Association for Public Policy Analysis & Management

While the labor market relevance of a wide range of non-cognitive skills is well established, the development of these skills is less studied. Our project analyzes whether vocational education and training, in particular Germanic apprenticeship training lasting 3 to 4 years, is an effective means for developing non-cognitive skills of adolescences, and which mechanisms drive this development. Therefore, we have collected a panel data set running from 2009 to 2015 and including repeated measures of non-cognitive skills (e.g., grit, Big Five personality traits), risk attitude, time preferences, and IQ.

In summary, using this panel data set of 150 individuals in Switzerland, we show that grit is rather malleable during young adulthood (between age 16 and 22) with a substantial variety in changes which indicates a heterogeneous development over time and contradicts a simple age effect. Comparing one grit score derived in 2009 to the same score derived in 2015, we find that the mean change is equal to .5 standard deviations of the baseline score, that the cumulative distribution functions of the two scores clearly differ, and that the later score stochastically dominates the earlier one. However, the changes seem to be quite heterogeneous and a significant number of individuals experiences reductions in grit, which contradicts the thesis that grit is just increasing with age. Moreover, finding rather low correlations between the two measures over time (Pearson correlation of .25***, rank correlation of .29***), we show similar patterns of malleability and heterogeneous development. Further, to reveal the mechanisms driving this heterogeneous development, we explain individual changes in grit by baseline characteristics (gender, age, mother’s education, high school math grade) and differing interventions during apprenticeship training, which happen between the two grit measures (training occupation, size of training firm, class size, …).

Our results carry implications for studies investigating the returns to latent non-cognitive skills. If non-cognitive skills are unstable during young adulthood, we need to learn more about the nature of potential changes to model growth processes when investigating the returns to these skills.