There is a recent and interesting puzzling result in the labor economic literature in which social programs -i.e. educational or labor market programs – that exhibit positive impacts on labor market outcomes seem, contrary to expectations, not to affect measures of non-cognitive skills (Card et al., 2011; Calero et al., 2014). 

There are two main alternative explanations for this conundrum.

(i) It may be that socio-emotional skills simply do not account for the labor market impacts that are found.

(ii) Socio-emotional skills are measured along with noise, mainly due the fact that measures of non-cognitive skills are generally self-reported. Recent evidence supports this argument (Heckman and Kautz, 2014).

This study provides further evidence consistent with the second possibility.


Programs that aim to foster soft-skills affects participants’ emotional regulation.


In addition, emotional dispositions bias self-reported measures of non-cognitive skills (Querengasser and Schindler, 2014).


This implies that impact on non-cognitive skills are not trivially detectable.

Despite the increasing relevance of emotional dispositions on measurements (e.g. Almlund et al., 2011; Heckman and Kautz, 2012), previous studies usually do not consider emotion dispositions since are hard to measure.

Hence, this study seeks to contribute in two dimensions.

1. Using novel data, it provides a rigorous causal estimate of the impact of a program to foster entrepreneurial skills for youth in a formal educational setting in a developing country context.

2. Using lab-in-the-field experiments, it assesses the impact of this program on emotional regulation.

  • Consistent with the puzzle, the program had significant impact on educational outcomes —in both dropouts and SAT-like test registration— yet no impact on the expected mechanisms, such us socio-emotional skills or creativity measures.
  • However, I find significant impact on emotional regulation —i.e. both emotional state and emotional responsiveness— from neurophysiological recordings.
    • First, the program has a negative impact on emotional state. This result is consistent with the puzzle and with the idea that emotional disposition bias self-reported measures of non-cognitive skill.
    • Second, program participation also reduces individuals’ emotional reaction to negative stimuli. Following the approach/withdrawal theory of emotions as motivation (e.g. Davidson et al., 1990; Harmon-Jones et al., 2010; Lempert and Phelps, 2014), the program participation makes individuals more resilient since the context of high vulnerability of the intervention.
  • Finally, the results also may inform researchers about evaluation of similar programs to develop socio-emotional skills in general, and to foster entrepreneurial skills in particular.