than men. In another recent study carried out in Spain, Iriondo Múgica (2022) analysed the impact of horizontal and vertical mismatches on salaries. The results showed that the two mismatches produced an important reduction in salary and represented a waste of resources in education. However, the reduction in the salary is lower in the horizontal than in the vertical mismatch. Their recommendations implied putting more emphasis on labour market demands to improve career prospects for university graduates.

3 Data

The analysis in this study is based on the Labour Insertion Survey for Recent University Graduates (EILU) developed by the National Statistics Institute (INE, 2019). This survey aims to provide information on the employment situation of university and master’s degree graduates and their integration into the labour market. Their target population sample was young people after 4 years of their graduation, and the database used corresponds to the survey conducted in 2019.

This survey was administered to the whole territory of Spain by the National Statistics Institute. The information provided in the survey contains socio-demographic and socio-economic characteristics of graduates, although the majority of it focuses on their first jobs after graduation and their current jobs four years after. Some information provided is objective, and verified with administrative sources like gender, age, place of residence, etc. Other information is subjective; based only on the individual answers.

The data were collected between July and December 2019 and that is why they correspond to graduates in the academic year 2013/2014. Our sample is made up of two subgroups represented by students who achieved an undergraduate degree (31.651) and those with a master’s degree (11.483).

Table 1 presents the explanatory variables used in the two models. There are only two continuous variables, Experience, defined at the bottom of Table 1 representing years of experience in the last job in years and ranging between 0 to 39, and Unobserved skills represent the omitted relevant variables. Further elaboration regarding this variable is provided later in the text. The remaining variables are dummy variables. They are presented in the main block of Table 1. This is at the same time split into two sub-blocks. The upper of these blocks presents the explanatory variables that are included in the two models (the binary and ordered logit) and the lower block presents the explanatory variables that are included in the ordered logit devoted to the horizontal mismatch only. All variables are presented with their label, description, codification and proportion in the data set.

Table 1 Description of variables mismatch vertical and horizontal
Full size table

The definition of the majority of the dummy variables is self-explanatory. Moreover, the variables Speaks 2, 3, 4 or more languages represent the effect of speaking two, three, four or more languages with respect to the reference category of speaking the native language only. The variables ICT_intermediate and ICT_expert show the effect of the ability to use a computer or other computing devices at an intermediate and expert level respectively concerning to basic level.

The variables Permanent_contract and Temporary_contract, both of them related to the current professional status of graduates, are compared with a trainee position (worked as a traineeFootnote 1 in their first job) that is the benchmark level. In the same line, the benchmark level in the field of knowledge of the degree is Science. The occupation Technical is the benchmark category for the variables of the second sub-block related to the type of occupation (Manager, Professional, Clerical support workers, Service and sales workers, Skilled agricultural, Craft and related trades workers, Machine operators, Elementary occupations.

Lastly, the challenges associated with omitting relevant variables that can lead to endogeneity issues, have been pivotal in both labour economics and the economics of education (Leuven & Oosterbeek, 2011). The lack of data on individual ability is a notable gap that could substantially affect our results, particularly in terms of job market entry. Without this data, we have adopted the method proposed by Chevalier and Lindley (2009) to overcome this issue. This method uses wage residuals to represent unobserved characteristics of individuals, alongside observed variables. These residuals help to identify certain wage influences not accounted for in the models examining vertical and horizontal job-education mismatches. In this paper, we use unobserved skills as proxies for unobserved characteristics, inferred from the wage regression residuals based on socio-economic and employment factors. This is one effective way to identify the hidden aspects of human capital that have a big influence on labour market outcomes.

4 Results and discussion

The methodology employed includes a binary logit model for the assessment of vertical mismatch and an ordinal logit model for examining horizontal mismatch. The theoretical foundations for these methods are detailed in Appendix 1. Tables 2 and 3 present the estimations of the binary and ordered logit models for the analysis of vertical and horizontal mismatches respectively.

Table 2 Binary logit on vertical mismatch estimates
Full size table
Table 3 Ordered logit on horizontal mismatch estimates
Full size table

The first column in Table 2 lists the names of the explanatory variables, while the second, third, and fourth columns display the estimated coefficients, the standard error, and the individual significance of each variable respectively. Regarding the estimates of the binary logit model representing the vertical mismatch presented in Table 2, it can be seen that generally, the vast majority of the explanatory variables are significant at a 5% level. The only exceptions are the variables representing the ability to use a computer or other computing devices at an intermediate level (ICT_intermediate) that is not significant and the gender variable (Female).

Similarly, Table 3 presents the estimates of the ordered logit model of the categories of horizontal mismatch. The majority of the explanatory variables are significant at a 5% significance level except for the dummy variables representing gender, nationality, studies abroad, private university, type of contract, and parents' educational level. The main conclusions of the results presented in Tables 2 and 3 are drawn on the interpretations based on the changes of the predicted probabilities defined in (5) for the vertical and (11) for the horizontal mismatch. The predicted probabilities used to compute these changes depend on a specific value of the explanatory variables. That is why we establish a benchmark individual defined as a woman, less than 30 years old, Spanish, with three years of working experience, who speaks two languages, with the ability to use a computer or other computing devices at an intermediate level. She completed part of her studies overseas, at a private university, with a degree in the field of Social and Legal Science, a master’s degree, a full-time job, a permanent contract and with type of occupation: Professional. Furthermore, the individual will have parents who have completed higher education, with the variable for unobserved skills adjusted to the median value.

Table 4 shows the changes in the predicted probability of a vertical mismatch for the benchmark individual as a result of a change in a particular explanatory variable obtained by (5). The benchmark individual's characteristics are displayed in the first column. The corresponding probability to be in vertical mismatch is 0.091 and it is presented in the first row of Table 4. The first row also shows this value graphically together with the corresponding confidence interval.

Table 4 Effects on the probability of being in vertical mismatch
Full size table

The second column presents the analysed change in one particular explanatory variable and the last column displays the predicted probability for the benchmark individual when one of the characteristics is changed. The dummy variables are changed from zero to one, or from one to zero depending on the values for the benchmark individual. The quantitative variable job experience is changed from 3 to 9, which means one standard deviation change from the median value, and the variable for unobserved skills was adjusted from its median to the median plus one standard deviation for analysing both vertical and horizontal mismatches (Table 1).

The graphical representation of the probabilities in the last column includes the confidence intervals of the probability. For example, if the gender of the benchmark individual changes from female to male the change in the predicted probability to be in vertical mismatch is according to (5) defined as 0.095—0.091 = 0.004. It can be easily seen in the graphical representation of these predicted probabilities and their corresponding confidence intervals in Table 4 that this tiny change is smaller than the confidence intervals and cannot be considered relevant. That is why we draw our conclusions on the highest differences that indicate a significant difference.

First, we focus on the responses to RQ1 and RQ2, which are related to the Tables 4 and 5. In attempting to address RQ1, an examination of the final column in Table 4 reveals that the characteristics exerting the greatest influence on the probabilities of experiencing a vertical mismatch relative to the benchmark individual include: university type, field of study, pursuit of master's degree studies, employment status (full-time), contract type, parental educational level, and potentially other unobserved skills.

Table 5 Effects on the probability of being in horizontal mismatch
Full size table

Similarly, to tackle RQ2, Table 5 offers information similar to what is shown in Table 4. However, in this instance, Table 5 presents the estimation of the ordered logit model, which examines the likelihood of a horizontal mismatch. The only discrepancy is that Table 5 displays the predicted probabilities for the three analysed categories: no horizontal mismatch, weak horizontal mismatch, and strong horizontal mismatch. Notably, the variables exerting the greatest influence on the likelihood of experiencing a strong horizontal mismatch include various fields of study, current job occupations, and primarily, the possession of a master’s degree.

These findings underscore the relevance of university type in assessing the likelihood of a vertical mismatch. Attending a private university correlates with a decreased probability of experiencing a vertical mismatch. This observation aligns with the conclusions drawn by Verdú and Davia (2018), who suggested that differences between universities may be influenced by academic and institutional factors. Additionally, findings from Flores (2020) indicate that private education may offer more practical benefits compared to public institutions. Specifically, graduates from private institutions may have enhanced career prospects post-graduation. Similar conclusions were drawn by Brunello and Cappellari (2008) in a comparable study conducted in Italy.

Addressing in more detail RQ1, a significant finding concerning vertical mismatch suggests that pursuing degrees in more experimental sciences (Engineering, Architecture, Health Science, and Science) appears to reduce the likelihood of experiencing vertical mismatch. Another significant factor influencing vertical mismatch, as highlighted in Table 4, is holding a full-time job. Engaging in part-time work increases the likelihood of experiencing a vertical mismatch. This finding is consistent with Wolbers (2003), who noted that full-time employment enhances the likelihood of achieving a well-matched job. Another relevant attribute for explaining vertical mismatch is the type of employment contract. Having a permanent contract appears to increase the likelihood of being overqualified for a position. This observation aligns with findings by Pérez Navarro (2021), who suggested that graduates in internships are more likely to work in roles related to their studies compared to those with permanent contracts. Additionally, a higher educational level of parents emerges as a significant factor in reducing the probability of vertical mismatch. This outcome is also supported by Barone and Ortiz (2011), who demonstrated that graduates from families with lower academic backgrounds are more susceptible to experiencing a vertical mismatch.

However, concerning Table 5 and addressing RQ2, a similar conclusion can be drawn regarding horizontal mismatch. Specifically, graduates in the social sciences, arts, and humanities are more prone to being overqualified for job positions compared to graduates in engineering, architecture, pure sciences, or health sciences. Furthermore, graduates in health sciences seem less inclined to pursue careers outside their fields of study compared to graduates in the arts, humanities, and pure sciences. These findings align with those presented by Pérez Navarro (2021) and Corominas et al. (2010), who identified a higher level of mismatch between university-acquired skills and workplace requirements for Spanish students, consistent with our findings. Moreover, different types of occupations emerge as significant determinants with considerable impact on the likelihood of experiencing horizontal mismatch. According to the data presented in the right-hand side block of Table 5, individuals in occupations outside the professional category (i.e., lower-level occupations) exhibit an increased probability of experiencing horizontal mismatch.

Following the analysis and discussion of RQ1 and RQ2, attention now turns to RQ3 and RQ4. Our focus shifts to the variable 'Master’s degree' as depicted in both Tables 4 and 5. Notably, this variable emerges as the second most influential factor impacting the likelihood of vertical mismatch (Table 4) and shows the most pronounced effect on the probability of horizontal mismatch (Table 5).

It appears that holding a master’s degree has the opposite effect on the two types of mismatch. Having a master’s degree decreases the probability of vertical mismatch, but it increases the probability of weak or strong horizontal mismatch. To analyse the effect of having a master’s degree, we use the continuous explanatory variable Experience to analyse its effects graphically. Tables 4 and 5 present the changes in probabilities for a change of job experience from 3 to 9 years only. Figures 1 and 2 present the effects of this variable on the predicted probabilities for the whole range that is from 0 to 39 years. These predictions are presented in the two figures with their corresponding 95% confidence intervals. These probabilities are presented for a benchmark individual with and without a master’s degree to analyse its effect in detail.

Fig. 1
figure 1

Experience's effects on the probability of vertical mismatch: Individuals with and without master’s degree

Full size image
Fig. 2
figure 2

Experience's effects on the probability of horizontal mismatch with and without a master’s degree

Full size image

The insights provided by the previous responses to RQ3 can be further elaborated upon with the aid of Fig. 1, which illustrates the influence of both years of experience and the possession of a master’s degree on the likelihood of vertical mismatch. The results suggest that obtaining a master’s degree significantly reduces the probability of vertical mismatch. The likelihood of being overqualified for a job increases with the lack of work experience, which is exacerbated if the graduate does not have a master’s degree. This implies that more years of experience decrease the probability of vertical mismatch. However, obtaining a master’s degree consistently tends to reduce the probability of overqualification. A similar result was reported by Baquero and Ruesga (2019). They suggested that pursuing a master’s degree has positive effects on job placement in terms of salary and suitability of the role in both the short and long term, and it also increases the probability of being suitably qualified for a job. Based on these results, possessing a master’s degree in Spain has positive effects on reducing the number of overqualified graduates. Concerning this, Alba-Ramírez (1993) suggested that vertical mismatch is mainly a short-term problem of the Spanish population that decreases in the medium and long term.

One of the biggest problems related to over-qualification in Spain is the low inadequate of the education level salary. This difference between real and expected levels of salary is around 8% to 13.6% according to Iriondo Múgica (2022) and McGuinness and Sloane (2011) who call this situation “a waste of resources”. In this regard, and according to Fig. 1, having a master’s degree in Spain reduces the probability of over-qualification and, therefore, it could contribute to a better and adequate integration of graduates into the labour market.

Figure 2 shows the effect of years of experience and the impact of having a master’s degree on the probability of being in horizontal mismatch, that is, working in an area of expertise different from the studied field. It presents only the two extreme values of the explained variable of the model presented in (6)-(10) and Table 3, specifically “no horizontal mismatch” and “strong horizontal mismatch”. It can be observed that the effect of having a master’s degree runs in the opposite direction to the effect identified in the vertical mismatch.

A strong horizontal mismatch means that graduates work in a completely different area than the one they studied. Figure 2 extends the response to RQ4 graphically, indicating that the likelihood of experiencing a strong horizontal mismatch is greater for students holding a master’s degree compared to those without one. The difference in probability of being in a strong horizontal mismatch with and without a master’s degree remains approximately the same for all values of working experience presented in axis \(X\) in Fig. 2. That means that gaining more working experience does not help to reduce the probability of being in a strong horizontal mismatch. This is an expected result as with increasing working experience a change of job to a different field becomes costlier.

It is supported, for example, by Budría and Moro-Egido (2008) who suggested that the capabilities of graduates are heterogeneous in terms of skills. For this reason, after studying for a master’s degree, several graduates are equipped with tools that would allow them to work in different disciplines. In the same vein, Alam et al. (2020) claimed that a master’s degree program aims to produce high-level employment skills to boost a country's economy. In this regard, a master’s degree could be a tool to diversify the skill set of graduates. Concerning that Iriondo Múgica (2022) concluded that the salary penalty for graduates in the vertical mismatch is greater than a possible salary penalty due to the horizontal mismatch as long as the educational qualification level of the workers is adequate, that is when there is no vertical mismatch. In these cases, the salary of graduates who work outside their field of study can be even higher than those who do. Based on these results, studying for a master’s degree in Spain positively affects the employment match of graduates four years after graduation because it decreases the probability of overqualification and provides graduates with skills required to work in various areas.

However, it is important to consider mechanisms that make easier access for new graduates into the labour market in the short term after graduation. Baquero and Ruesga (2019) suggested that to encourage the earlier and correct labour market integration of recent master’s graduates, greater efforts should be made to implement programs that permit easy contact between companies and graduates. One strategy could be the promotion of internships and closer collaborations between universities and companies. This proposal gains particular significance considering that, for Spanish students, access to internships during their studies represents the most valued aspect when selecting a master's program (Cabrera et al., 2022).

In the current context, given that internships can emerge as a crucial component for enhancing employment opportunities for graduates, it is fundamental to delve deeper into the knowledge about the impact of internships on educational mismatches as information becomes available. One alternative could be to analyse the impact of the various modalities of internships, considering a more precise definition of "good internships" (O’Higgins & Caro, 2021). According to Tzanakou et al. (2021), internships can be categorized into three main groups: those integrated into academic programs, often a requirement for graduation; those driven by labour market policies aimed at improving employability, frequently supported by governmental backing; and those undertaken directly in the labour market, outside of educational or governmental frameworks. Another relevant alternative for assessing the impact of internships could be to differentiate between paid and unpaid internships, as paid internships might reflect a greater commitment and a better organizational structure of the companies, which seemingly translates into greater effectiveness as a bridge to stable employment (O’Higgins & Caro, 2021).

Alam et al. (2020) emphasized the importance of increasing collaboration between industries and universities to reduce the educational mismatch gaps, particularly considering the rapid technological advances that are currently taking place. Master's degree could be useful for helping students and companies adjust to such fast technological developments. To this end, it could be of interest for companies and students to have an agreement allowing internships. First, this will make the future work placement of graduates easier, and second, it will make human capital with the skills required for the job available to companies. Dual vocational training programs are a successful and growing example of this form of training in Spain. Under these programs, students spend part of their learning time at educational institutes as well as working directly in companies under a mutual agreement (Vocational Training Observatory, 2022).

5 Conclusion

This research examines the likelihood of job-education mismatches, both vertical and horizontal, among Spanish graduates, assessing the prevalence of these mismatches four years after graduation. Additionally, the study evaluates the impact that completing a master's degree program may have on such mismatches. To achieve this goal the Labour Insertion Survey for Recent University Graduates carried on in 2019 in Spain is used. The methods used are a binary logit model for the vertical mismatch, and an ordered logit model for the horizontal one. The study also presents how the predicted probability of both of the mismatches changes for an individual with and without master’s degree studies.

Analysing the vertical mismatch, we come to the conclusion that various characteristics of Spanish students can raise the probability that they will be overqualified for a job position. The characteristics that seem to increase this likelihood the most are attending a public university, and not pursuing a master’s degree. Moreover, it appears that studying certain scientific disciplines, such as health, pure science, engineering, and architecture, tends to lower the probability of this vertical mismatch. Nevertheless, the most important conclusion is that studying for a master’s degree has a positive impact on reducing the overqualification of Spanish students and this impact becomes even bigger as they gain more work experience. The findings may be interpreted as evidence that higher educational attainment acts as a signalling mechanism in the job market, a theory proposed by Spence (1973). Employers may use the educational qualifications and work experience of candidates as indicators of their expected productivity. Therefore, individuals with recent master's degrees might have an advantage in the job market by capitalizing on these educational signals. However, it's also crucial to consider that the professional success of graduates and the beneficial effects of higher education are influenced by the individual's behaviours, innate abilities, and personal preferences in education, particularly in terms of how they value knowledge and their career ambitions, as suggested by Oosterbeek and van Ophem (2000) and Lazear (1977).

Focusing on the horizontal mismatch, those with lower levels type of occupation, those who perform part-time jobs, and those with studies in humanities, arts, or pure sciences have a higher probability of finding employment in fields unrelated to their fields of study, that is to be in a strong horizontal mismatch. On the contrary, not having a master’s degree and majoring in a discipline linked to health science decreases the probability of a strong horizontal mismatch.

Focusing on the main causes of the mismatches, Barone and Ortiz (2011) suggested that one of the main causes is the oversupply of graduates that creates so-called credential inflation (already mentioned by Thurow, 1975). They concluded that improvements in the university admission processes can reduce this credential inflation. In the same line, Müller and Gangl, (2003) advocated for a more stratified educational system resulting in a greater match of the university graduates and labour market necessities that could lead to a reduction of the excessive demand for higher education. This approach would generally be tied to promoting and improving alternative educational levels, such as vocational education and training, which has already shown success in lessening the demand for university degrees for example in Germany or the Czech Republic (population under 35 years with tertiary education levels of 30% and 44% respectively (Eurostat, 2020a)). A more optimistic view on this problem is described by Becker (1980) who stated that overqualification can be corrected by the market in the short term.

This is, probably, not the case in Spain as, according to our results, both the vertical as well as horizontal mismatches were present in 2019 when the analysed data were collected. This issue should gain higher priority for Spanish policy makers as the overqualification rate in Spain is the highest among all the EU countries (Eustat, 2022).

In this line, according to the results of our study, studying for a master’s degree seems to have a positive impact on improving the match between graduates and the labour market. It reduces the probability that graduates may be overqualified for a position (vertical mismatch), while also providing them with skills allowing them to work in other fields (horizontal mismatch).

This result is consistent with Spence (1973) but it may also resonate with Rubb’s (2003) theory, which posits a wage premium for individuals whose educational achievements surpass the explicit requirements of their job roles. Rubb (2003) highlights that roles requiring at least a bachelor's degree tend to be associated with more effective human capital management than roles that do not require such qualifications. Considering this, our analysis of Master's degree holders, who are typically employed in positions that demand at least a bachelor's degree, suggests that companies that manage their human capital well are likely to experience fewer vertical mismatches for employees with a Master's degree. Moreover, in cases where a horizontal mismatch arises, it's plausible that these individuals could benefit from the wage premium described by Rubb (2003), attributed to the sophisticated human capital management practices implemented by their employers.

In spite of the fact that this type of education seems to have an important effect on the two mismatches, the adequate provision of such programmes remains in Spain an issue at the national level. This is because offered master’s degrees seem to be distributed across different field inadequately. Table 6 shows the distributions of master’s degree provision among different fields in Spain between 2013 and 2019. As can be seen, this distribution has practically remained the same over these years. Furthermore, apparently, in Spain there is a lower specialisation in STEM fields (Science, Technology, Engineering and Mathematics), comparing with the rest of EU (CYD Foundation, 2023). This fact can be easily seen by comparing the share of master’s graduates in STEM areas compared to other close EU countries. More specifically, the number of STEM graduates per thousand inhabitants is 5.2 in Spain, 11.5 in France, 7.2 in Portugal and 7.1 in Italy. Even the EU countries average 7.3 is much higher than the Spanish figure (Eurostat, 2020b).

Table 6 Distribution of master’s degree by field of study in Spain
Full size table

Moreover, educational mismatches are related according to Somers et al. (2019) to the degree of general or specific skills of the employees, and, similarly, Robst (2007) suggested that health or STEM careers tend to provide their graduates with more specific knowledge, which reduces the likelihood of a horizontal mismatch. In this line Salas-Velasco (2021) pointed out that educational mismatches are less likely in fields such as health because its knowledge is not easily transferable to other activities in other more general fields. Thus, these graduates are less likely to seek employment in other areas because of this limited transferability.

Our findings suggest that expanding the range of master’s degrees in fields like engineering, architecture, health science, and science in Spain may help mitigate vertical mismatches, and that increasing master’s degrees specifically in health science could address horizontal mismatches.

The expansion of master’s degree offerings is particularly relevant in the context of the Bologna reform, which has been implemented by Spanish universities since around 2005. This reform replaced the traditional “Licenciatura” and “Diplomatura” degrees with degree structures in line with the rest of Europe, including Bachelor's, Master's, and Doctorate degrees. The intention was to make education more applicable to the labour market. Despite these efforts, there is still a notable gap between the skills of graduates and the jobs available, which poses ongoing challenges within the Bologna process and has led to difficult labour market conditions for young people in Spain (Elias, 2010).

Other European countries have observed similar trends, with the European Commission's Directorate-General for Education, Youth, Sport, and Culture (2018) noting a decline in the financial advantages of higher education degrees. The discrepancy between the qualifications of recent graduates and job requirements persists, indicating a need for policy reform. The Bologna initiative suggests incremental improvements to strengthen the connection between higher education and the job market, such as encouraging work placements within higher education programs European Commission/EACEA/Eurydice (2018).

Spanish government has implemented some measures to reduce the educational mismatches by the use of strengthening of cooperation between universities and private sector and better addressing of the on-the-job training. Examples of a tighter cooperation with the private sector is the dual vocational training (Ministry of Education & Vocational Training, 2022) and the Industrial Doctorate Programme (AEI, 2022).

However, one of the main limitations of the cooperation between universities and private companies in Spain has been the lack of funding of host trainees or trainee researchers, mainly in small and medium-sized companies (OECD, 2018). One of the critical points of this issues is that the vocational training programmes as well as the Industrial Doctorates require several years of training that imply high level of funding.

Considering that the outcomes of our study identified positive effects of master’s studies on reduction of educational mismatches, an alternative solution could be encouraging collaboration between universities and private companies based on internships of master’s degree students. These short-term internships would probably need lower funding than the medium- and long-term vocational training programmes and the Industrial Doctorates. Furthermore, this type of collaboration could also lead to better alignment of educational programmes and needs of the labour market.

In their educational journey, Spanish graduates are increasingly valuing internships as integral to securing employment that aligns with their qualifications and fosters long-term career growth (Morejón Cabrera & Mariel, 2023). Seeking to bridge the divide between their academic achievements and practical experience, students are turning to internships as a means to differentiate themselves in the competitive job market.

Our study culminates with three policy recommendations. Initially, we emphasize the importance of forging partnerships between educational institutions and the corporate sector. Such alliances would benefit from programs that bridge the gap between companies and Master's graduates, providing avenues such as internships or cooperative agreements that allow students to accumulate work experience alongside their academic pursuits. This synergy is expected to facilitate graduates' seamless entry into the labor force, offering employers a workforce endowed with relevant skills. This echoes the findings of Acemoglu and Autor (2011), as well as Bassanini et al. (2005), who recognize the critical impact of such collaborative efforts in reducing the disparities between educational outcomes and employment requirements.

Secondly, we advocate for the development of Master's programs with a strong practical component. The inclusion of internships or industry partnerships within these programs is advised to reconcile education with real-world demands, thereby bridging the initial experience gap that graduates often face. Card et al. (2018) support this view, highlighting the role of practical experience in developing skills that are directly applicable to the job market.

Lastly, we call for an enhancement of skill diversity within educational programs. We recommend that Master's programs be designed to foster a broad spectrum of skills, enabling graduates to transition across various professional arenas and reduce the risk of horizontal mismatches. Belfield and Levin (2002) have also stressed the importance of transferable skills and adaptability, which aligns with our study's assertion of their necessity in today's dynamic job market.

For future research it would be interesting to distinguish the effect of master’s degrees by considering its specific type, own master’s degrees and official master’s degrees. Own master’s are training programs created by the Universities, and the legitimacy and reputation of these programmes are predicated on the standing of the university that awards them. While official master's degrees are also offered by the University, but these are also regulated by the National Agency for Quality Assessment and Accreditation of Spain (UNIR, 2022).

It would be notably beneficial to examine if job education mismatches were more prevalent among the pre-Bologna “Licenciatura” degrees compared to the educational structures that followed the reform. This investigation is timely and pertinent, given the Bologna Process's transformation of Spain’s higher education from the traditional 4–6-year “Licenciatura” programs to shorter 3–4-year Bachelor's programs, which are now typically followed by 1–2-year Master's programs. Currently, there lacks definitive evidence to assess if the newly structured Master's programs are considered by employers to be sufficient in enhancing education or if they are on par with the comprehensive nature of the former “Licenciatura” degrees.