300 likes | 416 Views
Life Skills, Employability and Training for Disadvantaged Youth: Evidence from a Randomized Evaluation Design in the Dominican Republic. 10th International Seminar Itaú Economic Evaluation of Social Projects Pablo Ibarraran, October 29 th 2013. Overview.
E N D
Life Skills, Employability and Training for Disadvantaged Youth: Evidence from a Randomized Evaluation Design in the Dominican Republic 10th International Seminar Itaú Economic Evaluation of Social Projects Pablo Ibarraran, October 29th 2013
Overview • Results of RCT of a youth-training program in the Dominican Republic: “Juventud y Empleo” on: • Labor market outcomes • Behavior, perceptions and expectations • Non-cognitive skills • Motivation for further research in areas related to non-cognitive skills: • Measurement • Training • Labor market impacts
Youth Employment in the Dominican Republic • Big challenge, but not necessarily more than in other countries. • Large percentage not working nor studying: 34% of youth aged 18-29. • Youth unemployment rate = 2.3 * Adult unemployment rate. • Large percentage of youth employed in low quality jobs: Youth informality = 2 * Adult informality rate.
Juventud y Empleo • Based on the Joven-type of program, first with RCT evaluation component in LAC. • Targets disadvantaged youth (school dropouts, ages 16-19, not studying or working) • Combines classroom training with subsequent internship period of on-the-job work experience. • Role of the private sector: identification of training needs, provision of training sessions, offering of internships.
First Impact Evaluation • Carried out by Card et al. with data from a cohort that applied to receive training in early 2004 and was followed up in mid 2005. • Found little impact on employment; though sizable impacts on wages and formality for men. • Randomized design was potentially compromised by failure to include non shows and dropouts in the follow up survey. • Sample size was limited (n=1,345).
Revamped Program and Evaluation • Closer work with the training centers and firms to develop tailored courses that train youth for real vacancies. • Strengthening of the non-cognitive skills training (based on private sector claims). • Randomized assignment on a larger sample for each course (20 treatments and 15 controls). • More complete survey instrument with 15 modules.
Limited evidence on non-cognitive skills • Trend to include of non-cognitive skills components in training programs. • Evidence on labor market success determinants shows that employers value certain behaviors that are linked to high productivity workers • There is no clear evidence on how to best measure NCS empirically. • Knowledge gap in Latin America: • Measurement of non-cognitive skills. • Relation between non-cognitive skills and labor market outcomes of youth.
Randomization Process • The training centers (TC) identify potential participants from a group of eligible applicants • TC submit data on 35 eligible and interested youngsters for each course. • The Ministry receives the information from TC, verifies eligibility and randomly assigns 20 to the treatment group and 15 to the control group. • Training centers replaced up to five slots of non shows and dropouts with control group members.
Assignment and Participation Status • Classification based on administrative data. • Number of “never takers” equals number of “always takers” as course size was kept constant.
Execution 307 courses 350 Follow-up 300 250 26 Months after the end of the last course 200 150 100 50 0 Timeline
Follow Up Survey • Carried out 18 to 24 months after graduation. • Based on a random sample of 5,000 out of 10,309 individuals initially registered. • 3,250 in treatment group and 1,750 in control group. • Minimum power of 0.8 to detect an 8% effect in employment. • 70% success rate assumed in completion of follow up surveys.
Follow Up Survey • More than 80% of sample located for follow up survey (identical for the treatment (80.8%) and control (80.4%) groups. • Verification of the participation status. • Training centers enjoyed some control over who took the course. • Replacements that declined to participate classified in control group.
Realized Treatment and Control Groups • Compliance was low in the control group: • 37% contacted by training center and accepted course. • 23% took course and completed the internship. • The lottery still had a strong impact on the probability of participating and constitutes a strong instrument for the treatment.
Intention to Treat Effects • Estimation of ITT, using OLS with std errors clustered at the course level and TC fixed effects. • We center on three impacts (with heterogeneity) • Labor market • Youth behavior and life style, perceptions and expectations • Non-cognitive skills • Correlation with labor market performance
Instruments to Measure Non-Cognitive Skills • Social and Personal Competencies Scale (CPS): • Six basic competencies: leadership, behavior in situations of conflict, self-esteem, ability to relate with others, order, empathy, and communication skills. • Rosenberg Scale: to measure self-esteem. • Grit Scale: • Measures determination and strength of mind. • Four main categories: persistency of effort, enthusiasm about long term goals, consistency of interests, and ambition.
Are Impacts on NCS Relevant? • Statistically significant positive impacts on various measures of non-cognitive development. • 0.11 sd in total CPS score (similar for other categories) • 0.11 sd in the Rosenberg Scale in men. • 0.08 sd in the Grit Scale (similar for other categories). • What does this mean? • Literature on relationship between non-cognitive skills and professional success relatively new. • Still uncertain which specific competencies relate to participation in the labor market. • Limited evidence on the magnitude of the changes required to impact employability.
Are Impacts on NCS Relevant? • Reference to literature in cognitive and non-cognitive skills development: • 0.4 standard deviation per year in educational attainment in primary education (Hill et al. 2007). • 0.13 standard deviations in non-cognitive skills assessments with participation in sports clubs (Felfe et al. 2011).
Non-Cognitive Skills Unknowns We need evidence on: • The specific skills that are positively related to labor market participation. • Assessment of the magnitude of the change required to impact employability. • Replication of effective training program designs that can be easily scaled up. • Design of relevant instruments to measure (and interpret) non-cognitive skills development.
Summing Up • Labor market outcomes: • Job Quality for men: 18% national; 52% Santo Domingo. • Monthly earnings for those employed: 9% overall; 17% Santo Domingo. • No statistically significant impact on employment. • Youth behavior and expectations: • Drop in pregnancy: 28% overall; 43% teenagers. • Positive impact on expectations about the future. • Non-cognitive skills: • Positive impact on the CPS, Rosenberg (for men) and Grit Scales: 0.8 - 0.11 std. dev.