The role of cognitive and socio-emotional skills in labor markets

Economists have long considered educational levels to be the main indicator of skills. However, it is now well recognized that common proxies such as years of education do not appropriately reflect cross-country differences in skills acquisition, or how they impact the labor market [1]. International assessments (such as the OECD’s Programme for International Student Assessment (PISA) and Programme for the International Assessment of Adult Competencies (PIACC) tests) of students’ and adults’ skills show great disparities across similar educational levels, suggesting that years of education attained only partially reflects people’s abilities.

Instead, people should be distinguished by the broad sets of abilities, accumulated over one’s lifetime, that shape their decisions and success in the labor market [2]. In that sense, a large body of evidence has documented that direct measures of skills provide more adequate estimations of the differences in individuals’ potential productive capacity than is provided by the quantity of education they receive. Around the world, employers place high value on skills needed for emerging jobs (i.e. those associated with high use of technology, or involving substantial human interaction), which typically involve non-routine analytical tasks.

One noteworthy fact from the economic and psychology literature discussing skills is the plethora of definitions and taxonomies surrounding this concept. These days, the term “skills” is used broadly to include competencies, attitudes, beliefs, and behaviors that are malleable (modifiable) across an individual’s development and can be learned and improved through specific programs and policies. These broad sets of skills can be broadly divided into two main categories: (i) cognitive skills—aptitudes to perform mental tasks such as comprehension or reasoning; and (ii) socio-emotional skills—personality traits, behaviors, attitudes, and beliefs. Technical skills, associated with the specific knowledge needed to carry out one’s occupation, can be thought of as a subset of cognitive skills [2].

Cognitive skills are generally defined as intelligence or mental abilities. Two levels can be distinguished: (i) lower-order (basic) cognitive skills, which are foundational skills or basic academic knowledge such as literacy or numeracy; and (ii) higher-order (advanced) cognitive skills that involve more complex thinking, such as critical thinking or problem solving.

The term “socio-emotional skills” refers to a distinct set of skills that enable individuals to navigate interpersonal and social situations effectively. These skills encompass behaviors and attitudes that are consistent patterns of thoughts, feelings, and conduct (such as commitment, discipline, or the ability to work in a team) and personality traits (such as self-confidence, perseverance, and emotional stability), which are broad facets that are relatively stable over time [2]. The concept of socio-emotional skills was developed in the literature on educational psychology. In the economic literature, the term “socio-emotional skills” is often used interchangeably with terms such as “behavioral skills,” “life skills,” “non-cognitive skills,” or “soft skills.” Nonetheless, these terms differ slightly and therefore merit clarification. “Non-cognitive skills” refers to a broad range of behaviors, abilities, and traits that are not induced by intelligence or achievement. “Soft skills” and “life skills” usually include more technical skills such as language fluency and computer literacy. Psychologists argue that many of the abilities and traits that economists intend to capture with the term “non-cognitive skills” are in fact a result of cognition.

Whereas literacy, numeracy, and various forms of cognitive skills have traditionally been associated with gains in worker productivity, attention is now also focusing on the role of socio-emotional skills. This is due to a burgeoning demand for jobs that require skills related to dealing with personal and social situations, which is being created by new technologies and economic transformation across the globe. Indeed, the high value that surveys show is placed on socio-emotional and advanced cognitive skills by employers may reflect a decline in the relative number of jobs that require manual labor, and an increase in those that require non-routine analysis and independent reasoning.

A few caveats related to methodological shortcomings and data limitation must be highlighted in this context. First, studies may not offer the full picture of the effect of an individual’s true skills; results are conditional on the dimensions of the skill set measured in surveys, which may not reflect the full range of an individual’s abilities. Moreover, survey instruments are subject to measurement error of the skill construct [2]. For instance, test scores are sensitive to the amount of schooling completed at the time of the test and family background. Furthermore, the measures of ability are known to be very noisy. Thus, using test scores as an independent variable in regression model analysis could lead to measurement error bias.

Second, causal analyses of the impact of skills and traits on employment often suffer from endogeneity bias when including education in the analysis, casting doubt on the reliability of the estimations [4]. Schooling, cognitive skills, and socio-emotional skills are interrelated and influence each other. On the other hand, not controlling for schooling in linear estimations can lead to overestimations of the net effects of measures of skills on wages, capturing instead the whole effect of skills, independently of where they were formed (at school or elsewhere) [4]. Distinguishing, their respective effects is technically challenging and can lead to severe estimation bias when considered simultaneously in a standard regression model—an issue not addressed in the literature of the early 2000s.

Overall, the studies referred to in this article vary greatly in the reported magnitude of skills effects, samples, research questions, and statistical methodologies, which may challenge meaningful comparisons between them. In the general absence of randomized control-trial experiments, some authors have resorted to clever estimation techniques, such as involving non-parametric structural estimation to simulate distributions of unobserved heterogeneity in explaining test scores [4]. This method can mitigate measurement error concerns because it acknowledges the fact that skills are latent rather than observable, not depending on a single (and potentially poor) measure of skills.

Thus, the optimal approach to determine the association between a person’s skills (both cognitive and socio-emotional) and schooling or labor market outcomes would be based on longitudinal data. This data could allow researchers to link cognitive test scores and personality constructs from individuals’ youth with their outcomes as adults. Unfortunately, such data are rare, especially in low- and middle-income countries, and even fewer data sets provide direct assessments of skills. Thus, most of the robust evidence still comes from the US and other high-income countries, which limits generalizability to other countries.

Overall, the literature seems to suggest that both cognitive and socio-emotional skills are relevant to predict labor market success (in particular earnings and employment), with differences across population groups, industries, and occupations. In general, cognitive skills prove relevant to explain earnings, while socio-emotional skills seem particularly relevant to explain labor supply and schooling decisions. This implies school and vocational training programs should consider promoting both cognitive and socio-emotional skills to help improve a range of labor outcomes, such as improving job quality, or fostering higher labor market participation or tertiary education.

In that sense, it has long been established that early childhood development programs have the highest rate of return in shaping skills, and they often yield better outcomes than similar interventions later in life. These interventions lay the foundations for children that should be completed by later investments. Primary school is also an ideal context for shaping both cognitive and socio-emotional skills, as at this stage in life children possess both the neuro-biological capacity and psychosocial maturity to effectively practice and learn both types of skills. However, the elementary education curriculum in the majority of countries devotes rather limited resources to influencing character and traits. Schools continue to be judged solely by students’ performance in cognitive achievement tests; socio-emotional skills are often neglected, and teachers are not appropriately trained to foster the development of them.

Given the well-documented influence of family environments on skill formation, parenting and extra-curricular activities can play a significant role in fostering cognitive and socio-emotional skills. Finally, interventions targeting workers entering the labor force can effectively bolster their socio-emotional skills, thereby complementing the effects of improved labor market information and vocational counseling.

Unfortunately, the current evaluations used to determine which interventions are most effective at fostering both cognitive and socio-emotional skills mostly consider short-term program impacts. The few that do allow measurement of long-term impacts find mixed results; and some have been found to have no effect after a few years.

In sum, better measuring of skills profiles and better understanding of the interventions that permanently affect cognitive and socio-emotional skills would allow for more efficient policy design. In any case, further research is needed on the optimal combination of packages for different demographic and socio-economic population groups, particularly in low- and middle-income countries, for which the evidence is still scant.

The authors thank an anonymous referee and the IZA World of Labor editors for many helpful suggestions on earlier drafts. The authors also thank Wendy Cunningham and Miguel Sarzosa, who co-authored previous work which contains a larger number of background references for the material presented here, and has been used intensively in all major parts of this article (see Further reading).