This is the third in our series of posts by PhD students on the job market
Globally, youth unemployment is a major policy concern. Nowhere is this challenge more pronounced than in the African continent, home to one in five first time jobseekers. A common policy response to youth unemployment has been to invest in skills training. While these programs have proven effective for promoting employment in a few contexts, their job placement rates are often low, resulting in a mass of untapped talent.
One explanation for low placement rates is that supply-side information frictions may be a particularly significant barrier to entry for youth in low-income settings. Young jobseekers may lack knowledge on many aspects of the job search process, such as how to identify job openings, how to apply for jobs, and how to prepare for interviews. This limited information is commonly accompanied by unduly optimistic expectations of their work prospects, which lead them to turn away low-paying jobs in favor of greater opportunities that frequently do not materialize.
Meet Your Future: tailored, relevant, credible, and low-cost information
In my Job Market Paper we designed a mentorship program, which we called Meet Your Future (MYF), that connects soon-to-be graduates of vocational training institutes (VTIs) to successful young workers for personalized mentorship sessions. Our ultimate goal was to assist young jobseekers in forming realistic expectations of jobs available in the labor market, enhance their grasp of the search process, and eventually improve their initial match quality and, by extension, their career trajectory.
We evaluate the impacts of MYF using a randomized control trial. Specifically, we conduct a large-scale experiment with 1,112 vocational students poised to make the school-to-work transition in urban Uganda and track them for three years. The sample consists of 60% men and 40% women, mostly aged 19–21. They receive two years of training in one of thirteen specialized fields, with catering, plumbing, motor mechanics, and tailoring being the most popular. Absent the intervention, 21% of them has permanently left the labor force three months after graduating, resulting in underutilized talent. The rest undergo frequent transitions to, from, and between informal jobs. Ultimately, contrary to what trainees believe as they begin and complete vocational training, the likelihood that skilled workers will be employed in regular jobs is as low as 40% for those who are 22 to 23 years old nationwide.
Our primary method of data collection consists of deploying innovative questionnaires directed at both students and mentors. Specifically, we build a three-year panel of students consisting of six rounds of data collection beginning two years prior and following one year after the students’ graduation. We were able to re-interview 82% of students in the final follow-up, with attrition uncorrelated with treatment. We also build a two-year panel of mentors consisting of four rounds of data collection, three prior to the MYF roll-out and one after. In addition, we collect a post-intervention survey from students and mentors to measure immediate reactions. In a novel dimensional measurement, we capture voice recordings of the first interaction between students and mentors, allowing us to assess in detail the content of these engagements but also attributes that are often difficult to codify or are subject to measurement error, such as enthusiasm and curiosity. We collect this wealth of information to examine the inner workings of such mentorship links.
Young jobseekers are oblivious to labor market entry conditions and dynamics
The panel structure of our data allows us to compare each student's expected earnings with their realized earnings. Figure 1 shows expected and realized conditional monthly earnings in the control group. Similar to the existing literature, we find evidence of striking overoptimism regarding entry-level pay in our setting: 94% of the students overestimate their first-job earnings. On average, first-job realized earnings were just 14% of students' prior expectations. When their expectations are compared to their realized earnings after one year, the proportion rises to 65%, indicating that optimism about wages is prevalent, but especially pertinent to their first job, as students fail to account for the reality that many will be unpaid or low-paid. Likewise, only 21% of students claim they would accept an unpaid job as their first job, while the realized share of unpaid first jobs in the cohort is 52%. In addition, we highlight a novel fact: not only are new entrants overly optimistic about their starting salaries, but they also have a limited grasp of labor market dynamics and salary growth potential. Most crucially, students undervalue initial unpaid employment spells, failing to see that they are frequently stepping stones to secure employment.
Figure 1: Expected and realized monthly earnings in the control group
Access to mentors boosts employment outcomes and accelerates professional advancement
MYF is particularly successful in boosting employment outcomes. Access to mentors mitigates information frictions and improves employment outcomes. Three months after the school-to-work transition we identify large impacts on employment. Labor market participation is 5.7 p.p. higher (27% over a control mean of 79%) for treated students; these students obtain their first jobs more quickly and spend 33% more time utilizing human capital complementarities acquired via their vocational education (over a control mean of 52 hours per month). In addition, these first employment allow students to ascend the career ladder more rapidly. One year later, unconditional monthly earnings of treated students are 18% higher than those of control students, which correspond to an additional 6.15 USD per month. We estimate the IRR of this intervention on the order of 300%.
Mentorship improved outcomes through information about entry-level jobs and labor market dynamics, and not through job referrals, or building search capital.
We leverage our data on conversation topics to explore why our program was successful. Informed by the literature-identified supply-side frictions and the information on conversation content gathered from the recordings, we suggest four plausible mechanisms driving the effects of the intervention on labor market outcomes: job referrals; search tips; information about entry level conditions, and encouragement. To map the conversational material to our four mechanisms, we evaluate transcripts of the coaching sessions as well as a wealth of supplementary data characterizing the students' key takeaways (Panel A of Figure 2 presents the raw conversation content as computed using the Text data, while Panel B tells us what was learned by the students). We find that mentorship acted as a particularly salient information treatment: students altered their unduly optimistic assumptions about the health of the overall job market downward and their perception of the significance of early employment in determining future chances upward. In response, they reduce their reservations wages and decline fewer job offers. We do not identify direct job referrals or stronger search abilities as viable routes for the observed treatment effects, contrary to earlier empirical and theoretical studies.
Figure 2: content of conversations with mentors
Our findings highlight the role of distorted beliefs as an important channel by which information frictions decrease earnings and career advancement. They also emphasize the importance of balancing bad news with hope for better future outcomes, when correcting overly optimistic beliefs, in order to prevent discouragement, dropout from the labor force, and, particularly among skilled workers, human capital wastage. Finally, the program affordably increases the effectiveness of vocational training programs by a significant margin.
To confirm that learning about the entry level market conditions and the future value of today’s jobs are the two primary mechanisms via which MYF impacts job search behavior and labor market outcomes, we leverage the second randomization built into the research design, namely that of students to mentors. We do so by studying how each topic of conversation affects labor market outcomes. We combine Empirical Bayes and Instrumental Variables approaches and confirm our prior result: mentors providing mentees with information about entry level conditions and encouragement are the most effective ones.
Livia Alfonsi is a PhD Candidate at UC Berkeley.
Join the Conversation