Abstract
We use the Programme for the International Assessment of Adult Competencies (PIAAC) data to construct several measures of non-cognitive skills and to analyze the relationship between non-cognitive skills and earnings. We construct measures for non-cognitive skills based on previous research in the field, such as self-reports and non-response in the background questionnaire. We also take advantage of the computer based nature of the assessment to explore the effects of other constructs such as skipped questions in the test, average time per question, average time per correct answer and the number of keyboard and mouse movements before answering a question. We find that, even after controlling for PIAAC scores, previously proposed proxies of non-cognitive skills (self-reports and survey non-response) are significantly associated with individual earnings. We also use these measures of non-cognitive skills to propose a correction of the PIAAC scores as a measure of cognitive skills in the earnings equation. By doing so, we observe that the relationship between earnings and PIAAC scores might be partially mediated by non-cognitive skills. However, our results also show that PIAAC test scores remain highly significant and as the main determinant of earnings after the correction.
Highlights
Despite the important role that human capital occupies in economic thinking, its nonobservable nature has historically limited its measurement
Even after controlling for PIAAC scores, previously proposed proxies of non-cognitive skills are significantly associated with individual earnings
When using the additional test based measures we find that average time per question is not significantly correlated with individual earnings (except when included jointly with other non-self-reported measures and not controlling for PIAAC test scores [column (5)]
Summary
Despite the important role that human capital occupies in economic thinking, its nonobservable nature has historically limited its measurement. According to the conclusions of Borghans et al (2011), Duckworth et al (2011) and Segal (2012), scores on low-stakes achievement tests should not be interpreted as a measure of cognitive skills but as a broader notion of human capital that includes noncognitive skills. As emphasized in the psychology field (Wechsler 1940) and in the economics literature (Duckworth et al 2011; Segal 2012) performance on achievement tests depends on cognitive and on non-cognitive skills.3 It is not clear until what extent the association between PIAAC scores and earnings found by Hanushek et al (2015) is driven by cognitive or non-cognitive factors. The present paper is mainly intended to enrich the study of the relationship between PIAAC test scores and earnings started by Hanushek et al (2015) by providing measures of human capital that address the distinction between cognitive and non-cognitive skills.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
