Employers usually cannot observe the skills of labour market entrants directly, particularly during the first stage of the hiring process when applicants send in their CVs. As a result, applicants invest time, effort, or money to signal their skills to potential employers. These skill signals may include getting good grades in school and university, or learning a foreign language, but we know little about which skill investments pay off in finding employment when entering the labour market.
Research has established that labour market outcomes are associated with cognitive skills (Hanushek and Woessmann 2008, Hanushek et al. 2015), social skills (Deming 2017, Heinz and Schumacher 2017) and non-cognitive skills and personality traits such as conscientiousness, commitment, and perseverance (Almlund et al. 2011).
It is unclear, though, whether signals of these skills at the application stage have a causal impact on an applicant's employability at the start of a career. Analysing and isolating how skill signals such as grade point averages (GPA) affect labour market outcomes is difficult for at least two reasons:
- Skills are usually correlated. A person with a high GPA tends also to be better in other skill dimensions; the researcher does not observe all these dimensions.
- We do not know from observational data to what extent a potential employer observes the skill signals.
The choice experiment
We conducted a choice experiment among a representative sample of nearly 600 human resource (HR) managers in Germany (Piopiunik et al. 2018) to analyse the effects on employment of skill signals in three domains: cognitive skills, social skills, and maturity. The experiment mimicked the first stage of the application process in which HR managers decide who to invite for an interview by reviewing written applications
In the experiment, we simultaneously and independently randomise a broad range of skill signals on some fictitious CVs. HR managers participating in the online survey were asked to choose between two CVs that appeared side-by-side on the screen, indicating which applicant they would rather invite for a job interview at their firm. From this, we can estimate the causal effect of each skill on being invited for an interview.
In contrast to most studies using randomised resumés (e.g. Neumark 2016, Bertrand and Duflo 2016), our study is not concerned with discrimination based on innate characteristics such as gender and race. Instead, it examines the effects of the productivity signals in which individuals can invest.
We look at two distinct groups of labour-market entrants for whom relevant skills and requirements may vary: secondary-school graduates applying for an apprenticeship position, and university graduates with a BA in Business.
Cognitive skills were signalled through the school or university GPA, IT and foreign-language skills. Social skills were signalled through social volunteering and engaging in team sports. Maturity was signalled through age and long internships, as well as the high-school GPA in case of university graduates.
The experiment was followed by a short survey on HR managers’ characteristics and hiring preferences.
Which CV elements affected invitation decisions?
We find that skill signals in all three domains affect the probability of being invited for a job interview, but the specific signal that proves relevant differed across settings. Figure 1 shows point estimates and confidence intervals for the effects of the different signals separately for secondary-school and university graduates, as well as for males and females (note that gender was always fixed within CV choice sets).
Figure 1 Effects of CV skill signals on job interview invitations
Notes: Point estimates and 90% confidence intervals of the effects of different skill signals on being invited for a job interview in a choice experiment among HR managers. Bold labels indicate statistical significance at the 10% level. See Piopiunik et al. (2018) for details.
Cognitive skill signals
GPAs proved important for both genders and for both groups of labour market entrants, but with a stronger effect for university graduates than for secondary-school graduates. For university graduates, university GPA was the strongest considered skill signal. Extended IT skills, such as skills in HTML and Dreamweaver in addition to Microsoft Office, were important for secondary-school graduates, particularly among females. This was not true for university graduates, possibly because for these applicants IT skills are generally taken for granted. English skills were also rewarded only among secondary-school graduates. Therefore, signals seem to be important when skills are not generally expected.
By contrast, knowledge of a second foreign language (Spanish or French) was relevant only for university graduates and not for secondary-school graduates, and also only for females. This may indicate that skill signals increase employability where they are relevant and often used.
Social skill signals
Social skill signals also increased the probability of being invited for an interview for both groups of labour-market entrants. The relevant skill signal differs, however. Social volunteering was highly rewarded among secondary-school graduates – indeed, it was the most important skill signal for this group.
For university graduates, engaging in team sports was an effective signal of social skills. This suggests a role for information value and credibility of these signals. While social volunteering was a strong signal of social skills among young apprenticeship applicants, among university graduates it may be quite common, and may derive from strategic behaviour without reflecting actual underlying social skills. Playing team sports may thus prove to be the better signal of social skills among 23-year-olds, because it represents an individual choice and commitment at this stage.
Maturity skill signals
Maturity skill signals may reflect conscientiousness and reliability, and were also important for both groups of labour market entrants. For secondary-school graduates, the relevant signal was being born in the earlier calendar year in the school cohort (this effect was entirely driven by male applicants). The same was true for the relevance of high-school GPA among university graduates, which – conditional on university GPA – may also signal early maturity and focus. Long internships were rewarded among university graduates of both genders.
The effect of HR manager characteristics
The characteristics of the person who was choosing also mattered. For secondary-school graduates applying for an apprentice position, managing directors and older HR managers put less weight on school GPAs and more weight on IT skills, social volunteering, and experience through internships. Among university graduates, HR managers in large firms valued university GPAs more, possibly due to a more standardised selection process.
In a survey after the choice experiment, HR managers were also asked to rate the importance of different CV elements on a four-point scale, from very important to very unimportant. Figure 2 shows the shares of HR managers who found each characteristic “very important” or “rather important”. The survey results mostly mirror the experimental results.
Figure 2 The skills that matter to human resources managers
Notes: Shares of HR managers stating to find the indicated characteristic “very important”, or “rather important” for applicants in their firm. See Piopiunik et al. (2018) for details.
In regression analyses that interact the reported importance with the respective skill signals in the CV experiment, we find that the self-reported hiring priorities of HR managers are consistent with their decisions in the choice experiments, confirming the intended information value of the skill signals.
Signals of cognitive skills, social skills, and maturity affect employability
The results reveal important aspects about how signals of skills are processed and utilised in the labour market, and how they affect employability. The differences by gender and labour market entry age are consistent with varying relevance, expectedness, and credibility of the different skill signals in different contexts.
Furthermore, the way skill signals are processed differs by characteristics of the HR manager and the firm. Our results also suggest that signals with straightforward verifiability in real hiring situations – such as GPAs, internships, and age – tend to have higher returns than signals that are more costly to verify, such as language or team sports, in particular in large firms.
Overall, the results indicate that employers do observe and value skill signals on CVs. Skill signals in several domains causally affect the employment chances of labour market entrants.
Almlund, M, A Duckworth, J J Heckman, and T Kautz (2011), "Personality Psychology and Economics", in E A Hanushek, S Machin, L Woessmann (eds.), Handbook of the Economics of Education, Volume 3, North-Holland.
Bertrand, M and E Duflo (2016), "Field Experiments on Discrimination", NBER Working Paper 22014.
Deming, D J (2017), "The Growing Importance of Social Skills in the Labor Market", Quarterly Journal of Economics 132(4): 1593-1640.
Hanushek, E A, G Schwerdt, S Wiederhold, and L Woessmann (2015), "Returns to Skills around the World: Evidence from PIAAC", European Economic Review 73: 103-130.
Hanushek, E A and L Woessmann (2008), "The Role of Cognitive Skills in Economic Development", Journal of Economic Literature 46(3): 607-668.
Heinz, M, H Schumacher (2017), "Signaling Cooperation", European Economic Review 98: 199-216.
Neumark, D (2016), "Experimental Research on Labor Market Discrimination", NBER Working Paper 22022.
Piopiunik, M, G Schwerdt, L Simon, and L Woessmann (2018), "Skills, Signals, and Employability: An Experimental Investigation", CESifo Working Paper 6858.