Post Training Assistance in Skills Development Programs: Applicant Selection and Impact Evaluation

For wage employees, post-training will consist of the following: (a) Tailored job alerts sent to trainees via SMS for up to 3 months after the course; (b) Follow up calls made via a call centre to ensure trainee is aware of the job alerts; (c) On the job training in some cases.

For self-employed trades, assistance will comprise of: (a) 4-day course on entrepreneurship training. Training on how to start a business, calculating start up costs, coming up with a business plan etc; (b) Some preliminary sessions on entrepreneurship, also conducted during the earlier months of the course to mentally prepare students for entrepreneurship; (c) A ‘Livelihood Co-ordinator’ to assist in identifying sources of finance, writing a loan application, finding supplier/customer linkages; (d) a business plan sheet to plan activities required for the business.

2019-10-01

2020-08-31

For the post-training assistance, our primary outcomes of interest are: for entrepreneurs, the probability of applying for and getting a loan, success in finding buyers and suppliers and overall, starting a successful business in a trade. For job-seekers, the primary outcomes are with regard to probability of being employed, job tenure and job satisfaction.

For the selection of applicants: primary outcomes would be fraction of male/female trainees in a trade, average years of education, family characteristics as well as cognitive and non-cognitive abilities.

A successful business would be measured with regard to the sales, inputs, profits and satisfaction of the business. Cognitive and non-cognitive ability will be measured through standard tests.

For entrepreneurs: a secondary set of outcomes will be with respect to planning business activities (measured through their diaries, survey sheets), expanding their professional networks (upstream and downstream), co-ordination with other entrepreneurs, responses from the family. For job seekers, a secondary set of outcomes would be with respect to job search (applications, interviews, offers), job details (salary, location) as well as perceptions of the family with regard to (wage/self) employment.

Perception of the family will be captured through question regarding their satisfaction with the candidate's employment status, remittances received, willingness to support transitions. We will also be interested in disaggregating these outcomes by gender.

Post-training assistance: We propose to take the “batch” as the unit of randomization and randomly assign batches within a training center to either get some post-training assistance or no post-training assistance. For the treatment batches, all students within the batch will receive the following post-training assistance.

Selection into training program: As participants self-select into training programs, a crucial element of their performance is whether they are reaching the pool of applicants who would benefit the most. If ultimate job performance faces significant constraints on margins other than
training (like credit constraints, information frictions, social networks) then training programs might be inefficiently attracting potential trainees who are unconstrained, rather than those who could benefit the most from the skill training absent constraints. Our evaluation will take advantage of the mobilization strategy of YP to randomize across slums/villages, advertising and promoting different aspects of the training program. In particular, mobilization for a batch will follow the following three steps:
i. A sample of potential villages/slums for a trade that are targeted for mobilization are randomly allocated to either treatment or control.
ii. During mobilization in treatment areas, apart from the information on the training program, information on the availability of post-training assistance will be provided. In control areas, only information on the training program is provided and no information on post-training assistance is provided.
iii. To maintain credibility, students coming from treatment areas are allocated to treatment batches. Students coming from the control batches on the other hand are allocated randomly to either a treatment or a control batch.

We therefore have three types of students: (A) students from treatment areas; (B) students from control areas who are allocated to treatment batches; (C) students from control areas allocated to control batches.

Randomization will be done in office by a computer code.

The unit of randomization will be a trade-village/slum.

Yes

81 batches in 5 urban centres (Mumbai) and 66 batches in 5 rural centres with an average of 15 students per batch.

Approximately 2200 trainees.

Randomization will be done at the centre-trade level. 81 batches in urban centres and 66 batches in rural centres will be allocated equally among the three treatment arms.

We use administrative follow up data of 1469 trainees who completed their training with YP in 48 centers across rural and urban Maharashtra to conduct our power calculations. For the intervention on post-training assistance, we consider the probability of employment/starting a business as the primary outcome that we can measure in the data. Though outcomes on selection of trainees are not easy to measure, we consider two outcomes observed in the administrative data: (i) fraction of female trainees; (ii) average education (in years) of the trainees. There are 10 centers and our budget estimates cover approximately 150 batches, with an average of 15 trainees per batch. This results in 2200 trainees that would form our potential sample for the study. Assuming 𝛼 = 0.05 and 𝛽 = 0.2 (80% power), the minimum effect size (and percentage increase in parentheses) of the outcome variable are as follows: (a) fraction employed - 0.11 (10.86%); (b) fraction of females- 0.15 (9.92%); (c) 0.13 (3.07%).