The career services premium: how university job-search boosts graduate earnings

There's accumulating evidence of divergent graduate labour market outcomes (Pastore et al., 2021). Many graduates struggle to access the employment and earnings benefits traditionally associated with university degrees (Woessmann, 2016), leading to increasing interest of both scholars and policymakers in “the extent to which they are readily employable” (Mason et al., 2009). The European Commission (2017) has stated that the “employability of graduates is a European concern as in many member states, a significant share of tertiary graduates is unemployed or working in jobs for which they are overqualified” ^[1].

In response, universities have enhanced employability services, including job search support, guidance, career advice and placement, to smooth graduates' transition into the labour market. The prevalence of such programs varies largely across countries and academic traditions, and, more recently, they have been emphasised among European academic systems, including the Spanish higher education system through the 2023 LOSU reform (articles 33 and 37). Yet there is limited evidence on the labour market returns of such programs, and existing studies typically rely on correlational results (Scandurra et al., 2024), while others claim that evidence of these effects is sparse and atheoretical (Inceoglu et al., 2019).

We address three interrelated questions: (1) How do the earnings of graduates who entered the labour market through university career services differ from those of other graduates? (2) To what extent does gender influence earnings differentials associated with university career service entry routes? (3) Do university career services ensure graduates enjoy a wage premium, independent of their work experience? In line with human capital theory (Becker, 1964), we argue that using university career services can enhance graduates' skills and lead to higher earnings. Similarly, job search theory (Mortensen, 1986; Blau and Robins, 1990) suggests that early-career support reduces search costs and information asymmetry, thereby leading to better job matching and improved labour market outcomes. The existing literature argues that university careers services can facilitate transition (H1) (Pastore and Zimmermann, 2019). A particularly relevant question is whether these services may reduce gender wage differentials at the beginning of work experience. We hypothesise that returns to career services are higher for women because they facilitate matching, particularly in male-dominated sectors, which may help counter gender pay gaps (H2). Moreover, we expect career services to be more beneficial for individuals with limited work experience (H3).

Using data from a follow-up survey of university graduates from Catalan HEIs, conducted four years after graduation and covering the period 2008–2023, we assess whether obtaining a first job through university career services is associated with an earnings benefit. To account for the non-random sorting of graduates into different job entry routes, we employ an instrumental variable (IV) approach, exploiting graduates' exposure to university career services as an instrument.

While previous research has estimated the returns to internships and work placements, there remains a dearth of robust evidence on the impact of university career services. Studies show that mandatory internships raise earnings by 6–6.5% in Germany three years after graduation (Margaryan et al., 2022) and that UK sandwich-year placements increase salaries by 4% (Delis and Jones, 2023). Even larger wage premiums of up to 20% have been reported, but these estimates don't account for selection effects ^[2].

We contribute to the existing literature by providing evidence for Spain, which we believe is an interesting case study due to its persistent youth unemployment, high temporary employment (Di Paolo and Matano, 2022) and large share of skill mismatched workers (Barone and Ortiz, 2011; Dolado et al., 2013; Nieto and Ramos, 2017), coupled with an expansion of higher education participation that is among the largest in OECD economies (Gil-Hernandez et al., 2017).

Overall, our estimates reveal a wage premium of around 8.5% associated with entering the labour market via the university career service. Moreover, the heterogeneity analysis reveals important variations by gender and work experience. More precisely, female graduates appear to enjoy a larger wage premium of around 15.7%. For male graduates, we find instead a modest wage premium of approximately 3.4%, though this effect is not precisely estimated at conventional significance levels. The impact of career services varies notably with graduates' work experience. For those with no work experience, career services yield a substantial, statistically significant wage premium of approximately 17%. However, for graduates with up to two years of work experience, the effect becomes statistically insignificant.

When examining the intersection of gender and work experience, we find that among graduates with no experience, both men and women realise similar benefits from career services, approximately an 18.8% increase in wages for women and 14.1% for men. However, for those with up to two years of work experience, the wage premium becomes statistically insignificant for men, whereas it is around 14.5% for women, suggesting a differential impact across genders.

The main takeaway from this study is that university career services have a positive and substantial impact on graduates' earnings. Particularly for female graduates. While further research is needed to assess the external validity of the results, they imply that universities would be justified in devoting more resources to these services. This could have a positive effect on graduate labour market outcomes overall and towards reducing the gender wage gap.

Extensive research shows a wage premium for higher education graduates (for an overview Psacharopoulos and Patrinos, 2018), even after controlling for ability and unobserved traits (Dale and Krueger, 2002). However, this varies with macroeconomic conditions (Oreopoulos et al., 2012; Altonji et al., 2016), and institutional quality (Andrews et al., 2016). This article contributes by examining the returns to university career's services on graduate labour market outcomes. Research in this area remains limited, primarily based on small-scale case studies (Scandurra et al., 2024), scattered and atheoretical (Inceoglu et al., 2019). However, existing research points to two main benefits of employability activities: developing labour market-relevant skills and providing an observable signal of unobserved characteristics via a “stepping stone” working experience, reducing information asymmetry.

In the Spanish context, prior research on graduate labour market outcomes has focused predominantly on overeducation and skills mismatch rather than on career services per se. Studies using national graduate surveys have documented the difficulties Spanish graduates face in transitioning to quality employment (Dolado et al., 2013), including high rates of overeducation (Nieto and Ramos, 2017). For Catalonia specifically, research using the same AQU survey data has examined graduate employment outcomes across fields (Cortadas-Guasch, 2024) and the role of combining work with study (Di Paolo and Matano, 2022), yet no study has isolated the causal effect of university career services on earnings. At the European level, McGuinness et al. (2018), using cross-country graduate data, found that higher education work placements and institutional job-placement assistance substantially reduced initial labour market mismatch, while Klein and Weiss (2011) showed that mandatory internships in Germany benefited graduates from less-advantaged backgrounds.

Whilst outcomes vary by internship type and quality, internships and placements are consistently positively associated with earnings across multiple national contexts, including the United Kingdom (Delis and Jones, 2023; Luchinskaya and Tzanakou, 2025), the United States (Hunt and Scott, 2023) and Germany (Margaryan et al., 2022). Related studies confirm effects on the likelihood of being employed (Mason et al., 2009; Blackwell et al., 2000). Experimental evidence shows that internships facilitate a smoother transition into the labour market (Baert et al., 2021). In line with human capital theory (Becker, 1964), case studies suggest this can be attributed to the acquisition of professional skills (Divine et al., 2007; Busby, 2003; Margaryan et al., 2022). Likewise, experimental evidence supports the positive impact of employability activities embedded in the curriculum on career planning, knowledge and confidence (Bradley et al., 2022).

In line with screening and matching theories (Stiglitz, 1975; Jovanovic, 1979), internships provide a positive signal to employers (Nunley et al., 2016) and reduce information asymmetry by allowing employers to evaluate interns and interns to consider their prospects (Cook et al., 2004; Divine et al., 2007; Margaryan et al., 2022). In an experimental vignette study of Scottish social science graduates, Irwin et al. (2019) found that employers view external employability practices, particularly internships, as positive work experiences, especially when graduates are assigned to high-level roles. Lastly, Jackson and Dean (2023), using Australian graduate data, identified significant differences in employability-related activities across gender, age, disability and socio-economic status. Overall, the existing literature consistently supports the existence of significant gender differences (Blau and Kahn, 2017). Research shows that about 41% of the graduate wage gap is attributable to occupational segregation (Shauman, 2006), while work-pattern interruptions contribute to early-career wage gaps (Boye and Grönlund, 2018).

Despite extensive evidence on internships and placements, there remains little systemic analysis of how university career service activities affect graduate earnings. We address this gap by providing causal evidence on this relationship for Catalonia (Spain). Crucially, unlike internships and placements, which embed graduates in productive work settings and build human capital, the services studied here (guidance, counselling and job placement) function as search-enhancing intermediaries, not sites of skill formation. This distinction has been overlooked in the university-to-work transition literature, where causal estimates are sparse, focused on mandatory work-based learning (Margaryan et al., 2022; Baert et al., 2021) and silent on passive placement intermediation.

Two broad theoretical perspectives can account for the potential effect of university career services on graduate labour market outcomes. First, human capital theory (Becker, 1964; Mincer, 1958) suggests that employability activities may enhance graduates' productive capacity by developing job-relevant skills, career planning competencies and professional knowledge (Bradley et al., 2022; Divine et al., 2007). Under this interpretation, career services would improve earnings by augmenting the stock of human capital that graduates bring to the labour market. Second, job search and matching theories (Mortensen, 1986; Blau and Robins, 1990; Jovanovic, 1979) emphasise the role of information frictions in the labour market. From this perspective, career services operate primarily by reducing search costs and information asymmetry between graduates and employers, thereby improving the quality of initial job matches and shortening the transition from education to employment (Pastore and Zimmermann, 2019).

Although both theoretical accounts, i.e. human capital accumulation and the reduction of search frictions through improved job matching, may generate complementary wage effects that coexist in practice, we primarily interpret the expected career services premium through the lens of job search and matching theory. Unlike internships and placements, which embed graduates in productive work settings where skill acquisition is plausible, the career services examined here (such as guidance, counselling and job placement) are fundamentally short-term search-enhancing interventions. They connect graduates to vacancies, provide information about employer expectations and facilitate initial contact with firms. In line with screening theory (Stiglitz, 1975), career services may also function as an institutional channel that reduces employers' uncertainty about graduate quality, effectively acting as a matching intermediary rather than a site of human capital formation. This does not preclude modest human capital effects, for instance, improved interview skills or labour market knowledge, but we argue that the most plausible mechanism is the reduction of search frictions and the improvement of match quality at labour market entry.

This theoretical positioning generates testable predictions. If career services operate primarily through search enhancement rather than human capital accumulation, their returns should be most pronounced among graduates facing the greatest informational barriers. Accordingly, we hypothesise that: (H1) the earnings premium associated with career services is larger for graduates with no work experience, for whom labour market information is scarcest and search frictions most binding and (H2) the premium is larger for women, who may face additional asymmetries in accessing employer networks, particularly within male-dominated sectors.

The study analyses data from the IL-AQU graduate follow-up survey, conducted every 3 years by the Quality Assurance Agency of Catalonia (AQU), which tracks graduates who complete their studies within 4 years. The survey provides information on graduates' annual earnings, demographics and educational background. The variable of interest captures how respondents secured their first job after graduation, using 11 self-reported job-search routes (Table 1). For analytical purposes, we created a binary variable for university career services that takes the value 1 for those who access their first job through university career services facilitated activities, and 0 otherwise.

The outcome variable is the natural logarithm of gross annual earnings, deflated to 2008 prices using the Consumer Price Index provided by the Spanish National Institute of Statistics (INE). Figure 1 shows that graduates using university career services for their first job earn an average of €21,305 annually, 20% more than the €18,449 average for those using other methods, with slightly higher salary variation.

To ensure comparability and reduce confounding effects, we focus on recent graduates making direct school-to-work transitions. The sample includes respondents under age 30 who completed undergraduate studies across six triennial waves (2008–2023), excluding those enrolled in postgraduate education or who worked while studying. Pre-graduation work experience is excluded by design: since it operates through the same earnings channels as career services, retaining such graduates would confound the estimated premium. After dropping observations with missing values in key variables, the final sample comprises 5,919 graduates (Online Appendix Table A1). Online Appendix Table A2 reports the gender distribution of the sample by field of study, while Online Appendix Table A3 provides descriptive statistics of the instrument used in the IV analysis, i.e. the exposure rate, disaggregated by gender, field of study and survey year.

Table 2 presents descriptive statistics for the pooled sample. Approximately 7% secured their first job through university career services. Women comprise 53.6% of the sample, the average age is 27 years, and social sciences and legal studies is the most common field (47.9%). Parental education shows considerable variation, with 28.2% having both parents with primary education or lower, while 20.5% have both parents with higher education. The study analyses data from the IL-AQU graduate follow-up survey conducted every four years by the Quality Assurance Agency of Catalonia (AQU Catalunya).

Table 3 compares characteristics by job entry route. Career service users earn higher wages (log wage 9.995 vs. 9.812) and are disproportionately male (55.7 vs. 45.7%). Engineering graduates use career services most frequently (54.0 vs. 31.2%), while social sciences graduates use them less (33.7 vs. 49.0%). Career service users have higher parental education (25.3% with both parents having higher education vs. 20.1%) and slightly more work experience ^[3].⁴

These descriptive patterns suggest potential selection into career services based on field of study, gender and family background, motivating our IV approach to address endogeneity concerns.

Our objective is to assess whether access to the labour market through universities' career services provides an earnings premium for young graduates. The “treatment” is entry through university career services; the counterfactual assumes graduates would have used alternative strategies absent these services. We estimate:

Where log $wi$ is the deflated logarithm of annual gross earnings; $Ai$ indicates whether the individual has appealed to the university career services (i.e. career guidance, counselling or placement) or other routes (0); $Xi$ is a vector of gender, age and parents' education and access grade, which is a proxy for student ability (see Margaryan et al., 2022), and work experience. These are assumed to control differences in annual earnings attributable to gender, age, family background, ability and work experience, all of which are established determinants of labour market outcomes (Klein and Weiss, 2011; Di Paolo and Matano, 2022). Finally, $φf$⁠, $χu$ and $ψt$ represent field of study, university and survey year fixed effects, respectively, while $εi$ denotes the error term. The inclusion of university, field and year fixed effects ^[4] allows us to control for both time-varying and time-invariant institutional unobserved heterogeneity. This allows for a more nuanced understanding of potential confounding factors. The parameter $β$ captures the wage premium associated with having obtained the first job after graduation through university career services. It is important to clarify the timing, as the treatment variable reflects whether the graduate's first post-graduation job was secured through career services, regardless of whether contact with those services began during the final year of study or shortly after graduation. The survey question on which it is based asks respondents to identify the route through which they found their first job, so the parameter captures the net earnings effect of labour market entry via career services rather than the effect of service usage at any point in time.

OLS estimates of $β$ would be causal only if graduates were randomly assigned to job entry routes. This is implausible, as graduates self-select based on characteristics unobserved in the data yet that independently influence earnings. Three confounders are of particular concern. First, intrinsic motivation and job-search effort, as graduates who actively seek out career services may be more motivated in their search and such unobserved drive, rather than the service itself, may account for part of the observed wage premium. Second, unobserved ability may play a role. Although we control for access grades as a proxy for academic aptitude, there may be unobserved differences in abilities that are valued in the labour market but not strongly correlated with academic success. Third, unobserved social capital and networks could bias results as graduates embedded in denser professional or family networks may both use career services more readily and command higher wages through those connections. In the presence of such confounders, OLS estimates of $β$ are likely upward-biased.

To address this, we instrument individual career service usage with peer exposure, i.e. the share of graduates at the same university, in the same field and in the same survey year who obtained their first job through career services:

Where, j indexes other students in individual i peer group (i.e. within the same university, field of study and survey year), $service_usej$ indicates career service use and $Ng$ is total peer group size ^[5].

Graduates within the same institution and cohort are likely to share information about their job search experiences. When more peers successfully use career services, this signals effectiveness through observable outcomes and reputation within peer networks. Drawing on social learning theory (Bandura, 1978), we posit that graduates make decisions about career service use not in isolation but by observing and learning from peers' experiences. This peer influence operates through multiple mechanisms: direct observation of successful job placements, information sharing about service quality and normalisation of service usage within academic cohorts. Peers' exposure should affect, only through its influence on service usage, not through alternative channels. This assumption is supported by the placebo tests in Table 4, which confirm that the exposure rate is orthogonal to key individual characteristics. The main residual threat of a direct peer effect on graduates' productive effort is unlikely, as the mechanism we posit (social learning about service effectiveness) operates through uptake decisions rather than skill formation (cf. Manski, 1993).

To assess the validity of the exclusion restriction, we conduct a series of placebo tests reported in Table 4. If the instrument (the peer exposure rate to career services) was correlated with individual characteristics that independently affect earnings, it could influence labour market outcomes through channels other than career service usage, thereby violating the exclusion restriction.

We therefore regress three key variables on the exposure rate, controlling for field of study, university and survey year fixed effects: (1) access grade, which proxies for academic ability, (2) parents' education, which captures family socioeconomic background and (3) work experience, which reflects the presence of labour market attachment. If the instrument were picking up sorting on ability, social origin or work experience we would expect statistically significant coefficients in these regressions. As Table 4 illustrates, none of the estimated coefficients is statistically significant, for either the full sample or for women and men separately. This implies that the exposure rate is orthogonal to observable characteristics, supporting the assumption that peer exposure affects graduates' earnings only through its influence on individual career service usage and not through alternative channels.

Table 5 presents reduced-form estimates that regress log earnings directly on exposure rates, with all controls. The exposure rate has a positive and significant effect on earnings for the full sample (0.008, p < 0.01) and women (0.016, p < 0.01), but not for men (0.003, not significant). This preliminary evidence suggests that peer exposure increases earnings through service usage, with heterogeneous effects by gender. Overall, the combination of placebo tests showing no correlations with key graduates' characteristics, meaningful instrument variation and significantly reduced-form relationships supports our identification strategy and validates the IV approach for estimating causal career service returns.

Two residual threats to the exclusion restriction deserve acknowledgement. First, cohorts with high peer uptake could share unobserved productivity traits or face better labour market conditions; our placebo tests address observable confounders but cannot fully rule out unobserved cohort heterogeneity. Second, if peer exposure directly affects graduates' job-search effort rather than only their uptake decision, the exclusion restriction is violated. We consider this second threat modest, as the social learning mechanism operates through information about service effectiveness, not skill formation.

Table 6 reports OLS estimates indicating that graduates accessing the labour market through career services report a wage premium of 9.2% (exp(0.088) - 1) for the full sample, 13.6% and 5.2% for men. However, these estimates are likely biased because unobserved characteristics affect both career service use and earnings.

Table 7 presents IV estimates addressing selection bias. The first-stage results confirm strong instrument relevance. We report the Kleibergen–Paap (KP) F-statistic as the appropriate weak instrument diagnostic under heteroskedasticity-robust inference (Stock and Yogo, 2005). KP F-statistics are well above conventional weak instrument thresholds across all samples (940.8 full sample; 359.3 women; 602.3 men), decisively rejecting weak instrument concerns. The Anderson and Rubin (1949) test complements this by jointly assessing instrument validity and the significance of the endogenous regressor (career service usage) in a manner that is robust to weak instruments; it confirms that career service usage significantly affects graduates' annual earnings.

The second-stage estimates reveal an 8.5% wage premium (exp(0.082)-1) for the full sample with pronounced gender heterogeneity: women experience a 15.7% premium, while men show no significant effect. The Anderson–Rubin test confirms instrument validity and treatment significance for the full sample (F = 10.0, p = 0.002) and women (F = 16.9, p < 0.001), but not men (F = 0.88, p = 0.35) ^[6]. Combined with strong first-stage diagnostics, this suggests genuinely weak effects for men rather than instrument failure.

Under an IV framework, the estimated coefficients represent the Local Average Treatment Effect (LATE), capturing the causal effect for compliers, i.e. graduates whose career service usage responds to peer exposure. In our setting, the likely compliers are graduates who are receptive to peer influence and who would use career services when sufficiently exposed to peers who have done so but would not otherwise. This group plausibly excludes both those who would always seek out career services regardless of peer behaviour (always-takers) and those who would never use them (never-takers). The LATE interpretation implies that our estimates may not generalise to the full population of graduates but are internally valid for the policy-relevant margin of individuals whose behaviour is responsive to institutional and social cues. Always-takers may differ from compliers (e.g. being more motivated), implying the Average Treatment Effect (ATE) could be smaller than our LATE. However, compliers are graduates responsive to social cues about service effectiveness, precisely the population targeted by awareness campaigns. The LATE is therefore a policy-relevant estimate for expanding uptake. Extrapolating to never-takers lies beyond the scope of the present design.

IV estimates are slightly lower than OLS (8.5 vs. 9.2% overall; 15.7 vs. 13.6% for women), suggesting modest positive selection bias. Career service users have marginally higher unobserved earnings potential, but substantial causal benefits remain after accounting for selection.

In this section, we examine heterogeneity in the career service premium by post-graduation labour market attachment. Graduates are stratified by years of employment accumulated since graduation: those with no post-graduation experience (Table 8) and those with one to two years (Table 9). Within each stratum, the complier-based LATE is estimated separately, allowing us to assess how the premium varies with graduates' level of post-graduation labour market integration. By restricting the sample, we eliminate the most consequential confounder: career service users may systematically differ from their peers in accumulated human capital and labour market networks acquired through employment. In this sub-group, both treated and untreated graduates enter the labour market on an equal footing in terms of professional experience, making it substantially more credible to attribute any observed wage differential to the direct and sole intervention of university career services. Consistent with Hypothesis 3, which predicts larger returns for graduates without work experience, the results yield a substantially larger wage premium of approximately 17% (exp(0.157)−1), compared to the 8.5% estimated for the full sample, with slightly higher effects for women (18.8%) than men (14.2%). Notably, the male coefficient becomes statistically significant, suggesting that career services particularly benefit those at the initial stage of labour market entry. First-stage F-statistics remain strong (141.4 for the full sample; 81.3 for women; 73.2 for men), confirming robust identification.

It should be noted that subsample analyses, by reducing the number of observations, entail wider confidence intervals and reduced statistical precision. Accordingly, point estimates for smaller subgroups should be interpreted with caution, and the absence of statistical significance in some specifications does not necessarily imply the absence of an effect.

These findings have important implications for both educational institutions and career development strategies. They highlight the tangible economic value of targeted career support, especially for recent graduates entering the job market without any professional experience ^[7].

One to two years of post-graduation employment represents a meaningful intermediate category: graduates have begun to accumulate work experience but have not yet built the employer networks and sector knowledge that reduce reliance on career services. Graduates with more years of post-graduation experience approach the age ceiling of the analytic sample and are excluded from this comparison.

Table 9 examines one or two years of work experience and examines graduates' wage returns. Effects become more nuanced: the full-sample estimate is positive, although not statistically significant. Women retain a significant 14.5% premium, while men report a negative, insignificant effect (−6.5%). This suggests that career services provide diminishing returns as graduates accumulate experience, though women continue to benefit regardless of experience level. Family background effects are stronger in this subsample as graduates with both parents with higher education earn 15.1% more, with larger effects for women (19.2%) than for men (11.2%).

The results in Tables 8 and 9 reveal a clear and theoretically consistent pattern: the wage premium attributable to university career services is concentrated among graduates with little or no work experience and declines markedly as labour market exposure increases. For graduates with no work experience, career services generate a substantial, precisely estimated wage premium of 17%, with broadly symmetric effects for women (18.8%) and men (14.2%). By contrast, among graduates with one or two years of work experience, the full-sample estimate falls to approximately 2.3% and ceases to be statistically significant, while male graduates exhibit a negative, albeit insignificant, point estimate of −6.5%. Women remain the exception: they retain a significant premium of around 14.5% even with some work experience, indicating that the matching frictions that career services help resolve are more persistent for female graduates than for male graduates. This gradient is consistent with the theoretical prediction that career services operate primarily as a search-enhancing mechanism, whose marginal value is highest when graduates lack the employer contacts and sector-specific knowledge that prior employment typically provides. The findings thus reinforce the primary role of career services in facilitating school-to-work transitions rather than augmenting productive skills, and they underscore the sensitivity of inexperienced graduates and women to search frictions in the Catalan graduate labour market.

Table 10 examines whether gender effects vary by field. Social and legal studies consistently show significant positive effects (21.2% overall, similar for women and men), likely reflecting less-structured school-to-work transitions in these fields. For women in engineering, the point estimate is positive but not significant, suggesting that career services may help some women access male-dominated occupations, though the evidence remains inconclusive.

These results are broadly consistent with the internship literature, which provides the closest available point of comparison for causal estimates of placement-related interventions. Margaryan et al. (2022) find 6% earnings return from mandatory internships in Germany three years after graduation, while Delis and Jones (2023) report a 4% salary premium for UK sandwich-year placements. Our estimates for graduates without prior work experience substantially exceed these benchmarks, which is plausible given that career services operate at the point of labour market entry, when search frictions are most acute and employer matching is most consequential for subsequent wage trajectories. The larger premium estimated here likely reflects both the greater vulnerability of inexperienced graduates and the compounded role career services play in this sub-group, where they substitute for the labour market networks and employer exposure that prior employment would otherwise provide.

It should be noted that differences in point estimates across gender and work-experience subgroups are not subject to formal coefficient equality tests. Reduced sample sizes widen confidence intervals; the negative estimate for men with 1–2 years of experience (−6.5%) is consistent with zero, not indicative of a harmful effect. The heterogeneity patterns are consistent with the search-friction interpretation but do not constitute a direct test of it.

The results identify a stronger earnings premium for women consistent with H2. Female graduates in our sample are concentrated in fields characterised by less-formalised recruitment channels, in particular social and legal studies, where employer networks are likely to be less institutionalised and informational barriers to quality employment higher. Career services may therefore provide a more consequential matching advantage for women, enabling access to vacancies that would otherwise be harder to reach. In addition, male graduates are disproportionately concentrated in engineering, a field in which firms have well-established direct recruitment pipelines to universities, reducing the marginal value of career service intermediation. The gender asymmetry in returns is interpretable as evidence that career services are most effective where labour market networks are weakest and where gender-specific barriers to initial employer contact are most binding. These findings are consistent with research showing that gender occupational segregation accounts for a substantial share of early-career wage gaps among graduates (Blau and Kahn, 2017).

As a robustness check, we implement an alternative IV strategy that interacts the exposure rate with the ratio of career service workers (full-time equivalents) to graduates by university in 2003. This captures institutional capacity to provide career services. Due to data availability, the sample is restricted to four Barcelona universities (3,658 observations). Online Appendix Table A4 shows results consistent with our main findings: an overall premium of 8.9%, 14.9% for women and 6.3% for men. For graduates with no work experience, premiums are 16.3% overall and 20.7% for men, while those with 1–2 years' experience show an 8.6% overall premium and 20.6% for women. First-stage F-statistics remain strong (514.6 overall), confirming these results are not driven by our specific instrument choice.

This article examines the earnings premium associated with accessing first jobs through university career services. After controlling for graduates' sorting into job entry routes, the results indicate a wage premium of around 8.5%. A noteworthy finding is that female graduates consistently report larger wage premia. Furthermore, graduates with no work experience report higher wage premia (14.1–18.7%).

We do not directly observe the mechanism generating the premium. Our interpretation – that career services primarily reduce search frictions and improve match quality – is consistent with the heterogeneous effects in Section 5 but cannot be validated without richer data on firm quality or occupational sorting. Future research with employer–graduate linked records could provide a direct test.

These results underscore the critical role of university career services in facilitating graduate labour market entry. The findings highlight the impact of these services in helping university graduates navigate their initial career paths and secure valuable entry-level opportunities. The estimated wage premium of 8.5% somewhat exceeds the 4–6.5% benchmark in the European causal literature on internships and placements (Delis and Jones, 2023; Margaryan et al., 2022). This is not unexpected: unlike internships and placements, which are undertaken during the degree and benefit the full graduate population, university career services are used selectively at the point of labour market entry, and our IV strategy identifies the effect for those graduates whose job entry route is shaped by peer exposure. This interpretation is further supported by the pronounced heterogeneity documented below. From a descriptive standpoint, Eurostat data indicate that across EU member states, approximately 82.3% of recent tertiary graduates are employed within three years of graduation in 2024, yet rates of vertical mismatch, i.e. working in jobs below their qualification level, remain substantial, exceeding 35% in Spain vs. 21.5% in the EU-27 (Eurostat, 2024). In this context, an intervention that raises graduate earnings by around 8.5% represents a meaningful correction of a persistently inefficient labour market transition. While this study does not provide evidence on cost structure or scalability, career services are already embedded in university systems serving large graduate cohorts, suggesting expansions may plausibly be achievable at low marginal cost per student, relative to the overall cost of university provision. Direct cost-effectiveness assessment and long-run career tracking remain important directions for future research.

The research contributes to a growing body of evidence highlighting the economic value of targeted university support. By demonstrating the tangible earnings benefits for graduates with minimal professional experience, the study provides compelling evidence for continued investment in and development of university career services. While acknowledging certain analytical limitations that constrain broad generalisation, the study provides robust evidence to consistently validate the positive effects. The observed benefits of comprehensive, multi-year university career services are demonstrably significant and positively affect a substantial proportion of the graduate population.

The study acknowledges potential limitations and areas for future research. Further research is needed to test the external validity of the findings in other economies with lower youth unemployment, lower turnover and reduced skills mismatches. The Catalan labour market has specific institutional features such as high rates of temporary employment, bilingual education and a university system dominated by public generalist institutions, which may limit the direct transferability of the estimates. Results are most likely to generalise to Southern European contexts with similarly structured higher education systems and comparable school-to-work transition patterns. Factors such as graduates' access to alternative job-search resources and the potential need for specific competencies to effectively use career services warrant further investigation. Systematic assessment across short-, medium- and long-term horizons is recommended to comprehensively evaluate the contributions of these interventions and their potential unintended consequences.

The views expressed are purely those of the authors and may not in any circumstances be regarded as stating an official position of the European Commission. Rosario Scandurra acknowledges support as a Ramón y Cajal fellow (RYC2022-038527-I), funded by MICIU/AEI/10.13039/501100011033 and FSE+. The authors express their gratitude to AQU Catalunya for providing the data for scientific purposes and for its support. Based on work and reflections from the Employability in Programme Development project (2020-1-UK01-KA203-079171) supported by KA-2 Erasmus Action. The authors acknowledge networking support by COST Action CA23112 – Critical Perspectives on Career and Career Guidance (COCAG), funded by the European Cooperation in Science and Technology. The authors thank the participants in the Transition in Youth network for their comments on previous versions of the article. The authors benefited from using of artificial intelligence large language models (Anthropic Claude, version Opus 4.7) in the preparation of this manuscript. These were used to review text and provided suggestions on clarity, brevity, grammar and spelling.

The supplementary material for this article can be found online.