Related research

The literature on labour market mismatch is quite extensive (Reference Leuven, Oosterbeek, Hanushek, Machin and WoessmannLeuven and Oosterbeek, 2011; Reference PouliakasPouliakas, 2013). The well-established conclusions emerging from this literature highlight that mismatched workers endure wage penalties and have lower job satisfaction, higher turnover and absenteeism, and lower participation in training. For employers, costs associated with qualification and skills mismatch may take the form of higher recruitment costs, lower productivity and lower product quality.

Much of the existing research is based on the concept of educational or qualification mismatch, which is defined using educational credentials as a reference point (Reference Carroll and TaniCarroll and Tani, 2013; Reference McGuinness and BennettMcGuinness and Bennett, 2007). More and more authors, however, are using measures of deficits/surpluses in skills or competences (Reference Green and McIntoshGreen and McIntosh, 2007; Reference McGuinness and SloaneMcGuinness and Sloane, 2011; Reference Mavromaras, McGuinness and O’LearyMavromaras et al., 2010). The resulting evidence suggests that, quite unexpectedly, educational and skills mismatches are not strongly correlated, indicating that they are distinct empirical phenomena that need to be studied separately. The literature reports negative wage effects stemming from both forms of mismatch, with an increasingly robust evidence that the worst situation is to be overskilled and overeducated.

Several articles (Reference Green and ZhuGreen and Zhu, 2010; Reference McGuinness and SloaneMcGuinness and Sloane, 2011; Reference Mavromaras, McGuinness and O’LearyMavromaras et al., 2013) have analysed the impact of qualification and skills mismatch on job satisfaction. Examining the impact of mismatch on job satisfaction along with earnings is especially relevant because it can reveal whether or not this represents a voluntary status. Workers, for instance, may forego higher wages in favour of other, more satisfying, job attributes, such as job security or work–life balance. Therefore, several authors argued that finding negative effects of mismatch on earnings but positive effects on satisfaction would be suggestive of an intended choice of being mismatched. On the contrary, obtaining negative impacts of mismatch on both satisfaction and earnings indicates that being mismatched is driven by constrains in the labour market. The results of this incipient literature show that, in general, mismatch represents an involuntary situation (i.e. it has a negative effect on both labour market outcomes). Moreover, qualification mismatch per se is not strongly correlated with job satisfaction, whereas underutilisation of acquired skills has more serious consequences, particularly when accompanied by educational mismatch.

Doctoral education has attracted increasing attention in specialised academic journals in recent years, as well as among national and international policy management spheres (EUA, 2010). However, the evidence on the labour market situation of PhD holders, while growing, is still quite limited. More importantly, very few articles have focused on the determinants and the effects of job-education mismatch specifically among PhD holders. Regarding the determinants of mismatch, drawing on data from the US Survey of Doctoral Recipients (SDR), Reference Bender, Heywood, Freeman and GoroffBender and Heywood (2009) stress the importance of employment sector and work activity as determinant of having a job related to the PhD. In a subsequent article, Reference Bender and HeywoodBender and Heywood (2011) exploited the panel structure of the SDR data and found that job changes from the academia to business or government increase the likelihood of mismatch, but mostly at the early stage of the career. On the contrary, the main activity carried out in the workplace affects mismatch also in advanced phases of the working life of PhD graduates. However, both articles are silent about the potential role of academic attributes, which could indeed act as determinants of mismatch either directly or indirectly (through their effect on job characteristics). More recently, Reference GaetaGaeta (2015) analysed overeducation and overskilling among Italian PhD graduates, confirming the importance of sector and work activity as conditioning factors of both forms of mismatch. He also included dummies for on-time PhD completion, attending extra courses and pre-doctoral mobility as controls, and found that the latter two variables negatively correlate with mismatch. We consider that examining the effect of academic characteristic is of special importance, since it captures individual heterogeneity in human capital accumulation that is not fully accounted for PhD programme controls. For this reason, our models contain a large list of academic attributes that could affect mismatch. Moreover, we implement a stepwise inclusion of control variables, which enables analysing the extent to which academic characteristics affect mismatch indirectly because of their effect on job-related controls that in turn determine the risk of being mismatched in the labour market.

The evidence about the effect of mismatch among PhD holders is also limited. Reference Bender, Heywood, Freeman and GoroffBender and Heywood (2009, Reference Bender and Heywood2011) estimated the wage penalties due to job-education mismatch among US PhD holders. They show that having a job that is not related to the PhD has a negative impact on earnings, which is robust to the inclusion of individual fixed effects. Moreover, in the second article, they also performed separate estimates of the wage penalty associated with mismatch by field of study and at different career stages, and found worse effects for those with a PhD in Hard Sciences and, to a lesser extent, in Social Sciences, as well as for those at an advanced stage of their career.

Reference Canal Domínguez and Rodríguez GutiérrezCanal Domínguez and Rodríguez Gutiérrez (2013) studied wage differences by area of study and employment sector among Spanish PhD holders. Although job mismatch was not their specific focus, they included variables capturing job–PhD relatedness and qualification requirements as controls. The results suggest that working in a job that requires higher education levels (post-doctoral, PhD, graduate or undergraduate) is associated with higher earnings, relative to occupations that do not require higher education qualifications, especially outside the academic sector.

Regarding job satisfaction, Reference Bender and HeywoodBender and Heywood (2006) analysed gender differences in job satisfaction among US PhDs across sector (again using SDR data). They also controlled for a proxy of job–PhD relatedness and found that being in occupations closely related to the PhD improves job satisfaction in all sectors (academy, government and business) and for both males and females. Consistent results have been presented in the analysis of job satisfaction of their subsequent work specifically focused on mismatch among PhDs (Reference Bender, Heywood, Freeman and GoroffBender and Heywood, 2009). Since their dependent variable refers to overall job satisfaction, they are unable to examine whether or not mismatch has a different effect on satisfaction with different facets of employment, such as monetary and non-monetary elements.

Overall, to the best of our knowledge, this is the first article providing novel evidence about the impact of overeducation and overskilling on two key labour market outcomes of recent PhDs: earnings and job satisfaction.

Data and descriptive statistics

Our article is based on data from a recent cohort of PhD graduates from public universities in Catalonia, Spain. The number of PhDs awarded by Catalan universities increased from 968 during the 1997–1998 academic year to 1781 in 2010–2011 (an increase of 84%). As can be seen in Figure 1, the increase in doctoral awards in Catalonia has followed the same general tendency observed for Spain as a whole.

Figure 1. PhD theses approved in Spain and Catalonia between 1997–1998 and 2008–2009.

Source: Spanish National Statistical Institute (INE).

The rising flow of PhD holders in recent decades, however, has not caused a major supply shock in the Spanish labour market. In 2011, the doctoral graduation rate was just 1.1% of the population in the reference age cohort, considerably below the OECD mean of 1.6% and just slightly up from the rate of 0.9% reported for Spain in 2011 (OECD, 2013: 94). In 2009, there were 6.7 doctorate holders per thousand population aged 25–64 years in Spain, contrasting with 14 in Germany, 13.5 in the United States and 7.6 in Finland (OECD, 2013: 96).

The data used in the empirical analysis in the present study come from a 2011 survey on the early labour market experiences of PhD holders, implemented by the Quality Assurance Agency for the Catalan University System (AQU). ^{Footnote 2} The target population consisted of all Spanish-born individuals who completed a PhD in one of the seven Catalan public universities in 2006 or 2007. The entire population was composed of 1824 individuals and 1225 answered the questionnaire, which corresponds to a fairly high response rate of 67.2%. Graduates were contacted 4 years after PhD completion. Therefore, our analysis concerns the short-term mismatch situation of our sample of PhD holders. It could be argued that analysing this early period in the professional careers of PhD graduates may be misleading as they are still adjusting to a situation that may evolve towards a better match. While theoretically this would appear to be likely, the scant evidence available suggests that the incidence and the negative effects of mismatch on earnings of US PhDs are more pronounced among those in late careers, but are also present among recent PhD graduates (Reference Bender and HeywoodBender and Heywood, 2011), which makes our early career analysis particularly interesting.

The dataset contains basic socio-demographic data, information on academic attributes and the doctoral programme followed, as well as detailed information on current employment. We restricted the sample to those individuals who were in a full-time job at the time of the survey and were aged 40 years or younger when they started their PhD. ^{Footnote 3}

The main variables of interest are those concerning the job (mis)match status of PhD holders, taken from two specific questions from the AQU survey. In the first question, respondents were asked about the educational entry requirements for their current job. Four possibilities were considered: (1) a PhD degree, (2) a specific undergraduate degree (i.e. the degree held by the individual), (3) any undergraduate degree, and (4) no qualification requirements. In the second question, respondents were asked whether their PhD-specific skills were necessary in their current job. In this article, we adopt an approach similar to that of Reference Dolton and SillesDolton and Silles (2008). We classify individuals as overeducated if they stated that their PhD was not necessary for securing their current job and as overskilled if they considered that the skills acquired with the PhD are not necessary for performing the job. ^{Footnote 4} Table 1 shows the marginal and joint distribution of these two distinct dimensions of mismatch.

Table 1. Cross-tabulation of qualification and skills mismatch.

The data indicate that just over 72% of individuals in our sample were adequately matched in terms of skills and that just 53% were adequately matched in terms of education. These figures reflect a considerable level of overeducation and overskilling in our cohort of doctors. It is difficult to validate these data due to a lack of comparable information, but they are quite similar to results reported for Spain by Reference AuriolAuriol (2010) and Reference Canal Domínguez and Rodríguez GutiérrezCanal Domínguez and Rodríguez Gutiérrez (2013).

Cross-tabulation of education and skills mismatch measures shows that the probability of being well-matched in terms of skills was significantly higher for PhD graduates in occupations that required a PhD, meaning that these two distinct facets of (mis)match are likely to be interrelated. Indeed, the correlation between the two mismatch indicators is 0.51, which is well above the correlation indicated in articles using similar mismatch measures for college graduates. Looking at the extremes, 45% of our sample can be considered adequately education/skill-matched, while 26% were in jobs that seemed to require neither a PhD degree nor the skills acquired during doctoral studies (overeducated and overskilled).

What were the characteristics of our graduates? Did mismatch status reflect differences in observed individual, academic and labour market characteristics? ^{Footnote 5} Overall, it appears that mismatch status was clearly related to the academic and professional profile of the individual. Well-matched PhD holders were more likely to be younger males with a clear academic orientation, as they were more represented among those who developed their PhD thesis within a research group and completed the doctorate in at most 6 years. They also tended to have shorter job tenure and were more inclined to work in academia, research institutes or private firms doing research and development (R&D) work. The profile of graduates who were both overskilled and overeducated was similar but they had a significantly longer tenure and were more likely to have started working as adjunct professors or research assistants at a university before completing their PhD. It is interesting to note that strongly mismatched graduates were more likely to work either in the government or private sector.

The risk of mismatch also varied markedly across different types of PhD programmes. PhD graduates in the areas of humanities and social sciences were generally less likely to be well-matched, while the opposite was observed for PhD holders in hard sciences. The evidence for the area of health points out that mismatch was an important problem among PhD graduates in these fields, especially for those holding a PhD in Medicine. The incidence of mismatch was moderate among PhDs in technical studies, with the exception of Architecture.

Table A2 in the Appendix also reports descriptive information about potential wage and job satisfaction penalties associated with mismatch status. With respect to raw differences in gross annual earnings, adequately matched doctors were slightly more represented in the modal category (between €30,000 and €40,000). Interestingly, graduates earning more than €50,000 (the top-coding category) were significantly more likely to be mismatched, particularly in terms of skills. This is possibly related to the fact that a higher proportion of these graduates worked in the better paid private sector. The survey also provided information about perceived satisfaction. On average, job satisfaction was rather high (5.7 on a 1–7 scale) and the respondents were especially satisfied with promotion opportunities, but less happy with job content and pay. As expected, those who were overeducated and even more so those who were overskilled were significantly less satisfied with the match between their skills and the work they were doing.

The correlates of qualification and skills mismatch

In this section, we examine the variables conditioning mismatch. We estimated two Seemingly Unrelated Bivariate Probit equations that model the probability of being overskilled and of being overeducated. This allowed us to check for significant differences in the conditional association between the explanatory variables and each of the two types of mismatch. Table 2 shows the average changes in the predicted probabilities for four different specifications of the two equations. The baseline specification (1) contains socio-demographic variables, academic credentials and indicators for pre- and post-doctoral mobility. Model (2) adds the type and region of work, model (3) job attributes, and model (4) information on the main activities performed at work. ^{Footnote 6} This stepwise inclusion of controls was designed to observe whether and how academic and job-related attributes separately affect the likelihood of being overeducated and/or overskilled. Additionally, each model contains fixed PhD-type and university effects to capture factors shared by graduates with similar PhDs across the seven universities. As expected, the estimated correlation between the residuals of the two equations was positive and significant in all cases, pointing to the presence of common unobserved determinants of overskilling and overeducation.

Table 2. Probabilities of overqualification/overeducation – average marginal effects from Bivariate Probit models.

All the estimations include fixed effects for PhD type and university (not shown). Standard errors (in parenthesis) are clustered at the PhD programme level. The average marginal effects for indicator variables are average discrete changes in the predicted probabilities.

*** Significant at 1%; **significant at 5%; *significant at 10%.

Two immediate points emerge from the results of our analysis. First, the effect of academic attributes seems to be mainly indirect, since it loses importance once we control for the full set of work characteristics. Second, the stepwise inclusion of different groups of regressors did not generate striking changes in the estimated coefficients as we moved from one model to the next in terms of direction and statistical significance, meaning that the general picture of the conditioning factors of mismatch among PhD holders in our series remained largely unchanged.

Female graduates were more likely (around 5 percentage points (p.p) higher) to be overskilled than male graduates with similar characteristics, but we did not observe any gender differences in terms of overeducation. This finding could be a cause for concern as it might mean that while the formal process of accessing the labour market is similar for men and women, there may be subsequent (discriminatory) filtering of women into jobs or tasks requiring fewer skills. Age appeared to be unrelated with overskilling, whereas it had a negative effect on overeducation only when job characteristics were controlled for (probably reflecting cohort effects in education requirements).

Academic experience variables were divided into three groups: source of PhD funding, academic attributes that characterise PhD studies and research mobility. Ideally, these variables would capture the quality of training received and, therefore, signal the best doctoral graduates, who, in a normal, well-functioning labour market, should be better matched. ^{Footnote 7} However, if PhD programmes are, at least to some extent, designed to attract and channel students into the academic profession, these variables could also be capturing a sorting process into specific labour markets. The coefficients estimated show that the academic experience has a modest impact on the probability of being overskilled, but are markedly relevant for explaining the probability of being overeducated.

Funding one’s PhD through any of the three main mechanisms analysed did not seem to be associated with dramatic differences in mismatch status. It should be noted, however, that scholarships were generally associated with a lower incidence of both overeducation and overskilling, but this beneficial effect tended to be lost once job characteristics were controlled for. This means, at least to some extent, that individual profiles simply affect occupational choices, which, in turn, determine the chances of being mismatched or not. For the small group of students working in jobs unrelated to their PhDs during their doctoral studies (about 4% of the sample), we observed a higher risk of being affected by both forms of mismatch. These students might represent the least able students (those unable to secure a more favourable way of funding their PhD), but it is also possible that their desire to obtain a PhD was driven more by a ‘consumption’ motivation than by academic/professional goals.

The results for academic attributes capturing individual performance indicate that PhD duration had no impact on the probability of being mismatched. However, working in a research group while pursuing one’s PhD favoured access to jobs requiring a PhD degree, but its positive effect on the likelihood of the acquired skills being fully utilised disappeared after controlling for employment sector. This observation clearly points to the value of using research infrastructures as a stepping stone towards an academic career requiring a PhD.

Pre-doctoral research mobility and, most notably, post-doctoral mobility in national or international institutions decrease the probability of being mismatched by about 25 p.p. (relative to no post-doctoral mobility). Mobility experiences allow acquiring productive skills and personal maturity that most likely are rather difficult to reproduce in the student’s own institution during the regular training period. Moreover, research mobility, particularly after completing a PhD might also have a positive impact on job quality resulting from increased networking opportunities. This is indeed confirmed by the fact that although the size of the effect of research mobility diminishes after controlling for job-related variables, it still remains important (especially for overskilling).

Moving to the results obtained in the models where job-related variables are included as additional covariates, we must stress that the employment sector was of fundamental importance for explaining the likelihood of being overeducated and, in particular, of being overskilled. Working in the private sector and even more so in the public sector (i.e. government, public administration and other public non-academic institutions) substantially increased the chances of being mismatched and, even though this penalty was reduced after controlling for the main activities at work, it remained sizeable and significant. On the contrary, those who worked in research institutes were not more likely to be overskilled and were just slightly more likely to be overeducated when job attributes and main activities were maintained constant. These results are remarkable, especially considering that working outside an academic-research environment has a very large impact on the probability of being overskilled. They suggest that the problem of mismatch among PhD holders is not just related to a lack of recognition of the PhD credential outside the university (i.e. overeducation), but also point to a more fundamental problem related to a clear underutilisation of skills by many employers.

The estimates of work location suggest that working outside Spain was associated with a lower probability of being mismatched. This could reflect either positive sorting of PhD holders who migrated after completing their PhD or a higher supply of suitable jobs in destinations to which Spanish PhD graduates were likely to migrate, namely, Northern Europe and the United States. The estimated marginal effects for the additional job-related controls highlighted a sizeable positive effect of job tenure on overeducation. This result might be explained by the cohort nature of our data and also perhaps by the fact that some graduates in our sample may have started their current job before completing their PhD. However, seniority seemed to be unrelated to the likelihood of overskilling. In our sample, those with a permanent contract were slightly more likely to be adequately matched in terms of qualifications than others. In agreement with evidence reported for college graduates, we found that working in a medium–large firm (250–500 workers) had a beneficial effect on the probability of being adequately matched.

Finally, model (4) controlled for the main activities at work for those employed in non-university settings. Therefore, the effect of these variables in this model has to be interpreted as the impact of job task variation once the ‘average’ effect of employment sector has been controlled for. As expected, working in R&D reduced the likelihood of being mismatched in a consistent way for both indicators. This means that working outside a university per se was not synonymous with mismatch. Rather, the higher risk of being mismatched depended on whether or not PhD skills were utilised in (non-academic) jobs. In other words, working outside academia, but in a research-related job, would compensate for the higher average likelihood of mismatch among private and public sector workers. This was also confirmed by the positive conditional relationship of performing technical support tasks on both forms of mismatch, as well as the increased likelihood of those working in management and health care being mismatched. However, this last point should not be interpreted in terms of having ‘too many skills’ but rather in terms of having skills that are not useful for a particular job.

Overall, our results confirm the crucial importance of key job characteristics such as employment sector and job-tasks as conditioning factors of mismatch among PhD recipients, as highlighted by Reference Bender, Heywood, Freeman and GoroffBender and Heywood (2009, Reference Bender and Heywood2011) and Reference GaetaGaeta (2015). However, we have underscored the impact of academic variables on the probability of suffering some form of mismatch. These variables appeared to operate indirectly (although not exclusively so), since their effect on mismatch was mediated by job characteristics. That is, academic attributes affected the kind of job PhD holders obtained after completing their studies, which in turn determined the chances of being or not being mismatched in the labour market.

Were mismatched doctors penalised?

Mismatch and earnings

In this section, we examine the potential labour market penalty associated with job mismatch in our cohort of recent doctors. We estimated an extended earnings equation that includes several academic attributes and job characteristics as control variables, following an ‘assignment’ viefw of the labour market in which both individual human capital and academic and job characteristics determine earnings. Table 3 shows the estimates from the augmented earnings regression. ^{Footnote 8}

Following Reference Mavromaras, McGuinness and O’LearyMavromaras et al. (2013), we included the four job match statuses in our model: adequately matched, overskilled, overeducated, and overskilled and overeducated. Our results indicate that being overskilled or overeducated was not statistically associated with earnings, suggesting that PhD recipients in these two categories did not earn less than their adequately matched counterparts. On the contrary, compared to well-matched graduates, those who were both overeducated and overskilled faced a wage penalty of about 12%. These results are similar to those obtained for college graduates. Even though we used a different measure of mismatch, our results are also consistent with findings obtained by Reference Bender, Heywood, Freeman and GoroffBender and Heywood (2009, Reference Bender and Heywood2011) for US PhD holders, as well as with the evidence for Spain reported by Reference Canal Domínguez and Rodríguez GutiérrezCanal Domínguez and Rodríguez Gutiérrez (2013). This might be taken as further evidence that unless accompanied by a certain degree of skills mismatch, overeducation does not have a detrimental effect on job match among PhD graduates

Table 3. Mismatch and annual gross earnings (in logs) – Interval Regression.

Robust standard errors in italics.

*** Significant at 1%; **significant at 5%; *significant at 10%.

The estimates of the control variables in our model are quite standard and are just briefly discussed. The results show a significant ceteris paribus gender difference in annual earnings in favour of male doctors. As expected, earnings rose with age at job entry and job tenure and graduates who obtained their PhD while working in a job related to their studies earned more, reflecting human capital accumulation through experience. However, a longer time between finishing college and starting one’s PhD studies had a negative effect. Moreover, graduates who took more than 6 years to finish their PhD studies endured a wage penalty.

There was a sizeable positive earnings differential in favour of PhD holders working in the private sector compared to the university sector, but no significant differences were detected among those working in research institutes or in the public sector. The public sector dummy coefficient is however significantly higher and statistically different from zero when main activities at work are excluded from the model, ^{Footnote 9} suggesting that PhD recipients who worked in the public sector earned more than those who worked in a university only if they performed certain activities that were better remunerated, such as management and health care work.

As expected, we also found positive earnings returns to having a permanent contract and working in medium–large firms. Moreover, PhD recipients who performed management and health-related tasks were better paid than others. The estimates from the PhD-type fixed effects revealed that those with a PhD in biology earn more than those who studied humanities, sociology, political science or communication, but less than those who studied economics, business, chemistry, medicine or computer and information engineering.

Mismatch and job satisfaction

In this section, we analyse the conditional association between job mismatch and job satisfaction. We consider perceived overall job satisfaction as an aggregate indicator of all relevant aspects of the job. Moreover, we also analyse perceived satisfaction with four distinct facets of the job, namely, promotion opportunities, earnings, job content and job-skills match. Running separate regressions for overall satisfaction and job-domains satisfaction would provide a better insight into the channel through which mismatch affects the level of utility derived from the job. Given the ordinal nature of the variables in question, we applied the standard Ordered Probit approach.

Table 4 shows the marginal effect of each mismatch indicator on the probability of being very satisfied (the highest category) with the job as a whole and with each job domain. ^{Footnote 10} It appears that mismatched PhD holders are not less satisfied with earnings and career prospects than their well-matched peers. However, being mismatched is significantly associated with a lower probability of being very satisfied with job content and job-skills match, two domains that reflect intrinsic and non-monetary aspects of the job. Specifically, graduates in the overeducated category are less satisfied with job content and job-skills match, although educational requirements per se appear to have a lower impact on job satisfaction than skills utilisation. Indeed, skills underutilisation makes PhD holders significantly less likely to be very satisfied with job content and job-skills match. Moreover, the satisfaction loss for being both overeducated and overskilled is very similar to that associated with being overskilled only, indicating that underutilisation of skills is significantly more damaging to job satisfaction than disregard of the attained qualification.

Table 4. Mismatch and job satisfaction.

Average marginal effects on the probability of being very satisfied obtained from Ordered Probit models. Each model includes controls for gender, age, time between bachelor completion and PhD enrolment, PhD funding, PhD duration >6 years, PhD-type and university fixed effects, type of job, job location, current job tenure, permanent contract, firm size, main activities and annual earnings categories (complete estimates are reported in Table 3A in the Appendix). Robust standard errors in italics.

*** Significant at 1%; **significant at 5%; *significant at 10%.

A more general view of the relationship between job mismatch and job satisfaction among PhD holders can be obtained from the estimates of the overall satisfaction equation. Overall job satisfaction represents an aggregate of job domain satisfaction and very probably includes additional domains to the four we considered (Reference Van Praag and Ferrer-i-CarbonellVan Praag and Ferrer-i-Carbonell, 2007). Consistent with the results for job content and job-skill match satisfaction, being overeducated has only a modest negative effect on satisfaction with the job as a whole (which is imprecisely estimated). By contrast, job satisfaction is significantly lower when overeducation and overskilling are combined and is lower still among PhD graduates who are overskilled but not overeducated (although this estimate is less precise because of the few observations in this category). It might be that overeducated doctors enjoy other features of their job that would compensate for underutilisation of skills (compensating differentials). Alternatively, it is possible that the requirement of a PhD to get the job might have falsely raised the graduate’s expectations regarding the need for his or her skills. Whatever the case, our results are in line with those of Reference Bender and HeywoodBender and Heywood (2006, Reference Bender, Heywood, Freeman and Goroff2009), indicating that mismatch reduces job satisfaction. Our analysis of job-domains satisfaction reveals that most of the effect of mismatch is due to the lack of suitable non-monetary job characteristics, rather than monetary aspects. In conclusion, it is possible to claim that overskilling is unlikely to be voluntary.